├── .gitignore ├── DSP.py ├── LICENSE ├── README.md ├── STL3D ├── GestureradarV1.2.SLDPRT └── GestureradarV1.2.STL ├── UI_interface.py ├── colortrans.py ├── config ├── IWR1843_cfg.cfg └── IWR6843_cfg.cfg ├── dsp ├── ZoomFFT.py ├── __init__.py ├── angle_estimation.py ├── cfar.py ├── compensation.py ├── doppler_processing.py ├── music.py ├── noise_removal.py ├── range_processing.py └── utils.py ├── firmware ├── xwr1843_mmw_demo.bin ├── xwr6443_mmw_demo.bin ├── xwr6843AOP_mmw_demo.bin ├── xwr6843ISK_mmw_demo.bin └── xwr6843ODS_mmw_demo.bin ├── gesture_icons ├── 0.jpg ├── 1.jpg ├── 2.jpg ├── 3.jpg ├── 4.jpg ├── 5.jpg ├── 6.jpg └── 7.jpg ├── globalvar.py ├── img ├── 1.png ├── 2.png ├── 3.png ├── 4.gif ├── 5.mp4 ├── 6.jpg └── 7.jpg ├── iwr6843_tlv └── detected_points.py ├── libs ├── UDPCAPTUREADCRAWDATA.dll ├── UDPCAPTUREADCRAWDATA.lib └── libtest.so ├── main.py ├── radar_config.py └── real_time_process.py /.gitignore: -------------------------------------------------------------------------------- 1 | # Byte-compiled / optimized / DLL files 2 | __pycache__/ 3 | *.py[cod] 4 | *$py.class 5 | 6 | 7 | # Distribution / packaging 8 | .Python 9 | build/ 10 | develop-eggs/ 11 | dist/ 12 | downloads/ 13 | eggs/ 14 | .eggs/ 15 | lib/ 16 | lib64/ 17 | parts/ 18 | sdist/ 19 | var/ 20 | wheels/ 21 | *.egg-info/ 22 | .installed.cfg 23 | *.egg 24 | 25 | # PyInstaller 26 | # Usually these files are written by a PyInstaller build script 27 | *.manifest 28 | *.spec 29 | 30 | # Installer logs 31 | pip-log.txt 32 | pip-delete-this-directory.txt 33 | 34 | # Unit test / coverage reports 35 | htmlcov/ 36 | .tox/ 37 | .coverage 38 | .coverage.* 39 | .cache 40 | nosetests.xml 41 | coverage.xml 42 | *.cover 43 | .hypothesis/ 44 | .pytest_cache/ 45 | 46 | # Translations 47 | *.mo 48 | *.pot 49 | 50 | # Django stuff: 51 | *.log 52 | local_settings.py 53 | db.sqlite3 54 | 55 | # Flask stuff: 56 | instance/ 57 | .webassets-cache 58 | 59 | # Scrapy stuff: 60 | .scrapy 61 | 62 | # Sphinx documentation 63 | docs/_build/ 64 | 65 | # PyBuilder 66 | target/ 67 | 68 | # Jupyter Notebook 69 | .ipynb_checkpoints 70 | 71 | # pyenv 72 | .python-version 73 | 74 | # celery beat schedule file 75 | celerybeat-schedule 76 | 77 | # SageMath parsed files 78 | *.sage.py 79 | 80 | # Environments 81 | .env 82 | .venv 83 | env/ 84 | venv/ 85 | ENV/ 86 | env.bak/ 87 | venv.bak/ 88 | 89 | # Spyder project settings 90 | .spyderproject 91 | .spyproject 92 | 93 | # Rope project settings 94 | .ropeproject 95 | 96 | # mkdocs documentation 97 | /site 98 | 99 | # mypy 100 | .mypy_cache/ 101 | 102 | # PyCharm 103 | .idea/ 104 | *.iml 105 | *.iws 106 | *.ipr 107 | *.iws 108 | .idea_modules/ 109 | 110 | # VS Code 111 | .vscode/ 112 | *.code-workspace 113 | 114 | # Project specific 115 | dataset/ 116 | visualization/ 117 | *.npy 118 | *.pth 119 | *.pt 120 | *.pkl 121 | *.h5 122 | *.hdf5 123 | *.model 124 | # *.bin 125 | 126 | # Windows specific 127 | Thumbs.db 128 | ehthumbs.db 129 | Desktop.ini 130 | 131 | # macOS specific 132 | .DS_Store 133 | .AppleDouble 134 | .LSOverride 135 | Icon 136 | ._* 137 | .DocumentRevisions-V100 138 | .fseventsd 139 | .Spotlight-V100 140 | .TemporaryItems 141 | .Trashes 142 | .VolumeIcon.icns 143 | .com.apple.timemachine.donotpresent 144 | -------------------------------------------------------------------------------- /DSP.py: -------------------------------------------------------------------------------- 1 | # version: 1.0 2 | 3 | import numpy as np 4 | from collections import deque 5 | import dsp 6 | from dsp.doppler_processing import doppler_processing 7 | import dsp.range_processing as range_processing 8 | import dsp.angle_estimation as Angle_dsp 9 | import dsp.utils as utils 10 | import dsp.compensation as Compensation 11 | from dsp.utils import Window 12 | import globalvar as gl 13 | 14 | 15 | rti_queue = deque(maxlen=12) 16 | rdi_queue = deque(maxlen=12) 17 | rai_queue = deque(maxlen=12) 18 | rei_queue = deque(maxlen=12) 19 | 20 | gesturetimecnt = 0 21 | # 用于计数手悬停事件 22 | 23 | gesturetimecnt2 = 0 24 | 25 | NUM_TX = 3 26 | NUM_RX = 4 27 | VIRT_ANT = 4 28 | VIRT_ANT1 = 1 29 | # Data specific parameters 30 | NUM_CHIRPS = 64 31 | NUM_ADC_SAMPLES = 64 32 | RANGE_RESOLUTION = .0488 33 | DOPPLER_RESOLUTION = 0.0806 34 | NUM_FRAMES = 300 35 | 36 | # DSP processing parameters 37 | SKIP_SIZE = 4 # 忽略边缘角度的目标 38 | ANGLE_RES = 2 # 角度分辨率 39 | ANGLE_RANGE = 90 # 监督范围 40 | ANGLE_FFT_BINS= 64 41 | ANGLE_BINS = (ANGLE_RANGE * 2) // ANGLE_RES + 1 42 | BINS_PROCESSED = 64 43 | 44 | numRangeBins = NUM_ADC_SAMPLES 45 | numDopplerBins = NUM_CHIRPS 46 | 47 | # 计算分辨率 48 | range_resolution, bandwidth = dsp.range_resolution(NUM_ADC_SAMPLES) 49 | doppler_resolution = dsp.doppler_resolution(bandwidth) 50 | 51 | # Start DSP processing 52 | range_azimuth = np.zeros((int(ANGLE_BINS), BINS_PROCESSED)) 53 | range_elevation = np.zeros((int(ANGLE_BINS), BINS_PROCESSED)) 54 | azimuth_elevation = np.zeros((ANGLE_FFT_BINS, ANGLE_FFT_BINS, NUM_ADC_SAMPLES)) 55 | 56 | num_vec, steering_vec = Angle_dsp.gen_steering_vec(ANGLE_RANGE, ANGLE_RES, VIRT_ANT) #theta跨度 theta分辨率 Vrx天线信号的数量 57 | 58 | def doppler_fft(x,window_type_2d=None): 59 | 60 | fft2d_in = np.transpose(x, axes=(1, 0)) # rangbin chirps 61 | if window_type_2d: 62 | fft2d_in = utils.windowing(fft2d_in, window_type_2d, axis=1) 63 | 64 | # 这里用zoom——FTT，是不是更加有效提取有效频率信息呢 65 | fft2d_out = np.fft.fft(fft2d_in, axis=1)#frame rangbin dopplerbin 66 | fft2d_log_abs = np.log2(np.abs(fft2d_out)) 67 | # numADCSamples, numChirpsPerFrame 68 | det_matrix_vis = np.fft.fftshift(fft2d_log_abs, axes=1) 69 | return det_matrix_vis 70 | 71 | # 该函数主要是画三个图 72 | # RTI 73 | # DTI 74 | # ATI 75 | 76 | framecnt = 0 77 | def RDA_Time(adc_data, window_type_1d=None, clutter_removal_enabled=True, CFAR_enable=False, axis=-1): 78 | 79 | global gesturetimecnt, framecnt 80 | # 转换成(num_chirps_per_frame, num_rx_antennas, num_adc_samples) 81 | adc_data = np.transpose(adc_data, [0, 2, 1]) 82 | # radar_cube = range_processing(adc_data, window_type_1d, 2) 83 | # 2倍抽取 降低dmax 84 | # radar_cube = range_processing(np.concatenate([adc_data[:,:,0:64:2],adc_data[:,:,0:64:2]], axis=2), window_type_1d, 2) 85 | radar_cube = range_processing(2*adc_data[:,:,0:64:2], window_type_1d, 2) 86 | 87 | if clutter_removal_enabled: 88 | radar_cube = Compensation.clutter_removal(radar_cube,axis=0) 89 | 90 | 91 | # 距离多普勒图 92 | range_doppler_fft, aoa_input = doppler_processing(radar_cube, 93 | num_tx_antennas=3, 94 | interleaved=False, 95 | clutter_removal_enabled=False, #前面已经做了 96 | window_type_2d=Window.HANNING, 97 | accumulate = False) 98 | # (numRangeBins, numVirtualAntennas, num_doppler_bins) 99 | 100 | rdi_abs = np.transpose(np.fft.fftshift(np.abs(range_doppler_fft), axes=2), [0, 2, 1]) 101 | rdi_abs = np.flip(rdi_abs, axis=0) 102 | rdi_queue.append(rdi_abs) 103 | # 16个frame叠加返回 104 | rdi_framearray = np.array(rdi_queue)#frame chirps adcnum numVirtualAntennas 105 | 106 | # 角度图： 107 | # azimuth_elevation[0,0:4,:] = range_doppler_fft[:,[10,8,6,4],0].T 108 | # azimuth_elevation[1,0:4,:] = range_doppler_fft[:,[11,9,7,5],0].T 109 | # azimuth_elevation[2,0:2,:] = range_doppler_fft[:,[2,0],0].T 110 | # azimuth_elevation[3,0:2,:] = range_doppler_fft[:,[3,0],0].T 111 | # aei_raw = np.fft.fft2(azimuth_elevation, axes=[0, 1]) 112 | # aei_raw = np.log2(np.abs(aei_raw)) 113 | # 。range_doppler_fft 114 | # 距离时间图 115 | det_matrix = radar_cube[:, 0, :] 116 | 117 | #用距离图作为判断 118 | # [4:36,]是指4~36这个rangbin区间，有true的隔宿大于26个 119 | Iscapture = gl.get_value('IsRecognizeorCapture') 120 | # if(np.sum(det_matrix[2:36,:]>3e3)>26) and Iscapture:16 121 | if(np.sum(det_matrix[:,36:62]>3e3)>14) : 122 | if Iscapture: 123 | gesturetimecnt = gesturetimecnt + 1 124 | # print("有手势%d" %gesturetimecnt) 125 | 126 | if(gesturetimecnt>=2) and Iscapture: 127 | framecnt = framecnt + 1 128 | # framecnt是用来相当延迟多少帧再截图到judgegesture文件夹中 129 | if framecnt>=8: 130 | # print("进来了") 131 | # datetime.now() 132 | # a = datetime.now() #获得当前时间 133 | if gl.get_value('timer_2s'): 134 | gl.set_value('usr_gesture',True) 135 | gl.set_value('timer_2s',False) 136 | # print("有手势") 137 | framecnt = 0 138 | gesturetimecnt=0 139 | 140 | 141 | rti_queue.append(det_matrix) 142 | rti_framearray = np.array(rti_queue)#frame chirps adcnum 143 | rti_array = np.reshape(rti_framearray, (1, -1, 64))#chirps adcnum 144 | # (num_chirps_per_frame, num_range_bins, num_rx_antennas) 145 | rti_array_out = np.transpose(rti_array, [1, 2, 0]) 146 | 147 | # 微多普勒时间图 148 | micro_doppler_data = np.zeros((rti_framearray.shape[0], rti_framearray.shape[1], rti_framearray.shape[2]), dtype=np.float64) 149 | micro_doppler_data_out = np.zeros((16,64), dtype=np.float64) 150 | for i, frame in enumerate(rti_framearray): 151 | # --- Show output 152 | det_matrix_vis = doppler_fft(frame,window_type_2d=Window.HANNING) 153 | micro_doppler_data[i,:,:] = det_matrix_vis 154 | 155 | 156 | rti_array_out = np.flip(np.abs(rti_array_out), axis=1) 157 | # 158 | rti_array_out[rti_array_out<3e3]=0 159 | # # 用RDI图判断信噪比强度64*64 160 | # if(np.sum(rti_array_out[0:1024:16,:,:]<100)>4090): 161 | # SNR = False 162 | # else: 163 | # SNR = True 164 | 165 | micro_doppler_data_out = micro_doppler_data.sum(axis=1) 166 | 167 | micro_doppler_data_out[micro_doppler_data_out<20]=0 168 | 169 | return rti_array_out, rdi_framearray, micro_doppler_data_out 170 | 171 | 172 | 173 | def Range_Angle(data, padding_size=None, clutter_removal_enabled=True, window_type_1d = Window.HANNING,Music_enable = False): 174 | # (0:TX1-RX1,1:TX1-RX2,2:TX1-RX3,3:TX1-RX4,| 4:TX2-RX1,5:TX2-RX2,6:TX2-RX3,7:TX2-RX4,| 8:TX3-RX1,9:TX3-RX2,10:TX3-RX3,11:TX3-RX4) 175 | # data = np.fft.fft2(data[:, :, [1,0,9,8]], s=[padding_size[0], padding_size[1]], axes=[0, 1]) 176 | # global SNR 177 | 178 | # 转换成(num_chirps_per_frame, num_rx_antennas, num_adc_samples) 179 | adc_data = np.transpose(data, [0, 2, 1]) 180 | # radar_cube = dsp.zoom_range_processing(adc_data, 0.1, 0.5, 1, 0, adc_data.shape[2]) 181 | radar_cube = range_processing(2*adc_data[:,:,0:64:3], window_type_1d, 2) 182 | 183 | if clutter_removal_enabled: 184 | radar_cube = Compensation.clutter_removal(radar_cube,axis=0) 185 | 186 | # np.save('data.npy',radar_cube) 187 | frame_SNR = np.log(np.sum(np.abs(radar_cube[:,:])))-14.7 188 | if(np.abs(frame_SNR)<1.8): 189 | frame_SNR = 0 190 | # print(frame_SNR) 191 | # --- capon beamforming 192 | beamWeights = np.zeros((VIRT_ANT, BINS_PROCESSED), dtype=np.complex_) 193 | 194 | # Note that when replacing with generic doppler estimation functions, radarCube is interleaved and 195 | # has doppler at the last dimension. 196 | # 方位角 197 | for i in range(BINS_PROCESSED): 198 | if Music_enable: 199 | range_azimuth[:,i] = dsp.aoa_music_1D(steering_vec, radar_cube[:, [10,8,6,4], i].T, num_sources=1) 200 | else: #4,6,8,10# 201 | range_azimuth[:,i], beamWeights[:,i] = dsp.aoa_capon(radar_cube[:, [7,4,3,0], i].T, steering_vec, magnitude=True) 202 | # range_azimuth[:,i], beamWeights[:,i] = dsp.aoa_capon_new(radar_cube[:, [10,8,6,4], i].T,radar_cube[:, [10,8,6,4], i+1].T, steering_vec, magnitude=True) 203 | # 俯仰角 204 | for i in range(BINS_PROCESSED): 205 | if Music_enable: 206 | range_elevation[:,i] = dsp.aoa_music_1D(steering_vec, radar_cube[:, [1,0,9,8], i].T, num_sources=1) 207 | else: 208 | # radar_cube[:, [1,0,9,8], i].T*[[1],[-1],[1],[-1]] 不用这个，和导向矢量有关系把 209 | range_elevation[:,i], beamWeights[:,i] = dsp.aoa_capon(radar_cube[:, [7,6,11,10], i].T*[[1],[-1],[1],[-1]], steering_vec, magnitude=True) 210 | 211 | rdi_ab1 = np.flip(np.abs(range_azimuth), axis=1) 212 | # rdi_ab2 = np.fft.fftshift(range_elevation, axes=0) 213 | rdi_ab2 = np.flip(np.abs(range_elevation), axis=1) 214 | rdi_ab1 = np.minimum(rdi_ab1,rdi_ab1.max()/2) 215 | rdi_ab2 = np.minimum(rdi_ab2,rdi_ab2.max()/2) 216 | # 把不在手势范围的目标去除 217 | rdi_ab1[:,40:90] = 0 218 | rdi_ab2[:,40:90] = 0 219 | # rdi_ab1[:,13:19] = 0.1*rdi_ab1[:,13:19] 220 | # rdi_ab2[:,13:19] = 0.1*rdi_ab2[:,13:19] 221 | # rdi_ab1[:5,:] = 0 222 | # rdi_ab2[-5:,:] = 0 223 | # 加权 信噪比 224 | rdi_ab1 = rdi_ab1 / rdi_ab1.max() * frame_SNR 225 | rdi_ab2 = rdi_ab2 / rdi_ab2.max() * frame_SNR 226 | 227 | rai_queue.append(rdi_ab1) 228 | rei_queue.append(rdi_ab2) 229 | # 16个frame叠加返回 230 | rai_framearray = np.array(rai_queue)#frame chirps adcnum 231 | rei_framearray = np.array(rei_queue)#frame chirps adcnum 232 | 233 | return rai_framearray, rei_framearray 234 | 235 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | Creative Commons Attribution-NonCommercial 4.0 International Public License 2 | 3 | By using this software, you agree to the following terms: 4 | 5 | You are free to: 6 | - Share — copy and redistribute the material in any medium or format 7 | - Adapt — remix, transform, and build upon the material 8 | 9 | Under the following terms: 10 | - Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. 11 | - NonCommercial — You may not use the material for commercial purposes. 12 | 13 | No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits. 14 | 15 | Full license text: https://creativecommons.org/licenses/by-nc/4.0/legalcode 16 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # RadarStream 2 | 3 | RadarStream is a real-time RAWDATA acquisition, processing, and visualization system for TI MIMO mmWave radar series. 4 | 5 | 6 | 7 | https://github.com/user-attachments/assets/7ce99b51-a1af-4025-8a84-ee580eb92d04 8 | 9 | Demo1: Real-time Motion Detection and Radar Feature Visualization 10 |
11 | 图片描述 12 |
Demo2: Real-time Gesture Recognition System
13 |
14 | 15 | ## Project Overview 16 | 17 | This system supports Texas Instruments' MIMO mmWave radar series for real-time raw data acquisition, processing, and visualization. In addition to the RF evaluation board, the DCA1000EVM is required for data capture. Currently, the system has been tested with: 18 | - IWR6843ISK 19 | - IWR6843ISK-OBS 20 | - IWR1843ISK 21 | 22 | If you encounter any issues while using this project, please feel free to submit a pull request. 23 | 24 | ## Features ✨ 25 | 26 | * **Real-time, Multi-threaded Radar Data Acquisition from TI MIMO mmWave Radar Sensors:** 27 | * Leveraging a **multi-threaded architecture 🧵** for data acquisition and processing. 28 | * To overcome Python's Global Interpreter Lock (GIL) and enable true multi-core processing, the data acquisition module is **wrapped in C 🚀**, ensuring near real-time, frame-loss-free data capture and handling. 29 | * **Multi-dimensional Feature Extraction:** 30 | * Range-Time Information (RTI) 31 | * Doppler-Time Information (DTI) 32 | * Range-Doppler Information (RDI) 33 | * Range-Azimuth Information (RAI) 34 | * Range-Elevation Information (REI) 35 | * **Interactive Visualization Interface** 36 | 37 | ## Requirements 38 | 39 | - Python 3.6+ 40 | - PyQt5 41 | - PyQtGraph 42 | - NumPy 43 | - PyTorch 44 | - Matplotlib 45 | - Serial 46 | 47 | ## Hardware Requirements 48 | 49 | - TI MIMO mmWave Radar Sensor (tested with IWR6843ISK and IWR6843ISK-OBS) 50 | - DCA1000 EVM (essential for raw data capture) 51 | - PC with Windows OS 52 | 53 | ## Firmware Requirements 54 | The firmware must be selected from the `mmwave_industrial_toolbox_4_10_1\labs\Out_Of_Box_Demo\prebuilt_binaries/` directory inside any version of the mmwave_industrial_toolbox. 55 | There is no strict requirement to use version 4.10.1. 56 | 57 | ## Setup and Installation 58 | 59 | 1. Clone this repository 60 | 2. Install the required dependencies: 61 | ``` 62 | pip install pyqt5 pyqtgraph numpy torch matplotlib pyserial 63 | ``` 64 | 3. Connect the mmWave radar sensor and DCA1000 EVM to your computer (only need a 5V 3A DC power wire, a Ethernet Cable, and a micro USB wire) 65 | 4. Configure the network IPv4 settings (referencing the IPv4 configuration process from using mmWaveStudio for the DCA1000 EVM) 66 | 67 | Two different acquisition methods are shown here: one figure displays Raspberry Pi 4B acquisition, while the other demonstrates Windows-based acquisition. However, the Raspberry Pi acquisition has very few frames during real-time processing and display, making it prone to data loss. (not recommended to use Raspberry Pi for acquisition) 68 | 69 |

70 | 71 | 72 | 73 | 74 |

75 | 76 | ## 3D Printed Mount 77 | 78 | The repository includes STL files for a 3D printed structure designed to mount and secure the DCA1000EVM board. 79 | 80 | **Note:** You will need some M3 size nylon standoffs and screws for assembly. 81 | 82 |

83 | 84 |

85 | 86 | ## Usage 87 | 88 | 1. Run the main application: 89 | ``` 90 | python main.py 91 | ``` 92 | 2. Select the appropriate COM port for the radar CLI interface 93 | 3. Choose a radar configuration file 94 | 4. Click "Send Config" to initialize the radar 95 | 5. Use the interface to: 96 | - Visualize radar data in real-time 97 | - Capture training data for machine learning models 98 | 99 | 100 | ## Project Structure 101 | - `config/`: Configuration files for different radar settings 102 | - `gesture_icons/`: Gesture icons for visualization 103 | - `libs/`: Library files for radar communication 104 | - `STL3D`: 3D printed mount STL files 105 | - `main.py`: Main application entry point 106 | - `real_time_process.py`: Real-time data processing 107 | - `radar_config.py`: Radar configuration utilities 108 | - `iwr6843_tlv/`: TLV protocol implementation for IWR6843 109 | - `dsp/`: Digital signal processing modules 110 | - `UI_interface.py`: PyQt5 user interface 111 | 112 | 113 | ## Citation 114 | 115 | If this project helps your research, please consider citing our papers that are closely related to this tool: 116 | 117 | ``` 118 | 119 | ``` 120 | 121 | 122 | ## Acknowledgements 123 | 124 | This project references and builds upon: 125 | - [real-time-radar](https://github.com/AndyYu0010/real-time-radar) by AndyYu0010 126 | - [OpenRadar](https://github.com/PreSenseRadar/OpenRadar) - specifically the DSP module 127 | 128 | ## TODO 129 | 130 | Future improvements planned for this project: 131 | - [ ] Validate compatibility with more RF evaluation boards 132 | - [ ] Migrate from PyQt5 to PySide6 133 | - [ ] Make the API in the libs folder more flexible 134 | -------------------------------------------------------------------------------- /STL3D/GestureradarV1.2.SLDPRT: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/STL3D/GestureradarV1.2.SLDPRT -------------------------------------------------------------------------------- /STL3D/GestureradarV1.2.STL: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/STL3D/GestureradarV1.2.STL -------------------------------------------------------------------------------- /UI_interface.py: -------------------------------------------------------------------------------- 1 | # -*- coding: utf-8 -*- 2 | 3 | # Form implementation generated from reading ui file 'GestureRecognize.ui' 4 | # 5 | # Created by: PyQt5 UI code generator 5.12.3 6 | # 7 | # WARNING! All changes made in this file will be lost! 8 | 9 | from PyQt5 import QtCore, QtGui, QtWidgets 10 | from PyQt5.QtCore import pyqtSignal, Qt 11 | from PyQt5.QtWidgets import QComboBox 12 | import sys 13 | from pyqtgraph import GraphicsLayoutWidget 14 | import globalvar as gl 15 | import pyqtgraph as pg 16 | import os 17 | import matplotlib.cm 18 | import time 19 | import sys 20 | import os 21 | from PyQt5.QtGui import * 22 | from PyQt5.QtWidgets import * 23 | gl._init() 24 | gl.set_value('usr_gesture', False) 25 | 26 | 27 | radarRFconfigpathfile = 'config' 28 | modelpathfile = 'save_model' 29 | 30 | 31 | # ----新的combox 单机框任何地方都有响应 ---- # 32 | class ClickedComboBox(QComboBox): 33 | arrowClicked = pyqtSignal() 34 | 35 | def showPopup(self): # 重写showPopup函数 36 | super(ClickedComboBox, self).showPopup() 37 | self.arrowClicked.emit() 38 | 39 | # 最小化窗口 40 | class Qt_pet(QWidget): 41 | 42 | def __init__(self,MainWindow): 43 | super(Qt_pet, self).__init__() 44 | self.MainWindow = MainWindow 45 | self.dis_file = "gesture_icons/" 46 | self.windowinit() 47 | 48 | self.pos_first = self.pos() 49 | # self.timer.timeout.connect(self.img_update) 50 | 51 | def img_update(self,img_path): 52 | self.img_path = img_path 53 | self.qpixmap = QPixmap(self.img_path).scaled(256, 256) 54 | self.lab.setPixmap(self.qpixmap) 55 | 56 | def windowinit(self): 57 | self.x = 800 58 | self.y = 600 59 | self.setGeometry(self.x, self.y, 256, 256) 60 | self.img_path = 'gesture_icons/7.jpg' 61 | self.lab = QLabel(self) 62 | self.qpixmap = QPixmap(self.img_path).scaled(256, 256) 63 | self.lab.setPixmap(self.qpixmap) 64 | self.setWindowFlags(Qt.FramelessWindowHint | Qt.WindowStaysOnTopHint | Qt.SubWindow) 65 | self.setAutoFillBackground(False) 66 | self.setAttribute(Qt.WA_TranslucentBackground, True) 67 | # self.show() 68 | 69 | def mousePressEvent(self, QMouseEvent): 70 | if QMouseEvent.button() == Qt.LeftButton: 71 | self.pos_first = QMouseEvent.globalPos() - self.pos() 72 | QMouseEvent.accept() 73 | self.setCursor(QCursor(Qt.OpenHandCursor)) 74 | if QMouseEvent.button() == Qt.RightButton: 75 | self.MainWindow.show() 76 | self.hide() 77 | 78 | def mouseMoveEvent(self, QMouseEvent): 79 | if Qt.LeftButton: 80 | self.move(QMouseEvent.globalPos() - self.pos_first) 81 | # print(self.pos()) 82 | self.x, self.y = self.pos().x, self.pos().y 83 | QMouseEvent.accept() 84 | 85 | 86 | 87 | class Ui_MainWindow(object): 88 | def setupUi(self, MainWindow): 89 | MainWindow.setObjectName("MainWindow") 90 | MainWindow.resize(1280, 800) 91 | pg.setConfigOption('background', '#f0f0f0') 92 | pg.setConfigOption('foreground', 'd') 93 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 94 | sizePolicy.setHorizontalStretch(5) 95 | sizePolicy.setVerticalStretch(0) 96 | sizePolicy.setHeightForWidth(MainWindow.sizePolicy().hasHeightForWidth()) 97 | MainWindow.setSizePolicy(sizePolicy) 98 | MainWindow.setMinimumSize(QtCore.QSize(0, 0)) 99 | self.centralwidget = QtWidgets.QWidget(MainWindow) 100 | self.centralwidget.setObjectName("centralwidget") 101 | self.horizontalLayout_4 = QtWidgets.QHBoxLayout(self.centralwidget) 102 | self.horizontalLayout_4.setObjectName("horizontalLayout_4") 103 | self.tabWidget = QtWidgets.QTabWidget(self.centralwidget) 104 | self.tabWidget.setObjectName("tabWidget") 105 | self.tab = QtWidgets.QWidget() 106 | self.tab.setObjectName("tab") 107 | self.verticalLayout_3 = QtWidgets.QVBoxLayout(self.tab) 108 | self.verticalLayout_3.setObjectName("verticalLayout_3") 109 | self.splitter = QtWidgets.QSplitter(self.tab) 110 | self.splitter.setEnabled(True) 111 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Expanding, QtWidgets.QSizePolicy.Preferred) 112 | sizePolicy.setHorizontalStretch(0) 113 | sizePolicy.setVerticalStretch(0) 114 | sizePolicy.setHeightForWidth(self.splitter.sizePolicy().hasHeightForWidth()) 115 | self.splitter.setSizePolicy(sizePolicy) 116 | self.splitter.setAutoFillBackground(False) 117 | self.splitter.setOrientation(QtCore.Qt.Horizontal) 118 | self.splitter.setObjectName("splitter") 119 | self.groupBox_11 = QtWidgets.QGroupBox(self.splitter) 120 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Expanding, QtWidgets.QSizePolicy.Expanding) 121 | sizePolicy.setHorizontalStretch(5) 122 | sizePolicy.setVerticalStretch(15) 123 | sizePolicy.setHeightForWidth(self.groupBox_11.sizePolicy().hasHeightForWidth()) 124 | self.groupBox_11.setSizePolicy(sizePolicy) 125 | self.groupBox_11.setMinimumSize(QtCore.QSize(0, 0)) 126 | self.groupBox_11.setMaximumSize(QtCore.QSize(2400, 8000)) 127 | self.groupBox_11.setObjectName("groupBox_11") 128 | self.verticalLayout_4 = QtWidgets.QVBoxLayout(self.groupBox_11) 129 | self.verticalLayout_4.setContentsMargins(6, 6, 6, 6) 130 | self.verticalLayout_4.setObjectName("verticalLayout_4") 131 | self.groupBox_2 = QtWidgets.QGroupBox(self.groupBox_11) 132 | self.groupBox_2.setObjectName("groupBox_2") 133 | self.verticalLayout = QtWidgets.QVBoxLayout(self.groupBox_2) 134 | self.verticalLayout.setContentsMargins(6, 6, 6, 6) 135 | self.verticalLayout.setObjectName("verticalLayout") 136 | self.gridLayout_3 = QtWidgets.QGridLayout() 137 | self.gridLayout_3.setObjectName("gridLayout_3") 138 | self.label_15 = QtWidgets.QLabel(self.groupBox_2) 139 | self.label_15.setObjectName("label_15") 140 | self.gridLayout_3.addWidget(self.label_15, 3, 0, 1, 1) 141 | self.label_16 = QtWidgets.QLabel(self.groupBox_2) 142 | self.label_16.setObjectName("label_16") 143 | self.gridLayout_3.addWidget(self.label_16, 3, 1, 1, 1) 144 | self.label_36 = QtWidgets.QLabel(self.groupBox_2) 145 | self.label_36.setObjectName("label_36") 146 | self.gridLayout_3.addWidget(self.label_36, 4, 0, 1, 1) 147 | self.label_14 = QtWidgets.QLabel(self.groupBox_2) 148 | self.label_14.setObjectName("label_14") 149 | self.gridLayout_3.addWidget(self.label_14, 1, 1, 1, 1) 150 | self.label_43 = QtWidgets.QLabel(self.groupBox_2) 151 | self.label_43.setObjectName("label_43") 152 | self.gridLayout_3.addWidget(self.label_43, 6, 0, 1, 1) 153 | self.label_13 = QtWidgets.QLabel(self.groupBox_2) 154 | self.label_13.setObjectName("label_13") 155 | self.gridLayout_3.addWidget(self.label_13, 1, 0, 1, 1) 156 | self.label_35 = QtWidgets.QLabel(self.groupBox_2) 157 | self.label_35.setObjectName("label_35") 158 | self.gridLayout_3.addWidget(self.label_35, 2, 1, 1, 1) 159 | self.comboBox_7 = ClickedComboBox(self.groupBox_2) 160 | self.comboBox_7.setObjectName("comboBox_7") 161 | self.gridLayout_3.addWidget(self.comboBox_7, 0, 1, 1, 1) 162 | self.label_45 = QtWidgets.QLabel(self.groupBox_2) 163 | self.label_45.setAlignment(QtCore.Qt.AlignCenter) 164 | self.label_45.setObjectName("label_45") 165 | self.gridLayout_3.addWidget(self.label_45, 5, 0, 1, 2) 166 | self.comboBox_9 = QtWidgets.QComboBox(self.groupBox_2) 167 | self.comboBox_9.setEnabled(False) 168 | self.comboBox_9.setEditable(False) 169 | self.comboBox_9.setObjectName("comboBox_9") 170 | self.gridLayout_3.addWidget(self.comboBox_9, 7, 1, 1, 1) 171 | self.label_12 = QtWidgets.QLabel(self.groupBox_2) 172 | self.label_12.setObjectName("label_12") 173 | self.gridLayout_3.addWidget(self.label_12, 0, 0, 1, 1) 174 | self.label_44 = QtWidgets.QLabel(self.groupBox_2) 175 | self.label_44.setEnabled(False) 176 | self.label_44.setObjectName("label_44") 177 | self.gridLayout_3.addWidget(self.label_44, 7, 0, 1, 1) 178 | self.label_17 = QtWidgets.QLabel(self.groupBox_2) 179 | self.label_17.setObjectName("label_17") 180 | self.gridLayout_3.addWidget(self.label_17, 2, 0, 1, 1) 181 | self.label_37 = QtWidgets.QLabel(self.groupBox_2) 182 | self.label_37.setObjectName("label_37") 183 | self.gridLayout_3.addWidget(self.label_37, 4, 1, 1, 1) 184 | self.comboBox_8 = ClickedComboBox(self.groupBox_2) 185 | self.comboBox_8.setObjectName("comboBox_8") 186 | self.gridLayout_3.addWidget(self.comboBox_8, 6, 1, 1, 1) 187 | self.pushButton_11 = QtWidgets.QPushButton(self.groupBox_2) 188 | self.pushButton_11.setObjectName("pushButton_11") 189 | self.pushButton_12 = QtWidgets.QPushButton(self.groupBox_2) 190 | self.pushButton_12.setObjectName("pushButton_12") 191 | self.gridLayout_3.addWidget(self.pushButton_11, 8, 1, 1, 1) 192 | self.gridLayout_3.addWidget(self.pushButton_12, 8, 0, 1, 1) 193 | self.verticalLayout.addLayout(self.gridLayout_3) 194 | self.verticalLayout_4.addWidget(self.groupBox_2) 195 | self.groupBox_7 = QtWidgets.QGroupBox(self.groupBox_11) 196 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 197 | sizePolicy.setHorizontalStretch(0) 198 | sizePolicy.setVerticalStretch(2) 199 | sizePolicy.setHeightForWidth(self.groupBox_7.sizePolicy().hasHeightForWidth()) 200 | self.groupBox_7.setSizePolicy(sizePolicy) 201 | self.groupBox_7.setObjectName("groupBox_7") 202 | self.verticalLayout_7 = QtWidgets.QVBoxLayout(self.groupBox_7) 203 | self.verticalLayout_7.setContentsMargins(6, 6, 6, 6) 204 | self.verticalLayout_7.setObjectName("verticalLayout_7") 205 | self.horizontalLayout = QtWidgets.QHBoxLayout() 206 | self.horizontalLayout.setObjectName("horizontalLayout") 207 | self.label_5 = QtWidgets.QLabel(self.groupBox_7) 208 | self.label_5.setObjectName("label_5") 209 | self.horizontalLayout.addWidget(self.label_5) 210 | self.comboBox = ClickedComboBox(self.groupBox_7) 211 | self.comboBox.setObjectName("comboBox") 212 | self.horizontalLayout.addWidget(self.comboBox) 213 | self.verticalLayout_7.addLayout(self.horizontalLayout) 214 | self.groupBox_3 = QtWidgets.QGroupBox(self.groupBox_7) 215 | self.groupBox_3.setFlat(True) 216 | self.groupBox_3.setCheckable(True) 217 | self.groupBox_3.setChecked(False) 218 | self.groupBox_3.setObjectName("groupBox_3") 219 | self.verticalLayout_5 = QtWidgets.QVBoxLayout(self.groupBox_3) 220 | self.verticalLayout_5.setContentsMargins(2, 6, 2, 6) 221 | self.verticalLayout_5.setObjectName("verticalLayout_5") 222 | self.horizontalLayout_3 = QtWidgets.QHBoxLayout() 223 | self.horizontalLayout_3.setObjectName("horizontalLayout_3") 224 | self.label_2 = QtWidgets.QLabel(self.groupBox_3) 225 | self.label_2.setObjectName("label_2") 226 | self.horizontalLayout_3.addWidget(self.label_2) 227 | self.comboBox_2 = ClickedComboBox(self.groupBox_3) 228 | self.comboBox_2.setObjectName("comboBox_2") 229 | self.horizontalLayout_3.addWidget(self.comboBox_2) 230 | self.verticalLayout_5.addLayout(self.horizontalLayout_3) 231 | self.pushButton_15 = QtWidgets.QPushButton(self.groupBox_3) 232 | self.pushButton_15.setObjectName("pushButton_15") 233 | self.pushButton_15.setCheckable(True) 234 | self.verticalLayout_5.addWidget(self.pushButton_15) 235 | self.verticalLayout_7.addWidget(self.groupBox_3) 236 | self.groupBox_4 = QtWidgets.QGroupBox(self.groupBox_7) 237 | self.groupBox_4.setFlat(True) 238 | self.groupBox_4.setCheckable(True) 239 | self.groupBox_4.setChecked(False) 240 | self.groupBox_4.setObjectName("groupBox_4") 241 | self.verticalLayout_6 = QtWidgets.QVBoxLayout(self.groupBox_4) 242 | self.verticalLayout_6.setContentsMargins(2, 6, 2, 6) 243 | self.verticalLayout_6.setObjectName("verticalLayout_6") 244 | self.horizontalLayout_5 = QtWidgets.QHBoxLayout() 245 | self.horizontalLayout_5.setObjectName("horizontalLayout_5") 246 | self.label_3 = QtWidgets.QLabel(self.groupBox_4) 247 | self.label_3.setObjectName("label_3") 248 | self.horizontalLayout_5.addWidget(self.label_3) 249 | self.lineEdit_6 = QtWidgets.QLineEdit(self.groupBox_4) 250 | self.lineEdit_6.setObjectName("lineEdit_6") 251 | self.horizontalLayout_5.addWidget(self.lineEdit_6) 252 | self.comboBox_3 = QtWidgets.QComboBox(self.groupBox_4) 253 | self.comboBox_3.setObjectName("comboBox_3") 254 | self.comboBox_3.addItem("") 255 | self.comboBox_3.addItem("") 256 | self.comboBox_3.addItem("") 257 | self.comboBox_3.addItem("") 258 | self.comboBox_3.addItem("") 259 | self.comboBox_3.addItem("") 260 | self.comboBox_3.addItem("") 261 | self.horizontalLayout_5.addWidget(self.comboBox_3) 262 | self.verticalLayout_6.addLayout(self.horizontalLayout_5) 263 | self.pushButton = QtWidgets.QPushButton(self.groupBox_4) 264 | self.pushButton.setCheckable(True) 265 | self.pushButton.setObjectName("pushButton") 266 | self.verticalLayout_6.addWidget(self.pushButton) 267 | self.verticalLayout_7.addWidget(self.groupBox_4) 268 | self.verticalLayout_4.addWidget(self.groupBox_7) 269 | self.groupBox = QtWidgets.QGroupBox(self.groupBox_11) 270 | self.groupBox.setObjectName("groupBox") 271 | self.verticalLayout_2 = QtWidgets.QVBoxLayout(self.groupBox) 272 | self.verticalLayout_2.setSizeConstraint(QtWidgets.QLayout.SetDefaultConstraint) 273 | self.verticalLayout_2.setContentsMargins(5, 5, 5, 5) 274 | self.verticalLayout_2.setSpacing(0) 275 | self.verticalLayout_2.setObjectName("verticalLayout_2") 276 | self.textEdit = QtWidgets.QTextEdit(self.groupBox) 277 | self.textEdit.setFrameShape(QtWidgets.QFrame.StyledPanel) 278 | self.textEdit.setFrameShadow(QtWidgets.QFrame.Sunken) 279 | self.textEdit.setTabStopWidth(80) 280 | self.textEdit.setObjectName("textEdit") 281 | self.verticalLayout_2.addWidget(self.textEdit) 282 | self.verticalLayout_4.addWidget(self.groupBox) 283 | self.groupBox_9 = QtWidgets.QGroupBox(self.splitter) 284 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 285 | sizePolicy.setHorizontalStretch(10) 286 | sizePolicy.setVerticalStretch(15) 287 | sizePolicy.setHeightForWidth(self.groupBox_9.sizePolicy().hasHeightForWidth()) 288 | self.groupBox_9.setSizePolicy(sizePolicy) 289 | self.groupBox_9.setObjectName("groupBox_9") 290 | self.gridLayout = QtWidgets.QGridLayout(self.groupBox_9) 291 | self.gridLayout.setContentsMargins(11, 11, 11, 11) 292 | self.gridLayout.setHorizontalSpacing(7) 293 | self.gridLayout.setObjectName("gridLayout") 294 | self.label = QtWidgets.QLabel(self.groupBox_9) 295 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 296 | sizePolicy.setHorizontalStretch(0) 297 | sizePolicy.setVerticalStretch(0) 298 | sizePolicy.setHeightForWidth(self.label.sizePolicy().hasHeightForWidth()) 299 | self.label.setSizePolicy(sizePolicy) 300 | self.label.setAlignment(QtCore.Qt.AlignCenter) 301 | self.label.setObjectName("label") 302 | self.gridLayout.addWidget(self.label, 0, 0, 1, 1) 303 | self.label_7 = QtWidgets.QLabel(self.groupBox_9) 304 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 305 | sizePolicy.setHorizontalStretch(0) 306 | sizePolicy.setVerticalStretch(0) 307 | sizePolicy.setHeightForWidth(self.label_7.sizePolicy().hasHeightForWidth()) 308 | self.label_7.setSizePolicy(sizePolicy) 309 | self.label_7.setAlignment(QtCore.Qt.AlignCenter) 310 | self.label_7.setObjectName("label_7") 311 | self.gridLayout.addWidget(self.label_7, 0, 1, 1, 1) 312 | self.label_8 = QtWidgets.QLabel(self.groupBox_9) 313 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 314 | sizePolicy.setHorizontalStretch(0) 315 | sizePolicy.setVerticalStretch(0) 316 | sizePolicy.setHeightForWidth(self.label_8.sizePolicy().hasHeightForWidth()) 317 | self.label_8.setSizePolicy(sizePolicy) 318 | self.label_8.setAlignment(QtCore.Qt.AlignCenter) 319 | self.label_8.setObjectName("label_8") 320 | self.gridLayout.addWidget(self.label_8, 0, 2, 1, 1) 321 | self.graphicsView_4 = GraphicsLayoutWidget(self.groupBox_9) 322 | self.graphicsView_4.setEnabled(True) 323 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 324 | sizePolicy.setHorizontalStretch(0) 325 | sizePolicy.setVerticalStretch(1) 326 | sizePolicy.setHeightForWidth(self.graphicsView_4.sizePolicy().hasHeightForWidth()) 327 | self.graphicsView_4.setSizePolicy(sizePolicy) 328 | self.graphicsView_4.setMinimumSize(QtCore.QSize(255, 255)) 329 | self.graphicsView_4.setMaximumSize(QtCore.QSize(255, 255)) 330 | self.graphicsView_4.setObjectName("graphicsView_4") 331 | self.gridLayout.addWidget(self.graphicsView_4, 3, 1, 1, 1) 332 | self.graphicsView_2 = GraphicsLayoutWidget(self.groupBox_9) 333 | self.graphicsView_2.setEnabled(True) 334 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 335 | sizePolicy.setHorizontalStretch(0) 336 | sizePolicy.setVerticalStretch(1) 337 | sizePolicy.setHeightForWidth(self.graphicsView_2.sizePolicy().hasHeightForWidth()) 338 | self.graphicsView_2.setSizePolicy(sizePolicy) 339 | self.graphicsView_2.setMinimumSize(QtCore.QSize(255, 255)) 340 | self.graphicsView_2.setMaximumSize(QtCore.QSize(255, 255)) 341 | self.graphicsView_2.setObjectName("graphicsView_2") 342 | self.gridLayout.addWidget(self.graphicsView_2, 1, 1, 1, 1) 343 | self.label_9 = QtWidgets.QLabel(self.groupBox_9) 344 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 345 | sizePolicy.setHorizontalStretch(0) 346 | sizePolicy.setVerticalStretch(0) 347 | sizePolicy.setHeightForWidth(self.label_9.sizePolicy().hasHeightForWidth()) 348 | self.label_9.setSizePolicy(sizePolicy) 349 | self.label_9.setAlignment(QtCore.Qt.AlignCenter) 350 | self.label_9.setObjectName("label_9") 351 | self.gridLayout.addWidget(self.label_9, 2, 1, 1, 1) 352 | self.graphicsView = GraphicsLayoutWidget(self.groupBox_9) 353 | self.graphicsView.setEnabled(True) 354 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 355 | sizePolicy.setHorizontalStretch(0) 356 | sizePolicy.setVerticalStretch(1) 357 | sizePolicy.setHeightForWidth(self.graphicsView.sizePolicy().hasHeightForWidth()) 358 | self.graphicsView.setSizePolicy(sizePolicy) 359 | self.graphicsView.setMinimumSize(QtCore.QSize(255, 255)) 360 | self.graphicsView.setMaximumSize(QtCore.QSize(255, 255)) 361 | self.graphicsView.setObjectName("graphicsView") 362 | self.gridLayout.addWidget(self.graphicsView, 1, 0, 1, 1) 363 | self.graphicsView_3 = GraphicsLayoutWidget(self.groupBox_9) 364 | self.graphicsView_3.setEnabled(True) 365 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 366 | sizePolicy.setHorizontalStretch(0) 367 | sizePolicy.setVerticalStretch(1) 368 | sizePolicy.setHeightForWidth(self.graphicsView_3.sizePolicy().hasHeightForWidth()) 369 | self.graphicsView_3.setSizePolicy(sizePolicy) 370 | self.graphicsView_3.setMinimumSize(QtCore.QSize(255, 255)) 371 | self.graphicsView_3.setMaximumSize(QtCore.QSize(255, 255)) 372 | self.graphicsView_3.setObjectName("graphicsView_3") 373 | self.gridLayout.addWidget(self.graphicsView_3, 1, 2, 1, 1) 374 | self.graphicsView_5 = QtWidgets.QLabel(self.groupBox_9) 375 | self.graphicsView_5.setStyleSheet('border-width: 2px;border-style: solid;border-color: rgb(255, 170, 0);background-color: rgb(180,180,180);') 376 | self.graphicsView_5.setEnabled(True) 377 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 378 | sizePolicy.setHorizontalStretch(0) 379 | sizePolicy.setVerticalStretch(1) 380 | sizePolicy.setHeightForWidth(self.graphicsView_5.sizePolicy().hasHeightForWidth()) 381 | self.graphicsView_5.setSizePolicy(sizePolicy) 382 | self.graphicsView_5.setMinimumSize(QtCore.QSize(255, 255)) 383 | self.graphicsView_5.setMaximumSize(QtCore.QSize(255, 255)) 384 | self.graphicsView_5.setObjectName("graphicsView_5") 385 | self.gridLayout.addWidget(self.graphicsView_5, 3, 2, 1, 1) 386 | self.label_11 = QtWidgets.QLabel(self.groupBox_9) 387 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 388 | sizePolicy.setHorizontalStretch(0) 389 | sizePolicy.setVerticalStretch(0) 390 | sizePolicy.setHeightForWidth(self.label_11.sizePolicy().hasHeightForWidth()) 391 | self.label_11.setSizePolicy(sizePolicy) 392 | self.label_11.setAlignment(QtCore.Qt.AlignCenter) 393 | self.label_11.setObjectName("label_11") 394 | self.gridLayout.addWidget(self.label_11, 2, 0, 1, 1) 395 | self.label_10 = QtWidgets.QLabel(self.groupBox_9) 396 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 397 | sizePolicy.setHorizontalStretch(0) 398 | sizePolicy.setVerticalStretch(0) 399 | sizePolicy.setHeightForWidth(self.label_10.sizePolicy().hasHeightForWidth()) 400 | self.label_10.setSizePolicy(sizePolicy) 401 | self.label_10.setAlignment(QtCore.Qt.AlignCenter) 402 | self.label_10.setObjectName("label_10") 403 | self.gridLayout.addWidget(self.label_10, 2, 2, 1, 1) 404 | self.graphicsView_6 = GraphicsLayoutWidget(self.groupBox_9) 405 | self.graphicsView_6.setEnabled(True) 406 | sizePolicy = QtWidgets.QSizePolicy(QtWidgets.QSizePolicy.Preferred, QtWidgets.QSizePolicy.Preferred) 407 | sizePolicy.setHorizontalStretch(0) 408 | sizePolicy.setVerticalStretch(1) 409 | sizePolicy.setHeightForWidth(self.graphicsView_6.sizePolicy().hasHeightForWidth()) 410 | self.graphicsView_6.setSizePolicy(sizePolicy) 411 | self.graphicsView_6.setMinimumSize(QtCore.QSize(255, 255)) 412 | self.graphicsView_6.setMaximumSize(QtCore.QSize(255, 255)) 413 | self.graphicsView_6.setObjectName("graphicsView_6") 414 | self.gridLayout.addWidget(self.graphicsView_6, 3, 0, 1, 1) 415 | self.gridLayout.setRowStretch(0, 1) 416 | self.gridLayout.setRowStretch(1, 4) 417 | self.gridLayout.setRowStretch(2, 1) 418 | self.gridLayout.setRowStretch(3, 4) 419 | self.verticalLayout_3.addWidget(self.splitter) 420 | 421 | # 比例 控制 splitter 内 控件的比例 422 | self.splitter.addWidget(self.groupBox_9)#'你的第一个子widget' 423 | self.splitter.addWidget(self.groupBox_11)#'你的第二个子widget' 424 | self.splitter.setSizes([960,320])#直接写入数字列表 425 | 426 | self.tabWidget.addTab(self.tab, "") 427 | self.tab_2 = QtWidgets.QWidget() 428 | self.tab_2.setObjectName("tab_2") 429 | self.verticalLayout_8 = QtWidgets.QVBoxLayout(self.tab_2) 430 | self.verticalLayout_8.setObjectName("verticalLayout_8") 431 | self.label_4 = QtWidgets.QLabel(self.tab_2) 432 | self.label_4.setAlignment(QtCore.Qt.AlignCenter) 433 | # self.label_4.setPixmap(QtGui.QPixmap("visualization/depict.jpg")) 434 | self.label_4.setObjectName("label_4") 435 | self.verticalLayout_8.addWidget(self.label_4) 436 | self.tabWidget.addTab(self.tab_2, "") 437 | self.horizontalLayout_4.addWidget(self.tabWidget) 438 | MainWindow.setCentralWidget(self.centralwidget) 439 | self.menubar = QtWidgets.QMenuBar(MainWindow) 440 | self.menubar.setGeometry(QtCore.QRect(0, 0, 1079, 22)) 441 | self.menubar.setObjectName("menubar") 442 | self.menu = QtWidgets.QMenu(self.menubar) 443 | self.menu.setObjectName("menu") 444 | MainWindow.setMenuBar(self.menubar) 445 | self.statusbar = QtWidgets.QStatusBar(MainWindow) 446 | self.statusbar.setObjectName("statusbar") 447 | MainWindow.setStatusBar(self.statusbar) 448 | self.actionload = QtWidgets.QAction(MainWindow) 449 | self.actionload.setObjectName("actionload") 450 | self.menu.addSeparator() 451 | self.menu.addAction(self.actionload) 452 | self.menu.addSeparator() 453 | self.menubar.addAction(self.menu.menuAction()) 454 | 455 | self.retranslateUi(MainWindow) 456 | self.tabWidget.setCurrentIndex(0) 457 | self.slot_init() 458 | QtCore.QMetaObject.connectSlotsByName(MainWindow) 459 | 460 | def retranslateUi(self, MainWindow): 461 | _translate = QtCore.QCoreApplication.translate 462 | MainWindow.setWindowTitle(_translate("MainWindow", "MainWindow")) 463 | self.groupBox_11.setTitle(_translate("MainWindow", "配置")) 464 | self.groupBox_2.setTitle(_translate("MainWindow", "雷达配置")) 465 | self.label_15.setText(_translate("MainWindow", "最大距离：")) 466 | self.label_16.setText(_translate("MainWindow", "80m")) 467 | self.label_36.setText(_translate("MainWindow", "最大速度：")) 468 | self.label_14.setText(_translate("MainWindow", "3.75cm")) 469 | self.label_43.setText(_translate("MainWindow", "CLIPort:")) 470 | self.label_13.setText(_translate("MainWindow", "距离分辨率：")) 471 | self.label_35.setText(_translate("MainWindow", "2m/s")) 472 | self.comboBox_7.setToolTip(_translate("MainWindow", "配置文件放到config文件夹下")) 473 | self.label_45.setText(_translate("MainWindow", "其他待续...")) 474 | self.label_12.setText(_translate("MainWindow", "配置文件：")) 475 | self.label_44.setText(_translate("MainWindow", "DataPort:")) 476 | self.label_17.setText(_translate("MainWindow", "速度分辨率：")) 477 | self.label_37.setText(_translate("MainWindow", "20m/s")) 478 | self.pushButton_11.setText(_translate("MainWindow", "send")) 479 | self.pushButton_12.setText(_translate("MainWindow", "Exit")) 480 | self.groupBox_7.setTitle(_translate("MainWindow", "采集/识别")) 481 | self.label_5.setText(_translate("MainWindow", "color：")) 482 | self.groupBox_3.setTitle(_translate("MainWindow", "识别")) 483 | self.label_2.setText(_translate("MainWindow", "model：")) 484 | self.pushButton_15.setText(_translate("MainWindow", "recognize")) 485 | self.groupBox_4.setTitle(_translate("MainWindow", "采集")) 486 | self.label_3.setText(_translate("MainWindow", "scene：")) 487 | self.lineEdit_6.setText(_translate("MainWindow", "chaotic_dataset")) 488 | self.comboBox_3.setItemText(0, _translate("MainWindow", "Back")) 489 | self.comboBox_3.setItemText(1, _translate("MainWindow", "Dblclick")) 490 | self.comboBox_3.setItemText(2, _translate("MainWindow", "Down")) 491 | self.comboBox_3.setItemText(3, _translate("MainWindow", "Front")) 492 | self.comboBox_3.setItemText(4, _translate("MainWindow", "Left")) 493 | self.comboBox_3.setItemText(5, _translate("MainWindow", "Right")) 494 | self.comboBox_3.setItemText(6, _translate("MainWindow", "Up")) 495 | self.pushButton.setText(_translate("MainWindow", "capture")) 496 | self.groupBox.setTitle(_translate("MainWindow", "printlog")) 497 | self.groupBox_9.setTitle(_translate("MainWindow", "雷达数据实时显示")) 498 | self.label.setText(_translate("MainWindow", "距离-时间图")) 499 | self.label_7.setText(_translate("MainWindow", "多普勒-时间图")) 500 | self.label_8.setText(_translate("MainWindow", "距离-俯仰角度图")) 501 | self.label_9.setText(_translate("MainWindow", "距离-方位角度图")) 502 | self.label_11.setText(_translate("MainWindow", "距离-多普勒图")) 503 | self.label_10.setText(_translate("MainWindow", "手势输出")) 504 | self.tabWidget.setTabText(self.tabWidget.indexOf(self.tab), _translate("MainWindow", "real-time system")) 505 | # self.label_4.setText(_translate("MainWindow", "TextLabel")) 506 | self.tabWidget.setTabText(self.tabWidget.indexOf(self.tab_2), _translate("MainWindow", "waiting...")) 507 | self.menu.setTitle(_translate("MainWindow", "菜单")) 508 | self.actionload.setText(_translate("MainWindow", "mini")) 509 | self.actionload.setToolTip(_translate("MainWindow", "mini窗口")) 510 | 511 | def printlog(self,textEdit,string,fontcolor='green'): 512 | textEdit.moveCursor(QtGui.QTextCursor.End) 513 | gettime = time.strftime("%H:%M:%S", time.localtime()) 514 | textEdit.append(""+str(gettime)+"-->"+string+"") 515 | 516 | def slot_init(self): 517 | self.printlog(self.textEdit, 'Welcome!',fontcolor='green') 518 | self.getcolorlist() 519 | self.comboBox.currentIndexChanged.connect(self.setcolor) 520 | self.comboBox_7.arrowClicked.connect(self.configpath) #配置文件 521 | self.comboBox_2.arrowClicked.connect(self.modelpath) #选择识别模型 522 | self.groupBox_3.clicked.connect(lambda:self.IsRecognizeorCapture(box_name = 'box_3')) 523 | self.groupBox_4.clicked.connect(lambda:self.IsRecognizeorCapture(box_name = 'box_4')) 524 | 525 | self.pushButton_15.clicked.connect(lambda:self.Iscapture(self.pushButton_15,'recognizing','recognize')) 526 | self.pushButton.clicked.connect(lambda:self.Iscapture(self.pushButton,'capturing','capture')) 527 | 528 | def IsRecognizeorCapture(self,box_name): 529 | if box_name == 'box_3' and self.groupBox_3.isChecked(): 530 | self.groupBox_4.setChecked(False) 531 | self.pushButton.setChecked(False) 532 | self.Iscapture(self.pushButton,'capturing','capture') 533 | elif box_name == 'box_4'and self.groupBox_4.isChecked(): 534 | self.groupBox_3.setChecked(False) 535 | self.pushButton_15.setChecked(False) 536 | self.Iscapture(self.pushButton_15,'recognizing','recognize') 537 | elif box_name == 'box_3' and self.groupBox_3.isChecked()==False: 538 | gl.set_value('IsRecognizeorCapture',False) 539 | # self.printlog(self.textEdit, 'IsRecognizeorCapture:False',fontcolor='green') 540 | self.graphicsView_5.setPixmap(QtGui.QPixmap("gesture_icons/7.jpg")) 541 | elif box_name == 'box_4' and self.groupBox_4.isChecked()==False: 542 | gl.set_value('IsRecognizeorCapture',False) 543 | # self.printlog(self.textEdit, 'IsRecognizeorCapture:False',fontcolor='green') 544 | self.graphicsView_5.setPixmap(QtGui.QPixmap("gesture_icons/7.jpg")) 545 | 546 | def Iscapture(self,btn,text1,text2): 547 | if btn.isChecked(): 548 | gl.set_value('IsRecognizeorCapture',True) 549 | # self.printlog(self.textEdit, 'IsRecognizeorCapture:True',fontcolor='green') 550 | btn.setText(text1) 551 | else: 552 | gl.set_value('IsRecognizeorCapture',False) 553 | self.printlog(self.textEdit, 'IsRecognizeorCapture:False',fontcolor='green') 554 | self.graphicsView_5.setPixmap(QtGui.QPixmap("gesture_icons/7.jpg")) 555 | btn.setText(text2) 556 | 557 | def modelpath(self): 558 | self.comboBox_2.clear() 559 | self.comboBox_2.addItem("--select--") 560 | list = [] 561 | if (os.path.exists(modelpathfile)): 562 | files = os.listdir(modelpathfile) 563 | for file in files: 564 | list.append(modelpathfile+'/'+file) 565 | self.comboBox_2.addItems(list) 566 | 567 | def configpath(self): 568 | self.comboBox_7.clear() 569 | self.comboBox_7.addItem("--select--") 570 | list = [] 571 | if (os.path.exists(radarRFconfigpathfile)): 572 | files = os.listdir(radarRFconfigpathfile) 573 | for file in files: 574 | list.append(radarRFconfigpathfile+'/'+file) 575 | self.comboBox_7.addItems(list) 576 | 577 | def getcolorlist(self): 578 | values=matplotlib.cm.cmap_d.keys() 579 | self.comboBox.addItem("--select--") 580 | self.comboBox.addItem("customize") 581 | for value in values: 582 | self.comboBox.addItem(value) 583 | 584 | def setcolor(self): 585 | if(self.comboBox.currentText()!='--select--' and self.comboBox.currentText()!=''): 586 | self.printlog(self.textEdit, 'selected color:'+self.comboBox.currentText(),fontcolor='blue') 587 | 588 | 589 | 590 | 591 | if __name__ == '__main__': 592 | app = QtWidgets.QApplication(sys.argv) 593 | MainWindow = QtWidgets.QMainWindow() 594 | MainWindow.show() 595 | ui = Ui_MainWindow() 596 | ui.setupUi(MainWindow) 597 | # subWin = Qt_pet() 598 | sys.exit(app.exec_()) -------------------------------------------------------------------------------- /colortrans.py: -------------------------------------------------------------------------------- 1 | # MIT License 2 | # 3 | # Copyright (c) 2020 Sebastian Höfer 4 | # 5 | # Permission is hereby granted, free of charge, to any person obtaining a copy 6 | # of this software and associated documentation files (the "Software"), to deal 7 | # in the Software without restriction, including without limitation the rights 8 | # to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | # copies of the Software, and to permit persons to whom the Software is 10 | # furnished to do so, subject to the following conditions: 11 | # 12 | # The above copyright notice and this permission notice shall be included in all 13 | # copies or substantial portions of the Software. 14 | # 15 | # THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | # AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | # SOFTWARE. 22 | 23 | 24 | """ 25 | This is an example how to use an ImageView with Matplotlib Colormaps (cmap). 26 | The function 'cmapToColormap' converts the Matplotlib format to the internal 27 | format of PyQtGraph that is used in the GradientEditorItem. The function 28 | itself has no dependencies on Matplotlib! Hence the weird if clauses with 29 | 'hasattr' instead of 'isinstance'. 30 | """ 31 | 32 | 33 | import numpy as np 34 | from pyqtgraph.Qt import QtCore, QtGui 35 | import pyqtgraph 36 | import matplotlib 37 | import collections 38 | 39 | 40 | def cmapToColormap(cmap, nTicks=64): 41 | """ 42 | Converts a Matplotlib cmap to pyqtgraphs colormaps. No dependency on matplotlib. 43 | Parameters: 44 | *cmap*: Cmap object. Imported from matplotlib.cm.* 45 | *nTicks*: Number of ticks to create when dict of functions is used. Otherwise unused. 46 | """ 47 | 48 | # Case #1: a dictionary with 'red'/'green'/'blue' values as list of ranges (e.g. 'jet') 49 | # The parameter 'cmap' is a 'matplotlib.colors.LinearSegmentedColormap' instance ... 50 | if hasattr(cmap, '_segmentdata'): 51 | colordata = getattr(cmap, '_segmentdata') 52 | if ('red' in colordata) and isinstance(colordata['red'], collections.Sequence): 53 | 54 | # collect the color ranges from all channels into one dict to get unique indices 55 | posDict = {} 56 | for idx, channel in enumerate(('red', 'green', 'blue')): 57 | for colorRange in colordata[channel]: 58 | posDict.setdefault(colorRange[0], [-1, -1, -1])[idx] = colorRange[2] 59 | 60 | indexList = list(posDict.keys()) 61 | indexList.sort() 62 | # interpolate missing values (== -1) 63 | for channel in range(3): # R,G,B 64 | startIdx = indexList[0] 65 | emptyIdx = [] 66 | for curIdx in indexList: 67 | if posDict[curIdx][channel] == -1: 68 | emptyIdx.append(curIdx) 69 | elif curIdx != indexList[0]: 70 | for eIdx in emptyIdx: 71 | rPos = (eIdx - startIdx) / (curIdx - startIdx) 72 | vStart = posDict[startIdx][channel] 73 | vRange = (posDict[curIdx][channel] - posDict[startIdx][channel]) 74 | posDict[eIdx][channel] = rPos * vRange + vStart 75 | startIdx = curIdx 76 | del emptyIdx[:] 77 | for channel in range(3): # R,G,B 78 | for curIdx in indexList: 79 | posDict[curIdx][channel] *= 255 80 | 81 | rgb_list = [[i, posDict[i]] for i in indexList] 82 | 83 | # Case #2: a dictionary with 'red'/'green'/'blue' values as functions (e.g. 'gnuplot') 84 | elif ('red' in colordata) and isinstance(colordata['red'], collections.Callable): 85 | indices = np.linspace(0., 1., nTicks) 86 | luts = [np.clip(np.array(colordata[rgb](indices), dtype=np.float), 0, 1) * 255 \ 87 | for rgb in ('red', 'green', 'blue')] 88 | rgb_list = zip(indices, list(zip(*luts))) 89 | 90 | # If the parameter 'cmap' is a 'matplotlib.colors.ListedColormap' instance, with the attributes 'colors' and 'N' 91 | elif hasattr(cmap, 'colors') and hasattr(cmap, 'N'): 92 | colordata = getattr(cmap, 'colors') 93 | # Case #3: a list with RGB values (e.g. 'seismic') 94 | if len(colordata[0]) == 3: 95 | indices = np.linspace(0., 1., len(colordata)) 96 | scaledRgbTuples = [(rgbTuple[0] * 255, rgbTuple[1] * 255, rgbTuple[2] * 255) for rgbTuple in colordata] 97 | rgb_list = zip(indices, scaledRgbTuples) 98 | 99 | # Case #4: a list of tuples with positions and RGB-values (e.g. 'terrain') 100 | # -> this section is probably not needed anymore!? 101 | elif len(colordata[0]) == 2: 102 | rgb_list = [(idx, (vals[0] * 255, vals[1] * 255, vals[2] * 255)) for idx, vals in colordata] 103 | 104 | # Case #X: unknown format or datatype was the wrong object type 105 | else: 106 | raise ValueError("[cmapToColormap] Unknown cmap format or not a cmap!") 107 | 108 | # Convert the RGB float values to RGBA integer values 109 | return list([(pos, (int(r), int(g), int(b), 255)) for pos, (r, g, b) in rgb_list]) 110 | 111 | 112 | def pg_get_cmap(cmap): 113 | if(cmap=='customize'): 114 | position = np.arange(64) 115 | position = position / 64 116 | position[0] = 0 117 | position = np.flip(position) 118 | colors = [[62, 38, 168, 255], [63, 42, 180, 255], [65, 46, 191, 255], [67, 50, 202, 255], [69, 55, 213, 255], 119 | [70, 60, 222, 255], [71, 65, 229, 255], [70, 71, 233, 255], [70, 77, 236, 255], [69, 82, 240, 255], 120 | [68, 88, 243, 255], 121 | [68, 94, 247, 255], [67, 99, 250, 255], [66, 105, 254, 255], [62, 111, 254, 255], [56, 117, 254, 255], 122 | [50, 123, 252, 255], 123 | [47, 129, 250, 255], [46, 135, 246, 255], [45, 140, 243, 255], [43, 146, 238, 255], [39, 150, 235, 255], 124 | [37, 155, 232, 255], 125 | [35, 160, 229, 255], [31, 164, 225, 255], [28, 129, 222, 255], [24, 173, 219, 255], [17, 177, 214, 255], 126 | [7, 181, 208, 255], 127 | [1, 184, 202, 255], [2, 186, 195, 255], [11, 189, 188, 255], [24, 191, 182, 255], [36, 193, 174, 255], 128 | [44, 195, 167, 255], 129 | [49, 198, 159, 255], [55, 200, 151, 255], [63, 202, 142, 255], [74, 203, 132, 255], [88, 202, 121, 255], 130 | [102, 202, 111, 255], 131 | [116, 201, 100, 255], [130, 200, 89, 255], [144, 200, 78, 255], [157, 199, 68, 255], [171, 199, 57, 255], 132 | [185, 196, 49, 255], 133 | [197, 194, 42, 255], [209, 191, 39, 255], [220, 189, 41, 255], [230, 187, 45, 255], [239, 186, 53, 255], 134 | [248, 186, 61, 255], 135 | [254, 189, 60, 255], [252, 196, 57, 255], [251, 202, 53, 255], [249, 208, 50, 255], [248, 214, 46, 255], 136 | [246, 220, 43, 255], 137 | [245, 227, 39, 255], [246, 233, 35, 255], [246, 239, 31, 255], [247, 245, 27, 255], [249, 251, 20, 255]] 138 | colors = np.flip(colors, axis=0) 139 | color_map = pyqtgraph.ColorMap(position, colors) 140 | return color_map 141 | else: 142 | 143 | pos, rgba_colors = zip(*cmapToColormap(cmap)) 144 | # Set the colormap 145 | pgColormap = pyqtgraph.ColorMap(pos, rgba_colors) 146 | return pgColormap 147 | 148 | 149 | 150 | # Start Qt event loop unless running in interactive mode. 151 | if __name__ == '__main__': 152 | import sys 153 | import matplotlib.cm 154 | 155 | if (sys.flags.interactive != 1) or not hasattr(QtCore, 'PYQT_VERSION'): 156 | app = QtGui.QApplication([]) 157 | 158 | # Create window with ImageView widget 159 | win = pyqtgraph.GraphicsLayoutWidget() 160 | win.resize(800,800) 161 | win.show() 162 | 163 | imv = pyqtgraph.ImageItem() 164 | 165 | view = win.addViewBox() 166 | view.addItem(imv) 167 | win.setCentralWidget(view) 168 | win.show() 169 | win.setWindowTitle('Example: Matplotlib colormaps in pyqtgraph ImageView') 170 | 171 | # Create random data set with noisy signals 172 | data = np.random.normal(size=(200, 200)) 173 | gradient = np.linspace(0., 10., 200) 174 | data += gradient[:, np.newaxis] 175 | 176 | 177 | # Convert a matplotlib colormap into a list of (tickmark, (r,g,b)) tuples 178 | pos, rgba_colors = zip(*cmapToColormap(matplotlib.cm.coolwarm)) 179 | # Set the colormap 180 | pgColormap = pyqtgraph.ColorMap(pos, rgba_colors) 181 | imv.setLookupTable(pgColormap.getLookupTable()) 182 | 183 | # Display the data 184 | imv.setImage(data) 185 | 186 | QtGui.QApplication.instance().exec_() -------------------------------------------------------------------------------- /config/IWR1843_cfg.cfg: -------------------------------------------------------------------------------- 1 | % *************************************************************** 2 | % Created for SDK ver:03.05 3 | % Created using Visualizer ver:3.5.0.0 4 | % Frequency:77 5 | % Platform:xWR18xx 6 | % Scene Classifier:best_range 7 | % Azimuth Resolution(deg):15 8 | % Range Resolution(m):0.47 9 | % Maximum unambiguous Range(m):50 10 | % Maximum Radial Velocity(m/s):18.28 11 | % Radial velocity resolution(m/s):2.29 12 | % Frame Duration(msec):50 13 | % RF calibration data:None 14 | % Range Detection Threshold (dB):15 15 | % Doppler Detection Threshold (dB):15 16 | % Range Peak Grouping:enabled 17 | % Doppler Peak Grouping:enabled 18 | % Static clutter removal:disabled 19 | % Angle of Arrival FoV: Full FoV 20 | % Range FoV: Full FoV 21 | % Doppler FoV: Full FoV 22 | % *************************************************************** 23 | sensorStop 24 | flushCfg 25 | 26 | # 1:frame based chirps, 2:continuous chirp, 3:adv frame config [1/3] 27 | dfeDataOutputMode 1 28 | 29 | #* <0(cascading)>[15][x][0] 30 | #***[para 1 2] rx/tx Channel En <0x1111 means four devices enable> 31 | channelCfg 15 5 0 32 | 33 | #* [2] 34 | # [1/2] 35 | #[para 1] bit of one data 36 | #[para 2] complex or real 37 | adcCfg 2 1 38 | 39 | #* [-1] 40 | # [0] 41 | # [1] 42 | # [1] 43 | # [1] 44 | adcbufCfg -1 0 1 1 1 45 | 46 | # =========================================================================== 47 | 48 | #* 49 | # [0] [0] 50 | # ***[para 8] 51 | # ***[para 10]Samples num 52 | # ***[para 11]Samples rate 53 | # ***[para 12]CornerFreq1 <0:175kHz 1:235kHz 2:350kHz 3:700kHz> 54 | # ***[para 13]CornerFreq2 <0:175kHz 1:235kHz 2:350kHz 3:700kHz> 55 | # 56 | # 57 | # 58 | # 59 | profileCfg 0 77 8 7 18.64 0 0 30 1 133 12499 0 0 30 60 | 61 | #* 62 | # [0] [0] [0] 63 | # [0] 64 | chirpCfg 0 0 0 0 0 0 0 1 65 | chirpCfg 1 1 0 0 0 0 0 4 66 | #* 67 | # [should be 4x] 68 | # [1] 69 | #***[para 3]number of loops(num of chirps in a frame) 70 | #***[para 4]num of frames range(0~65535)--0 means infinite 71 | #***[para 5]frame periodicity in ms (float values allowed) 72 | frameCfg 0 1 16 0 50 1 0 73 | 74 | # =========================================================================== 75 | 76 | #* [0] 77 | lowPower 0 0 78 | 79 | # [-1] 80 | # 81 | # 82 | # [0] 83 | # [0] 84 | # [0] 85 | # [0] 86 | guiMonitor -1 1 1 0 0 0 1 87 | 88 | # Must be two lines 89 | cfarCfg -1 0 2 8 4 3 0 15 1 90 | cfarCfg -1 1 0 4 2 3 1 15 1 91 | 92 | # <> [0] ... 93 | multiObjBeamForming -1 1 0.5 94 | # <> [0] 95 | clutterRemoval -1 0 96 | # <> [0] ... 97 | calibDcRangeSig -1 0 -5 8 256 98 | # <> [0] 99 | extendedMaxVelocity -1 0 100 | 101 | #* [-1] [0] 102 | # [1] [0] 103 | lvdsStreamCfg -1 0 1 0 104 | 105 | # 106 | compRangeBiasAndRxChanPhase 0.0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 107 | # [0] ... 108 | measureRangeBiasAndRxChanPhase 0 1.5 0.2 109 | 110 | CQRxSatMonitor 0 3 4 31 0 111 | CQSigImgMonitor 0 65 4 112 | # [0] [0] 113 | analogMonitor 0 0 114 | 115 | # [-1] 116 | # 117 | aoaFovCfg -1 -90 90 -90 90 118 | 119 | # [-1] <0:range,1:Doppler> 120 | # 121 | cfarFovCfg -1 0 0 49.99 122 | cfarFovCfg -1 1 -18.28 18.28 123 | 124 | calibData 0 0 0 125 | 126 | sensorStart 127 | -------------------------------------------------------------------------------- /config/IWR6843_cfg.cfg: -------------------------------------------------------------------------------- 1 | % *************************************************************** 2 | % Created for SDK ver:03.04 3 | % Created using Visualizer ver:3.5.0.0 4 | % Frequency:60 5 | % Platform:xWR68xx 6 | % Scene Classifier:best_range_res 7 | % Azimuth Resolution(deg):15 8 | % Range Resolution(m):0.044 9 | % Maximum unambiguous Range(m):9.02 10 | % Maximum Radial Velocity(m/s):1 11 | % Radial velocity resolution(m/s):0.13 12 | % Frame Duration(msec):100 13 | % RF calibration data:None 14 | % Range Detection Threshold (dB):15 15 | % Doppler Detection Threshold (dB):15 16 | % Range Peak Grouping:enabled 17 | % Doppler Peak Grouping:enabled 18 | % Static clutter removal:disabled 19 | % Angle of Arrival FoV: Full FoV 20 | % Range FoV: Full FoV 21 | % Doppler FoV: Full FoV 22 | % *************************************************************** 23 | sensorStop 24 | flushCfg 25 | dfeDataOutputMode 1 26 | channelCfg 15 7 0 27 | adcCfg 2 1 28 | adcbufCfg -1 0 1 1 1 29 | profileCfg 0 60 200 7 40 0 0 100 1 64 2000 0 0 158 30 | chirpCfg 0 0 0 0 0 0 0 1 31 | chirpCfg 1 1 0 0 0 0 0 2 32 | chirpCfg 2 2 0 0 0 0 0 4 33 | frameCfg 0 2 64 0 110 1 0 34 | lowPower 0 0 35 | guiMonitor -1 1 1 0 0 0 1 36 | cfarCfg -1 0 2 8 4 3 0 20 1 37 | cfarCfg -1 1 0 4 2 3 1 15 1 38 | multiObjBeamForming -1 1 0.5 39 | clutterRemoval -1 1 40 | calibDcRangeSig -1 0 -5 8 256 41 | extendedMaxVelocity -1 0 42 | bpmCfg -1 0 0 1 43 | lvdsStreamCfg -1 0 1 0 44 | compRangeBiasAndRxChanPhase 0.0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1 0 45 | measureRangeBiasAndRxChanPhase 0 1.5 0.2 46 | CQRxSatMonitor 0 3 5 121 0 47 | CQSigImgMonitor 0 127 4 48 | analogMonitor 0 0 49 | aoaFovCfg -1 -90 90 -90 90 50 | cfarFovCfg -1 0 0 0.5 51 | cfarFovCfg -1 1 -3 3.00 52 | calibData 0 0 0 53 | sensorStart 54 | -------------------------------------------------------------------------------- /dsp/ZoomFFT.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | import numpy as np 14 | from matplotlib import pyplot as plt 15 | from numpy import pi 16 | from numpy.fft import fft, fftfreq, fftshift 17 | from scipy import signal 18 | import logging 19 | import sys 20 | 21 | 22 | class ZoomFFT: 23 | """This class is an implementation of the Zoom Fast Fourier Transform (ZoomFFT). 24 | 25 | The zoom FFT (Fast Fourier Transform) is a signal processing technique used to 26 | analyse a portion of a spectrum at high resolution. The steps to apply the zoom 27 | FFT to this region are as follows: 28 | 29 | 1. Frequency translate to shift the frequency range of interest down to near 30 | 0 Hz (DC) 31 | 2. Low pass filter to prevent aliasing when subsequently sampled at a lower 32 | sample rate 33 | 3. Re-sample at a lower rate 34 | 4. FFT the re-sampled data (Multiple blocks of data are needed to have an FFT of 35 | the same length) 36 | 37 | The resulting spectrum will now have a much smaller resolution bandwidth, compared 38 | to an FFT of non-translated data. 39 | 40 | """ 41 | 42 | def __init__(self, low_freq, high_freq, fs, signal=None): 43 | """Initialize the ZoomFFT class. 44 | 45 | Args: 46 | low_freq (int): Lower frequency limit 47 | high_freq (int): Upper frequency limit 48 | fs (int): sampling Frequency 49 | signal (np.ndarray): Signal to perform the ZoomFFT on 50 | 51 | """ 52 | self.low_freq = low_freq 53 | self.high_freq = high_freq 54 | self.fs = fs 55 | 56 | if (low_freq < 0) or (high_freq > fs) or ((high_freq - low_freq) > fs): 57 | raise Exception("invalid inputs. Program Terminated! ") 58 | 59 | if len(signal): 60 | self.signal = signal 61 | self.length = len(signal) 62 | else: 63 | # the default now is a sine signal, for demo purpose 64 | pass 65 | 66 | def set_signal(self, signal): 67 | """Sets given signal as a member variable of the class. 68 | 69 | e.g. ZoomFFT.create_signal(generate_sinewave(a, b, c) + generate_sinewave(d, e, f)) 70 | 71 | Args: 72 | signal (np.ndarray): Signal to perform the ZoomFFT on 73 | 74 | """ 75 | self.signal = signal 76 | 77 | def sinewave(self, f, length, amplitude=1): 78 | """Generates a sine wave which could be used as a part of the signal. 79 | 80 | Args: 81 | f (int): Frequency of the sine wave 82 | length (int): Number of data points in the sine wave 83 | amplitude (int): Amplitude of the sine wave 84 | 85 | Returns: 86 | x (np.ndarray): Generated sine wave with the given parameters. 87 | """ 88 | self.length = length 89 | x = amplitude * np.sin(2 * pi * f / self.fs * np.arange(length)) 90 | return x 91 | 92 | def compute_fft(self): 93 | """Computes the Fast Fourier Transform (FFT) of the signal. 94 | 95 | Returns: 96 | X (np.ndarray): A frequency-shifted, unscaled, FFT of the signal. 97 | """ 98 | try: 99 | X = fft(self.signal) 100 | X = np.abs(fftshift(X)) # unscaled 101 | return X 102 | except NameError: 103 | print("signal not defined. Program terminated!") 104 | except: 105 | print("Unexpected error:", sys.exc_info()[0]) 106 | raise 107 | 108 | def plot_fft(self, d=None): 109 | """Plots the Fast Fourier Transform (FFT) of the signal. 110 | 111 | Args: 112 | d (int): Sample spacing (inverse of the sampling rate) 113 | 114 | """ 115 | try: 116 | d = 1 / self.fs if d is None else d 117 | X = self.compute_fft() 118 | freq = fftfreq(self.length, d) 119 | 120 | self.original_sample_range = 1 / (self.length * d) 121 | 122 | fig1, ax1 = plt.subplots() 123 | ax1.stem(fftshift(freq), X / self.length) 124 | ax1.set_xlabel('Frequency (Hz)', fontsize=12) 125 | ax1.set_ylabel('Magnitude', fontsize=12) 126 | ax1.set_title('FFT Two-sided spectrum', fontsize=12) 127 | ax1.grid() 128 | 129 | plt.show() 130 | except: 131 | print("Unexpected error:", sys.exc_info()[0]) 132 | raise 133 | 134 | def compute_zoomfft(self, resample_number=None): 135 | """Computes the Zoom Fast Fourier Transform (ZoomFFT) of the signal. 136 | 137 | Args: 138 | resample_number (int): The number of samples in the resampled signal. 139 | 140 | Returns: 141 | Xd (np.ndarray): A frequency-shifted, unscaled, ZoomFFT of the signal. 142 | bw_factor (int): Bandwidth factor 143 | fftlen (int): Length of the ZoomFFT output 144 | Ld (int): for internal use 145 | F (int): for internal use 146 | """ 147 | try: 148 | bw_of_interest = self.high_freq - self.low_freq 149 | 150 | if self.length % bw_of_interest != 0: 151 | logging.warning("length of signal should be divisible by bw_of_interest. Zoom FFT Spectrum may distort!") 152 | input("Press Enter to continue...") 153 | 154 | fc = (self.low_freq + self.high_freq) / 2 155 | bw_factor = np.floor(self.fs / bw_of_interest).astype(np.uint8) 156 | 157 | # mix the signal down to DC, and filter it through the FIR decimator 158 | ind_vect = np.arange(self.length) 159 | y = self.signal * np.exp(-1j * 2 * pi * ind_vect * fc / self.fs) 160 | 161 | resample_number = bw_of_interest / self.original_sample_range if resample_number is None else resample_number 162 | 163 | resample_range = bw_of_interest / resample_number 164 | 165 | if resample_range != self.original_sample_range: 166 | logging.warning("resample resolution != original sample resolution. Zoom FFT Spectrum may distort!") 167 | input("Press Enter to continue...") 168 | 169 | xd = signal.resample(y, np.int(resample_number)) 170 | 171 | fftlen = len(xd) 172 | Xd = fft(xd) 173 | Xd = np.abs(fftshift(Xd)) # unscaled 174 | 175 | Ld = self.length / bw_factor 176 | fsd = self.fs / bw_factor 177 | F = fc + fsd / fftlen * np.arange(fftlen) - fsd / 2 178 | return Xd, bw_factor, fftlen, Ld, F 179 | except NameError: 180 | print("signal not defined. Program terminated!") 181 | except: 182 | print("Unexpected error:", sys.exc_info()[0]) 183 | raise 184 | 185 | def plot_zoomfft(self, resample_number=None): 186 | """Plots the Zoom Fast Fourier Transform (ZoomFFT) of the signal. 187 | 188 | Args: 189 | resample_number (int): The number of samples in the resampled signal. 190 | 191 | """ 192 | try: 193 | bw_of_interest = self.high_freq - self.low_freq 194 | resample_number = bw_of_interest / self.original_sample_range if resample_number is None else resample_number 195 | Xd, bw_factor, fftlen, Ld, F = self.compute_zoomfft(resample_number) 196 | 197 | fig1, ax1 = plt.subplots() 198 | 199 | ax1.stem(F, Xd / Ld, linefmt='C1-.', markerfmt='C1s') 200 | ax1.grid() 201 | ax1.set_xlabel('Frequency (Hz)', fontsize=12) 202 | ax1.set_ylabel('Magnitude', fontsize=12) 203 | ax1.set_title('Zoom FFT Spectrum. Mixer Approach.', fontsize=12) 204 | fig1.subplots_adjust(hspace=0.35) 205 | plt.show() 206 | except: 207 | print("Unexpected error:", sys.exc_info()[0]) 208 | raise 209 | -------------------------------------------------------------------------------- /dsp/__init__.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | from .angle_estimation import * 14 | from .cfar import * 15 | from .compensation import * 16 | from .doppler_processing import * 17 | from .range_processing import * 18 | from .utils import * 19 | from .noise_removal import * 20 | from .music import * -------------------------------------------------------------------------------- /dsp/cfar.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | import numpy as np 14 | from scipy.ndimage import convolve1d 15 | 16 | """ Various cfar algorithm types 17 | 18 | From https://www.mathworks.com/help/phased/ug/constant-false-alarm-rate-cfar-detectors.html 19 | |-----------------------------------------------------------------------------------------------------------| 20 | | Algorithm | Typical Usage | 21 | |-----------------------------------------------------------------------------------------------------------| 22 | | Cell-averaging CFAR | Most situations | 23 | | Greatest-of cell-averaging CFAR | When it is important to avoid false alarms at the edge of clutter | 24 | | Smallest-of cell-averaging CFAR | When targets are closely located | 25 | | Order statistic CFAR | Compromise between greatest-of and smallest-of cell averaging | 26 | |-----------------------------------------------------------------------------------------------------------| 27 | 28 | """ 29 | 30 | 31 | def ca(x, *argv, **kwargs): 32 | """Detects peaks in signal using Cell-Averaging CFAR (CA-CFAR). 33 | 34 | Args: 35 | x (~numpy.ndarray): Signal. 36 | *argv: See mmwave.dsp.cfar.ca\_ 37 | **kwargs: See mmwave.dsp.cfar.ca\_ 38 | 39 | Returns: 40 | ~numpy.ndarray: Boolean array of detected peaks in x. 41 | 42 | Examples: 43 | >>> signal = np.random.randint(100, size=10) 44 | >>> signal 45 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 46 | >>> det = mm.dsp.ca(signal, l_bound=20, guard_len=1, noise_len=3) 47 | >>> det 48 | array([False, False, True, False, False, False, False, True, False, 49 | True]) 50 | 51 | Perform a non-wrapping CFAR 52 | 53 | >>> signal = np.random.randint(100, size=10) 54 | >>> signal 55 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 56 | >>> det = mm.dsp.ca(signal, l_bound=20, guard_len=1, noise_len=3, mode='constant') 57 | >>> det 58 | array([False, True, True, False, False, False, False, True, True, 59 | True]) 60 | 61 | """ 62 | if isinstance(x, list): 63 | x = np.array(x) 64 | threshold, _ = ca_(x, *argv, **kwargs) 65 | ret = (x > threshold) 66 | return ret 67 | 68 | 69 | def ca_(x, guard_len=4, noise_len=8, mode='wrap', l_bound=4000): 70 | """Uses Cell-Averaging CFAR (CA-CFAR) to calculate a threshold that can be used to calculate peaks in a signal. 71 | 72 | Args: 73 | x (~numpy.ndarray): Signal. 74 | guard_len (int): Number of samples adjacent to the CUT that are ignored. 75 | noise_len (int): Number of samples adjacent to the guard padding that are factored into the calculation. 76 | mode (str): Specify how to deal with edge cells. Examples include 'wrap' and 'constant'. 77 | l_bound (float or int): Additive lower bound while calculating peak threshold. 78 | 79 | Returns: 80 | Tuple [ndarray, ndarray] 81 | 1. (ndarray): Upper bound of noise threshold. 82 | #. (ndarray): Raw noise strength. 83 | 84 | Examples: 85 | >>> signal = np.random.randint(100, size=10) 86 | >>> signal 87 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 88 | >>> threshold = mm.dsp.ca_(signal, l_bound=20, guard_len=1, noise_len=3) 89 | >>> threshold 90 | (array([70, 76, 64, 79, 81, 91, 74, 71, 70, 79]), array([50, 56, 44, 59, 61, 71, 54, 51, 50, 59])) 91 | 92 | Perform a non-wrapping CFAR thresholding 93 | 94 | >>> signal = np.random.randint(100, size=10) 95 | >>> signal 96 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 97 | >>> threshold = mm.dsp.ca_(signal, l_bound=20, guard_len=1, noise_len=3, mode='constant') 98 | >>> threshold 99 | (array([44, 37, 41, 65, 81, 91, 67, 51, 34, 46]), array([24, 17, 21, 45, 61, 71, 47, 31, 14, 26])) 100 | 101 | """ 102 | if isinstance(x, list): 103 | x = np.array(x) 104 | assert type(x) == np.ndarray 105 | 106 | kernel = np.ones(1 + (2 * guard_len) + (2 * noise_len), dtype=x.dtype) / (2 * noise_len) 107 | kernel[noise_len:noise_len + (2 * guard_len) + 1] = 0 108 | 109 | noise_floor = convolve1d(x, kernel, mode=mode) 110 | threshold = noise_floor + l_bound 111 | 112 | return threshold, noise_floor 113 | 114 | 115 | def caso(x, *argv, **kwargs): 116 | """Detects peaks in signal using Cell-Averaging Smallest-Of CFAR (CASO-CFAR). 117 | 118 | Args: 119 | x (~numpy.ndarray): Signal. 120 | *argv: See mmwave.dsp.cfar.caso\_ 121 | **kwargs: See mmwave.dsp.cfar.caso\_ 122 | 123 | Returns: 124 | ~numpy.ndarray: Boolean array of detected peaks in x. 125 | 126 | Examples: 127 | >>> signal = np.random.randint(100, size=10) 128 | >>> signal 129 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 130 | >>> det = mm.dsp.caso(signal, l_bound=20, guard_len=1, noise_len=3) 131 | >>> det 132 | array([False, False, True, False, False, False, False, True, True, 133 | True]) 134 | 135 | Perform a non-wrapping CFAR 136 | 137 | >>> signal = np.random.randint(100, size=10) 138 | >>> signal 139 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 140 | >>> det = mm.dsp.caso(signal, l_bound=20, guard_len=1, noise_len=3, mode='constant') 141 | >>> det 142 | array([False, True, True, False, False, False, False, True, True, 143 | True]) 144 | 145 | """ 146 | if isinstance(x, list): 147 | x = np.array(x) 148 | threshold, _ = caso_(x, *argv, **kwargs) 149 | ret = (x > threshold) 150 | return ret 151 | 152 | 153 | def caso_(x, guard_len=4, noise_len=8, mode='wrap', l_bound=4000): 154 | """Uses Cell-Averaging Smallest-Of CFAR (CASO-CFAR) to calculate a threshold that can be used to calculate peaks in a signal. 155 | 156 | Args: 157 | x (~numpy.ndarray): Signal. 158 | guard_len (int): Number of samples adjacent to the CUT that are ignored. 159 | noise_len (int): Number of samples adjacent to the guard padding that are factored into the calculation. 160 | mode (str): Specify how to deal with edge cells. 161 | l_bound (float or int): Additive lower bound while calculating peak threshold. 162 | 163 | Returns: 164 | Tuple [ndarray, ndarray] 165 | 1. (ndarray): Upper bound of noise threshold. 166 | #. (ndarray): Raw noise strength. 167 | 168 | Examples: 169 | >>> signal = np.random.randint(100, size=10) 170 | >>> signal 171 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 172 | >>> threshold = mm.dsp.caso_(signal, l_bound=20, guard_len=1, noise_len=3) 173 | >>> (threshold[0].astype(int), threshold[1].astype(int)) 174 | (array([69, 55, 49, 72, 72, 86, 69, 55, 49, 72]), array([49, 35, 29, 52, 52, 66, 49, 35, 29, 52])) 175 | 176 | Perform a non-wrapping CFAR thresholding 177 | 178 | >>> signal = np.random.randint(100, size=10) 179 | >>> signal 180 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 181 | >>> threshold = mm.dsp.caso_(signal, l_bound=20, guard_len=1, noise_len=3, mode='constant') 182 | >>> (threshold[0].astype(int), threshold[1].astype(int)) 183 | (array([69, 55, 49, 72, 72, 86, 69, 55, 49, 72]), array([49, 35, 29, 52, 52, 66, 49, 35, 29, 52])) 184 | 185 | """ 186 | if isinstance(x, list): 187 | x = np.array(x) 188 | 189 | l_window, r_window = _cfar_windows(x, guard_len, noise_len, mode) 190 | 191 | # Generate scaling based on mode 192 | l_window = l_window / noise_len 193 | r_window = r_window / noise_len 194 | if mode == 'wrap': 195 | noise_floor = np.minimum(l_window, r_window) 196 | elif mode == 'constant': 197 | edge_cells = guard_len + noise_len 198 | noise_floor = np.minimum(l_window, r_window) 199 | noise_floor[:edge_cells] = r_window[:edge_cells] 200 | noise_floor[-edge_cells:] = l_window[-edge_cells:] 201 | else: 202 | raise ValueError(f'Mode {mode} is not a supported mode') 203 | 204 | threshold = noise_floor + l_bound 205 | return threshold, noise_floor 206 | 207 | 208 | def cago(x, *argv, **kwargs): 209 | """Detects peaks in signal using Cell-Averaging Greatest-Of CFAR (CAGO-CFAR). 210 | 211 | Args: 212 | x (~numpy.ndarray): Signal. 213 | *argv: See mmwave.dsp.cfar.cago\_ 214 | **kwargs: See mmwave.dsp.cfar.cago\_ 215 | 216 | Returns: 217 | ~numpy.ndarray: Boolean array of detected peaks in x. 218 | 219 | Examples: 220 | >>> signal = np.random.randint(100, size=10) 221 | >>> signal 222 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 223 | >>> det = mm.dsp.cago(signal, l_bound=20, guard_len=1, noise_len=3) 224 | >>> det 225 | array([False, False, True, False, False, False, False, True, False, 226 | False]) 227 | 228 | Perform a non-wrapping CFAR 229 | 230 | >>> signal = np.random.randint(100, size=10) 231 | >>> signal 232 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 233 | >>> det = mm.dsp.cago(signal, l_bound=20, guard_len=1, noise_len=3, mode='constant') 234 | >>> det 235 | array([False, True, True, False, False, False, False, True, True, 236 | True]) 237 | 238 | """ 239 | if isinstance(x, list): 240 | x = np.array(x) 241 | threshold, _ = cago_(x, *argv, **kwargs) 242 | ret = (x > threshold) 243 | return ret 244 | 245 | 246 | def cago_(x, guard_len=4, noise_len=8, mode='wrap', l_bound=4000): 247 | """Uses Cell-Averaging Greatest-Of CFAR (CAGO-CFAR) to calculate a threshold that can be used to calculate peaks in a signal. 248 | 249 | Args: 250 | x (~numpy.ndarray): Signal. 251 | guard_len (int): Number of samples adjacent to the CUT that are ignored. 252 | noise_len (int): Number of samples adjacent to the guard padding that are factored into the calculation. 253 | mode (str): Specify how to deal with edge cells. 254 | l_bound (float or int): Additive lower bound while calculating peak threshold. 255 | 256 | Returns: 257 | Tuple [ndarray, ndarray] 258 | 1. (ndarray): Upper bound of noise threshold. 259 | #. (ndarray): Raw noise strength. 260 | 261 | Examples: 262 | >>> signal = np.random.randint(100, size=10) 263 | >>> signal 264 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 265 | >>> threshold = mm.dsp.cago_(signal, l_bound=20, guard_len=1, noise_len=3) 266 | >>> (threshold[0].astype(int), threshold[1].astype(int)) 267 | (array([72, 97, 80, 87, 90, 97, 80, 87, 90, 86]), array([52, 77, 60, 67, 70, 77, 60, 67, 70, 66])) 268 | 269 | Perform a non-wrapping CFAR thresholding 270 | 271 | >>> signal = np.random.randint(100, size=10) 272 | >>> signal 273 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 274 | >>> threshold = mm.dsp.cago_(signal, l_bound=20, guard_len=1, noise_len=3, mode='constant') 275 | >>> (threshold[0].astype(int), threshold[1].astype(int)) 276 | (array([69, 55, 49, 72, 90, 97, 69, 55, 49, 72]), array([49, 35, 29, 52, 70, 77, 49, 35, 29, 52])) 277 | 278 | """ 279 | if isinstance(x, list): 280 | x = np.array(x) 281 | 282 | l_window, r_window = _cfar_windows(x, guard_len, noise_len, mode) 283 | 284 | # Generate scaling based on mode 285 | l_window = l_window / noise_len 286 | r_window = r_window / noise_len 287 | if mode == 'wrap': 288 | noise_floor = np.maximum(l_window, r_window) 289 | elif mode == 'constant': 290 | edge_cells = guard_len + noise_len 291 | noise_floor = np.maximum(l_window, r_window) 292 | noise_floor[:edge_cells] = r_window[:edge_cells] 293 | noise_floor[-edge_cells:] = l_window[-edge_cells:] 294 | else: 295 | raise ValueError(f'Mode {mode} is not a supported mode') 296 | 297 | threshold = noise_floor + l_bound 298 | return threshold, noise_floor 299 | 300 | 301 | def os(x, *argv, **kwargs): 302 | """Performs Ordered-Statistic CFAR (OS-CFAR) detection on the input array. 303 | 304 | Args: 305 | x (~numpy.ndarray): Noisy array to perform cfar on with log values 306 | *argv: See mmwave.dsp.cfar.os\_ 307 | **kwargs: See mmwave.dsp.cfar.os\_ 308 | 309 | 310 | Returns: 311 | ~numpy.ndarray: Boolean array of detected peaks in x. 312 | 313 | Examples: 314 | >>> signal = np.random.randint(100, size=10) 315 | >>> signal 316 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 317 | >>> det = mm.dsp.os(signal, k=3, scale=1.1, guard_len=0, noise_len=3) 318 | >>> det 319 | array([False, True, True, False, False, False, False, True, False, 320 | True]) 321 | 322 | """ 323 | if isinstance(x, list): 324 | x = np.array(x) 325 | threshold, _ = os_(x, *argv, **kwargs) 326 | ret = (x > threshold) 327 | return ret 328 | 329 | 330 | def os_(x, guard_len=0, noise_len=8, k=12, scale=1.0): 331 | """Performs Ordered-Statistic CFAR (OS-CFAR) detection on the input array. 332 | 333 | Args: 334 | x (~numpy.ndarray): Noisy array to perform cfar on with log values 335 | guard_len (int): Number of samples adjacent to the CUT that are ignored. 336 | noise_len (int): Number of samples adjacent to the guard padding that are factored into the calculation. 337 | k (int): Ordered statistic rank to sample from. 338 | scale (float): Scaling factor. 339 | 340 | Returns: 341 | Tuple [ndarray, ndarray] 342 | 1. (ndarray): Upper bound of noise threshold. 343 | #. (ndarray): Raw noise strength. 344 | 345 | Examples: 346 | >>> signal = np.random.randint(100, size=10) 347 | >>> signal 348 | array([41, 76, 95, 28, 25, 53, 10, 93, 54, 85]) 349 | >>> threshold = mm.dsp.os_(signal, k=3, scale=1.1, guard_len=0, noise_len=3) 350 | >>> (threshold[0].astype(int), threshold[1].astype(int)) 351 | (array([93, 59, 58, 58, 83, 59, 59, 58, 83, 83]), array([85, 54, 53, 53, 76, 54, 54, 53, 76, 76])) 352 | 353 | """ 354 | if isinstance(x, list): 355 | x = np.array(x, dtype=np.uint32) 356 | 357 | n = len(x) 358 | noise_floor = np.zeros(n) 359 | threshold = np.zeros(n, dtype=np.float32) 360 | cut_idx = -1 361 | 362 | # Initial CUT 363 | left_idx = list(np.arange(n - noise_len - guard_len - 1, n - guard_len - 1)) 364 | right_idx = list(np.arange(guard_len, guard_len + noise_len)) 365 | 366 | # All other CUTs 367 | while cut_idx < (n - 1): 368 | cut_idx += 1 369 | 370 | left_idx.pop(0) 371 | left_idx.append((cut_idx - 1) % n) 372 | 373 | right_idx.pop(0) 374 | right_idx.append((cut_idx + guard_len + noise_len) % n) 375 | 376 | window = np.concatenate((x[left_idx], x[right_idx])) 377 | window.partition(k) 378 | noise_floor[cut_idx] = window[k] 379 | threshold[cut_idx] = noise_floor[cut_idx] * scale 380 | 381 | return threshold, noise_floor 382 | 383 | 384 | def _cfar_windows(x, guard_len, noise_len, mode): 385 | if type(x) != np.ndarray: 386 | raise TypeError(f'Expected array-like input got {type(x)}') 387 | 388 | # Create kernels 389 | r_kernel = np.zeros(1 + (2 * guard_len) + (2 * noise_len), dtype=x.dtype) 390 | r_kernel[:noise_len] = 1 391 | l_kernel = r_kernel[::-1] 392 | 393 | # Do initial convolutions 394 | l_window = convolve1d(x, l_kernel, mode=mode) 395 | r_window = convolve1d(x, r_kernel, mode=mode) 396 | 397 | return l_window, r_window 398 | 399 | 400 | WRAP_UP_LIST_IDX = lambda x, total: x if x >= 0 else x + total 401 | WRAP_DN_LIST_IDX = lambda x, total: x if x < total else x - total 402 | WRAP_DOPPLER_IDX = lambda x, num_doppler_bins: np.bitwise_and(x, num_doppler_bins - 1) 403 | DOPPLER_IDX_TO_SIGNED = lambda idx, fft_size: idx if idx < fft_size // 2 else idx - fft_size 404 | 405 | 406 | def peak_grouping(obj_raw, 407 | det_matrix, 408 | num_doppler_bins, 409 | max_range_idx, 410 | min_range_idx, 411 | group_in_doppler_direction, 412 | group_in_range_direction): 413 | """Performs peak grouping on detection Range/Doppler matrix. 414 | 415 | The function groups neighboring peaks into one. The grouping is done according to two input flags: 416 | group_in_doppler_direction and group_in_doppler_direction. For each detected peak the function checks if the peak is 417 | greater than its neighbors. If this is true, the peak is copied to the output list of detected objects. The 418 | neighboring peaks that are used for checking are taken from the detection matrix and copied into 3x3 kernel 419 | regardless of whether they are CFAR detected or not. Note: Function always reads 9 samples per detected object 420 | from L3 memory into local array tempBuff, but it only needs to read according to input flags. For example if only 421 | the group_in_doppler_direction flag is set, it only needs to read middle row of the kernel, i.e. 3 samples per 422 | target from detection matrix. 423 | 424 | Args: 425 | obj_raw (np.ndarray): (num_detected_objects, 3). detected objects from CFAR. 426 | det_matrix (np.ndarray): Range-doppler profile. shape is numRangeBins x num_doppler_bins. 427 | num_doppler_bins (int): number of doppler bins. 428 | max_range_idx (int): max range of detected objects. 429 | min_range_idx (int): min range of detected objects 430 | group_in_doppler_direction (int): flag to perform grouping along doppler direction. 431 | group_in_range_direction (int): flag to perform grouping along range direction. 432 | 433 | Returns: 434 | obj_out (np.ndarray): detected object after grouping. 435 | 436 | """ 437 | 438 | num_detected_objects = obj_raw.shape[0] 439 | 440 | num_obj_out = 0 441 | kernel = np.empty([9]) 442 | 443 | if (group_in_doppler_direction == 1) and (group_in_range_direction == 1): 444 | # Grouping both in Range and Doppler direction 445 | start_ind = 0 446 | step_ind = 1 447 | end_ind = 8 448 | elif (group_in_doppler_direction == 0) and (group_in_range_direction == 1): 449 | # Grouping only in Range direction 450 | start_ind = 1 451 | step_ind = 3 452 | end_ind = 7 453 | elif (group_in_doppler_direction == 1) and (group_in_range_direction == 0): 454 | # Grouping only in Doppler direction */ 455 | start_ind = 3 456 | step_ind = 1 457 | end_ind = 5 458 | else: 459 | # No grouping, copy all detected objects to the output matrix within specified min max range 460 | # num_detected_objects = min(num_detected_objects, MAX_OBJ_OUT) 461 | obj_out = obj_raw[obj_raw[:, RANGEIDX] <= max_range_idx and obj_raw[:, RANGEIDX] > min_range_idx] 462 | obj_out[:, DOPPLERIDX] = np.bitwise_and(obj_out[:, DOPPLERIDX], num_doppler_bins - 1) 463 | 464 | return obj_out 465 | 466 | # Start checking 467 | obj_out = np.zeros((num_obj_out, 3)) 468 | for i in range(num_detected_objects): 469 | detected_obj_flag = 0 470 | range_idx = obj_raw[i, 0] 471 | doppler_idx = obj_raw[i, 1] 472 | peak_val = obj_raw[i, 2] 473 | 474 | if (range_idx <= max_range_idx) and (range_idx >= min_range_idx): 475 | detected_obj_flag = 1 476 | 477 | # Fill local 3x3 kernel from detection matrix in L3 478 | start_idx = (range_idx - 1) * num_doppler_bins 479 | temp_ptr = det_matrix[start_idx:] 480 | row_start = 0 481 | row_end = 2 482 | 483 | if range_idx == min_range_idx: 484 | start_idx = range_idx * num_doppler_bins 485 | temp_ptr = det_matrix[start_idx:] 486 | row_start = 1 487 | kernel[0] = 0 488 | kernel[1] = 0 489 | kernel[2] = 0 490 | elif range_idx == max_range_idx: 491 | row_end = 1 492 | kernel[6] = 0 493 | kernel[7] = 0 494 | kernel[8] = 0 495 | 496 | for j in range(row_start, row_end + 1): 497 | for k in range(3): 498 | 499 | temp_idx = doppler_idx + (k - 1) 500 | 501 | if temp_idx < 0: 502 | temp_idx += num_doppler_bins 503 | elif temp_idx >= num_doppler_bins: 504 | temp_idx -= num_doppler_bins 505 | 506 | kernel[j * 3 + k] = temp_ptr[temp_idx] 507 | 508 | temp_ptr = temp_ptr[num_doppler_bins:] 509 | 510 | # Compare the detected object to its neighbors 511 | # Detected object is at index 4 512 | for k in range(start_ind, end_ind + 1, step_ind): 513 | if kernel[k] > kernel[4]: 514 | detected_obj_flag = 0 515 | 516 | if detected_obj_flag == 1: 517 | obj_out[num_obj_out, 0] = range_idx 518 | obj_out[num_obj_out, 1] = DOPPLER_IDX_TO_SIGNED(doppler_idx, num_doppler_bins) 519 | obj_out[num_obj_out, 2] = peak_val 520 | num_obj_out += 1 521 | 522 | if num_obj_out >= MAX_OBJ_OUT: 523 | break 524 | 525 | return num_obj_out, obj_out 526 | 527 | 528 | def peak_grouping_qualified(obj_raw, 529 | num_doppler_bins, 530 | max_range_idx, 531 | min_range_idx, 532 | group_in_doppler_direction, 533 | group_in_range_direction): 534 | """Performs peak grouping on list of CFAR detected objects. 535 | 536 | The function groups neighboring peaks into one. The grouping is done according to two input flags: 537 | group_in_doppler_direction and group_in_doppler_direction. For each detected peak the function checks if the peak is 538 | greater than its neighbors. If this is true, the peak is copied to the output list of detected objects. The 539 | neighboring peaks that are used for checking are taken from the list of CFAR detected objects, (not from the 540 | detection matrix), and copied into 3x3 kernel that has been initialized to zero for each peak under test. If the 541 | neighboring cell has not been detected by CFAR, its peak value is not copied into the kernel. Note: Function always 542 | search for 8 peaks in the list, but it only needs to search according to input flags. 543 | 544 | Args: 545 | obj_raw (np.ndarray): (num_detected_objects, 3). detected objects from CFAR. 546 | num_doppler_bins (int): number of doppler bins. 547 | max_range_idx (int): max range of detected objects. 548 | min_range_idx (int): min range of detected objects 549 | group_in_doppler_direction (int): flag to perform grouping along doppler direction. 550 | group_in_range_direction (int): flag to perform grouping along range direction. 551 | 552 | Returns: 553 | obj_out (np.ndarray): detected object after grouping. 554 | 555 | """ 556 | 557 | num_detected_objects = obj_raw.shape[0] 558 | 559 | if (group_in_doppler_direction == 1) and (group_in_range_direction == 1): 560 | # Grouping both in Range and Doppler direction 561 | start_ind = 0 562 | step_ind = 1 563 | end_ind = 8 564 | elif (group_in_doppler_direction == 0) and (group_in_range_direction == 1): 565 | # Grouping only in Range direction 566 | start_ind = 1 567 | step_ind = 3 568 | end_ind = 7 569 | elif (group_in_doppler_direction == 1) and (group_in_range_direction == 0): 570 | # Grouping only in Doppler direction */ 571 | start_ind = 3 572 | step_ind = 1 573 | end_ind = 5 574 | else: 575 | # No grouping, copy all detected objects to the output matrix within specified min max range 576 | num_detected_objects = min(num_detected_objects, MAX_OBJ_OUT) 577 | obj_out = obj_raw[(obj_raw['range_idx'][:num_detected_objects] <= max_range_idx) & 578 | (obj_raw['range_idx'][:num_detected_objects] > min_range_idx)] 579 | 580 | return obj_out 581 | 582 | # Start checking 583 | idx_obj_in_range = np.argwhere((obj_raw['range_idx'] <= max_range_idx) & 584 | (obj_raw['range_idx'] >= min_range_idx))[:, 0] 585 | 586 | obj_in_range = obj_raw[idx_obj_in_range] 587 | kernels = np.zeros((obj_in_range.shape[0], 9)) 588 | detected_obj_flag = np.ones(obj_in_range.shape[0]) 589 | 590 | # Populate the middle column. 591 | # Populate the 4th element. 592 | kernels[:, 4] = obj_in_range['peakVal'] 593 | 594 | # Populate the 1st element. 595 | obj_in_range_previous = obj_raw[idx_obj_in_range - 1] 596 | assert obj_in_range_previous.shape == obj_in_range.shape, "obj_in_range_previous indexing is wrong" 597 | idx_temp = ((obj_in_range_previous['range_idx']) == (obj_in_range['range_idx'] - 1)) & \ 598 | ((obj_in_range_previous['doppler_idx']) == (obj_in_range['doppler_idx'])) 599 | kernels[idx_temp, 1] = obj_in_range_previous['peakVal'][idx_temp] 600 | # 0th detected object has no left neighbor. 601 | kernels[idx_obj_in_range[idx_obj_in_range[:] == 0], 1] = 0 602 | 603 | # Populate the 7th element. 604 | obj_in_range_next = obj_raw[(idx_obj_in_range + 1) % num_detected_objects] 605 | assert obj_in_range_next.shape == obj_in_range.shape, "obj_in_range_next indexing is wrong" 606 | idx_temp = ((obj_in_range_next['range_idx']) == (obj_in_range['range_idx'] + 1)) & \ 607 | ((obj_in_range_next['doppler_idx']) == (obj_in_range['doppler_idx'])) 608 | kernels[idx_temp, 7] = obj_in_range_next['peakVal'][idx_temp] 609 | # last detected object, i.e. num_detected_objects-th has no left neighbor. 610 | kernels[idx_obj_in_range[idx_obj_in_range[:] == num_detected_objects], 7] = 0 611 | 612 | for i, idxDeteced in enumerate(idx_obj_in_range): 613 | doppler_idx = obj_in_range['doppler_idx'][i] 614 | range_idx = obj_in_range['range_idx'][i] 615 | # Fill the left column 616 | k_left = WRAP_UP_LIST_IDX(idxDeteced - 1, num_detected_objects) 617 | k_right = WRAP_DN_LIST_IDX(idxDeteced + 1, num_detected_objects) 618 | for _ in range(num_detected_objects): 619 | k_left_doppler_idx = obj_raw['doppler_idx'][k_left] 620 | k_left_range_idx = obj_raw['range_idx'][k_left] 621 | k_left_peak_val = obj_raw['peakVal'][k_left] 622 | if k_left_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx - 2, num_doppler_bins): 623 | break 624 | if k_left_range_idx == range_idx + 1 and k_left_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx - 1, 625 | num_doppler_bins): 626 | kernels[i, 6] = k_left_peak_val 627 | elif k_left_range_idx == range_idx and k_left_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx - 1, 628 | num_doppler_bins): 629 | kernels[i, 3] = k_left_peak_val 630 | elif k_left_range_idx == range_idx - 1 and k_left_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx - 1, 631 | num_doppler_bins): 632 | kernels[i, 0] = k_left_peak_val 633 | k_left = WRAP_UP_LIST_IDX(k_left - 1, num_detected_objects) 634 | 635 | k_right_doppler_idx = obj_raw['doppler_idx'][k_right] 636 | k_right_range_idx = obj_raw['range_idx'][k_right] 637 | k_right_peak_val = obj_raw['peakVal'][k_right] 638 | if k_right_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx - 2, num_doppler_bins): 639 | break 640 | if k_right_range_idx == range_idx + 1 and k_right_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx + 1, 641 | num_doppler_bins): 642 | kernels[i, 8] = k_right_peak_val 643 | elif k_right_range_idx == range_idx and k_right_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx + 1, 644 | num_doppler_bins): 645 | kernels[i, 5] = k_right_peak_val 646 | elif k_right_range_idx == range_idx - 1 and k_right_doppler_idx == WRAP_DOPPLER_IDX(doppler_idx + 1, 647 | num_doppler_bins): 648 | kernels[i, 2] = k_right_peak_val 649 | k_right = WRAP_DN_LIST_IDX(k_right + 1, num_detected_objects) 650 | 651 | detected_obj_flag[np.argwhere(np.max(kernels[:, start_ind:end_ind:step_ind]) != kernels[:, 4])] = 0 652 | obj_out = obj_in_range[detected_obj_flag[:] == 1] 653 | 654 | if obj_out.shape[0] > MAX_OBJ_OUT: 655 | obj_out = obj_out[:MAX_OBJ_OUT, ...] 656 | 657 | return obj_out 658 | -------------------------------------------------------------------------------- /dsp/compensation.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | import math 14 | from .utils import * 15 | 16 | 17 | def _generate_dft_sin_cos_table(dft_length): 18 | """Generate SIN/COS table for doppler compensation reference. 19 | 20 | Generate SIN/COS table. Also generates Sine/Cosine at half, one thrid and two thirds the bin value. This is a helper 21 | function only called by add_doppler_compensation(). 22 | 23 | Args: 24 | dft_len: (int) dft_len Length of the DFT. It is used as numDopperBins, which is numCHirpsPerFrame/numTxAntenns. 25 | 26 | Returns: 27 | dft_sin_cos_table (np.ndarray): ndarray in complex format with generated sine (image) cosine (real) table. 28 | bins (np.ndarray): Sin/Cos at half, one thrid and two thirds the bin. 29 | """ 30 | dft_sin_cos_table = np.arange(dft_length, dtype=np.float32) 31 | dft_sin_cos_table = np.cos(2 * np.pi * dft_sin_cos_table / dft_length) + \ 32 | 1j * -np.sin(2 * np.pi * dft_sin_cos_table / dft_length) 33 | 34 | # 1/2, 1/3 and 2/3 bins 35 | bins = np.array([0.5, 1.0/3, 2.0/3]) 36 | bins = np.cos(2 * np.pi * bins / dft_length) - 1j * np.sin(2 * np.pi * bins / dft_length) 37 | 38 | return dft_sin_cos_table, bins 39 | 40 | def add_doppler_compensation(input_data, 41 | num_tx_antennas, 42 | doppler_indices=None, 43 | num_doppler_bins=None): 44 | """Compensation of Doppler phase shift in the virtual antennas. 45 | 46 | Compensation of Doppler phase shift on the virtual antennas (corresponding to second or third Tx antenna chirps). 47 | Symbols corresponding to virtual antennas, are rotated by half of the Doppler phase shift measured by Doppler FFT 48 | for 2 Tx system and 1/3 and 2/3 of the Doppler phase shift for 3 Tx system. The phase shift read from the table 49 | using half or 1/3 of the object Doppler index value. If the Doppler index is odd, an extra half of the bin phase 50 | shift is added. 51 | 52 | The original function is called per detected objects. This functions is modified to directly compensate the 53 | azimuth_in matrix (numDetObj, num_angle_bins) 54 | 55 | Args: 56 | input_data (ndarray): (range, num_antennas, doppler) Radar data cube that needs to be compensated. It can be the input 57 | of azimuth FFT after CFAR or the intermediate right before beamforming. 58 | num_tx_antennas (int): Number of transmitters. 59 | num_doppler_bins (int): (Optional) Number of doppler bins in the radar data cube. If given, that means the doppler 60 | indices are signed and needs to be converted to unsigned. 61 | doppler_indices (ndarray): (Optional) Doppler index of the object with the shape of (num_detected_objects). If given, 62 | that means we only compensate on selected doppler bins. 63 | 64 | Return: 65 | input_data (ndarray): Original input data with the columns related to virtual receivers got compensated.原始数据 66 | 67 | Example: 68 | >>> # If the compensation is done right before naive azimuth FFT and objects is detected already. you need to 69 | >>> # feed in the doppler_indices 70 | >>> dataIn = add_doppler_compensation(dataIn, 3, doppler_indices, 128) 71 | """ 72 | num_antennas = input_data.shape[1] 73 | if num_tx_antennas == 1: 74 | return input_data 75 | elif num_tx_antennas > 3: 76 | raise ValueError("the specified number of transimitters is currently not supported") 77 | 78 | # Call the gen function above to generate the tables. 79 | azimuth_mod_coefs, bins = _generate_dft_sin_cos_table(int(num_doppler_bins)) 80 | 81 | # Convert signed doppler indices to unsigned and divide Doppler index by 2. 82 | if doppler_indices is not None: 83 | if num_doppler_bins is not None: 84 | doppler_compensation_indices = doppler_indices & (num_doppler_bins - 1) 85 | doppler_compensation_indices[doppler_compensation_indices[:] >= (num_doppler_bins / 2)] -= num_doppler_bins 86 | doppler_compensation_indices = doppler_compensation_indices // 2 87 | doppler_compensation_indices[doppler_compensation_indices[:] < 0] += num_doppler_bins 88 | exp_doppler_compensation = azimuth_mod_coefs[doppler_compensation_indices] 89 | else: 90 | exp_doppler_compensation = azimuth_mod_coefs 91 | 92 | # Add half bin rotation if Doppler index was odd 93 | if num_tx_antennas == 2: 94 | exp_doppler_compensation[(doppler_indices[:] % 2) == 1] *= bins[0] 95 | else: 96 | exp_doppler_compensation[(doppler_indices[:] % 3) == 1] *= bins[1] 97 | exp_doppler_compensation[(doppler_indices[:] % 3) == 2] *= bins[2] 98 | 99 | # Expand the dim so that the broadcasting below will work. 100 | exp_doppler_compensation = np.expand_dims(exp_doppler_compensation, axis=1) 101 | 102 | # Rotate 103 | azimuth_values = input_data[:, (num_antennas/num_tx_antennas):, :] 104 | for azi_val in azimuth_values: 105 | Re = exp_doppler_compensation.real * azi_val.imag - exp_doppler_compensation.imag * azi_val.real 106 | Im = exp_doppler_compensation.imag * azi_val.imag + exp_doppler_compensation.real * azi_val.real 107 | input_data[:, (num_antennas/num_tx_antennas):, :] = Re + 1j * Im 108 | 109 | return input_data 110 | 111 | 112 | def rx_channel_phase_bias_compensation(rx_channel_compensations, input, num_antennas): 113 | """Compensation of rx channel phase bias. 114 | 115 | Args: 116 | rx_channel_compensations: rx channel compensation coefficient. 117 | input: complex number. 118 | num_antennas: number of symbols. 119 | """ 120 | azimuth_values = input[:num_antennas] 121 | rx_channel_compensations_values = rx_channel_compensations[:num_antennas] 122 | 123 | Re = rx_channel_compensations_values * (azimuth_values.imag - azimuth_values.real) 124 | Im = rx_channel_compensations_values * (azimuth_values.imag + azimuth_values.real) 125 | input[:num_antennas] = Re + 1j * Im 126 | 127 | return 128 | 129 | 130 | def near_field_correction(idx, 131 | detected_objects, 132 | start_range_index, 133 | end_range_index, 134 | azimuth_input, 135 | azimuth_output, 136 | num_angle_bins, 137 | num_rx_antennas, 138 | range_resolution): 139 | """Correct phase error as the far-field plane wave assumption breaks. 140 | 141 | Calculates near field correction for input detected index (corresponding 142 | to a range position). Referring to top level doxygen @ref 143 | nearFieldImplementation, this function performs the Set 1 rotation with the 144 | correction and adds to Set 0 in place to produce result in Set 0 of the 145 | azimuth_output. 146 | 147 | This correction is done per detected objects from CFAR detection 148 | 149 | Args: 150 | idx: index of the detected objects in detected_objects. 151 | detected_objects: detected objects matrix with dimension of 100 x 6, where it holds at most 100 objects and 6 members are 152 | rangeIdx, dopplerIdx, peakVal, x, y and z. It is configured as a structured array. 153 | start_range_index: start range index of near field correction. 154 | end_range_index: end range index of near field correction. 155 | azimuth_input: complex array of which length is num_angle_bins+numVirtualAntAzim, where numVirtualAntAzim = 4, 8 or 12 156 | depending on how many Txs are used. 157 | 158 | Returns: 159 | None. azimuth_output is changed in-place. 160 | """ 161 | 162 | LAMBDA_77GHz_MILLIMETER = 3e8 / 77e9 163 | MMWDEMO_TWO_PI_OVER_LAMBDA = 2.0 * math.pi / LAMBDA_77GHz_MILLIMETER 164 | 165 | # Sanity check and check if nearFieldCorrection is necessary. 166 | assert idx >= 0 and idx < MAX_OBJ_OUT, "idx is out of bound!" 167 | rangeIdx = detected_objects['rangeIdx'][idx] 168 | if rangeIdx < start_range_index or rangeIdx >= end_range_index: 169 | print("{} is out of the nearFieldCorrection range".format(rangeIdx)) 170 | return 171 | 172 | # num_angle_bins = 64 173 | azimuth_input[:num_angle_bins] = 0 174 | azimuth_input[num_rx_antennas: num_rx_antennas + num_rx_antennas] = azimuth_input[num_angle_bins:] 175 | 176 | # azimuth_output has length of 2*num_angle_bins. 177 | azimuth_output[num_angle_bins:] = np.fft.fft(azimuth_input, n=num_angle_bins) 178 | 179 | # #define MMWDEMO_NEAR_FIELD_A (0) 180 | # B can be changed to position the desired reference (boresight) in the geometry */ 181 | # #define MMWDEMO_NEAR_FIELD_B (LAMBDA_77GHz_MILLIMETER) //((((2 + 0.75) * LAMBDA_77GHz_MILLIMETER)) + 8.7) 182 | # #define MMWDEMO_NEAR_FIELD_C (2 * LAMBDA_77GHz_MILLIMETER) 183 | # 8.7 mm is the actual (approximate) measurement of distance between tx1 and rx4, 184 | # measured using a wooden scale that has a resolution of 1 mm 185 | # #define MMWDEMO_NEAR_FIELD_D (MMWDEMO_NEAR_FIELD_C + 8.7) 186 | # #define MMWDEMO_NEAR_FIELD_E (MMWDEMO_NEAR_FIELD_D + 1.5 * LAMBDA_77GHz_MILLIMETER) 187 | geometry_points = {"A": 0, 188 | "B": LAMBDA_77GHz_MILLIMETER, 189 | "C": 2 * LAMBDA_77GHz_MILLIMETER, 190 | "D": 2 * LAMBDA_77GHz_MILLIMETER + 8.7, 191 | "E": (2 + 1.5) * LAMBDA_77GHz_MILLIMETER + 8.7} 192 | 193 | # AB, CB, DB, EB 194 | geometry_lines = np.array([geometry_points["A"] - geometry_points["B"], 195 | geometry_points["C"] - geometry_points["B"], 196 | geometry_points["D"] - geometry_points["B"], 197 | geometry_points["E"] - geometry_points["B"]]) 198 | 199 | geometry_lines_square = geometry_lines * geometry_lines 200 | 201 | range_in_millimeter = (detected_objects['rangeIdx'][idx] * range_resolution - range_resolution) * 1000 202 | range_squared = range_in_millimeter * range_in_millimeter 203 | theta_incrementation = 2.0 / num_angle_bins 204 | 205 | for i in range(num_angle_bins): 206 | theta = i * theta_incrementation if i < num_angle_bins / 2 else (i - num_angle_bins) * theta_incrementation 207 | 208 | tx1 = np.sqrt(range_squared + geometry_lines_square[1] - range_in_millimeter * theta * geometry_lines[1] * 2) 209 | rx4 = np.sqrt(range_squared + geometry_lines_square[2] - range_in_millimeter * theta * geometry_lines[2] * 2) 210 | tx2 = np.sqrt(range_squared + geometry_lines_square[0] - range_in_millimeter * theta * geometry_lines[0] * 2) 211 | rx1 = np.sqrt(range_squared + geometry_lines_square[3] - range_in_millimeter * theta * geometry_lines[3] * 2) 212 | 213 | if range > 0: 214 | psi = MMWDEMO_TWO_PI_OVER_LAMBDA * ((tx2 + rx1) - (rx4 + tx1)) - np.pi * theta 215 | corrReal = np.cos(psi) 216 | corrImag = np.sin(-psi) 217 | 218 | out1CorrReal = azimuth_output[num_angle_bins + i].real * corrReal + \ 219 | azimuth_output[num_angle_bins + i].imag * corrImag 220 | out1CorrImag = azimuth_output[num_angle_bins + i].imag * corrReal + \ 221 | azimuth_output[num_angle_bins + i].real * corrImag 222 | 223 | azimuth_output[i] = (azimuth_output[i].real + out1CorrReal) + \ 224 | (azimuth_output[i].imag + out1CorrImag) * 1j 225 | 226 | return 227 | 228 | # 清除positive_bin_idx：negative_bin_idx之间的静态目标 229 | def dc_range_signature_removal(fft_out1_d, 230 | positive_bin_idx, 231 | negative_bin_idx, 232 | calib_dc_range_sig_cfg, 233 | num_tx_antennas, 234 | num_chirps_per_frame): 235 | """Compensation of DC range antenna signature. 236 | 237 | Antenna coupling signature dominates the range bins close to the radar. These are the bins in the range FFT output 238 | located around DC. This feature is under user control in terms of enable/disable and start/end range bins through a 239 | CLI command called calibDcRangeSig. During measurement (when the CLI command is issued with feature enabled), each 240 | of the specified range bins for each of the virtual antennas are accumulated over the specified number of chirps 241 | and at the end of the period, the average is computed for each bin/antenna combination for removal after the 242 | measurement period is over. Note that the number of chirps to average must be power of 2. It is assumed that no 243 | objects are present in the vicinity of the radar during this measurement period. After measurement is done, the 244 | removal starts for all subsequent frames during which each of the bin/antenna average estimate is subtracted from 245 | the corresponding received samples in real-time for subsequent processing. 246 | 247 | This function has a measurement phase while calib_dc_range_sig_cfg.counter is less than the preferred value and calibration 248 | phase afterwards. The original function is performed per chirp. Here it is modified to be called per frame. 249 | 250 | Args: 251 | fft_out1_d: (num_chirps_per_frame, num_rx_antennas, numRangeBins). Output of 1D FFT. 252 | positive_bin_idx: the first positive_bin_idx range bins (inclusive) to be compensated. 253 | negative_bin_idx: the last -negative_bin_idx range bins to be compensated. 254 | calib_dc_range_sig_cfg: a simple class for calibration configuration's storing purpose. 255 | num_tx_antennas: number of transmitters. 256 | num_chirps_per_frame: number of total chirps per frame. 257 | 258 | Returns: 259 | None. fft_out1_d is modified in-place. 260 | """ 261 | if not calib_dc_range_sig_cfg.counter: 262 | calib_dc_range_sig_cfg.mean.fill(0) 263 | 264 | # Calibration 265 | if calib_dc_range_sig_cfg.counter < calib_dc_range_sig_cfg.num_frames * num_tx_antennas: 266 | # Accumulate 267 | calib_dc_range_sig_cfg.mean[0, :positive_bin_idx + 1] = np.sum( 268 | fft_out1_d[0::2, :, :positive_bin_idx + 1], 269 | axis=(0, 1)) 270 | calib_dc_range_sig_cfg.mean[0, positive_bin_idx + 1:] = np.sum(fft_out1_d[0::2, :, negative_bin_idx:], 271 | axis=(0, 1)) 272 | 273 | calib_dc_range_sig_cfg.mean[1, :positive_bin_idx + 1] = np.sum( 274 | fft_out1_d[1::2, :, :positive_bin_idx + 1], 275 | axis=(0, 1)) 276 | calib_dc_range_sig_cfg.mean[1, positive_bin_idx + 1:] = np.sum(fft_out1_d[1::2, :, negative_bin_idx:], 277 | axis=(0, 1)) 278 | 279 | calib_dc_range_sig_cfg.counter += 1 280 | 281 | if calib_dc_range_sig_cfg.counter == (calib_dc_range_sig_cfg.num_frames * num_tx_antennas): 282 | # Divide 283 | num_avg_chirps = calib_dc_range_sig_cfg.num_frames * num_chirps_per_frame 284 | calib_dc_range_sig_cfg.mean /= num_avg_chirps 285 | 286 | else: 287 | # fft_out1_d -= mean 288 | fft_out1_d[0::2, :, :positive_bin_idx + 1] -= calib_dc_range_sig_cfg.mean[0, :positive_bin_idx + 1] 289 | fft_out1_d[0::2, :, positive_bin_idx + 1:] -= calib_dc_range_sig_cfg.mean[0, positive_bin_idx + 1:] 290 | fft_out1_d[1::2, :, :positive_bin_idx + 1] -= calib_dc_range_sig_cfg.mean[1, :positive_bin_idx + 1] 291 | fft_out1_d[1::2, :, positive_bin_idx + 1:] -= calib_dc_range_sig_cfg.mean[1, positive_bin_idx + 1:] 292 | 293 | 294 | def clutter_removal(input_val, axis=0): 295 | """Perform basic static clutter removal by removing the mean from the input_val on the specified doppler axis. 296 | 297 | Args: 298 | input_val (ndarray): Array to perform static clutter removal on. Usually applied before performing doppler FFT. 299 | e.g. [num_chirps, num_vx_antennas, num_samples], it is applied along the first axis. 300 | axis (int): Axis to calculate mean of pre-doppler. 301 | 302 | Returns: 303 | ndarray: Array with static clutter removed. 304 | 305 | """ 306 | # Reorder the axes 307 | reordering = np.arange(len(input_val.shape)) 308 | reordering[0] = axis 309 | reordering[axis] = 0 310 | input_val = input_val.transpose(reordering) 311 | 312 | # Apply static clutter removal 313 | mean = input_val.transpose(reordering).mean(0) 314 | output_val = input_val - mean 315 | 316 | return output_val.transpose(reordering) 317 | -------------------------------------------------------------------------------- /dsp/doppler_processing.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | import numpy as np 14 | from . import compensation 15 | from . import utils 16 | 17 | 18 | def doppler_resolution(band_width, start_freq_const=77, ramp_end_time=62, idle_time_const=100, num_loops_per_frame=128, 19 | num_tx_antennas=3): 20 | """Calculate the doppler resolution for the given radar configuration. 21 | 22 | Args: 23 | start_freq_const (int): Frequency chirp starting point. 24 | ramp_end_time (int): Frequency chirp end point. 25 | idle_time_const (int): Idle time between chirps. 26 | band_width (float): Radar config bandwidth. 27 | num_loops_per_frame (int): The number of loops in each frame. 28 | num_tx_antennas (int): The number of transmitting antennas (tx) on the radar. 29 | 30 | Returns: 31 | doppler_resolution (float): The doppler resolution for the given radar configuration. 32 | 33 | """ 34 | 35 | light_speed_meter_per_sec = 299792458 36 | 37 | center_frequency = start_freq_const * 1e9 + band_width / 2 38 | chirp_interval = (ramp_end_time + idle_time_const) * 1e-6 39 | doppler_resolution = light_speed_meter_per_sec / ( 40 | 2 * num_loops_per_frame * num_tx_antennas * center_frequency * chirp_interval) 41 | 42 | return doppler_resolution 43 | 44 | 45 | def separate_tx(signal, num_tx, vx_axis=1, axis=0): 46 | """Separate interleaved radar data from separate TX along a certain axis to account for TDM radars. 47 | 48 | Args: 49 | signal (ndarray): Received signal. 50 | num_tx (int): Number of transmit antennas. 51 | vx_axis (int): Axis in which to accumulate the separated data. 52 | axis (int): Axis in which the data is interleaved. 53 | 54 | Returns: 55 | ndarray: Separated received data in the 56 | 57 | """ 58 | # Reorder the axes 59 | reordering = np.arange(len(signal.shape)) 60 | reordering[0] = axis 61 | reordering[axis] = 0 62 | signal = signal.transpose(reordering) 63 | 64 | out = np.concatenate([signal[i::num_tx, ...] for i in range(num_tx)], axis=vx_axis) 65 | 66 | return out.transpose(reordering) 67 | 68 | 69 | def doppler_processing(radar_cube, 70 | num_tx_antennas=2, 71 | clutter_removal_enabled=False, 72 | interleaved=False, 73 | window_type_2d=None, 74 | accumulate=True): 75 | """Perform 2D FFT on the radar_cube. 76 | 77 | Interleave the radar_cube, perform optional windowing and 2D FFT on the radar_cube. Optional antenna couping 78 | signature removal can also be performed right before 2D FFT. In constrast to the original TI codes, CFAR and peak 79 | grouping are intentionally separated with 2D FFT for the easiness of debugging. 80 | 81 | Args: 82 | radar_cube (ndarray): Output of the 1D FFT. If not interleaved beforehand, it has the shape of 83 | (numChirpsPerFrame, numRxAntennas, numRangeBins). Otherwise, it has the shape of 84 | (numRangeBins, numVirtualAntennas, num_doppler_bins). It is assumed that after interleaving the doppler 85 | dimension is located at the last axis. 86 | num_tx_antennas (int): Number of transmitter antennas. This affects how interleaving is performed. 87 | clutter_removal_enabled (boolean): Flag to enable naive clutter removal. 88 | interleaved (boolean): If the input radar_cube is interleaved before passing in. The default radar_cube is not 89 | interleaved, i.e. has the shape of (numChirpsPerFrame, numRxAntennas, numRangeBins). The interleaving 90 | process will transform it such that it becomes (numRangeBins, numVirtualAntennas, num_doppler_bins). Note 91 | that this interleaving is only applicable to TDM radar, i.e. each tx emits the chirp sequentially. 92 | window_type_2d (mmwave.dsp.utils.Window): Optional windowing type before doppler FFT. 93 | accumulate (boolean): Flag to reduce the numVirtualAntennas dimension. 94 | 95 | Returns: 96 | detMatrix (ndarray): (numRangeBins, num_doppler_bins) complete range-dopper information. Original datatype is 97 | uint16_t. Note that azimuthStaticHeatMap can be extracted from zero-doppler index for 98 | visualization. 99 | aoa_input (ndarray): (numRangeBins, numVirtualAntennas, num_doppler_bins) ADC data reorganized by vrx instead of 100 | physical rx. 101 | """ 102 | 103 | if interleaved: 104 | # radar_cube is interleaved in the first dimension (for 2 tx and 0-based indexing, odd are the chirps from tx1, 105 | # and even are from tx2) so it becomes ( , numVirtualAntennas, numADCSamples), where 106 | # numChirpsPerFrame = num_doppler_bins * num_tx_antennas as designed. 107 | # Antennas associated to tx1 (Ping) are 0:4 and to tx2 (Pong) are 5:8. 108 | fft2d_in = separate_tx(radar_cube, num_tx_antennas, vx_axis=1, axis=0) 109 | else: 110 | fft2d_in = radar_cube 111 | 112 | # (Optional) Static Clutter Removal 113 | if clutter_removal_enabled: 114 | fft2d_in = compensation.clutter_removal(fft2d_in, axis=0) 115 | 116 | # transpose to (numRangeBins, numVirtualAntennas, num_doppler_bins) 117 | fft2d_in = np.transpose(fft2d_in, axes=(2, 1, 0)) # numADCSamples, numRxAntennas, numChirpsPerFrame, 118 | 119 | # Windowing 16x32 120 | if window_type_2d: 121 | fft2d_in = utils.windowing(fft2d_in, window_type_2d, axis=2) 122 | 123 | # It is assumed that doppler is at the last axis. 124 | # FFT 32x32 125 | fft2d_out = np.fft.fft(fft2d_in) 126 | aoa_input = fft2d_out 127 | 128 | # Save zero-Doppler as azimuthStaticHeatMap, watch out for the bit shift in 129 | # original code. 130 | 131 | # Log_2 Absolute Value 132 | fft2d_log_abs = np.log2(np.abs(fft2d_out)) 133 | 134 | # Accumulate 135 | if accumulate: 136 | return np.sum(fft2d_log_abs, axis=1), aoa_input 137 | else: 138 | # return fft2d_log_abs, aoa_input 139 | return fft2d_out, aoa_input 140 | 141 | 142 | def doppler_estimation(radar_cube, 143 | beam_weights, 144 | num_tx_antennas=2, 145 | clutter_removal_enabled=False, 146 | interleaved=False, 147 | window_type_2d=None): 148 | """Perform doppler estimation on the weighted sum of range FFT output across all virtual antennas. 149 | 150 | In contrast to directly computing doppler FFT from the output of range FFT, this function combines it across all 151 | the virtual receivers first using the weights generated from beamforming. Then FFT is performed and argmax is taken 152 | across each doppler axis to return the indices of max doppler values. 153 | 154 | Args: 155 | radar_cube (ndarray): Output of the 1D FFT with only ranges on detected objects. If not interleaved beforehand, 156 | it has the shape of (numChirpsPerFrame, numRxAntennas, numDetObjs). Otherwise, it has the shape of 157 | (numDetObjs, numVirtualAntennas, num_doppler_bins). It is assumed that after interleaving the doppler 158 | dimension is located at the last axis. 159 | beam_weights (ndarray): Weights to sum up the radar_cube across the virtual receivers. It is from the 160 | beam-forming and has the shape of (numVirtualAntennas, numDetObjs) 161 | num_tx_antennas (int): Number of transmitter antennas. This affects how interleaving is performed. 162 | clutter_removal_enabled (boolean): Flag to enable naive clutter removal. 163 | interleaved (boolean): If the input radar_cube is interleaved before passing in. The default radar_cube is not 164 | interleaved, i.e. has the shape of (numChirpsPerFrame, numRxAntennas, numDetObjs). The interleaveing process 165 | will transform it such that it becomes (numDetObjs, numVirtualAntennas, num_doppler_bins). Note that this 166 | interleaving is only appliable to TDM radar, i.e. each tx emits the chirp sequentially. 167 | window_type_2d (string): Optional windowing type before doppler FFT. 168 | 169 | Returns: 170 | doppler_est (ndarray): (numDetObjs) Doppler index for each detected objects. Positive index means moving away 171 | from radar while negative index means moving towards the radar. 172 | """ 173 | fft2d_in = None 174 | if not interleaved: 175 | num_doppler_bins = radar_cube.shape[0] / num_tx_antennas 176 | # radar_cube is interleaved in the first dimension (for 2 tx and 0-based indexing, odd are the chirps from tx1, 177 | # and even are from tx2) so it becomes (num_doppler_bins, numVirtualAntennas, numADCSamples), where 178 | # numChirpsPerFrame = num_doppler_bins * num_tx_antennas as designed. 179 | # Antennas associated to tx1 (Ping) are 0:4 and to tx2 (Pong) are 5:8. 180 | if num_tx_antennas == 2: 181 | fft2d_in = np.concatenate((radar_cube[0::2, ...], radar_cube[1::2, ...]), axis=1) 182 | elif num_tx_antennas == 3: 183 | fft2d_in = np.concatenate((radar_cube[0::3, ...], radar_cube[1::3, ...], radar_cube[2::3, ...]), axis=1) 184 | 185 | # transpose to (numRangeBins, numVirtualAntennas, num_doppler_bins) 186 | fft2d_in = np.transpose(fft2d_in, axes=(2, 1, 0)) 187 | else: 188 | num_doppler_bins = radar_cube.shape[2] 189 | 190 | # (Optional) Static Clutter Removal 191 | if clutter_removal_enabled: 192 | fft2d_in = compensation.clutter_removal(fft2d_in) 193 | 194 | # Weighted sum across all virtual receivers. 195 | fft2d_in = np.einsum('ijk,jk->ik', fft2d_in, beam_weights) 196 | 197 | # Windowing 16x32 198 | if window_type_2d: 199 | fft2d_in = utils.windowing(fft2d_in, window_type_2d, axis=1) 200 | 201 | # It is assumed that doppler is at the last axis. 202 | # FFT 32x32 203 | doppler_est = np.fft.fft(fft2d_in) 204 | doppler_est = np.argmax(doppler_est, axis=1) 205 | doppler_est[doppler_est[:] >= num_doppler_bins] -= num_doppler_bins * 2 206 | 207 | return doppler_est 208 | -------------------------------------------------------------------------------- /dsp/music.py: -------------------------------------------------------------------------------- 1 | import numpy as np 2 | import numpy.linalg as LA 3 | from .angle_estimation import cov_matrix 4 | 5 | def _noise_subspace(covariance, num_sources): 6 | """helper function to get noise_subspace. 7 | """ 8 | if covariance.ndim != 2 or covariance.shape[0] != covariance.shape[1]: 9 | raise ValueError("covariance matrix should be a 2D square matrix.") 10 | if num_sources >= covariance.shape[0]: 11 | raise ValueError("number of sources should be less than number of receivers.") 12 | _, v = LA.eigh(covariance) 13 | 14 | return v[:, :-num_sources] 15 | 16 | def aoa_music_1D(steering_vec, rx_chirps, num_sources): 17 | """Implmentation of 1D MUltiple SIgnal Classification (MUSIC) algorithm on ULA (Uniformed Linear Array). 18 | 19 | Current implementation assumes covariance matrix is not rank deficient and ULA spacing is half of the wavelength. 20 | .. math:: 21 | P_{} (\\theta) = \\frac{1}{a^{H}(\\theta) \mathbf{E}_\mathrm{n}\mathbf{E}_\mathrm{n}^H a(\\theta)} 22 | where :math:`E_{n}` is the noise subpace and :math:`a` is the steering vector. 23 | 24 | 25 | Args: 26 | steering_vec (~np.ndarray): steering vector with the shape of (FoV/angel_resolution, num_ant). 27 | FoV/angel_resolution is usually 181. It is generated from gen_steering_vec() function. 28 | rx_chirps (~np.ndarray): Ouput of the 1D range FFT. The shape is (num_ant, num_chirps_per_frame). 29 | num_sources (int): Number of sources in the scene. Needs to be smaller than num_ant for ULA. 30 | 31 | Returns: 32 | (~np.ndarray): the spectrum of the MUSIC. Objects should be holes for the equation and thus sharp peaks. 33 | """ 34 | num_antennas = rx_chirps.shape[0] 35 | assert num_antennas == steering_vec.shape[1], "Mismatch between number of receivers in " 36 | if num_antennas < num_sources: 37 | raise ValueError("number of sources shoule not exceed number ") 38 | 39 | R = cov_matrix(rx_chirps) 40 | noise_subspace = _noise_subspace(R, num_sources) 41 | v = noise_subspace.T.conj() @ steering_vec.T 42 | spectrum = np.reciprocal(np.sum(v * v.conj(), axis=0).real) 43 | 44 | return spectrum 45 | 46 | def aoa_root_music_1D(steering_vec, rx_chirps, num_sources): 47 | """Implmentation of 1D root MUSIC algorithm on ULA (Uniformed Linear Array). 48 | 49 | The root MUSIC follows the same equation as the original MUSIC, only to solve the equation instead of perform 50 | matrix multiplication. 51 | This implementations referred to the github.com/morriswmz/doatools.py 52 | 53 | Args: 54 | steering_vec (~np.ndarray): steering vector with the shape of (FoV/angel_resolution, num_ant). 55 | FoV/angel_resolution is usually 181. It is generated from gen_steering_vec() function. 56 | rx_chirps (~np.ndarray): Ouput of the 1D range FFT. The shape is (num_ant, num_chirps_per_frame). 57 | num_sources (int): Number of sources in the scene. Needs to be smaller than num_ant for ULA. 58 | 59 | Returns: 60 | (~np.ndarray): the spectrum of the MUSIC. Objects should be holes for the equation and thus sharp peaks. 61 | """ 62 | num_antennas = rx_chirps.shape[0] 63 | assert num_antennas == steering_vec.shape[1], "Mismatch between number of receivers in " 64 | if num_antennas < num_sources: 65 | raise ValueError("number of sources shoule not exceed number ") 66 | 67 | R = cov_matrix(rx_chirps) 68 | noise_subspace = _noise_subspace(R, num_sources) 69 | v = noise_subspace @ noise_subspace.T.conj() 70 | coeffs = np.zeros(num_antennas-1, dtype=np.complex64) 71 | for i in range(1, num_antennas): 72 | coeffs[i - 1] += np.sum(np.diag(v, i)) 73 | coeffs = np.hstack((coeffs[::-1], np.sum(np.diag(v)), coeffs.conj())) 74 | 75 | z = np.roots(coeffs) 76 | z = np.abs(z[z <= 1.0]) 77 | if len(z) < num_sources: 78 | return None 79 | z.sort() 80 | z = z[-num_sources:] 81 | 82 | # Assume to be half wavelength spacing 83 | sin_vals = np.angle(z) / np.pi 84 | locations = np.rad2deg(np.arcsin(sin_vals)) 85 | 86 | return locations 87 | 88 | def aoa_spatial_smoothing(covariance_matrix, num_subarrays, forward_backward=False): 89 | """Perform spatial smoothing on the precomputed covariance matrix. 90 | 91 | Spatial smoothing is to decorrelate the coherent sources. It is performed over covariance matrix. 92 | This implementations referred to the github.com/morriswmz/doatools.py 93 | 94 | Args: 95 | covariance_matrx (~np.ndarray): Covariance matrix of input signal. 96 | num_subarrays (int): Number of subarrays to perform the spatial smoothing. 97 | forward_backward (bool): If True, perform backward smoothing as well. 98 | 99 | Returns: 100 | (~np.ndarray): Decorrelated covariance matrix. 101 | """ 102 | num_receivers = covariance_matrix.shape[0] 103 | assert num_subarrays >=1 and num_subarrays <= num_receivers, "num_subarrays is wrong" 104 | 105 | # Forward pass 106 | result = covariance_matrix[:num_receivers-num_subarrays+1, :num_receivers-num_subarrays+1].copy() 107 | for i in range(1, num_subarrays): 108 | result += covariance_matrix[i:i+num_receivers-num_subarrays+1, i:i+num_receivers-num_subarrays+1] 109 | result /= num_subarrays 110 | if not forward_backward: 111 | return result 112 | 113 | # Adds backward pass 114 | if np.iscomplexobj(result): 115 | return 0.5 * (result + np.flip(result).conj()) 116 | else: 117 | return 0.5 * (result + np.flip(result)) 118 | 119 | def aoa_esprit(steering_vec, rx_chirps, num_sources, displacement): 120 | """ Perform Estimation of Signal Parameters via Rotation Invariance Techniques (ESPIRIT) for Angle of Arrival. 121 | 122 | ESPRIT exploits the structure in the signal subspace. 123 | 124 | Args: 125 | steering_vec (~np.ndarray): steering vector with the shape of (FoV/angel_resolution, num_ant). 126 | FoV/angel_resolution is usually 181. It is generated from gen_steering_vec() function. 127 | rx_chirps (~np.ndarray): Ouput of the 1D range FFT. The shape is (num_ant, num_chirps_per_frame). 128 | num_sources (int): Number of sources in the scene. Needs to be smaller than num_ant for ULA. 129 | displacement (int): displacmenet between two subarrays. 130 | 131 | Returns: 132 | (~np.ndarray): the spectrum of the ESPRIT. Objects should be holes for the equation and thus sharp peaks. 133 | """ 134 | num_antennas = rx_chirps.shape[0] 135 | if displacement > num_antennas/2 or displacement <= 0: 136 | raise ValueError("The separation between two subarrays can only range from 1 to half of the original array size.") 137 | 138 | subarray1 = rx_chirps[:num_antennas - displacement] 139 | subarray2 = rx_chirps[displacement:] 140 | assert subarray1.shape == subarray2.shape, "separating subarrays encounters error." 141 | 142 | R1 = cov_matrix(subarray1) 143 | R2 = cov_matrix(subarray2) 144 | _, v1 = LA.eigh(R1) 145 | _, v2 = LA.eigh(R2) 146 | 147 | E1 = v1[:, -num_sources:] 148 | E2 = v2[:, -num_sources:] 149 | C = np.concatenate((E1.T.conj(), E2.T.conj()), axis=0) @ np.concatenate((E1, E2), axis=1) 150 | _, Ec = LA.eigh(C) 151 | Ec = Ec[::-1, :] 152 | 153 | phi = -Ec[:num_antennas, num_antennas:] @ LA.inv(Ec[num_antennas:, num_antennas:]) 154 | w, _ = LA.eig(phi) 155 | 156 | sin_vals = np.angle(w) / np.pi 157 | locations = np.rad2deg(np.arcsin(sin_vals)) 158 | 159 | return locations -------------------------------------------------------------------------------- /dsp/noise_removal.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | import numpy as np 14 | 15 | 16 | def peak_grouping_along_doppler(det_obj_2d, 17 | det_matrix, 18 | num_doppler_bins): 19 | """Perform peak grouping along the doppler direction only. 20 | This is a temporary remedy for the slow and old implementation of peak_grouping_qualified() function residing in 21 | dsp.py currently. Will merge this back to there to enable more generic peak grouping. 22 | """ 23 | num_det_objs = det_obj_2d.shape[0] 24 | range_idx = det_obj_2d['rangeIdx'] 25 | doppler_idx = det_obj_2d['dopplerIdx'] 26 | kernel = np.zeros((num_det_objs, 3), dtype=np.float32) 27 | kernel[:, 0] = det_matrix[range_idx, doppler_idx - 1] 28 | kernel[:, 1] = det_obj_2d['peakVal'].astype(np.float32) 29 | kernel[:, 2] = det_matrix[range_idx, (doppler_idx + 1) % num_doppler_bins] 30 | detectedFlag = (kernel[:, 1] > kernel[:, 0]) & (kernel[:, 1] > kernel[:, 2]) 31 | return det_obj_2d[detectedFlag] 32 | 33 | 34 | def range_based_pruning(det_obj_2d_raw, 35 | snr_thresh, 36 | peak_val_thresh, 37 | max_range, 38 | min_range, 39 | range_resolution): 40 | """Filter out the objects out of the range and not sufficing SNR/peakVal requirement. 41 | 42 | Filter out the objects based on the two following conditions: 43 | 1. Not within [min_range and max_range]. 44 | 2. Does not satisfy SNR/peakVal requirement, where it requires higher standard when closer and lower when further. 45 | """ 46 | det_obj_2d = det_obj_2d_raw[(det_obj_2d_raw['rangeIdx'] >= min_range) & \ 47 | (det_obj_2d_raw['rangeIdx'] <= max_range)] 48 | snr_idx1 = (det_obj_2d['SNR'] > snr_thresh[0, 1]) & (det_obj_2d['rangeIdx'] * range_resolution < snr_thresh[0, 0]) 49 | snr_idx2 = (det_obj_2d['SNR'] > snr_thresh[1, 1]) & \ 50 | (det_obj_2d['rangeIdx'] * range_resolution < snr_thresh[1, 0]) & \ 51 | (det_obj_2d['rangeIdx'] * range_resolution >= snr_thresh[0, 0]) 52 | snr_idx3 = (det_obj_2d['SNR'] > snr_thresh[2, 1]) & (det_obj_2d['rangeIdx'] * range_resolution > snr_thresh[1, 0]) 53 | snr_idx = snr_idx1 | snr_idx2 | snr_idx3 54 | 55 | peak_val_idx = np.logical_not((det_obj_2d['peakVal'] < peak_val_thresh[0, 1]) & \ 56 | (det_obj_2d['rangeIdx'] * range_resolution < peak_val_thresh[0, 0])) 57 | combined_idx = snr_idx & peak_val_idx 58 | det_obj_2d = det_obj_2d[combined_idx] 59 | 60 | return det_obj_2d 61 | 62 | 63 | def prune_to_peaks(det_obj2_d_raw, 64 | det_matrix, 65 | num_doppler_bins, 66 | reserve_neighbor=False): 67 | """Reduce the CFAR detected output to local peaks. 68 | 69 | Reduce the detected output to local peaks. If reserveNeighbor is toggled, will also return the larger neighbor. For 70 | example, given an array [2, 1, 5, 3, 2], default method will return [2, 5] while reserve neighbor will return 71 | [2, 5, 3]. The neighbor has to be a larger neighbor of the two immediate ones and also be part of the peak. the 1st 72 | element "1" in the example is not returned because it's smaller than both sides so that it is not part of the peak. 73 | 74 | Args: 75 | det_obj2_d_raw (np.ndarray): The detected objects structured array which contains the range_idx, doppler_idx, 76 | peakVal and SNR, etc. 77 | det_matrix (np.ndarray): Output of doppler FFT with virtual antenna dimensions reduced. It has the shape of 78 | (num_range_bins, num_doppler_bins). 79 | num_doppler_bins (int): Number of doppler bins. 80 | reserve_neighbor (boolean): if toggled, will return both peaks and the larger neighbors. 81 | 82 | Returns: 83 | cfar_det_obj_index_pruned (np.ndarray): Pruned version of cfar_det_obj_index. 84 | cfar_det_obj_SNR_pruned (np.ndarray): Pruned version of cfar_det_obj_SNR. 85 | """ 86 | 87 | range_idx = det_obj2_d_raw['rangeIdx'] 88 | doppler_idx = det_obj2_d_raw['dopplerIdx'] 89 | next_idx = doppler_idx + 1 90 | next_idx[doppler_idx == num_doppler_bins - 1] = 0 91 | prev_idx = doppler_idx - 1 92 | prev_idx[doppler_idx == 0] = num_doppler_bins - 1 93 | 94 | prev_val = det_matrix[range_idx, prev_idx] 95 | current_val = det_matrix[range_idx, doppler_idx] 96 | next_val = det_matrix[range_idx, next_idx] 97 | 98 | if reserve_neighbor: 99 | next_next_idx = next_idx + 1 100 | next_next_idx[next_idx == num_doppler_bins - 1] = 0 101 | prev_prev_idx = prev_idx - 1 102 | prev_prev_idx[prev_idx == 0] = num_doppler_bins - 1 103 | 104 | prev_prev_val = det_matrix[range_idx, prev_prev_idx] 105 | next_next_val = det_matrix[range_idx, next_next_idx] 106 | is_neighbor_of_peak_next = (current_val > next_next_val) & (current_val > prev_val) 107 | is_neighbor_of_peak_prev = (current_val > prev_prev_val) & (current_val > next_val) 108 | 109 | pruned_idx = (current_val > prev_val) & (current_val > next_val) | is_neighbor_of_peak_next | is_neighbor_of_peak_prev 110 | else: 111 | pruned_idx = (current_val > prev_val) & (current_val > next_val) 112 | 113 | det_obj2_d_pruned = det_obj2_d_raw[pruned_idx] 114 | 115 | return det_obj2_d_pruned 116 | -------------------------------------------------------------------------------- /dsp/range_processing.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | import numpy as np 14 | from . import utils 15 | from . import ZoomFFT 16 | 17 | def range_resolution(num_adc_samples, dig_out_sample_rate=2500, freq_slope_const=60.012): 18 | """ Calculate the range resolution for the given radar configuration 19 | 20 | Args: 21 | num_adc_samples (int): The number of given ADC samples in a chirp 22 | dig_out_sample_rate (int): The ADC sample rate 23 | freq_slope_const (float): The slope of the freq increase in each chirp 24 | 25 | Returns: 26 | tuple [float, float]: 27 | range_resolution (float): The range resolution for this bin 28 | band_width (float): The bandwidth of the radar chirp config 29 | """ 30 | light_speed_meter_per_sec = 299792458 31 | freq_slope_m_hz_per_usec = freq_slope_const 32 | adc_sample_period_usec = 1000.0 / dig_out_sample_rate * num_adc_samples 33 | band_width = freq_slope_m_hz_per_usec * adc_sample_period_usec * 1e6 34 | range_resolution = light_speed_meter_per_sec / (2.0 * band_width) 35 | 36 | return range_resolution, band_width 37 | 38 | 39 | def range_processing(adc_data, window_type_1d=None, axis=-1): 40 | """Perform 1D FFT on complex-format ADC data. 41 | 42 | Perform optional windowing and 1D FFT on the ADC data. 43 | 44 | Args: 45 | adc_data (ndarray): (num_chirps_per_frame, num_rx_antennas, num_adc_samples). Performed on each frame. adc_data 46 | is in complex by default. Complex is float32/float32 by default. 47 | window_type_1d (mmwave.dsp.utils.Window): Optional window type on 1D FFT input. Default is None. Can be selected 48 | from Bartlett, Blackman, Hanning and Hamming. 49 | 50 | Returns: 51 | radar_cube (ndarray): (num_chirps_per_frame, num_rx_antennas, num_range_bins). Also called fft_1d_out 52 | """ 53 | # windowing numA x numB suggests the coefficients is numA-bits while the 54 | # input and output are numB-bits. Same rule applies to the FFT. 55 | fft1d_window_type = window_type_1d 56 | if fft1d_window_type: 57 | fft1d_in = utils.windowing(adc_data, fft1d_window_type, axis=axis) 58 | else: 59 | fft1d_in = adc_data 60 | 61 | # Note: np.fft.fft is a 1D operation, using higher dimension input defaults to slicing last axis for transformation 62 | radar_cube = np.fft.fft(fft1d_in, n = 64,axis=axis) 63 | # radar_cube = np.fft.fft(fft1d_in, axis=axis) 64 | 65 | return radar_cube 66 | 67 | 68 | def zoom_range_processing(adc_data, low_freq, high_freq, fs, d, resample_number): 69 | """Perform ZoomFFT on complex-format ADC data in a user-defined frequency range. 70 | 71 | Args: 72 | adc_data (ndarray): (num_chirps_per_frame, num_rx_antennas, num_adc_samples). Performed on each frame. adc_data 73 | is in complex by default. Complex is float32/float32 by default. 74 | low_freq (int): a user-defined number which specifies the lower bound on the range of frequency spectrum which 75 | the user would like to zoom on 76 | high_freq (int): a user-defined number which specifies the higher bound on the range of frequency spectrum which 77 | the user would like to zoom on 78 | fs (int) : sampling rate of the original signal 79 | d (int): Sample spacing (inverse of the sampling rate) 80 | resample_number (int): The number of samples in the re-sampled signal. 81 | 82 | Returns: 83 | zoom_fft_spectrum (ndarray): (num_chirps_per_frame, num_rx_antennas, resample_number). 84 | """ 85 | # adc_data shape: [num_chirps_per_frame, num_rx_antennas, num_range_bins] 86 | num_chirps_per_frame = adc_data.shape[0] 87 | num_rx_antennas = adc_data.shape[1] 88 | # num_range_bins = adc_data.shape[2] 89 | 90 | zoom_fft_spectrum = np.zeros(shape=(num_chirps_per_frame, num_rx_antennas, resample_number)) 91 | 92 | for i in range(num_chirps_per_frame): 93 | for j in range(num_rx_antennas): 94 | zoom_fft_inst = ZoomFFT.ZoomFFT(low_freq, high_freq, fs, adc_data[i, j, :]) 95 | zoom_fft_inst.compute_fft() 96 | zoom_fft_spectrum[i, j, :] = zoom_fft_inst.compute_zoomfft() 97 | 98 | return zoom_fft_spectrum 99 | 100 | 101 | def zoom_fft_visualize(zoom_fft_spectrum, antenna_idx, range_bin_idx): 102 | '''to be implemented''' 103 | pass 104 | -------------------------------------------------------------------------------- /dsp/utils.py: -------------------------------------------------------------------------------- 1 | # Copyright 2019 The OpenRadar Authors. All Rights Reserved. 2 | # Licensed under the Apache License, Version 2.0 (the "License"); 3 | # you may not use this file except in compliance with the License. 4 | # You may obtain a copy of the License at 5 | # http://www.apache.org/licenses/LICENSE-2.0 6 | # Unless required by applicable law or agreed to in writing, software 7 | # distributed under the License is distributed on an "AS IS" BASIS, 8 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 9 | # See the License for the specific language governing permissions and 10 | # limitations under the License. 11 | # ============================================================================== 12 | 13 | import numpy as np 14 | try: 15 | from enum import Enum 16 | except ImportError: 17 | print("enum only exists in Python 3.4 or newer") 18 | 19 | try: 20 | class Window(Enum): 21 | BARTLETT = 1 22 | BLACKMAN = 2 23 | HAMMING = 3 24 | HANNING = 4 25 | except NameError: 26 | class Window: 27 | BARTLETT = 1 28 | BLACKMAN = 2 29 | HAMMING = 3 30 | HANNING = 4 31 | 32 | RANGEIDX = 0 33 | DOPPLERIDX = 1 34 | PEAKVAL = 2 35 | 36 | MAX_OBJ_OUT = 100 37 | 38 | def windowing(input, window_type, axis=0): 39 | """Window the input based on given window type. 40 | 41 | Args: 42 | input: input numpy array to be windowed. 43 | 44 | window_type: enum chosen between Bartlett, Blackman, Hamming, Hanning and Kaiser. 45 | 46 | axis: the axis along which the windowing will be applied. 47 | 48 | Returns: 49 | 50 | """ 51 | 52 | # 输入为(num_chirps_per_frame, num_adc_samples, num_rx_antennas) 53 | # 要转(num_chirps_per_frame, num_rx_antennas, num_adc_samples) 54 | 55 | window_length = input.shape[axis] 56 | if window_type == Window.BARTLETT: 57 | window = np.bartlett(window_length) 58 | elif window_type == Window.BLACKMAN: 59 | window = np.blackman(window_length) 60 | elif window_type == Window.HAMMING: 61 | window = np.hamming(window_length) 62 | elif window_type == Window.HANNING: 63 | window = np.hanning(window_length) 64 | else: 65 | raise ValueError("The specified window is not supported!!!") 66 | output = input * window 67 | return output 68 | 69 | -------------------------------------------------------------------------------- /firmware/xwr1843_mmw_demo.bin: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/firmware/xwr1843_mmw_demo.bin -------------------------------------------------------------------------------- /firmware/xwr6443_mmw_demo.bin: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/firmware/xwr6443_mmw_demo.bin -------------------------------------------------------------------------------- /firmware/xwr6843AOP_mmw_demo.bin: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/firmware/xwr6843AOP_mmw_demo.bin -------------------------------------------------------------------------------- /firmware/xwr6843ISK_mmw_demo.bin: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/firmware/xwr6843ISK_mmw_demo.bin -------------------------------------------------------------------------------- /firmware/xwr6843ODS_mmw_demo.bin: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/firmware/xwr6843ODS_mmw_demo.bin -------------------------------------------------------------------------------- /gesture_icons/0.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/0.jpg -------------------------------------------------------------------------------- /gesture_icons/1.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/1.jpg -------------------------------------------------------------------------------- /gesture_icons/2.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/2.jpg -------------------------------------------------------------------------------- /gesture_icons/3.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/3.jpg -------------------------------------------------------------------------------- /gesture_icons/4.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/4.jpg -------------------------------------------------------------------------------- /gesture_icons/5.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/5.jpg -------------------------------------------------------------------------------- /gesture_icons/6.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/6.jpg -------------------------------------------------------------------------------- /gesture_icons/7.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/gesture_icons/7.jpg -------------------------------------------------------------------------------- /globalvar.py: -------------------------------------------------------------------------------- 1 | #!/usr/bin/python 2 | # -*- coding: utf-8 -*- 3 | 4 | def _init(): 5 | global _global_dict 6 | _global_dict = {} 7 | 8 | def set_value(name, value): 9 | _global_dict[name] = value 10 | 11 | def get_value(name, defValue=None): 12 | try: 13 | return _global_dict[name] 14 | except KeyError: 15 | return defValue 16 | -------------------------------------------------------------------------------- /img/1.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/img/1.png -------------------------------------------------------------------------------- /img/2.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/img/2.png -------------------------------------------------------------------------------- /img/3.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/img/3.png -------------------------------------------------------------------------------- /img/4.gif: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/img/4.gif -------------------------------------------------------------------------------- /img/5.mp4: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/img/5.mp4 -------------------------------------------------------------------------------- /img/6.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/img/6.jpg -------------------------------------------------------------------------------- /img/7.jpg: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/img/7.jpg -------------------------------------------------------------------------------- /iwr6843_tlv/detected_points.py: -------------------------------------------------------------------------------- 1 | import serial 2 | import numpy as np 3 | import time 4 | import struct 5 | 6 | DEBUG=False 7 | MAGIC_WORD_ARRAY = np.array([2, 1, 4, 3, 6, 5, 8, 7]) 8 | MAGIC_WORD = b'\x02\x01\x04\x03\x06\x05\x08\x07' 9 | MSG_AZIMUT_STATIC_HEAT_MAP = 8 10 | 11 | 12 | class IWR6843AOP_TLV: 13 | def __init__(self, sdk_version=3.4, cli_baud=115200,data_baud=921600, num_rx=4, num_tx=3, 14 | verbose=False, connect=True, mode=0,cli_loc='COM9',data_loc='COM10',config_file=""): 15 | super(IWR6843AOP_TLV, self).__init__() 16 | self.connected = False 17 | self.verbose = verbose 18 | self.mode = mode 19 | if connect: 20 | # self.cli_port = serial.Serial(cli_loc, cli_baud) 21 | # self.data_port = serial.Serial(data_loc, data_baud) 22 | self.connected = True 23 | self.sdk_version = sdk_version 24 | self.num_rx_ant = num_rx 25 | self.num_tx_ant = num_tx 26 | self.num_virtual_ant = num_rx * num_tx 27 | self.config_file = config_file 28 | # if mode == 0: 29 | # self._initialize(self.config_file) 30 | 31 | def _configure_radar(self, config): 32 | for i in config: 33 | self.cli_port.write((i + '\n').encode()) 34 | # print(i) 35 | time.sleep(0.01) 36 | 37 | def _initialize(self, config_file): 38 | config = [line.rstrip('\r\n') for line in open(config_file)] 39 | if self.connected: 40 | pass 41 | # self._configure_radar(config) 42 | 43 | self.config_params = {} # Initialize an empty dictionary to store the configuration parameters 44 | 45 | for i in config: 46 | 47 | # Split the line 48 | split_words = i.split(" ") 49 | 50 | # Hard code the number of antennas, change if other configuration is used 51 | num_rx_ant = 4 52 | num_tx_ant = 3 53 | 54 | # Get the information about the profile configuration 55 | if "profileCfg" in split_words[0]: 56 | start_freq = int(split_words[2]) 57 | idle_time = int(split_words[3]) 58 | ramp_end_time = float(split_words[5]) 59 | freq_slope_const = int(split_words[8]) 60 | num_adc_samples = int(split_words[10]) 61 | num_adc_samples_round_to2 = 1 62 | 63 | while num_adc_samples > num_adc_samples_round_to2: 64 | num_adc_samples_round_to2 = num_adc_samples_round_to2 * 2 65 | 66 | dig_out_sample_rate = int(split_words[11]) 67 | 68 | # Get the information about the frame configuration 69 | elif "frameCfg" in split_words[0]: 70 | 71 | chirp_start_idx = int(split_words[1]) 72 | chirp_end_idx = int(split_words[2]) 73 | num_loops = int(split_words[3]) 74 | num_frames = int(split_words[4]) 75 | frame_periodicity = float(split_words[5]) 76 | 77 | # Combine the read data to obtain the configuration parameters 78 | num_chirps_per_frame = (chirp_end_idx - chirp_start_idx + 1) * num_loops 79 | self.config_params["numDopplerBins"] = num_chirps_per_frame / num_tx_ant 80 | self.config_params["numRangeBins"] = num_adc_samples_round_to2 81 | self.config_params["rangeResolutionMeters"] = round((3e8 * dig_out_sample_rate * 1e3) / ( 82 | 2 * freq_slope_const * 1e12 * num_adc_samples),2) 83 | self.config_params["rangeIdxToMeters"] = (3e8 * dig_out_sample_rate * 1e3) / ( 84 | 2 * freq_slope_const * 1e12 * self.config_params["numRangeBins"]) 85 | self.config_params["dopplerResolutionMps"] =round( 3e8 / ( 86 | 2 * start_freq * 1e9 * (idle_time + ramp_end_time) * 1e-6 * self.config_params[ 87 | "numDopplerBins"] * num_tx_ant),2) 88 | self.config_params["maxRange"] = round((300 * 0.9 * dig_out_sample_rate) / (2 * freq_slope_const * 1e3),2) 89 | self.config_params["maxVelocity"] =round( 3e8 / ( 90 | 4 * start_freq * 1e9 * (idle_time + ramp_end_time) * 1e-6 * num_tx_ant),2) 91 | return self.config_params 92 | 93 | 94 | def close(self): 95 | """End connection between radar and machine 96 | 97 | Returns: 98 | None 99 | 100 | """ 101 | self.cli_port.write('sensorStop\n'.encode()) 102 | self.cli_port.close() 103 | self.data_port.close() 104 | 105 | def _read_buffer(self): 106 | """ 107 | 108 | Returns: 109 | 110 | """ 111 | byte_buffer = self.data_port.read(self.data_port.in_waiting) 112 | 113 | return byte_buffer 114 | 115 | def _parse_header_data(self, byte_buffer, idx): 116 | """Parses the byte buffer for the header of the data 117 | 118 | Args: 119 | byte_buffer: Buffer with TLV data 120 | idx: Current reading index of the byte buffer 121 | 122 | Returns: 123 | Tuple [Tuple (int), int] 124 | 125 | """ 126 | magic, idx = self._unpack(byte_buffer, idx, order='>', items=1, form='Q') 127 | (version, length, platform, frame_num, cpu_cycles, num_obj, num_tlvs), idx = self._unpack(byte_buffer, idx, 128 | items=7, form='I') 129 | subframe_num, idx = self._unpack(byte_buffer, idx, items=1, form='I') 130 | return (version, length, platform, frame_num, cpu_cycles, num_obj, num_tlvs, subframe_num), idx 131 | 132 | def _parse_header_tlv(self, byte_buffer, idx): 133 | """ Parses the byte buffer for the header of a tlv 134 | 135 | """ 136 | (tlv_type, tlv_length), idx = self._unpack(byte_buffer, idx, items=2, form='I') 137 | return (tlv_type, tlv_length), idx 138 | 139 | def _parse_msg_detected_points(self, byte_buffer, idx): 140 | """ Parses the information of the detected points message 141 | 142 | """ 143 | (x,y,z,vel), idx = self._unpack(byte_buffer, idx, items=4, form='f') 144 | 145 | return (x,y,z,vel), idx 146 | 147 | def _parse_msg_detected_points_side_info(self,byte_buffer, idx): 148 | (snr,noise), idx = self._unpack(byte_buffer, idx, items=2, form='H') 149 | return (snr,noise),idx 150 | 151 | def _parse_msg_azimut_static_heat_map(self, byte_buffer, idx): 152 | """ Parses the information of the azimuth heat map 153 | 154 | """ 155 | try: 156 | (imag, real), idx = self._unpack(byte_buffer, idx, items=2, form='H') 157 | return (imag, real), idx 158 | except TypeError: 159 | pass 160 | 161 | 162 | def _process_azimut_heat_map(self, byte_buffer): 163 | """ 164 | 热图 165 | """ 166 | idx = byte_buffer.index(MAGIC_WORD) 167 | header_data, idx = self._parse_header_data(byte_buffer, idx) 168 | # print(header_data,idx) 169 | (tlv_type, tlv_length), idx = self._parse_header_tlv(byte_buffer, idx) 170 | # print(tlv_type, tlv_length,idx) 171 | azimuth_map = np.zeros((self.num_virtual_ant, self.config_params['numRangeBins'], 2),dtype=np.int16) 172 | # azimuth_map = np.zeros((7, self.config_params['numRangeBins'], 2),dtype=np.int16) 173 | for bin_idx in range(self.config_params['numRangeBins']): 174 | # for ant in range(7): 175 | for ant in range(self.num_virtual_ant): 176 | azimuth_map[ant][bin_idx][:], idx = self._parse_msg_azimut_static_heat_map(byte_buffer, idx) 177 | return azimuth_map 178 | 179 | def _process_detected_points(self, byte_buffer): 180 | """ 181 | 点云 182 | """ 183 | idx = byte_buffer.index(MAGIC_WORD) 184 | header_data, idx = self._parse_header_data(byte_buffer, idx) 185 | # print(header_data,idx) 186 | 187 | num_tlvs=header_data[6] 188 | 189 | #### tvl1 #### 190 | (tlv_type, tlv_length), idx = self._parse_header_tlv(byte_buffer, idx) 191 | num_points=int(tlv_length/16) 192 | data=np.zeros((num_points,6),dtype=np.float) 193 | for i in range(num_points): 194 | ( x, y, z,vel), idx = self._parse_msg_detected_points(byte_buffer, idx) 195 | data[i][0]=x 196 | data[i][1]=y 197 | data[i][2]=z 198 | data[i][3]=vel 199 | 200 | (tlv_type, tlv_length), idx = self._parse_header_tlv(byte_buffer, idx) 201 | for i in range(num_points): 202 | (snr,noise), idx = self._parse_msg_detected_points_side_info(byte_buffer, idx) 203 | data[i][4]=snr 204 | data[i][5]=noise 205 | 206 | return data 207 | @staticmethod 208 | def _unpack(byte_buffer, idx, order='', items=1, form='I'): 209 | """Helper function for parsing binary byte data 210 | 211 | Args: 212 | byte_buffer: Buffer with data 213 | idx: Curex in the buffer 214 | order: Little endian or big endian 215 | items: Number of items to be extracted 216 | form: Data type to be extracted 217 | 218 | Returns:rent ind 219 | Tuple [Tuple (object), int] 220 | 221 | """ 222 | size = {'H': 2, 'h': 2, 'I': 4, 'Q': 8, 'f': 4} 223 | try: 224 | data = struct.unpack(order + str(items) + form, byte_buffer[idx:idx + (items * size[form])]) 225 | if len(data) == 1: 226 | data = data[0] 227 | return data, idx + (items * size[form]) 228 | except: 229 | return None 230 | 231 | -------------------------------------------------------------------------------- /libs/UDPCAPTUREADCRAWDATA.dll: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/libs/UDPCAPTUREADCRAWDATA.dll -------------------------------------------------------------------------------- /libs/UDPCAPTUREADCRAWDATA.lib: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/libs/UDPCAPTUREADCRAWDATA.lib -------------------------------------------------------------------------------- /libs/libtest.so: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/Tkwer/RadarStream/f6fe0fbf7d4a20cb1198779c97d2c3f69e7bfc17/libs/libtest.so -------------------------------------------------------------------------------- /main.py: -------------------------------------------------------------------------------- 1 | from real_time_process import UdpListener, DataProcessor 2 | from radar_config import SerialConfig 3 | from radar_config import DCA1000Config 4 | from queue import Queue 5 | import pyqtgraph as pg 6 | from PyQt5 import QtWidgets 7 | from pyqtgraph.Qt import QtCore, QtGui, QtWidgets 8 | import time 9 | import torch 10 | import sys 11 | import numpy as np 12 | from serial.tools import list_ports 13 | import iwr6843_tlv.detected_points as readpoint 14 | import globalvar as gl 15 | # import models.predict as predict 16 | # from models.model import CNet, FeatureFusionNet 17 | import os 18 | os.environ['KMP_DUPLICATE_LIB_OK']='True' 19 | import matplotlib.pyplot as plt 20 | from colortrans import pg_get_cmap 21 | 22 | # ----------------------------------------------- 23 | from UI_interface import Ui_MainWindow, Qt_pet 24 | # ----------------------------------------------- 25 | 26 | 27 | datasetfile = 'dataset' 28 | datasetsencefile = ' ' 29 | gesturedict = { 30 | '0':'backward', 31 | '1':'dbclick', 32 | '2':'down', 33 | '3':'front', 34 | '4':'Left', 35 | '5':'Right', 36 | '6':'up', 37 | '7':'NO' 38 | } 39 | 40 | cnt = 0 41 | 42 | _flagdisplay = False 43 | 44 | def loadmodel(): 45 | global model 46 | if (modelfile.currentText()!='--select--'and modelfile.currentText()!=''): 47 | model_info = torch.load(modelfile.currentText(),map_location='cpu') 48 | # TODO: 49 | model = [] 50 | model.load_state_dict(model_info['state_dict']) 51 | printlog('加载'+modelfile.currentText()+'模型成功!',fontcolor='blue') 52 | else: 53 | printlog("请加载模型!",fontcolor='red') 54 | 55 | def cleartjpg(): 56 | view_gesture.setPixmap(QtGui.QPixmap("gesture_icons/"+str(7)+".jpg")) 57 | subWin.img_update("gesture_icons/"+str(7)+".jpg") 58 | 59 | def Judge_gesture(a,b,c,d,e): 60 | global _flagdisplay 61 | if model: 62 | # TODO: 63 | fanhui = [] #predict.predictGesture(model,d,b,e,c,a) 64 | view_gesture.setPixmap(QtGui.QPixmap("gesture_icons/"+str(fanhui)+".jpg")) 65 | subWin.img_update("gesture_icons/"+str(fanhui)+".jpg") 66 | QtCore.QTimer.singleShot(2000, cleartjpg) 67 | _flagdisplay = True 68 | printlog("输出:" + gesturedict[str(fanhui)],fontcolor='blue') 69 | return gesturedict[str(fanhui)] 70 | 71 | def update_figure(): 72 | global img_rdi, img_rai, img_rti, img_rei, img_dti 73 | global idx,cnt 74 | 75 | img_rti.setImage(RTIData.get().sum(2)[0:1024:16,:], levels=[0, 1e4]) 76 | # img_rdi.setImage(RDIData.get()[:, :, 0].T, levels=[30, 50]) 77 | img_rdi.setImage(RDIData.get().sum(0)[:, :, 0].T,levels=[2e4, 4e5]) 78 | # img_rei.setImage(REIData.get().T,levels=[0, 3]) 79 | img_rei.setImage(REIData.get()[4:12,:,:].sum(0).T,levels=[0, 8]) 80 | img_dti.setImage(DTIData.get(),levels=[0, 1000]) 81 | # img_rai.setImage(RAIData.get().sum(0).T, levels=[1.2e3, 4e6]) 82 | # img_rai.setImage(RAIData.get()[0,:,:].T, levels=[8e3, 2e4]) 83 | # img_rai.setImage(RAIData.get(),levels=[0, 3]) 84 | img_rai.setImage(RAIData.get()[4:12,:,:].sum(0),levels=[0, 8]) 85 | 86 | 87 | if gl.get_value('usr_gesture'): 88 | RT_feature = RTIData.get().sum(2)[0:1024:16,:] 89 | DT_feature = DTIData.get() 90 | RDT_feature = RDIData.get()[:, :, :, 0] 91 | ART_feature = RAIData.get() 92 | ERT_feature = REIData.get() 93 | 94 | # if Recognizebtn.isChecked(): 95 | if Recognizebtn.isChecked(): 96 | # 识别 97 | 98 | time_start = time.time() # 记录开始时间 99 | result = Judge_gesture(RT_feature,DT_feature,RDT_feature, 100 | ART_feature,ERT_feature) 101 | time_end = time.time() # 记录结束时间 102 | time_sum = time_end - time_start # 计算的时间差为程序的执行时间，单位为秒/s 103 | printlog('识别时间:'+str(time_sum)+'s, '+'识别结果:'+result,fontcolor='blue') 104 | 105 | 106 | elif CaptureDatabtn.isChecked() and datasetsencefile != '': 107 | idx=idx+1 108 | # 收集 109 | np.save(datasetsencefile+'/RT_feature_'+str(idx).zfill(5)+'.npy',RT_feature) 110 | np.save(datasetsencefile+'/DT_feature_'+str(idx).zfill(5)+'.npy',DT_feature) 111 | np.save(datasetsencefile+'/RDT_feature_'+str(idx).zfill(5)+'.npy',RDT_feature) 112 | np.save(datasetsencefile+'/ART_feature_'+str(idx).zfill(5)+'.npy',ART_feature) 113 | np.save(datasetsencefile+'/ERT_feature_'+str(idx).zfill(5)+'.npy',ERT_feature) 114 | printlog('采集到特征:'+datasetfilebox.currentText()+'-'+str(idx).zfill(5),fontcolor='blue') 115 | 116 | gl.set_value('usr_gesture', False) 117 | 118 | QtCore.QTimer.singleShot(1, update_figure) 119 | 120 | 121 | def printlog(string,fontcolor): 122 | logtxt.moveCursor(QtGui.QTextCursor.End) 123 | gettime = time.strftime("%H:%M:%S", time.localtime()) 124 | logtxt.append(""+str(gettime)+"-->"+string+"") 125 | 126 | def getradarparameters(): 127 | if radarparameters.currentIndex() > -1 and radarparameters.currentText() != '--select--': 128 | radarparameters.setToolTip(radarparameters.currentText()) 129 | configParameters = readpoint.IWR6843AOP_TLV()._initialize(config_file = radarparameters.currentText()) 130 | rangeResolutionlabel.setText(str(configParameters["rangeResolutionMeters"])+'cm') 131 | dopplerResolutionlabel.setText(str(configParameters["dopplerResolutionMps"])+'m/s') 132 | maxRangelabel.setText(str(configParameters["maxRange"])+'m') 133 | maxVelocitylabel.setText(str(configParameters["maxVelocity"])+'m/s') 134 | 135 | def openradar(config,com): 136 | global radar_ctrl 137 | radar_ctrl = SerialConfig(name='ConnectRadar', CLIPort=com, BaudRate=115200) 138 | radar_ctrl.StopRadar() 139 | radar_ctrl.SendConfig(config) 140 | processor.start() 141 | processor.join(timeout=1) 142 | update_figure() 143 | 144 | def updatacomstatus(cbox): 145 | port_list = list(list_ports.comports()) 146 | cbox.clear() 147 | for i in range(len(port_list)): 148 | cbox.addItem(str(port_list[i][0])) 149 | 150 | def setserialport(cbox, com): 151 | global CLIport_name 152 | global Dataport_name 153 | if cbox.currentIndex() > -1: 154 | port = cbox.currentText() 155 | if com == "CLI": 156 | CLIport_name = port 157 | 158 | else: 159 | Dataport_name = port 160 | 161 | def sendconfigfunc(): 162 | global CLIport_name 163 | global Dataport_name 164 | if len(CLIport_name) != 0 and radarparameters.currentText() != '--select--': 165 | openradar(radarparameters.currentText(),CLIport_name) 166 | printlog(string = '发送成功', fontcolor='green') 167 | else: 168 | printlog(string = '发送失败', fontcolor='red') 169 | 170 | 171 | def setintervaltime(): 172 | gl.set_value('timer_2s', True) 173 | QtCore.QTimer.singleShot(2000, setintervaltime) 174 | 175 | # cnt 用来计数 200ms*cnt，代表显示多长时间 176 | cnt = 0 177 | def setdisplaygestureicontime(): 178 | global _flagdisplay, cnt 179 | if _flagdisplay==True: 180 | cnt = cnt + 1 181 | if cnt>4: 182 | cnt = 0 183 | view_gesture.setPixmap(QtGui.QPixmap("gesture_icons/"+str(7)+".jpg")) 184 | subWin.img_update("gesture_icons/"+str(7)+".jpg") 185 | _flagdisplay=False 186 | QtCore.QTimer.singleShot(200, setdisplaygestureicontime) 187 | 188 | def setcolor(): 189 | if(color_.currentText()!='--select--' and color_.currentText()!=''): 190 | if color_.currentText() == 'customize': 191 | pgColormap = pg_get_cmap(color_.currentText()) 192 | else: 193 | cmap=plt.cm.get_cmap(color_.currentText()) 194 | pgColormap = pg_get_cmap(cmap) 195 | lookup_table = pgColormap.getLookupTable(0.0, 1.0, 256) 196 | img_rdi.setLookupTable(lookup_table) 197 | img_rai.setLookupTable(lookup_table) 198 | img_rti.setLookupTable(lookup_table) 199 | img_dti.setLookupTable(lookup_table) 200 | img_rei.setLookupTable(lookup_table) 201 | 202 | def get_filelist(dir,Filelist): 203 | newDir=dir 204 | #注意看dir是文件名还是路径＋文件名！！！！！！！！！！！！！！ 205 | if os.path.isfile(dir): 206 | dir_ = os.path.basename(dir) 207 | if (dir_[:2] == 'DT') and (dir_[-4:] == '.npy'): 208 | Filelist[0].append(dir) 209 | elif (dir_[:2] == 'RT') and (dir_[-4:] == '.npy'): 210 | Filelist[1].append(dir) 211 | elif (dir_[:3] == 'RDT') and (dir_[-4:] == '.npy'): 212 | Filelist[2].append(dir) 213 | elif (dir_[:3] == 'ART') and (dir_[-4:] == '.npy'): 214 | Filelist[3].append(dir) 215 | elif (dir_[:3] == 'ERT') and (dir_[-4:] == '.npy'): 216 | Filelist[4].append(dir) 217 | elif os.path.isdir(dir): 218 | for s in os.listdir(dir): 219 | newDir=os.path.join(dir,s) 220 | get_filelist(newDir,Filelist) 221 | return Filelist 222 | 223 | def savedatasetsencefile(): 224 | global datasetsencefile,start_captureidx,idx 225 | datasetsencefile = datasetfile+'/'+ whodatafile.text()+'/'+datasetfilebox.currentText() 226 | if not os.path.exists(datasetsencefile): #判断是否存在文件夹如果不存在则创建为文件夹 227 | os.makedirs(datasetsencefile) 228 | 229 | featurelist = get_filelist(datasetsencefile, [[] for i in range(5)]) 230 | start_captureidx = len(featurelist[0]) 231 | idx = start_captureidx 232 | 233 | 234 | def show_sub(): 235 | subWin.show() 236 | MainWindow.hide() 237 | 238 | 239 | 240 | def application(): 241 | global color_,radarparameters,maxVelocitylabel,maxRangelabel,dopplerResolutionlabel,rangeResolutionlabel,logtxt 242 | global Recognizebtn,CaptureDatabtn,view_gesture,modelfile,datasetfilebox,whodatafile 243 | global img_rdi, img_rai, img_rti, img_rei, img_dti,ui 244 | global subWin,MainWindow 245 | app = QtWidgets.QApplication(sys.argv) 246 | MainWindow = QtWidgets.QMainWindow() 247 | MainWindow.show() 248 | ui = Ui_MainWindow() 249 | 250 | ui.setupUi(MainWindow) 251 | subWin = Qt_pet(MainWindow) 252 | 253 | # 改了D:\Applications\anaconda3\Lib\site-packages\pyqtgraph\graphicsItems\ViewBox 254 | # 里的ViewBox.py第919行padding = self.suggestPadding(ax)改成padding = 0 255 | view_rdi = ui.graphicsView_6.addViewBox() 256 | ui.graphicsView_6.setCentralWidget(view_rdi)#去边界 257 | view_rai = ui.graphicsView_4.addViewBox() 258 | ui.graphicsView_4.setCentralWidget(view_rai)#去边界 259 | view_rti = ui.graphicsView.addViewBox() 260 | ui.graphicsView.setCentralWidget(view_rti)#去边界 261 | view_dti = ui.graphicsView_2.addViewBox() 262 | ui.graphicsView_2.setCentralWidget(view_dti)#去边界 263 | view_rei = ui.graphicsView_3.addViewBox() 264 | ui.graphicsView_3.setCentralWidget(view_rei)#去边界 265 | 266 | view_gesture = ui.graphicsView_5 267 | view_gesture.setPixmap(QtGui.QPixmap("gesture_icons/7.jpg")) 268 | 269 | sendcfgbtn = ui.pushButton_11 270 | exitbtn = ui.pushButton_12 271 | Recognizebtn = ui.pushButton_15 272 | CaptureDatabtn = ui.pushButton 273 | 274 | color_ = ui.comboBox 275 | modelfile = ui.comboBox_2 276 | datasetfilebox = ui.comboBox_3 277 | radarparameters = ui.comboBox_7 278 | Cliportbox = ui.comboBox_8 279 | 280 | logtxt = ui.textEdit 281 | whodatafile = ui.lineEdit_6 282 | changepage = ui.actionload 283 | 284 | 285 | rangeResolutionlabel = ui.label_14 286 | dopplerResolutionlabel = ui.label_35 287 | maxRangelabel = ui.label_16 288 | maxVelocitylabel = ui.label_37 289 | 290 | # --------------------------------------------------- 291 | # lock the aspect ratio so pixels are always square 292 | # view_rai.setAspectLocked(True) 293 | # view_rti.setAspectLocked(True) 294 | img_rdi = pg.ImageItem(border=None) 295 | img_rai = pg.ImageItem(border=None) 296 | img_rti = pg.ImageItem(border=None) 297 | img_dti = pg.ImageItem(border=None) 298 | img_rei = pg.ImageItem(border=None) 299 | 300 | # Colormap 301 | pgColormap = pg_get_cmap('customize') 302 | lookup_table = pgColormap.getLookupTable(0.0, 1.0, 256) 303 | img_rdi.setLookupTable(lookup_table) 304 | img_rai.setLookupTable(lookup_table) 305 | img_rti.setLookupTable(lookup_table) 306 | img_dti.setLookupTable(lookup_table) 307 | img_rei.setLookupTable(lookup_table) 308 | 309 | view_rdi.addItem(img_rdi) 310 | view_rai.addItem(img_rai) 311 | view_rti.addItem(img_rti) 312 | view_dti.addItem(img_dti) 313 | view_rei.addItem(img_rei) 314 | 315 | 316 | Cliportbox.arrowClicked.connect(lambda:updatacomstatus(Cliportbox)) 317 | Cliportbox.currentIndexChanged.connect(lambda:setserialport(Cliportbox, com = 'CLI')) 318 | color_.currentIndexChanged.connect(setcolor) 319 | modelfile.currentIndexChanged.connect(loadmodel) 320 | radarparameters.currentIndexChanged.connect(getradarparameters) 321 | datasetfilebox.currentIndexChanged.connect(savedatasetsencefile) 322 | whodatafile.editingFinished.connect(savedatasetsencefile) 323 | sendcfgbtn.clicked.connect(sendconfigfunc) 324 | Recognizebtn.clicked.connect(setintervaltime) 325 | # Recognizebtn.clicked.connect(setdisplaygestureicontime) 326 | CaptureDatabtn.clicked.connect(setintervaltime) 327 | changepage.triggered.connect(show_sub) 328 | # 2022/2/24 添加小型化控件 不能正常退出了 329 | exitbtn.clicked.connect(app.instance().exit) 330 | 331 | app.instance().exec_() 332 | 333 | try: 334 | if radar_ctrl.CLIPort: 335 | if radar_ctrl.CLIPort.isOpen(): 336 | radar_ctrl.StopRadar() 337 | except: 338 | pass 339 | 340 | 341 | if __name__ == '__main__': 342 | # Queue for access data 343 | BinData = Queue() # 原始数据队列 344 | 345 | # 时间信息 346 | RTIData = Queue() # 距离时间队列 347 | DTIData = Queue() # 多普勒时间队列 348 | 349 | # 连续过程信息 350 | RDIData = Queue() # 距离多普勒队列 351 | RAIData = Queue() # 距离方位角队列 352 | REIData = Queue() # 方位角俯仰角队列 353 | 354 | # Radar config parameters 355 | NUM_TX = 3 356 | NUM_RX = 4 357 | NUM_CHIRPS = 64 358 | NUM_ADC_SAMPLES = 64 359 | 360 | radar_config = [NUM_ADC_SAMPLES, NUM_CHIRPS, NUM_TX, NUM_RX] 361 | frame_length = NUM_ADC_SAMPLES * NUM_CHIRPS * NUM_TX * NUM_RX * 2 362 | 363 | # config DCA1000 to receive bin data 364 | dca1000_cfg = DCA1000Config('DCA1000Config',config_address = ('192.168.33.30', 4096), 365 | FPGA_address_cfg=('192.168.33.180', 4096)) 366 | 367 | collector = UdpListener('Listener', BinData, frame_length) 368 | processor = DataProcessor('Processor', radar_config, BinData, RTIData, DTIData, 369 | RDIData, RAIData, REIData) 370 | collector.start() 371 | 372 | application() 373 | 374 | dca1000_cfg.DCA1000_close() 375 | 376 | collector.join(timeout=1) 377 | 378 | print("Program close") 379 | sys.exit() 380 | -------------------------------------------------------------------------------- /radar_config.py: -------------------------------------------------------------------------------- 1 | # collect data from TI DCA1000 EVM 2 | 3 | import serial 4 | import time 5 | import socket 6 | 7 | # Radar EVM setting 8 | class SerialConfig(): 9 | def __init__(self, name, CLIPort, BaudRate): 10 | self.name = name 11 | self.CLIPort = serial.Serial(CLIPort, baudrate=BaudRate) 12 | 13 | def close(self): 14 | self.CLIPort.close() 15 | 16 | def SendConfig(self, ConfigFileName): 17 | for line in open(ConfigFileName): 18 | self.CLIPort.write((line.rstrip('\r\n') + '\n').encode()) 19 | print(f"Sent: {line.strip()}") 20 | 21 | # 等待并读取返回数据（带超时） 22 | start_time = time.time() 23 | timeout = 0.1 # 2秒超时 24 | response = b'' 25 | 26 | while time.time() - start_time < timeout: 27 | if self.CLIPort.in_waiting > 0: 28 | response += self.CLIPort.read(self.CLIPort.in_waiting) 29 | time.sleep(0.01) # 避免CPU占用过高 30 | 31 | print(f"Received: {response.decode(errors='ignore').strip()}") 32 | time.sleep(0.01) 33 | 34 | def StartRadar(self): 35 | self.CLIPort.write('sensorStart\n'.encode()) 36 | print('sensorStart\n') 37 | 38 | def StopRadar(self): 39 | self.CLIPort.write('sensorStop\n'.encode()) 40 | print('sensorStop\n') 41 | 42 | def DisconnectRadar(self): 43 | self.CLIPort.write('sensorStop\n'.encode()) 44 | self.CLIPort.close() 45 | 46 | # DCA1000 47 | class DCA1000Config(): 48 | def __init__(self, name, config_address, FPGA_address_cfg): 49 | self.name = name 50 | self.config_address = config_address 51 | self.FPGA_address_cfg = FPGA_address_cfg 52 | cmd_order = ['9', 'E', '3', 'B', '5', '6'] 53 | self.sockConfig = socket.socket(socket.AF_INET, socket.SOCK_DGRAM) 54 | self.sockConfig.bind(config_address) 55 | for k in range(5): 56 | # Send the command 57 | self.sockConfig.sendto(self.send_cmd(cmd_order[k]), FPGA_address_cfg) 58 | time.sleep(0.1) 59 | # Request data back on the config port 60 | msg, server = self.sockConfig.recvfrom(2048) 61 | # print('receive command:', msg.hex()) 62 | 63 | def DCA1000_close(self): 64 | self.sockConfig.sendto(self.send_cmd('6'), self.FPGA_address_cfg) 65 | self.sockConfig.close() 66 | 67 | def send_cmd(self, code): 68 | # command code list 69 | CODE_1 = (0x01).to_bytes(2, byteorder='little', signed=False) 70 | CODE_2 = (0x02).to_bytes(2, byteorder='little', signed=False) 71 | CODE_3 = (0x03).to_bytes(2, byteorder='little', signed=False) 72 | CODE_4 = (0x04).to_bytes(2, byteorder='little', signed=False) 73 | CODE_5 = (0x05).to_bytes(2, byteorder='little', signed=False) 74 | CODE_6 = (0x06).to_bytes(2, byteorder='little', signed=False) 75 | CODE_7 = (0x07).to_bytes(2, byteorder='little', signed=False) 76 | CODE_8 = (0x08).to_bytes(2, byteorder='little', signed=False) 77 | CODE_9 = (0x09).to_bytes(2, byteorder='little', signed=False) 78 | CODE_A = (0x0A).to_bytes(2, byteorder='little', signed=False) 79 | CODE_B = (0x0B).to_bytes(2, byteorder='little', signed=False) 80 | CODE_C = (0x0C).to_bytes(2, byteorder='little', signed=False) 81 | CODE_D = (0x0D).to_bytes(2, byteorder='little', signed=False) 82 | CODE_E = (0x0E).to_bytes(2, byteorder='little', signed=False) 83 | 84 | # packet header & footer 85 | header = (0xA55A).to_bytes(2, byteorder='little', signed=False) 86 | footer = (0xEEAA).to_bytes(2, byteorder='little', signed=False) 87 | 88 | # data size 89 | dataSize_0 = (0x00).to_bytes(2, byteorder='little', signed=False) 90 | dataSize_6 = (0x06).to_bytes(2, byteorder='little', signed=False) 91 | 92 | # data 93 | data_FPGA_config = (0x01020102031e).to_bytes(6, byteorder='big', signed=False) 94 | data_packet_config = (0xc005350c0000).to_bytes(6, byteorder='big', signed=False) 95 | 96 | # connect to DCA1000 97 | connect_to_FPGA = header + CODE_9 + dataSize_0 + footer 98 | read_FPGA_version = header + CODE_E + dataSize_0 + footer 99 | config_FPGA = header + CODE_3 + dataSize_6 + data_FPGA_config + footer 100 | config_packet = header + CODE_B + dataSize_6 + data_packet_config + footer 101 | start_record = header + CODE_5 + dataSize_0 + footer 102 | stop_record = header + CODE_6 + dataSize_0 + footer 103 | 104 | if code == '9': 105 | re = connect_to_FPGA 106 | elif code == 'E': 107 | re = read_FPGA_version 108 | elif code == '3': 109 | re = config_FPGA 110 | elif code == 'B': 111 | re = config_packet 112 | elif code == '5': 113 | re = start_record 114 | elif code == '6': 115 | re = stop_record 116 | else: 117 | re = 'NULL' 118 | # print('send command:', re.hex()) 119 | return re -------------------------------------------------------------------------------- /real_time_process.py: -------------------------------------------------------------------------------- 1 | import threading as th 2 | import numpy as np 3 | import DSP 4 | from dsp.utils import Window 5 | from ctypes import * 6 | 7 | # 这一块处理需要手动修改代码很不友好，希望有人能pull request 8 | dll = cdll.LoadLibrary('realtimeSystem/libs/UDPCAPTUREADCRAWDATA.dll') 9 | # dll = cdll.LoadLibrary('realtimeSystem/dll/libtest.so') 10 | 11 | a = np.zeros(1).astype(np.int) 12 | # 内存大小至少是frame_length的两倍 ，双缓冲区 13 | # 98304 的计算方法是例如你的配置如下： 14 | 15 | # adc_sample = 64 16 | # chirp = 64 17 | # tx_num = 3 18 | # rx_num = 4 19 | # frame_length = adc_sample * chirp * tx_num * rx_num * 2 = 98304 20 | b = np.zeros(98304*2).astype(c_short) 21 | 22 | # 转换为ctypes，这里转换后的可以直接利用ctypes转换为c语言中的int*，然后在c中使用 23 | a_ctypes_ptr = cast(a.ctypes.data, POINTER(c_int)) 24 | # 转换为ctypes，这里转换后的可以直接利用ctypes转换为c语言中的int*，然后在c中使用 25 | b_ctypes_ptr = cast(b.ctypes.data, POINTER(c_short)) 26 | 27 | 28 | class UdpListener(th.Thread): 29 | def __init__(self, name, bin_data, data_frame_length): 30 | th.Thread.__init__(self, name=name) 31 | self.bin_data = bin_data 32 | self.frame_length = data_frame_length 33 | 34 | def run(self): 35 | global a_ctypes_ptr, b_ctypes_ptr 36 | dll.captureudp(a_ctypes_ptr, b_ctypes_ptr, self.frame_length) 37 | 38 | 39 | class DataProcessor(th.Thread): 40 | def __init__(self, name, config, bin_queue, rti_queue, dti_queue, rdi_queue, rai_queue, rei_queue): 41 | 42 | th.Thread.__init__(self, name=name) 43 | self.adc_sample = config[0] 44 | self.chirp_num = config[1] 45 | self.tx_num = config[2] 46 | self.rx_num = config[3] 47 | self.bin_queue = bin_queue 48 | self.rti_queue = rti_queue 49 | self.dti_queue = dti_queue 50 | self.rdi_queue = rdi_queue 51 | self.rai_queue = rai_queue 52 | self.rei_queue = rei_queue 53 | 54 | def run(self): 55 | global frame_count 56 | frame_count = 0 57 | lastflar = 0 58 | while True: 59 | # 对应dll中的双缓冲区，0区和1区 60 | if(lastflar != a_ctypes_ptr[0]): 61 | # print(a_ctypes_ptr[0]) 62 | lastflar = a_ctypes_ptr[0] 63 | # self.bin_data.put(np.array(b_ctypes_ptr[98304*(1-a_ctypes_ptr[0]):98304*(2-a_ctypes_ptr[0])])) 64 | # data = self.bin_queue.get() 65 | data = np.array( 66 | b_ctypes_ptr[98304*(1-a_ctypes_ptr[0]):98304*(2-a_ctypes_ptr[0])]) 67 | data = np.reshape(data, [-1, 4]) 68 | data = data[:, 0:2:] + 1j * data[:, 2::] 69 | # [num_chirps*tx_num, wuli_antennas, num_samples] 70 | data = np.reshape( 71 | data, [self.chirp_num * self.tx_num, -1, self.adc_sample]) 72 | # [num_chirps*tx_num, num_samples, wuli_antennas] 73 | data = data.transpose([0, 2, 1]) 74 | # 192 = 64*3 记得改 75 | # TX1的：[num_chirps, num_samples, wuli_antennas] 76 | ch1_data = data[0: self.adc_sample*3: 3, :, :] 77 | # TX2的：[num_chirps, num_samples, wuli_antennas] 78 | ch2_data = data[1: self.adc_sample*3: 3, :, :] 79 | # TX3的：[num_chirps, num_samples, wuli_antennas] 80 | ch3_data = data[2: self.adc_sample*3: 3, :, :] 81 | # channel的排序方式：(0:TX1-RX1,1:TX1-RX2,2:TX1-RX3,3:TX1-RX4,| 4:TX2-RX1,5:TX2-RX2,6:TX2-RX3,7:TX2-RX4,| 8:TX3-RX1,9:TX3-RX2,10:TX3-RX3,11:TX3-RX4) 82 | data = np.concatenate([ch1_data, ch2_data, ch3_data], axis=2) 83 | 84 | frame_count += 1 85 | 86 | rti, rdi, dti = DSP.RDA_Time( 87 | data, window_type_1d=Window.HANNING, axis=1) 88 | 89 | # _, rdi = DSP.Range_Doppler(data, mode=2, padding_size=[128, 64]) 90 | rai, rei = DSP.Range_Angle( 91 | data, padding_size=[128, 64, 64]) 92 | self.rti_queue.put(rti) 93 | self.dti_queue.put(dti) 94 | self.rdi_queue.put(rdi) 95 | self.rai_queue.put(rai) 96 | self.rei_queue.put(rei) 97 | --------------------------------------------------------------------------------