├── test.py ├── LICENSE ├── README.md └── bj_delta.py /test.py: -------------------------------------------------------------------------------- 1 | from bj_delta import * 2 | 3 | 4 | print("Test 1") 5 | Rate1 = np.array([686.76, 309.58, 157.11, 85.95]) 6 | PSNR1 = np.array([40.28, 37.18, 34.24, 31.42]) 7 | Rate2 = np.array([893.34, 407.8, 204.93, 112.75]) 8 | PSNR2 = np.array([40.39, 37.21, 34.17, 31.24]) 9 | 10 | print("BD-PSNR: ", bj_delta(Rate1, PSNR1, Rate2, PSNR2, mode=0)) 11 | print("BD-RATE: ", bj_delta(Rate1, PSNR1, Rate2, PSNR2, mode=1)) 12 | 13 | print("Test 2") 14 | Rate1 = np.array([0.001, 0.005, 0.020, 0.100, 0.7500]) 15 | PSNR1 = np.array([18.02, 20.57, 23.09, 27.71, 35.74]) 16 | Rate2 = np.array([0.00103, 0.00507, 0.02098, 0.09638, 0.73051]) 17 | PSNR2 = np.array([24.81, 29.08, 33.09, 37.27, 43.12]) 18 | 19 | print("BD-PSNR: ", bj_delta(Rate1, PSNR1, Rate2, PSNR2, mode=0)) 20 | print("BD-RATE: ", bj_delta(Rate1, PSNR1, Rate2, PSNR2, mode=1)) 21 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) 2020 João Ascenso 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 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Bjontegaard metric computation 2 | 3 | ## Introduction 4 | Bjontegaard's metric computes the average gain in PSNR or the average saving in bitrate (in %) 5 | between two rate-distortion curves [1][2]. 6 | 7 | Similar to other packages [3] this function allows to compute the metric for more than 4 RD points, 8 | but in this case for the python language. 9 | 10 | ## Usage 11 | Check the test.py for some simple tests. 12 | ``` 13 | from bj_delta import * 14 | bd_psnr = bj_delta(Rate1, PSNR1, Rate2, PSNR2, mode=0)) 15 | bd_rate = bj_delta(Rate1, PSNR1, Rate2, PSNR2, mode=1)) 16 | ``` 17 | * Rate1,PSNR1: RD points for curve 1. 18 | * Rate2,PSNR2: RD points for curve 2. 19 | * Mode: 0 for average PSNR difference and 1 for average bitrate savings (%). 20 | * Returns the calculated Bjontegaard metric (BD-Rate or BD-PSNR). 21 | 22 | ## References 23 | [1] G. Bjontegaard, "Calculation of average PSNR differences between RD-curves", VCEG-M33, Austin, TX, USA, April 2001.
24 | [2] S. Pateux, J. Jung, "An excel add-in for computing Bjontegaard metric and its evolution", VCEG-AE07, Marrakech, MA, January 2007.
25 | [3] G. Valenzise, https://www.mathworks.com/matlabcentral/fileexchange/27798-bjontegaard-metric.
26 | -------------------------------------------------------------------------------- /bj_delta.py: -------------------------------------------------------------------------------- 1 | import numpy as np 2 | # BD-Rate and BD-PNSR computation 3 | # (c) Joao Ascenso (joao.ascenso@lx.it.pt) 4 | 5 | 6 | def bj_delta(R1, PSNR1, R2, PSNR2, mode=0): 7 | lR1 = np.log(R1) 8 | lR2 = np.log(R2) 9 | 10 | # find integral 11 | if mode == 0: 12 | # least squares polynomial fit 13 | p1 = np.polyfit(lR1, PSNR1, 3) 14 | p2 = np.polyfit(lR2, PSNR2, 3) 15 | 16 | # integration interval 17 | min_int = max(min(lR1), min(lR2)) 18 | max_int = min(max(lR1), max(lR2)) 19 | 20 | # indefinite integral of both polynomial curves 21 | p_int1 = np.polyint(p1) 22 | p_int2 = np.polyint(p2) 23 | 24 | # evaluates both poly curves at the limits of the integration interval 25 | # to find the area 26 | int1 = np.polyval(p_int1, max_int) - np.polyval(p_int1, min_int) 27 | int2 = np.polyval(p_int2, max_int) - np.polyval(p_int2, min_int) 28 | 29 | # find avg diff between the areas to obtain the final measure 30 | avg_diff = (int2-int1)/(max_int-min_int) 31 | else: 32 | # rate method: sames as previous one but with inverse order 33 | p1 = np.polyfit(PSNR1, lR1, 3) 34 | p2 = np.polyfit(PSNR2, lR2, 3) 35 | 36 | # integration interval 37 | min_int = max(min(PSNR1), min(PSNR2)) 38 | max_int = min(max(PSNR1), max(PSNR2)) 39 | 40 | # indefinite interval of both polynomial curves 41 | p_int1 = np.polyint(p1) 42 | p_int2 = np.polyint(p2) 43 | 44 | # evaluates both poly curves at the limits of the integration interval 45 | # to find the area 46 | int1 = np.polyval(p_int1, max_int) - np.polyval(p_int1, min_int) 47 | int2 = np.polyval(p_int2, max_int) - np.polyval(p_int2, min_int) 48 | 49 | # find avg diff between the areas to obtain the final measure 50 | avg_exp_diff = (int2-int1)/(max_int-min_int) 51 | avg_diff = (np.exp(avg_exp_diff)-1)*100 52 | return avg_diff 53 | --------------------------------------------------------------------------------