├── shadow1.jpg
├── shadow2.jpg
├── shadow3.jpg
├── Shadow Detection and Removal Based on YCbCr Color Space.pdf
├── .idea
    └── vcs.xml
└── shadow-removal-based-on-YCbCr-color-space.py


/shadow1.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/mmostipan/shadow-removal/HEAD/shadow1.jpg


--------------------------------------------------------------------------------
/shadow2.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/mmostipan/shadow-removal/HEAD/shadow2.jpg


--------------------------------------------------------------------------------
/shadow3.jpg:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/mmostipan/shadow-removal/HEAD/shadow3.jpg


--------------------------------------------------------------------------------
/Shadow Detection and Removal Based on YCbCr Color Space.pdf:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/mmostipan/shadow-removal/HEAD/Shadow Detection and Removal Based on YCbCr Color Space.pdf


--------------------------------------------------------------------------------
/.idea/vcs.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0" encoding="UTF-8"?>
2 | <project version="4">
3 |   <component name="VcsDirectoryMappings">
4 |     <mapping directory="$PROJECT_DIR$" vcs="Git" />
5 |   </component>
6 | </project>


--------------------------------------------------------------------------------
/shadow-removal-based-on-YCbCr-color-space.py:
--------------------------------------------------------------------------------
 1 | import numpy as np
 2 | import cv2
 3 | 
 4 | # read an image with shadow...
 5 | # and it converts to BGR color space automatically
 6 | or_img = cv2.imread('shadow4.jpg')
 7 | 
 8 | # covert the BGR image to an YCbCr image
 9 | y_cb_cr_img = cv2.cvtColor(or_img, cv2.COLOR_BGR2YCrCb)
10 | 
11 | # copy the image to create a binary mask later
12 | binary_mask = np.copy(y_cb_cr_img)
13 | 
14 | # get mean value of the pixels in Y plane
15 | y_mean = np.mean(cv2.split(y_cb_cr_img)[0])
16 | 
17 | # get standard deviation of channel in Y plane
18 | y_std = np.std(cv2.split(y_cb_cr_img)[0])
19 | 
20 | # classify pixels as shadow and non-shadow pixels
21 | for i in range(y_cb_cr_img.shape[0]):
22 |     for j in range(y_cb_cr_img.shape[1]):
23 | 
24 |         if y_cb_cr_img[i, j, 0] < y_mean - (y_std / 3):
25 |             # paint it white (shadow)
26 |             binary_mask[i, j] = [255, 255, 255]
27 |         else:
28 |             # paint it black (non-shadow)
29 |             binary_mask[i, j] = [0, 0, 0]
30 | 
31 | # Using morphological operation
32 | # The misclassified pixels are
33 | # removed using dilation followed by erosion.
34 | kernel = np.ones((3, 3), np.uint8)
35 | erosion = cv2.erode(binary_mask, kernel, iterations=1)
36 | 
37 | # sum of pixel intensities in the lit areas
38 | spi_la = 0
39 | 
40 | # sum of pixel intensities in the shadow
41 | spi_s = 0
42 | 
43 | # number of pixels in the lit areas
44 | n_la = 0
45 | 
46 | # number of pixels in the shadow
47 | n_s = 0
48 | 
49 | # get sum of pixel intensities in the lit areas
50 | # and sum of pixel intensities in the shadow
51 | for i in range(y_cb_cr_img.shape[0]):
52 |     for j in range(y_cb_cr_img.shape[1]):
53 |         if erosion[i, j, 0] == 0 and erosion[i, j, 1] == 0 and erosion[i, j, 2] == 0:
54 |             spi_la = spi_la + y_cb_cr_img[i, j, 0]
55 |             n_la += 1
56 |         else:
57 |             spi_s = spi_s + y_cb_cr_img[i, j, 0]
58 |             n_s += 1
59 | 
60 | # get the average pixel intensities in the lit areas
61 | average_ld = spi_la / n_la
62 | 
63 | # get the average pixel intensities in the shadow
64 | average_le = spi_s / n_s
65 | 
66 | # difference of the pixel intensities in the shadow and lit areas
67 | i_diff = average_ld - average_le
68 | 
69 | # get the ratio between average shadow pixels and average lit pixels
70 | ratio_as_al = average_ld / average_le
71 | 
72 | # added these difference
73 | for i in range(y_cb_cr_img.shape[0]):
74 |     for j in range(y_cb_cr_img.shape[1]):
75 |         if erosion[i, j, 0] == 255 and erosion[i, j, 1] == 255 and erosion[i, j, 2] == 255:
76 | 
77 |             y_cb_cr_img[i, j] = [y_cb_cr_img[i, j, 0] + i_diff, y_cb_cr_img[i, j, 1] + ratio_as_al,
78 |                                  y_cb_cr_img[i, j, 2] + ratio_as_al]
79 | 
80 | # covert the YCbCr image to the BGR image
81 | final_image = cv2.cvtColor(y_cb_cr_img, cv2.COLOR_YCR_CB2BGR)
82 | 
83 | cv2.imshow("im1", or_img)
84 | cv2.imshow("im2", final_image)
85 | cv2.waitKey(0)
86 | cv2.destroyAllWindows()
87 | 


--------------------------------------------------------------------------------