-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathnote_color_threshold.py
305 lines (264 loc) · 11.6 KB
/
note_color_threshold.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
'''
Author: MasterYip [email protected]
Date: 2022-11-05 14:11:07
LastEditors: MasterYip [email protected]
LastEditTime: 2023-06-09 21:18:05
FilePath: \comprehensive-coding\FastImgNotingDown\note_color_threshold.py
Description:
Copyright (c) 2022 by MasterYip [email protected], All Rights Reserved.
'''
#!/usr/bin/env python3
# coding=utf-8
import cv2
import numpy as np
from cv2 import (ADAPTIVE_THRESH_MEAN_C, COLOR_GRAY2RGB, COLOR_HSV2RGB,
COLOR_RGB2GRAY, COLOR_RGB2HSV, IMWRITE_PNG_COMPRESSION,
THRESH_BINARY, adaptiveThreshold, bitwise_not,
connectedComponents, connectedComponentsWithStats, cvtColor,
dilate, erode, imread, imwrite)
# from matplotlib.pyplot import imshow, subplot
from numpy import mean, ones, power, uint8
def imgInversionCheck(img, Thresh_dia=801):
'''
description: Check if the image is inverted.
param {*} self
param {*} img
param {*} Thresh_dia
return {*} True if the image is inverted.
'''
grey = cvtColor(img, COLOR_RGB2GRAY)
thresh = adaptiveThreshold(
grey, 255, ADAPTIVE_THRESH_MEAN_C, THRESH_BINARY,
Thresh_dia, 0)
mean_normal = mean(thresh) # Usually black when input is white
thresh = adaptiveThreshold(
bitwise_not(grey), 255, ADAPTIVE_THRESH_MEAN_C, THRESH_BINARY,
Thresh_dia, 0)
mean_inverse = mean(thresh)
return mean_inverse > mean_normal
def note_color_threshold_for_white_bkg(img, sat_remap_order=0.5, dilate_rad=0,
Thresh_bias=50, Thresh_dia=351, mode=0):
# pre processing
grey = cvtColor(img, COLOR_RGB2GRAY)
thresh = adaptiveThreshold(
grey, 255, ADAPTIVE_THRESH_MEAN_C, THRESH_BINARY,
Thresh_dia, Thresh_bias)
# (considering backgroud color, for cases like blacboard notes)
if mean(thresh) > 127:
thresh = bitwise_not(thresh)
# Dilate handwriting
if dilate_rad > 0:
kernel = ones((dilate_rad, dilate_rad), uint8)
thresh = dilate(thresh, kernel, iterations=1)
elif dilate_rad < 0:
kernel = ones((-dilate_rad, -dilate_rad), uint8)
thresh = erode(thresh, kernel, iterations=1)
if mode == 0: # Black & White
return bitwise_not(thresh)
elif mode == 1: # Recognize Color
# Filter out handwritings
num_objects, labels = connectedComponents(thresh, connectivity=8)
# Increase Saturation of Origin Pic
saturated_HSV = cvtColor(img, COLOR_RGB2HSV)
saturated_HSV[:, :, 1] = power(
saturated_HSV[:, :, 1]/255, sat_remap_order)*255
img_sat_remap = cvtColor(saturated_HSV, COLOR_HSV2RGB)
# filter out background
img_sat_remap[labels == 0] = [255, 255, 255]
# Combination (turn pixels which has low saturation to black)
thresh_RGB = cvtColor(thresh, COLOR_GRAY2RGB)
thresh_RGB = bitwise_not(thresh_RGB)
thresh_RGB[saturated_HSV[:, :, 1] >
100] = img_sat_remap[saturated_HSV[:, :, 1] > 100]
return thresh_RGB
def note_color_threshold_for_black_bkg(img, sat_remap_order=0.5, dilate_rad=0,
Thresh_bias=50, Thresh_dia=351, mode=0):
# pre processing
img = bitwise_not(img)
grey = cvtColor(img, COLOR_RGB2GRAY)
thresh = adaptiveThreshold(
grey, 255, ADAPTIVE_THRESH_MEAN_C, THRESH_BINARY,
Thresh_dia, Thresh_bias)
# (considering backgroud color, for cases like blacboard notes)
if mean(thresh) > 127:
thresh = bitwise_not(thresh)
# Dilate handwriting
if dilate_rad > 0:
kernel = ones((dilate_rad, dilate_rad), uint8)
thresh = dilate(thresh, kernel, iterations=1)
elif dilate_rad < 0:
kernel = ones((-dilate_rad, -dilate_rad), uint8)
thresh = erode(thresh, kernel, iterations=1)
if mode == 0: # Black & White
return bitwise_not(thresh)
elif mode == 1: # Recognize Color
# Filter out handwritings
num_objects, labels = connectedComponents(thresh, connectivity=8)
# Increase Saturation of Origin Pic
saturated_HSV = cvtColor(img, COLOR_RGB2HSV)
saturated_HSV[:, :, 1] = power(
saturated_HSV[:, :, 1]/255, sat_remap_order)*255
img_sat_remap = cvtColor(saturated_HSV, COLOR_HSV2RGB)
# filter out background
img_sat_remap[labels == 0] = [255, 255, 255]
# Combination (turn pixels which has low saturation to black)
thresh_RGB = cvtColor(thresh, COLOR_GRAY2RGB)
thresh_RGB = bitwise_not(thresh_RGB)
thresh_RGB[saturated_HSV[:, :, 1] >
100] = img_sat_remap[saturated_HSV[:, :, 1] > 100]
return thresh_RGB
def note_color_threshold(img, sat_remap_order=0.5,
dilate_rad=0, invert_threshold=127):
# pre processing
# (considering backgroud color, for cases like blacboard notes)
if mean(img) > invert_threshold: # 似乎需要更换更好的算法处理黑白 因为黑板亮度不好说（用直方图）
img = bitwise_not(img)
grey = cvtColor(img, COLOR_RGB2GRAY)
thresh = adaptiveThreshold(
grey, 255, ADAPTIVE_THRESH_MEAN_C, THRESH_BINARY, 351, 50)
# Dilate handwriting
kernel = ones((dilate_rad, dilate_rad), uint8)
thresh = dilate(thresh, kernel, iterations=1)
# Filter out handwritings
num_objects, labels = connectedComponents(thresh, connectivity=8)
# Increase Saturation of Origin Pic
saturated_HSV = cvtColor(img, COLOR_RGB2HSV)
saturated_HSV[:, :, 1] = power(
saturated_HSV[:, :, 1]/255, sat_remap_order)*255
img_sat_remap = cvtColor(saturated_HSV, COLOR_HSV2RGB)
# filter out background
img_sat_remap[labels == 0] = [255, 255, 255]
# Combination (turn pixels which has low saturation to black)
img_saturation = saturated_HSV[:, :, 1]
thresh_RGB = cvtColor(thresh, COLOR_GRAY2RGB)
thresh_RGB = bitwise_not(thresh_RGB)
thresh_RGB[img_saturation > 100] = img_sat_remap[img_saturation > 100]
return thresh_RGB
class BlackboardRecorder(object):
graph_display = False
def __init__(self, src_img):
'''
description: Preprocess src_img, including threshhold, Erode.\n
self.thresh - for White Bkg img\n
self.thresh_ivt - for Black Bkg img\n
return None
'''
self.img = src_img
self.erode_rad_rel = 0.005
self.dilate_rad_rel = 0.005
self.thresh_rad_rel = 1
thresh_bias = 0
# Thresh
thresh_rad = int(self.thresh_rad_rel *
min(self.img.shape[:1])/2)*2+1 # must be odd
self.grey = cvtColor(self.img, COLOR_RGB2GRAY)
# for White Bkg
self.thresh = adaptiveThreshold(self.grey, 255,
ADAPTIVE_THRESH_MEAN_C, THRESH_BINARY,
thresh_rad, thresh_bias)
# for Black Bkg
self.thresh_ivt = bitwise_not(self.thresh)
# Erode
erode_rad = int(self.erode_rad_rel*min(self.img.shape[:1]))+1
erode_kernel = ones((erode_rad, erode_rad), uint8) # 待改为圆形
self.thresh = erode(self.thresh, erode_kernel, iterations=1)
self.thresh_ivt = erode(self.thresh_ivt, erode_kernel, iterations=1)
dilate_rad = int(self.dilate_rad_rel*min(self.img.shape[:1]))+1
dilate_kernel = ones((dilate_rad, dilate_rad), uint8) # 待改为圆形
self.thresh = dilate(self.thresh, dilate_kernel, iterations=1)
self.thresh_ivt = dilate(self.thresh_ivt, dilate_kernel, iterations=1)
def find_possible_ROIs(self, thresh):
'''
description: find_possible_ROIs, temporarily support ONE target\n
params:\n
thresh\n
return (src_points, dst_points, (dst_img_w, dst_img_h))
'''
num_labels, labels, stats, centroids = connectedComponentsWithStats(
thresh, connectivity=4, ltype=None)
if self.graph_display:
subplot(1, 4, 1)
imshow(thresh)
# ===================有待改进
def find_blackboard(num_labels, labels, stats, centroids):
img_area = labels.shape[0]*labels.shape[1]
id_list = []
area_coef = 0.3
img_area_coef = 0.1
for i in range(1, stats.__len__()):
if (stats[i][4] > area_coef*stats[i][2]*stats[i][3] and
stats[i][2]*stats[i][3] > img_area_coef*img_area and
stats[i][2]*stats[i][3] < img_area):
id_list.append(i)
print(i, stats[i])
# 面积最大
# TODO:Need to be fixed
return id_list[stats[id_list, 4].argmax()]
id = find_blackboard(num_labels, labels, stats, centroids)
ROI = np.zeros(labels.shape, np.uint8)
ROI[labels == id] = 255
if self.graph_display:
subplot(1, 4, 2)
imshow(ROI)
def get_polygon(labels, id):
blank = np.zeros(labels.shape, dtype=np.uint8)
blank[labels == id] = 255
kernel = np.ones((5, 5), np.uint8)
blank = cv2.dilate(blank, kernel, iterations=1)
contours, hierarchy = cv2.findContours(
blank, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
c0 = contours[0]
# this para is important(decide the approximation,
# i.e. the polygon edge numbers)
coef = 0.005 # 0.05 default (Simpler when larger)
epsilon = coef * cv2.arcLength(c0, True)
approx = cv2.approxPolyDP(c0, epsilon, True)
return approx
approx = get_polygon(labels, id)
draw_img = self.img.copy()
res = cv2.drawContours(
draw_img, [approx], -1, (255, 0, 0), int(draw_img.shape[0]/20))
if self.graph_display:
subplot(1, 4, 3)
imshow(res)
# print(approx[:,0])
def find_angular_point(points):
ang_pt = points[:4].copy()
for pt in points:
if pt.sum() < ang_pt[0].sum():
ang_pt[0] = pt.copy()
if pt.sum() > ang_pt[2].sum():
ang_pt[2] = pt.copy()
if pt[0]-pt[1] < ang_pt[1][0]-ang_pt[1][1]:
ang_pt[1] = pt.copy()
if pt[0]-pt[1] > ang_pt[3][0]-ang_pt[3][1]:
ang_pt[3] = pt.copy()
return ang_pt
def wrap_size_est(ang_pt):
img_h = int((np.linalg.norm(ang_pt[0]-ang_pt[1]) +
np.linalg.norm(ang_pt[2]-ang_pt[3]))/2)
img_w = int((np.linalg.norm(ang_pt[1]-ang_pt[2]) +
np.linalg.norm(ang_pt[3]-ang_pt[1]))/2)
return (img_w, img_h)
src_points = find_angular_point(approx[:, 0])
print(src_points)
# target_img_shape = (img_w, img_h)
img_w, img_h = wrap_size_est(src_points)
print(img_w, img_h)
dst_points = np.array([[0, 0], [0, img_h], [img_w, img_h], [img_w, 0]])
return (src_points, dst_points, (img_w, img_h))
def homography_projection(self, src_img, src_pt, dst_pt, dst_shape):
H, _ = cv2.findHomography(src_pt, dst_pt)
img_src_warp = cv2.warpPerspective(src_img, H, dst_shape)
if self.graph_display:
subplot(1, 4, 4)
imshow(img_src_warp)
return img_src_warp
def note_color_thershold_test():
name = 'test.jpg'
img = imread(name)
thresh_color = note_color_threshold(img, 0.6, 2)
imwrite('thresh_color.png', thresh_color, [IMWRITE_PNG_COMPRESSION, 9])
if __name__ == '__main__':
bb = BlackboardRecorder()
print(bb['erode_rad_rel'])