-
Notifications
You must be signed in to change notification settings - Fork 8
/
model_fuse.py
389 lines (320 loc) · 12.1 KB
/
model_fuse.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
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
import numpy as np
import glob
import matplotlib.pyplot as plt
import cv2 as cv
import time
import os
def fill_and_delete(label):
gray_label = label[:,:,0].copy()
# print(np.unique(gray_label))
res1=cv.findContours(gray_label,mode=cv.RETR_EXTERNAL,method=cv.CHAIN_APPROX_NONE)
contours,idx1=res1
# print(len(contours))
for i in range(len(contours)):
area=cv.contourArea(contours[i])
cv.fillPoly(gray_label,[contours[i]],(255,255,255))
'''
填补空洞
'''
if area<=1000:
# print('正在填充面积小于100的区域')
cv.drawContours(gray_label,contours,i,0,cv.FILLED)
continue
res1=cv.findContours(gray_label,mode=cv.RETR_EXTERNAL,method=cv.CHAIN_APPROX_NONE)
contours1,idx1=res1
# print(len(contours1))
plt.imshow(gray_label)
cv.imwrite('gray.png',gray_label)
return gray_label,contours1
def dilate_process(h, w, contours, kernel, iter_time):
result = []
for j in range(len(contours)):
cur_img = np.zeros((h, w), dtype=np.uint8)
# print(len(contours))
cv.drawContours(cur_img, contours, j, 255, cv.FILLED)
dilate1 = cv.dilate(cur_img, kernel, iterations=iter_time)
res = cv.findContours(dilate1, mode=cv.RETR_EXTERNAL, method=cv.CHAIN_APPROX_NONE)
if len(res) == 2:
contours1, _ = res
else:
_, contours1, _ = res
result.append(contours1[0])
# print('膨胀阶段')
return result
def fill_small_target(img, contours):
fill_flag = False
for i in range(len(contours)):
area = cv.contourArea(contours[i])
cv.fillPoly(img, [contours[i]], (255, 255, 255))
if area <= 500:
fill_flag = True
# print('正在填充腐蚀之后面积小于1000的区域')
cv.drawContours(img, contours, i, 0, cv.FILLED)
continue
return img, fill_flag
def erode_process(img, kernel_size, iteration):
erode = img.copy()
kernel = np.ones((1, kernel_size), np.uint8)
erosion1 = cv.erode(erode, kernel, iterations=iteration)
contours1, _ = cv.findContours(erosion1, mode=cv.RETR_EXTERNAL, method=cv.CHAIN_APPROX_NONE)
if len(contours1) == 1:
# print('此处没有发生水平方向的重叠')
return None
else:
# print('此物体发生了水平方向的重叠')
erosion1, flag = fill_small_target(erosion1, contours1)
if not flag:
# print('没有可以填充的小物体存在')
h, w = img.shape
cnt = dilate_process(h, w, contours1, kernel, iteration)
return cnt
else:
contours1, _ = cv.findContours(erosion1, mode=cv.RETR_EXTERNAL, method=cv.CHAIN_APPROX_NONE)
if len(contours1) == 0:
return False
h, w = img.shape
cnt = dilate_process(h, w, contours1, kernel, iteration)
return cnt
def erode_process1(img, kernel_size, iteration):
erode = img.copy()
kernel = np.ones((kernel_size, 1), np.uint8)
erosion1 = cv.erode(erode, kernel, iterations=iteration)
contours1, _ = cv.findContours(erosion1, mode=cv.RETR_EXTERNAL, method=cv.CHAIN_APPROX_NONE)
if len(contours1) == 1:
# print('此处没有发生竖直方向的重叠')
return None
else:
# print('此物体发生了竖直方向的重叠')
erosion1, flag = fill_small_target(erosion1, contours1)
if not flag:
# print('没有可以填充的小物体存在')
h, w = img.shape
cnt = dilate_process(h, w, contours1, kernel, iteration)
'''
返回了多个目标的cnt
'''
return cnt
else:
contours1, _ = cv.findContours(erosion1, mode=cv.RETR_EXTERNAL, method=cv.CHAIN_APPROX_NONE)
h, w = img.shape
if len(contours1) == 0:
return False
cnt = dilate_process(h, w, contours1, kernel, iteration)
return cnt
def iou(bbox1, bbox2):
'''
bbox1---->[4,]
bbox2---->[N,4]
'''
wh = np.minimum(bbox1[2:], bbox2[:, 2:]) - np.maximum(bbox1[:2], bbox2[:, :2])
wh = np.maximum(wh, 0)
bbox1_area = (bbox1[2] - bbox1[0]) * (bbox1[3] - bbox1[1])
bbox2_area = (bbox2[:, 2] - bbox2[:, 0]) * (bbox2[:, 3] - bbox2[:, 1])
union_area = wh[:, 0] * wh[:, 1]
res_iou = union_area / (bbox1_area + bbox2_area - union_area)
return res_iou
def compute_iou(cnt1, cnt2):
all_bbox1 = []
for i in range(len(cnt1)):
x1, y1, w1, h1 = cv.boundingRect(cnt1[i])
x2 = x1 + w1
y2 = y1 + h1
all_bbox1.append([x1, y1, x2, y2])
all_bbox1 = np.array(all_bbox1)
all_bbox2 = []
for i in range(len(cnt2)):
x1, y1, w1, h1 = cv.boundingRect(cnt2[i])
x2 = x1 + w1
y2 = y1 + h1
all_bbox2.append([x1, y1, x2, y2])
all_bbox2 = np.array(all_bbox2)
valid_cnt1_index = []
valid_cnt2_index = []
all_mask = [False] * len(all_bbox2)
for i in range(len(all_bbox1)):
res_iou = iou(all_bbox1[i], all_bbox2)
iou_mask = res_iou > 0.7
all_mask += iou_mask
if not np.any(iou_mask):
valid_cnt1_index.append(i)
for i in range(len(all_mask)):
if all_mask[i]:
continue
valid_cnt2_index.append(i)
return valid_cnt1_index, valid_cnt2_index
def eroede_dilate_process(gray_label, contours1):
h, w = gray_label.shape
all_cnt = []
for i in range(len(contours1)):
plot_img = np.zeros((h, w), dtype=np.uint8)
cv.drawContours(plot_img, contours1, i, 255, cv.FILLED)
# plt.imshow(plot_img)
cur_cnt = erode_process(plot_img, 5, 5)
cur_cnt1 = erode_process1(plot_img, 5, 5)
if cur_cnt == False or cur_cnt1 == False:
continue
elif cur_cnt is None and cur_cnt1 is None:
all_cnt.append(contours1[i])
continue
# 没有新目标的产生,默认使用原始的cnt
elif cur_cnt is not None and cur_cnt1 is not None:
# dely_compute_iou
for k in cur_cnt:
all_cnt.append(k)
for k in cur_cnt1:
all_cnt.append(k)
'''
valid_1,valid_2 = compute_iou(cur_cnt,cur_cnt1)
if len(valid_1)>=1:
for m in valid_1:
all_cnt.append(cur_cnt[m])
if len(valid_2)>=1:
for n in valid_2:
all_cnt.append(cur_cnt1[n])
'''
continue
'''此处函数待验证是否有BUG以及是否实现所需功能'''
elif cur_cnt is not None:
for j in range(len(cur_cnt)):
all_cnt.append(cur_cnt[j])
continue
else:
for j in range(len(cur_cnt1)):
all_cnt.append(cur_cnt1[j])
# plt.show()
# plt.cla()
return all_cnt
def small_target(input_img,edge,epsilon):
approx = cv.approxPolyDP(edge,epsilon,True)
points = approx.reshape((-1, 2))
count=0
rate=0.002
while len(points)!=4:
epsilon = rate * cv.arcLength(edge, True)
rate=rate+0.002
approx = cv.approxPolyDP(edge,epsilon,True)
points = approx.reshape((-1, 2))
count+=1
if count>10:
break
if len(points)==4:
print("小目标的优化结果为4边形")
else:
print("小目标的优化方法为外接最小矩形")
rect = cv.minAreaRect(edge)
# 得到最小外接矩形的(中心(x,y), (宽,高), 旋转角度)
points = cv.boxPoints(rect)
return points
def big_building(img,edge,epsilon):
epsilon = 0.005 * cv.arcLength(edge, True)
approx = cv.approxPolyDP(edge,epsilon, True)
# points = approx.reshape((-1, 2))
return approx
def big_building1(img,edge,epsilon):
epsilon = 0.004 * cv.arcLength(edge, True)
approx = cv.approxPolyDP(edge,epsilon, True)
# points = approx.reshape((-1, 2))
return approx
def big_building2(img,edge,epsilon):
epsilon = 0.002 * cv.arcLength(edge, True)
approx = cv.approxPolyDP(edge,epsilon, True)
# points = approx.reshape((-1, 2))
return approx
def only_plt(input_img,all_cnt):
for i in range(len(all_cnt)):
cv.drawContours(input_img,all_cnt,i,255,cv.FILLED)
return input_img
def model_confuse(path, name=''):
'''
save_path = 'all_result' + '/' + dir_name
print(save_path)
os.makedirs(save_path, exist_ok=True)
'''
all_path = glob.glob(path + '/' + '*.png')
print(all_path)
if len(all_path) != 5:
print('no five images')
return
label = cv.imread(all_path[0])
gray_label, cnt = fill_and_delete(label)
all_cnt = eroede_dilate_process(gray_label, cnt)
label = np.zeros((label.shape))
l1 = only_plt(label.copy(), all_cnt)
label = cv.imread(all_path[1])
gray_label, cnt = fill_and_delete(label)
all_cnt = eroede_dilate_process(gray_label, cnt)
label = np.zeros((label.shape))
l2 = only_plt(label.copy(), all_cnt)
label = cv.imread(all_path[2])
gray_label, cnt = fill_and_delete(label)
all_cnt = eroede_dilate_process(gray_label, cnt)
label = np.zeros((label.shape))
l3 = only_plt(label.copy(), all_cnt)
label = cv.imread(all_path[3])
gray_label, cnt = fill_and_delete(label)
all_cnt = eroede_dilate_process(gray_label, cnt)
label = np.zeros((label.shape))
l4 = only_plt(label.copy(), all_cnt)
label = cv.imread(all_path[4])
gray_label, cnt = fill_and_delete(label)
all_cnt = eroede_dilate_process(gray_label, cnt)
label = np.zeros((label.shape))
l5 = only_plt(label.copy(), all_cnt)
final_label = l1 // 255 + l2 // 255 + l3 // 255 + l4 // 255 + l5 // 255
# cv.imwrite(save_path + '/' + 'sliding_bam1.png', l1)
# cv.imwrite(save_path + '/' + 'sliding_deep1.png', l2)
# cv.imwrite(save_path + '/' + 'sliding_scse1.png', l3)
# cv.imwrite(save_path + '/' + 'sliding_res341.png', l4)
# cv.imwrite(save_path + '/' + 'sliding_hrnet1.png', l5)
label = np.where(final_label >= 3, 255, 0)
label = np.array(label, np.uint8)
# print(label.dtype)
def only_plt1(input_img, all_cnt):
all_color = [[255, 0, 0], [0, 255, 0], [0, 0, 255], [255, 255, 0], [255, 0, 255], [0, 255, 255],
[255, 255, 255]]
for i in range(len(all_cnt)):
colors = all_color[i % 7]
cv.drawContours(image=input_img,
contours=[all_cnt[i]],
contourIdx=-1,
color=colors,
thickness=3)
return input_img
gray_label, cnt = fill_and_delete(label)
all_cnt = eroede_dilate_process(gray_label, cnt)
# plot_res = only_plt1(images, all_cnt)
label = np.zeros(gray_label.shape,np.uint8)
for i in range(len(all_cnt)):
cv.drawContours(label, all_cnt, i, 255, cv.FILLED)
# h, w, c = plot_res.shape
cv.imwrite(path + r'\{}_result.png'.format(name),label)
# cv.imwrite(save_path + '/' + 'sliding_plot_images.png', l1)
'''
top = 0
down = 512
left = 0
right = 512
while down <= h:
# left = 0
# right = 512
while right <= w:
cv.line(plot_res, (left, 0), (left, h), (0, 255, 0), 2, 8)
cv.line(plot_res, (right, 0), (right, h), (255, 0, 0), 2, 8)
left = right - int(512 * 0.5)
right = left + 512
cv.line(plot_res, (0, top), (w, top), (255, 255, 255), 2, 8)
cv.line(plot_res, (0, down), (w, down), (0, 0, 0), 2, 8)
top = down - int(512 * 0.5)
down = top + 512
print(top, down)
cv.imwrite(save_path + '/' + 'sliding_plot_images.png', plot_res)
'''
if __name__ == '__main__':
name = os.listdir('D:/res_image')
# 名称
for i in range(len(name)):
path = 'D:/res_image' +'/'+ name[i]
# 绝对路劲
# child path
model_confuse(path,name[i])
print('图片{}预测结束'.format(name[i]))