-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathStrength.py
398 lines (377 loc) · 16.6 KB
/
Strength.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
390
391
392
393
394
395
396
397
398
# -*- coding = utf-8 -*-
# @Time : 2021-11-01 10:57
# @Author: gla1ve
# @File: Strength.py.py
# @Software: PyCharm
# 人生苦短， 我用python(划掉) Java
import cv2
import numpy as np
import matplotlib.pyplot as plt
import random
from matplotlib.ticker import MultipleLocator
from heapq import nsmallest
from BMP import BMP # 前一个BMP是文件名，后一个BMP是类名
'''
upd: 2021-11-13
我认为我很多函数写的很乱，有的是直接修改self.pic，而有的是修改了tmp, 只是用cv2显示出来了而已
'''
class Strength:
def __init__(self, pic_path):
self.pic_path = pic_path
self.pic = cv2.imread(self.pic_path, cv2.IMREAD_COLOR).astype(np.uint8)
def ShowPic(self):
cv2.imshow('PKQ', self.pic)
cv2.waitKey(0) # 0一直显示，直到有键盘输入。也可以是其他数字，毫秒级
# 分三个通道绘制R,G,B的灰度直方图 这里bmp=1只是一个占位符, 如果flag是True,bmp会传入一个BMP类的对象
def ShowHist(self, bmp=1, flag=False): # 里面有些plt的用法，值得积累qwq
if flag is False:
bmp = BMP(self.pic_path)
blue, green, red = bmp.ToArray()
B = np.bincount(blue)
G = np.bincount(green)
R = np.bincount(red)
scale = MultipleLocator(50)
Bins = 150
# ax为两条坐标轴的实例
plt.rcParams['figure.figsize'] = (8, 10)
plt.subplots_adjust(wspace=0.2, hspace=0.5) # 设置子图的间隔
# 1 / 8
plt.subplot(421), plt.hist(blue, bins=Bins, density=True)
plt.title('Blue'), plt.ylabel("ratio")
# 2 / 8
plt.subplot(422), plt.plot(np.arange(B.shape[0]), B, 'b')
plt.xlim(0, 256), plt.title('Blue')
ax = plt.gca()
ax.xaxis.set_major_locator(scale)
# 3 / 8
plt.subplot(423), plt.hist(green, bins=Bins, density=True)
plt.title('Green'), plt.ylabel("ratio")
# 4 / 8
plt.subplot(424), plt.plot(np.arange(G.shape[0]), G, 'g')
plt.xlim(0, 256), plt.title('Green')
ax = plt.gca()
ax.xaxis.set_major_locator(scale)
# 5 / 8
plt.subplot(425), plt.hist(red, bins=Bins, density=True)
plt.title('Red'), plt.ylabel("ratio")
# 6 / 8
plt.subplot(426), plt.plot(np.arange(R.shape[0]), R, 'r')
plt.xlim(0, 256), plt.title('Red')
ax = plt.gca()
ax.xaxis.set_major_locator(scale)
# 7 / 8
mix = (blue / 3 + green / 3 + red / 3).astype(np.uint8)
M = np.bincount(mix)
plt.subplot(427), plt.hist(mix, bins=Bins, density=True)
plt.title('Avg'), plt.ylabel("ratio")
# 8 / 8
plt.subplot(428), plt.plot(np.arange(M.shape[0]), M, 'b')
plt.xlim(0, 256), plt.title('Avg')
ax = plt.gca()
ax.xaxis.set_major_locator(scale)
plt.show()
# 直方图均衡化 (仅限真彩图, 三个通道分开均衡化)
def Histogram_Equalization2(self):
bmp = BMP(self.pic_path)
# self.ShowHist(bmp=bmp, flag=True)
assert (bmp.IsRealColor is True)
blue, green, red = bmp.ToArray()
B = np.bincount(blue)
G = np.bincount(green)
R = np.bincount(red)
b_arr, g_arr, r_arr = np.zeros(256), np.zeros(256), np.zeros(256)
# 归一化后求前缀和
for i in range(B.shape[0]):
b_arr[i] = B[i]
for i in range(G.shape[0]):
g_arr[i] = G[i]
for i in range(R.shape[0]):
r_arr[i] = R[i]
for i in range(1, 256):
b_arr[i] += b_arr[i - 1]
g_arr[i] += g_arr[i - 1]
r_arr[i] += r_arr[i - 1]
b_arr = b_arr / bmp.biHeight / bmp.biWidth
b_arr = (b_arr * 255 + 0.5).astype(np.uint8)
g_arr = g_arr / bmp.biHeight / bmp.biWidth
g_arr = (g_arr * 255 + 0.5).astype(np.uint8)
r_arr = r_arr / bmp.biHeight / bmp.biWidth
r_arr = (r_arr * 255 + 0.5).astype(np.uint8)
for i in range(bmp.biHeight):
for j in range(bmp.biWidth):
bmp.data[i][j][0] = b_arr[int(bmp.data[i][j][0] + 0.5)]
bmp.data[i][j][1] = g_arr[int(bmp.data[i][j][1] + 0.5)]
bmp.data[i][j][2] = r_arr[int(bmp.data[i][j][2] + 0.5)]
# self.ShowHist(bmp=bmp, flag=True)
cv2.imwrite('img/Strength.bmp', bmp.data[:, :].astype(np.uint8))
cv2.imshow('After Histogram_Equalization', bmp.data[:, :].astype(np.uint8))
cv2.waitKey(0)
# 直方图均衡化 (仅限真彩图, 三个通道一起均衡化) 这个不太准，应该用下面那个分开的
def Histogram_Equalization(self):
bmp = BMP(self.pic_path)
# self.ShowHist(bmp=bmp, flag=True)
assert (bmp.IsRealColor is True)
blue, green, red = bmp.ToArray()
mix = (blue / 3 + green / 3 + red / 3 + 0.5).astype(np.uint8)
tmp = np.bincount(mix)
M = np.zeros(256)
# 归一化后求前缀和
for i in range(tmp.shape[0]):
M[i] = tmp[i]
for i in range(1, 256):
M[i] += M[i - 1]
M = M / bmp.biHeight / bmp.biWidth
M = (M * 255 + 0.5).astype(np.uint8)
for i in range(bmp.biHeight):
for j in range(bmp.biWidth):
for k in range(3):
bmp.data[i][j][k] = M[int(bmp.data[i][j][k] + 0.5)]
# self.ShowHist(bmp=bmp, flag=True)
# bmp.SaveBMP()
# bmp.ShowPic(BMP.default_save_path)
# plt.imshow(bmp.data / 255)
# plt.show()
# print(bmp.data[:5])
"""
上面的方法都不用了 终于摸索出来了。。 plt.imshow的时候：
要么强制转换成uint8, 要么就用浮点数进行/255归一化；
而cv2.imshow传数组的时候，会把RGB和在一起，所以bmp处理完之后，只能取前2维
"""
cv2.imwrite('img/Strength.bmp', bmp.data[:, :].astype(np.uint8))
cv2.imshow('After Histogram_Equalization', bmp.data[:, :].astype(np.uint8))
cv2.waitKey(0)
# 显示二值化后的图片 (这里只能用幻数了, 建议先Hist看一下应该设置成几qwq)
'''
python一切皆引用。 这里Img会直接修改self.pic， 所以使用二值化函数之后，原来的对象会变成黑白的
'''
def Binarization(self):
Img = self.pic
for i in range(Img.shape[0]):
for j in range(Img.shape[1]):
v = Img[i][j]
if any(v <= 100): # 幻数
v = 0
else:
v = 255
Img[i][j] = v
cv2.imwrite('img/Strength.bmp', Img)
cv2.imshow('Binarization', Img)
cv2.waitKey(0)
# 显示灰度级切片后的图片， 对原对象进行修改
def Grey_Cutter(self):
Img = self.pic
for i in range(Img.shape[0]):
for j in range(Img.shape[1]):
v = Img[i][j]
if any(v >= 120): # 幻数
v = 255
else:
v = 0
Img[i][j] = v
cv2.imwrite('img/Strength.bmp', Img)
cv2.imshow('Binarization', Img)
cv2.waitKey(0)
# 显示灰度化后的图片
def Grey_Pic(self):
Img = cv2.imread(self.pic_path, 0).astype(np.uint8)
cv2.imwrite('img/Strength.bmp', Img)
cv2.imshow('Grey', Img)
cv2.waitKey(0)
# 为图片添加随机噪声
'''
注意，我这里对图片添加椒盐、随机、高斯噪声，都不是对原对象进行修改，而是利用cv2将图片保存到某个地方。
'''
def Generate_Random_Noise(self, prob=0.02):
output = np.zeros(self.pic.shape, np.uint8)
for i in range(self.pic.shape[0]):
for j in range(self.pic.shape[1]):
rdn = random.random()
if rdn < prob:
output[i][j] = self.pic[i][j] + random.random() * 255
else:
output[i][j] = self.pic[i][j]
cv2.imwrite('img/Noised.bmp', output)
cv2.imshow('Salt_And_Pepper_Noise', output)
cv2.waitKey(0)
# 为图片添加椒盐噪声
def Generate_Salt_And_Pepper_Noise(self, prob=0.02):
output = np.zeros(self.pic.shape, np.uint8)
upb = 1 - prob
for i in range(self.pic.shape[0]):
for j in range(self.pic.shape[1]):
rdn = random.random()
if rdn < prob:
output[i][j] = 0
elif rdn > upb:
output[i][j] = 255
else:
output[i][j] = self.pic[i][j]
# output = np.asarray(output).astype(np.uint8)
cv2.imwrite('img/Noised.bmp', output)
cv2.imshow('Salt_And_Pepper_Noise', output)
cv2.waitKey(0)
# 为图片添加高斯噪声
'''
np.clip()用法：
整个数组的值限制在指定值a_min与a_max之间，对比a_min小的和比a_max大的值就重置为a_min,和a_max。
对于每个位置添加gauss噪声， 然后显示出来。
'''
def Generate_Gauss_Noise(self, mean=0, var=0.001):
image = np.array(self.pic / 255, dtype=float)
noise = np.random.normal(mean, var ** 0.5, image.shape)
out = image + noise
if out.min() < 0:
low_clip = -1.
else:
low_clip = 0.
out = np.clip(out, low_clip, 1.0)
out = np.uint8(out * 255)
cv2.imwrite('img/Noised.bmp', out)
cv2.imshow('Gauss_Noise', out)
cv2.waitKey(0)
# 高斯滤波[高斯模糊](调库实现的 cv2)
def Gaussian_Blur(self, n=3):
out = cv2.GaussianBlur(self.pic, (n, n), 0)
cv2.imwrite('img/Strength.bmp', out)
cv2.imshow('Gauss_Noise', out)
cv2.waitKey(0)
# 均值滤波
'''
注意这里的tmp是[0, 0, 0], 所以取均值的时候直接/8即可, 而不是np.average[这样会求出一个数作为均值，然后广播给img。这会
导致RGB三个通道的值是一样的，就变成灰图了]
注意在使用np.average median的时候, 要结合axis来使用。 否则就是对所有数字求了
'''
def Average_Filtering(self):
img = np.zeros(self.pic.shape, np.uint8)
dir = [[-1, -1], [-1, 0], [-1, 1], [0, -1], [0, 1], [1, -1], [1, 0], [1, 1]]
# 等价于: dir = [[x, y] for x in range(-1, 2) for y in range(-1, 2)]
for i in range(self.pic.shape[0]):
for j in range(self.pic.shape[1]):
if 0 < i < self.pic.shape[0] - 1 and 0 < j < self.pic.shape[1] - 1:
tmp = np.zeros_like(self.pic.shape)
for k in range(8):
tmp += self.pic[i + dir[k][0]][j + dir[k][1]]
img[i][j] = (tmp / 8).astype(np.uint8)
else:
img[i][j] = self.pic[i][j].astype(np.uint8)
cv2.imwrite('img/Strength.bmp', img)
cv2.imshow('Average_Filtering', img)
cv2.waitKey(0)
# 中值滤波
def Median_Filtering(self):
img = np.zeros(self.pic.shape, np.uint8)
dir = [[-1, -1], [-1, 0], [-1, 1], [0, -1], [0, 1], [1, -1], [1, 0], [1, 1]]
for i in range(self.pic.shape[0]):
for j in range(self.pic.shape[1]):
if 0 < i < self.pic.shape[0] - 1 and 0 < j < self.pic.shape[1] - 1:
tmp = []
for k in range(8):
tmp.append(self.pic[i + dir[k][0]][j + dir[k][1]])
img[i][j] = np.median(tmp, axis=0).astype(np.uint8)
else:
img[i][j] = self.pic[i][j].astype(np.uint8)
cv2.imwrite('img/Strength.bmp', img)
cv2.imshow('Median_Filtering', img)
cv2.waitKey(0)
# SNN, 对称近邻均值滤波, 取一个（2n+1)*(2n+1)的区域, 这里默认取的是5*5的。算一下，大约是512*512*5*5的.非常慢
'''
这里有一个问题： 对于真彩图像RGB三个通道来说，是分开看哪个离得近呢，还是合起来看哪个离得近？
e.g. 目标点[50, 50, 50], d1 = [200, 50, 50], d2 = [50, 200, 200], 是分开变成[50, 50, 50]，还是直接选d1?
'''
def SNN_Filtering(self, n=2):
N = n * 2 + 1
img = np.zeros(self.pic.shape, np.uint8)
for i in range(self.pic.shape[0]):
for j in range(self.pic.shape[1]):
if n <= i < self.pic.shape[0] - n and n <= j < self.pic.shape[1] - n:
tmp = np.zeros_like(self.pic.shape)
# 接下来对每一个点处理
for x in range(i - n, i + 1):
for y in range(j - n, j + n + 1):
if x == i and y == j:
break
d1, d2 = self.pic[x][y], self.pic[2 * i - x][2 * j - y]
for zz in range(3):
# print(d1[zz], d2[zz], self.pic[i][j][zz])
if np.abs(int(d1[zz]) - self.pic[i][j][zz]) < np.abs(int(d2[zz]) - self.pic[i][j][zz]):
tmp[zz] += d1[zz]
else:
tmp[zz] += d2[zz]
# exit(0)
# if np.abs(np.sum(d1 - self.pic[i][j])) < np.abs(np.sum(d2 - self.pic[i][j])):
# tmp += d1
# else:
# tmp += d2
img[i][j] = (tmp / (N * N // 2)).astype(np.uint8)
else:
img[i][j] = self.pic[i][j].astype(np.uint8)
cv2.imwrite('img/Strength.bmp', img)
cv2.imshow('SNN_Filtering', img)
cv2.waitKey(0)
# K-均值滤波 这里使用的是N = 5, K = 7
def KNN_Filtering(self):
dir = [[x, y] for x in range(-2, 3) for y in range(-2, 3)]
img = np.zeros(self.pic.shape, np.uint8)
for i in range(self.pic.shape[0]):
for j in range(self.pic.shape[1]):
if 2 <= i < self.pic.shape[0] - 2 and 2 <= j < self.pic.shape[1] - 2:
R, G, B = [], [], []
for k in range(25):
B.append(self.pic[i + dir[k][0]][j + dir[k][1]][0])
G.append(self.pic[i + dir[k][0]][j + dir[k][1]][1])
R.append(self.pic[i + dir[k][0]][j + dir[k][1]][2])
b = nsmallest(7, B, key=lambda x: abs(int(x) - self.pic[i][j][0]))
g = nsmallest(7, G, key=lambda x: abs(int(x) - self.pic[i][j][1]))
r = nsmallest(7, R, key=lambda x: abs(int(x) - self.pic[i][j][2]))
img[i][j] = np.asarray([np.average(b), np.average(g), np.average(r)]).astype(np.uint8)
else:
img[i][j] = self.pic[i][j]
cv2.imwrite('img/Strength.bmp', img)
cv2.imshow('KNN_Filtering', img)
cv2.waitKey(0)
# 二值图像的黑白点噪声滤波
def Black_And_White_Dot_Noise_Filtering(self):
dir = [[-1, -1], [-1, 0], [-1, 1], [0, -1], [0, 1], [1, -1], [1, 0], [1, 1]]
for i in range(self.pic.shape[0]):
for j in range(self.pic.shape[1]):
if 0 < i < self.pic.shape[0] - 1 and 0 < j < self.pic.shape[1] - 1:
tmp = np.zeros_like(self.pic.shape)
for k in range(8):
tmp += self.pic[i + dir[k][0]][j + dir[k][1]]
if any(np.abs(self.pic[i][j] - np.average(tmp / 8)) >= 127.5):
self.pic[i][j] = 255 - self.pic[i][j]
cv2.imwrite('img/Strength.bmp', self.pic)
cv2.imshow('Black_And_White_Dot_Noise_Filtering', self.pic)
cv2.waitKey(0)
if __name__ == '__main__':
# from heapq import nsmallest
# s = [[1, 2, 3, 4, 5, 6, 7], [2, 4, 5, 6, 8, 86]]
# print(nsmallest(3, s, key=lambda x: abs(x - 50)))
# exit(0)
save_path = r'E:\大三上学期课内\数字图像\图像处理图片\Lena\tmp\save.bmp'
path = 'img/Snow_blur.bmp'
b = Strength(path)
b.Grey_Pic()
# b.Histogram_Equalization2()
# b.Binarization()
# b.Grey_Cutter()
# b.Gaussian_Blur()
# b.Generate_Salt_And_Pepper_Noise()
# b.KNN_Filtering()
# b.Generate_Random_Noise(prob=0.5)
# b.Grey_Cutter()
# b.Generate_Gauss_Noise(var=0.01)
# b.SNN_Filtering(n=2)
# b.Black_And_White_Dot_Noise_Filtering()
# b.Median_Filtering()
# b.Average_Filtering()
# b.Grey_Cutter()
# b.ShowHist()
# b.Grey_Cutter()
# hist_cv = cv2.calcHist([img], [0], None, [256], [0, 256])
# print(hist_cv)
# plt.subplot(121), plt.imshow(img, 'gray')
# plt.subplot(122), plt.hist(hist_cv, density=True)
# plt.show()
# a = Strength(path)
# a.ShowPic()