-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathblendingPoisson_alpha.py
300 lines (264 loc) · 9.46 KB
/
blendingPoisson_alpha.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
# -*- coding: utf-8 -*-
"""
Created on Tue Mar 27 15:10:40 2018
@author: lilei0129
"""
from __future__ import print_function
import numpy as np
import scipy
import math
from scipy.sparse import linalg
from scipy.misc import toimage
from scipy.misc import imshow
import cv2
from PIL import Image
import time
np.set_printoptions(threshold=np.inf)
# config & input
start = time.time()
#def addTransparency(img, factor = 0.7 ): #增加alpha通道
# img = img.convert('RGBA')
# img_blender = Image.new('RGBA', img.size, (0,0,0,0))
# img = Image.blend(img_blender, img, factor)
# return img
def generateData(back,fore,mask):
I = []
J = []
S = []
B = []
count = 0
for i in xrange(int(back.shape[0])):
for j in xrange(int(back.shape[1])):
if mask.item(i,j) < 0.1: #black pixel, then insert 1 at that index
I.extend([count])
J.extend([count])
S.extend([1])
B.extend([back[i,j]]) #set b = background pixel value
else: #white pixel, insert gradient of i,j
J.extend([count-1, count+1, count-fore.shape[1], count+fore.shape[1], count])
I.extend([count, count, count, count, count])
S.extend([1, 1, 1, 1, -4])
# B.extend( [fore[i-1,j] + fore[i+1,j] + fore[i,j-1] + fore[i,j+1] - 4.0*fore[i,j]] )
'''
With Gradient Mixing
tmpforeB = fore[i-1,j] + fore[i+1,j] + fore[i,j-1] + fore[i,j+1] - 4.0*fore[i,j]
tmpbackB = back[i-1,j] + back[i+1,j] + back[i,j-1] + back[i,j+1] - 4.0*back[i,j]
B.extend( [0.5*tmpforeB + 0.5*tmpbackB] )
'''
tmpforeB = fore[i-1,j] + fore[i+1,j] + fore[i,j-1] + fore[i,j+1] - 4.0*fore[i,j]
tmpbackB = back[i-1,j] + back[i+1,j] + back[i,j-1] + back[i,j+1] - 4.0*back[i,j]
B.extend( [1.3*tmpforeB + 0*tmpbackB] )
count+=1
I = np.asarray(I) #column for SPARSE MATRIX A
J = np.asarray(J) #row for SPARSE MATRIX A
S = np.asarray(S) #data for SPARSE MATRIX A
B = np.asarray(B)#B for Ax=b
return I,J,S,B
"""泊松处理"""
def poissonProcess(backImg,foreImg,mask):
rows = backImg.shape[0] #545p
cols = backImg.shape[1] #429p
# channels = backImg.shape[2] #3 for BGR
#alls = rows * cols * channels
alls = rows * cols #total number of pixels in image
#split BGR
backB, backG, backR = cv2.split(backImg)
foreB, foreG, foreR = cv2.split(foreImg)
numRowsInA = alls # pixels(row) * pixels(col)
Ib,Jb,Sb,Bb = generateData(backB,foreB,mask)
Ig, Jg, Sg, Bg = generateData(backG,foreG,mask)
Ir, Jr, Sr, Br = generateData(backR,foreR,mask)
Ab = scipy.sparse.coo_matrix((Sb, (Ib, Jb)), shape=(numRowsInA, alls))
Ag = scipy.sparse.coo_matrix((Sg, (Ig, Jg)), shape=(numRowsInA, alls))
Ar = scipy.sparse.coo_matrix((Sr, (Ir, Jr)), shape=(numRowsInA, alls))
Ab = Ab.tocsc() # Convert A matrix to Compressed Sparse Row format
Ag = Ag.tocsc()
Ar = Ar.tocsc()
Rb = scipy.sparse.linalg.spsolve(Ab,Bb)
Rb = np.reshape(Rb, (rows,cols))
Rg = scipy.sparse.linalg.spsolve(Ag,Bg)
Rg = np.reshape(Rg, (rows,cols))
Rr = scipy.sparse.linalg.spsolve(Ar,Br)
Rr = np.reshape(Rr, (rows,cols))
merged = cv2.merge((Rb,Rg,Rr))
return merged
def slow_color(back,fore,mask,backImg_color,foreImg_color,wc=0.6):
##输入权重系数
backImg_color = np.array(backImg_color)
foreImg_color = np.array(foreImg_color)
length = len(foreImg_color)
##更改只读模式
back.flags.writeable = True
fore.flags.writeable = True
##获取长宽
image_Row = (fore.shape)[0]
image_Column = (fore.shape)[1]
##生成权重矩阵
WC = np.zeros([image_Row,image_Column])
WC_2 = np.ones([image_Row,image_Column])
###第一次生成处理
for Row in range(1,image_Row):
for Column in range(1,image_Column):
if mask[Row,Column] > 0.5:
WC[Row,Column] = wc
###先膨胀处理
kernel = np.ones((5,5),np.uint8)
mask = cv2.dilate(mask,kernel,iterations = 1)
###第二次腐蚀膨胀生成处理
erosion = mask
for i in range(length):
WC_test = WC
for Row in range(1,image_Row-1):
for Column in range(1,image_Column-1):
if erosion[Row-1,Column]- erosion[Row,Column] <0 or erosion[Row,Column-1]- erosion[Row,Column] < 0 :
WC_test[Row,Column] = WC[Row,Column]*foreImg_color[i]
elif erosion[Row+1,Column]- erosion[Row,Column] <0 or erosion[Row,Column+1]- erosion[Row,Column] < 0 :
WC_test[Row,Column] = WC[Row,Column]*foreImg_color[i]
kernel = np.ones((5,5),np.uint8)
erosion = cv2.erode(erosion,kernel,iterations = 1)
WC = WC_test
WC_3 = WC_2[:,:]-WC[:,:]
# fore[:,:,3] = (fore[:,:,3]*(WC_3[:,:]) + back[:,:,3]*(WC[:,:]))
fore[:,:,0] = (fore[:,:,0]*(WC_3[:,:]) + back[:,:,0]*(WC[:,:]))
fore[:,:,1] = fore[:,:,1]*(WC_3[:,:]) + back[:,:,1]*WC[:,:]
fore[:,:,2] = fore[:,:,2]*(WC_3[:,:]) + back[:,:,2]*WC[:,:]
return fore,WC
#
#for i in range (1,200):
#
# backImageName = '1/' + 'beijing2.jpg'
# ###载入背景图
# foreImageName = 'lady_image/lady_image/' +str(i)+'.jpg'
# ###载入前景图
# maskName = 'lady_image/lady_image_BI_canny/' + 'BI_'+ str(i)+'.jpg'
# ###载入二值图
# outputName = 'lady_image/result4/' + str(i)+'.jpg'
#
#
#
# backImg = cv2.imread(backImageName,cv2.IMREAD_UNCHANGED) / 255.0
# foreImg = cv2.imread(foreImageName,cv2.IMREAD_UNCHANGED) / 255.0
# mask = cv2.imread(maskName,0) / 255.0
#
# rows = backImg.shape[0] #545p
# cols = backImg.shape[1] #429p
# channels = backImg.shape[2] #3 for BGR
#
# #alls = rows * cols * channels
# alls = rows * cols #total number of pixels in image
#
# #split BGR
# backB, backG, backR = cv2.split(backImg)
# foreB, foreG, foreR = cv2.split(foreImg)
#
# #print (backImg)
#
#
# ''' TEST NDARRAY ITEMSET
# B = np.zeros(4)
# B.itemset(1,2)
# print B
# '''
#
# ''' TESTING R WITH SMALL NDARRAY
# print ("***** Testing R with Small NDARRAY *****")
# TestA = np.array([[0,0,1,1],[1,0,2,0],[1,1,0,1],[1,1,0,0]])
# TestB = np.array([15,12,22,16])
#
# print ("matrix A at [2,1]",end ="")
# print (TestA[2,1])
#
# TestA = scipy.sparse.coo_matrix(TestA) #convert np.array A to coo_matrix
# TestA = TestA.tocsc()
#
# print ("matrix A shape: ",end="")
# print (TestA.shape)
# print (TestA)
# print ("matrix B shape: %s" % TestB.shape)
# print (TestB)
# R = scipy.sparse.linalg.spsolve(TestA,TestB)
# print ("solution X type: %s" % type(R))
# print (R)
# print ("solution X shape: ",end="")
# print (R.shape)
# print ("\n\n")
# '''
#
# """
# Construct matrix A & B
# """
# # print ("***** Generating Matrices Ab, Ag, Ar *****")
#
# numRowsInA = alls # pixels(row) * pixels(col)
#
# Ib,Jb,Sb,Bb = generateData(backB,foreB,mask)
#
# Ig, Jg, Sg, Bg = generateData(backG,foreG,mask)
# Ir, Jr, Sr, Br = generateData(backR,foreR,mask)
#
# Ab = scipy.sparse.coo_matrix((Sb, (Ib, Jb)), shape=(numRowsInA, alls))
# Ag = scipy.sparse.coo_matrix((Sg, (Ig, Jg)), shape=(numRowsInA, alls))
# Ar = scipy.sparse.coo_matrix((Sr, (Ir, Jr)), shape=(numRowsInA, alls))
# Ab = Ab.tocsc() # Convert A matrix to Compressed Sparse Row format
# Ag = Ag.tocsc()
# Ar = Ar.tocsc()
#
# """
# extract final result from R
# Solve Ax = b for each of B,G,R
# """
#
# #print ("***** Solving X for AX = B *****")
# #R = scipy.sparse.linalg.cg(Ab, Bb)
# Rb = scipy.sparse.linalg.spsolve(Ab,Bb)
# Rb = np.reshape(Rb, (rows,cols))
# Rg = scipy.sparse.linalg.spsolve(Ag,Bg)
# Rg = np.reshape(Rg, (rows,cols))
# Rr = scipy.sparse.linalg.spsolve(Ar,Br)
# Rr = np.reshape(Rr, (rows,cols))
# merged = cv2.merge((Rb,Rg,Rr))
#
# """
# 利用原图增强真实感
#
# """
# '''
# #图片增加alpha通道
#
## backImg2 = cv2.imread(backImageName,cv2.IMREAD_UNCHANGED)
## backImg2 = cv2.cvtColor(backImg2, cv2.COLOR_RGB2BGRA)
## backImg2 = cv2.cvtColor(backImg2, cv2.COLOR_BGRA2RGBA)
#
## foreImg = np.copy(foreImg)
## foreImg = foreImg.astype(np.uint8)
## foreImg = cv2.cvtColor(foreImg, cv2.COLOR_RGB2RGBA)
##
## merged = np.copy(merged)
## merged = merged.astype(np.uint8)
## merged = cv2.cvtColor(merged, cv2.COLOR_RGB2RGBA)
# '''
#
#
## foreImg_color =np.array(range(5,10,2))*0.1
# foreImg_color = [0.6,0.7,0.8,0.9]
# backImg_color = foreImg_color[::-1]
# print_merged,WC= slow_color(foreImg,merged,mask,backImg_color,foreImg_color)
#
#
#
# #cv2.imshow("merged",merged)
#
## file=open('1/data2.txt','w')
## file.write(str(WC ));
## file.close()
# cv2.waitKey(0)
# cv2.destroyAllWindows()
# cv2.imwrite(outputName, print_merged*255)
## cv2.imwrite(outputName, print_merged*255)
# print (time.time() - start)
# """
# uncomment these lines after you generate the final result in matrix 'img'
# cv2.imshow('output', R);
# cv2.waitKey(0)
# cv2.imwrite(outputName, R * 255);
# """