forked from chenriwei/FaceTools
-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathaligment.py
177 lines (166 loc) · 6.39 KB
/
aligment.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
# -*- coding: utf-8 -*-
"""
Created on Fri May 29 20:01:59 2015
@author: Chenriwei
@brief: 根据检测到的点，对其人脸图像
"""
import cv2
import os
import numpy as np
import matplotlib.pylab as plt
import skimage
from skimage import transform as tf
from skimage import io
def compute_affine_transform(refpoints, points, w = None):
'''
计算仿射变换矩阵
'''
if w == None:#每个关键点的权重
w = [1] * (len(points) * 2)
assert(len(w) == 2*len(points))
y = []
for n, p in enumerate(refpoints):
y += [p[0]/w[n*2], p[1]/w[n*2+1]]
A = []
for n, p in enumerate(points):
A.extend([ [p[0]/w[n*2], p[1]/w[n*2], 0, 0, 1/w[n*2], 0], [0, 0, p[0]/w[n*2+1], p[1]/w[n*2+1], 0, 1/w[n*2+1]] ])
lstsq = cv2.solve(np.array(A), np.array(y), flags=cv2.DECOMP_SVD)
h11, h12, h21, h22, dx, dy = lstsq[1]
#err = 0#lstsq[1]
#R = np.array([[h11, h12, dx], [h21, h22, dy]])
# The row above works too - but creates a redundant dimension
R = np.array([[h11[0], h12[0], dx[0]], [h21[0], h22[0], dy[0]]])
return R#, err
def alignment(filename,points,ref_points):
'''
@brief: 根据检测到的点，对其人脸图像
'''
assert(len(points)==len(ref_points))
num_point=len(ref_points)/2
#参考图像的点
dst=np.empty((num_point,2),dtype=np.int)
k=0
for i in range(num_point):
for j in range(2):
dst[i][j]=ref_points[k]
k=k+1
#待对齐图像的点
src=np.empty((num_point,2),dtype=np.int)
k=0
for i in range(num_point):
for j in range(2):
src[i][j]=points[k]
k=k+1
#根据检测到的点，求其相应的仿射变换矩阵
tfrom=tf.estimate_transform('affine',dst,src)
#用opencv的试试,其只能采用三个点,计算矩阵M
# pts1 = np.float32([[src[0][0],src[0][1]],[src[1][0],src[1][1]],[src[2][0],src[2][1]]])
# pts2 = np.float32([[dst[0][0],dst[0][1]],[dst[1][0],dst[1][1]],[dst[2][0],dst[2][1]]])
# M = cv2.getAffineTransform(pts2,pts1)
#用最小二乘法的方法进行处理
pts3 = np.float32([[src[0][0],src[0][1]],[src[1][0],src[1][1]],[src[2][0],src[2][1]],[src[3][0],src[3][1]],[src[4][0],src[4][1]]])
pts4 = np.float32([[dst[0][0],dst[0][1]],[dst[1][0],dst[1][1]],[dst[2][0],dst[2][1]],[dst[3][0],dst[3][1]],[dst[4][0],dst[4][1]]])
N = compute_affine_transform(pts4,pts3)
#
im=skimage.io.imread(filename)
if im.ndim==3:
rows,cols,ch = im.shape
else:
rows,cols = im.shape
warpimage_cv2 = cv2.warpAffine(im,N,(cols,rows))
warpimage=tf.warp(im,inverse_map=tfrom)
return warpimage,warpimage_cv2
def align_all(filelist,imageRootPath,savePath):
'''
@breif:对其所有的人脸图像，需要选择参考的图像，默认为第一张
'''
fid=open(filelist)
for i in range(1):
line=fid.readline()
word=line.split()
print 'reference image:'+imageRootPath+word[0]
fid.close()
#随机选择一个参考图像和参考点
#todo：人工选择一张比较合适的图像作为参考图像，默认情况下，第一张作为参考图像
ref_points=np.empty((10,1),dtype=np.int)
points=np.empty((10,1),dtype=np.int)
filename=word[0]
for i in range(10):
ref_points[i]=int(word[i+1])
refimage=skimage.io.imread(imageRootPath+word[0])
if refimage.ndim==3:
rows,cols,ch = refimage.shape
else:
rows,cols = refimage.shape
#rows,cols,ch = refimage.shape
#为保留数据的完整性，重新扫描
fid=open(filelist,'r')
lines=fid.readlines()
for line in lines:
word=line.split()
filename=word[0]
for j in range(10):
points[j]=int(word[j+1])
#originimage=skimage.io.imread(imageRootPath+filename)
warpimage,warpimage_cv2=alignment(imageRootPath+filename,points,ref_points)
savename=savePath+filename
#处理多层文件，不能写入的问题，新建文件
dirname, basename = os.path.split(savename)
if not os.path.exists(dirname):
os.makedirs(dirname)
if warpimage_cv2.shape== refimage.shape:
skimage.io.imsave(savename,warpimage_cv2)
else:
if warpimage_cv2.ndim==3:
rows,cols,ch = warpimage_cv2.shape
skimage.io.imsave(savename,warpimage_cv2[0:rows,0:cols,:])
else:
rows,cols = warpimage_cv2.shape
skimage.io.imsave(savename,warpimage_cv2[0:rows,0:cols])
fid.close()
if __name__=='__main__':
imageRootPath=r'F:\Dataset\MORPH'
savePath=r'F:\MyDataset\MORPH\face_aligned_5_points'
fid=open('imageListWithPoints.list','r')
for i in range(2):
line=fid.readline()
word=line.split()
print 'reference image:',word
fid.close()
#随机选择一个参考图像和参考点
#todo：人工选择一张比较合适的图像作为参考图像，默认情况下，第一张作为参考图像
ref_points=np.empty((10,1),dtype=np.int)
points=np.empty((10,1),dtype=np.int)
filename=word[0]
for i in range(10):
ref_points[i]=int(word[i+1])
#while fid.readline()==
refimage=skimage.io.imread(imageRootPath+word[0])
plt.subplot(1,3,1)
plt.imshow(refimage)
plt.axis('off')
rows,cols,ch = refimage.shape
fid=open('imageListWithPoints.list','r')
lines=fid.readlines()
for line in lines:
#for i in range(1000):
#word=fid.readline().split()
word=line.split()
filename=word[0]
for j in range(10):
points[j]=int(word[j+1])
originimage=skimage.io.imread(imageRootPath+filename)
warpimage,warpimage_cv2=alignment(imageRootPath+filename,points,ref_points)
# plt.subplot(1,3,2)
# plt.imshow(warpimage)
# plt.axis('off')
# plt.subplot(1,3,3)
# plt.imshow(warpimage_cv2)
# plt.axis('off')
#warpimage=tf.resize(warpimage,(200,200))
#todo: 处理大小不同图像的输出
if warpimage_cv2.shape== refimage.shape:
skimage.io.imsave(savePath+filename,warpimage_cv2)
else:
skimage.io.imsave(savePath+filename,warpimage_cv2[0:rows,0:cols,:])
fid.close()