-
Notifications
You must be signed in to change notification settings - Fork 0
/
ocrReg.py
executable file
·291 lines (250 loc) · 12.3 KB
/
ocrReg.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
# -*- coding: utf-8 -*-
"""
Created on Thu Jun 14 20:39:52 2018
@author: Zhou WenZhang
"""
# -*- coding: utf-8 -*
import os
import json
import ocr
import cv2
import mahotas
import numpy as np
import tensorflow as tf
tf.app.flags.DEFINE_string('test_data_path', 'test_data', '')
tf.app.flags.DEFINE_string('reg_data_path', 'reg_data', '')
FLAGS = tf.app.flags.FLAGS
#获取外接矩形框
def bbox(points):
res = np.zeros((2,2),dtype=np.int32)
res[0,:] = np.min(points, axis=0)
res[1,:] = np.max(points, axis=0)
return res
#规则n*m表格分割
def getLC(im,ld_dt):
#水平投影
hhist = []
mark=0
start = 0
end = 0
for i in range(im.shape[0]):
nmi = np.sum(im[i,:]==255)
if nmi>0 and mark==0:
start = i
mark = 1
elif nmi==0 and mark==1:
end = i
mark=0
hhist.append([start,end])
#print(hhist)
#垂直投影
vhist = []
mark=0
start = 0
end = 0
for i in range(im.shape[1]):
nmi = np.sum(im[:,i]==255)
if nmi>0 and mark==0:
start = i
mark = 1
elif nmi==0 and mark==1:
end = i
mark=0
vhist.append([start,end])
nld_dt = []
for mi in range(len(ld_dt)):
m = ld_dt[mi]
for i in range(len(hhist)):
mkk = 0
for j in range(len(vhist)):
cx = (m[0,0]+m[1,0])/2
cy = (m[1,1]+m[0,1])/2
if cx>=vhist[j][0] and cx <= vhist[j][1] and cy>=hhist[i][0] and cy <= hhist[i][1]:
lm = {'e':m,'y':i,'x':j}
nld_dt.append(lm)
mkk=1
break
if mkk==1:
break
return len(hhist), len(vhist), nld_dt
mrc = ['{','}','|',':','[',']','!',',','\'','’','‘','_'] #需要过滤的异常识别结果
nondigit = ['⑦']
def main(argv=None):
#创建目录
isExists=os.path.exists(FLAGS.reg_data_path)
if not isExists:
os.makedirs(FLAGS.reg_data_path)
fin = open(FLAGS.reg_data_path+'/reg.txt','w') #记录识别结果txt
#开始读取文件目录图片识别并保存结果
for root, dirs, files in os.walk(FLAGS.test_data_path):
#进行排序,顺序识别和检测
file_dicts = {}
kindex = []
for fnx in range(len(files)):
if files[fnx].endswith('.DS_Store'):
continue
file = files[fnx]
kindex.append(int(os.path.splitext(file)[0]))
file_dicts[int(os.path.splitext(file)[0])] = file
kindex.sort()
#开始识别
for fnx in range(len(kindex)):
iix = kindex[fnx]
file = file_dicts[iix]
print(file)
if os.path.splitext(file)[1] == '.png' or os.path.splitext(file)[1] == '.jpg':
if os.path.exists(FLAGS.test_data_path+'/jsondata/'+os.path.splitext(file)[0]+".json"):
with open(FLAGS.test_data_path+'/jsondata/'+os.path.splitext(file)[0]+".json",'r') as load_f:
fin.writelines(file+'\n')
load_dict = json.load(load_f)
im = np.array(cv2.imread(FLAGS.test_data_path+'/'+file))
#对检测的文字外框进行修正
ld_dt = []
imbk = np.zeros((im.shape[0],im.shape[1]),dtype=np.uint8)
for inc in load_dict:
ps = []
for p in load_dict[inc]['coordinate']:
m = [int(load_dict[inc]['coordinate'][p]['x']),
int(load_dict[inc]['coordinate'][p]['y'])]
ps.append(m)
box = bbox(ps)
if box[0,1]<0:
box[0,1] = 0
if box[0,0]<0:
box[0,0] = 0
cv2.rectangle(imbk,(box[0,0],box[0,1]),(box[1,0],box[1,1]),(255,255,255),1)
ld_dt.append(box)
#表格判断
rs, cs, nld_dt = getLC(imbk,ld_dt)
xs = []
for ii in range(rs):
m = []
for jj in range(cs):
if jj==0:
m.append('')
else:
m.append(',')
xs.append(m)
number_rect_num = 0 #记录矩形框包含数字的信息
for inc in nld_dt:
box = inc['e']
yy = inc['y']
xx = inc['x']
#print(box)
#边框修正
lim = im[box[0,1]:box[1,1],box[0,0]:box[1,0],:]
thresh = cv2.cvtColor(lim,cv2.COLOR_BGR2GRAY)
T= mahotas.thresholding.otsu(thresh)
thresh[thresh >T] = 255#矩阵thresh中>T的值赋值为255
thresh[thresh <= T] = 0#矩阵thresh中<255的值赋值为0
#确认背景和前景
nm = np.sum(thresh==255)
bk = 255
if nm>(box[1,0]-box[0,0])*(box[1,1]-box[0,1])/2:
bk = 0
startx = 0
endx = 0
marks = 0
marke = 0
lt = box[1,0]-box[0,0]
for i in range(lt):
nmi = np.sum(thresh[:,i]==bk)
if nmi>0:
startx = box[0,0] + i-7
marks = 1
break
for i in range(lt):
nmi = np.sum(thresh[:,lt-i-1]==bk)
if nmi>0:
marke = 1
endx = box[1,0]-i + 6
break
if marks ==0:
startx = box[0,0]
if marke ==0:
endx = box[1,0]
starty = 0
endy = 0
marks = 0
marke = 0
lt = box[1,1]-box[0,1]
for i in range(lt):
nmi = np.sum(thresh[i,:]==bk)
if nmi>0:
starty = box[0,1] + i-4
marks = 1
break
for i in range(lt):
nmi = np.sum(thresh[lt-i-1:]==bk)
if nmi>0:
marke = 1
endy = box[1,1]-i + 4
break
if marks ==0:
starty = box[0,1]
if marke ==0:
endy = box[1,1]
recs = [[startx,starty,endx,starty,
startx,endy,endx,endy]]
#print(recs)
if endy-starty<lt/2:
starty = int(starty - (endx-startx-endy+starty)/2)
endy = int(endy + (endx-startx-endy+starty)/2)
#进行阈值
#lim = im[box[0,1]:box[1,1],box[0,0]:box[1,0],:]
result = ocr.charRec(img = im,text_recs=recs,adjust=False)
if len(result)>0:
astr =[]
mk=0
asc = []
#对异常识别结果进行过滤
for rt in range(len(result[0][1])):
if result[0][1][rt] in mrc: #异常的处理
continue
elif result[0][1][rt] is '.' and mk==0: #第一个'.'
continue
elif ord(result[0][1][rt]) == 9675 or ord(result[0][1][rt]) == 9312: #异常O
astr.append('0')
mk=1
elif result[0][1][rt]=='o' or result[0][1][rt]=='O':
o_0_mark = 0
if rt >0 and result[0][1][rt-1]>='0' and result[0][1][rt-1]<='9':
o_0_mark=1
if rt <len(result[0][1])-1 and result[0][1][rt+1]>='0' and result[0][1][rt+1]<='9':
o_0_mark=1
if o_0_mark==1:
astr.append('0')
else:
astr.append('O')
else:
astr.append(result[0][1][rt])
asc.append(ord(result[0][1][rt]))
mk=1
#判断是否为数字
if ''.join(astr).isdigit():
c_n = 0
for el in astr:
if ord(el)>200:
c_n=1
break
if c_n ==0:
number_rect_num = number_rect_num + 1
if xs[yy][xx] == ',' or xs[yy][xx] == '':
xs[yy][xx] = xs[yy][xx]+''.join(astr).lower()
else:
xs[yy][xx] = xs[yy][xx]+','+''.join(astr).lower()
#文本输出修正
for ii in range(rs):
mk = 0
for jj in range(cs):
if jj==0:
if xs[ii][jj]!='':
mk=1
fin.writelines(xs[ii][jj])
else:
fin.writelines(xs[ii][jj])
fin.writelines('\n')
fin.writelines('\n')
fin.close()
if __name__ == '__main__':
tf.app.run()