-
Notifications
You must be signed in to change notification settings - Fork 4
/
detection.py
151 lines (112 loc) · 4.35 KB
/
detection.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
# -*- coding: UTF-8 -*-
import cv2
import numpy as np
'''
detect the region to recognition
'''
def morph_process(gray): #elimate some region in order to detect the text
ret, binary = cv2.threshold(gray, 60, 255, cv2.THRESH_BINARY_INV)
h,w = binary.shape
#下半部分消除：
ele = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 5))
binary_up = binary[0:h * 2 / 5, 0:w]
binary_medium = binary[h * 2 / 5:h * 3 / 5, 0:w]
binary_down = binary[h * 3 / 5:, 0:w]
eroded_down = cv2.erode(binary_down, ele, iterations=1)
binary = np.concatenate((binary_up,binary_medium,eroded_down))
ele_up = cv2.getStructuringElement(cv2.MORPH_RECT, (30, 10))
ele_medium= cv2.getStructuringElement(cv2.MORPH_RECT, (40, 10))
ele_down = cv2.getStructuringElement(cv2.MORPH_RECT, (45, 10))
binary_up = binary[0:h*2/5, 0:w]
binary_medium = binary[h*2/5:h*3/5,0:w]
binary_down = binary[h*3/5:,0:w]
dilation_up = cv2.dilate(binary_up, ele_up, iterations=1)
dilation_medium = cv2.dilate(binary_medium, ele_medium, iterations=1)
dilation_down = cv2.dilate(binary_down, ele_down, iterations=1)
dilation = np.concatenate((dilation_up, dilation_medium,dilation_down), axis = 0)
return dilation
def findTextRegion(img):
region = []
# 1. 查找轮廓
image, contours, hierarchy = cv2.findContours(img, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
# 2. 筛选那些面积小的
for i in range(len(contours)):
cnt = contours[i]
# 计算该轮廓的面积
area = cv2.contourArea(cnt)
# 面积小的都筛选掉
if (area < 5000):
continue
# 面积大的也不用
if(area >40000):
continue
# 轮廓近似，作用很小
epsilon = 0.001 * cv2.arcLength(cnt, True)
approx = cv2.approxPolyDP(cnt, epsilon, True)
# 找到最小的矩形，该矩形可能有方向
rect = cv2.minAreaRect(cnt)
# box是四个点的坐标
box = cv2.boxPoints(rect)
box = np.int0(box)
# 计算高和宽
height = abs(box[0][1] - box[2][1])
width = abs(box[0][0] - box[2][0])
if (height *2 > width ): # 处理不扁的
continue
if(max(box[0][1] , box[2][1])> img.shape[0]*5/6): # 处理过于靠近底层的
continue
if(width < img.shape[1]/9): #处理过短的:
continue
if (width * 2 > img.shape[1]): # 处理过于宽的
continue
if(height<img.shape[0]/23): #处理过于细的
continue
if max(box[0][1], box[2][1]) > img.shape[0] * 4/5:
continue
if(max(box[2][0],box[0][0]) < img.shape[1]/4 and max(box[0][1] , box[2][1]) > img.shape[0]*1/2): #处理下左半部分
continue
if (max(box[2][0], box[0][0]) < img.shape[1] / 3 and min(box[0][1], box[2][1]) < img.shape[0] * 1/ 3): #处理上左
continue
if(min(box[0][1] , box[2][1]) < img.shape[0]/5): #处理过于靠上的字符
continue
if (min(box[0][0], box[2][0]) > img.shape[1]/2): # 中心往右的部分删除
continue
region.append(box)
minwidth = 100000
num = -1
index = 10000
for reg in region: #将宽度最短的去掉
num = num + 1
width = abs(reg[0][0] - reg[2][0])
if width < minwidth:
minwidth = width
index = num
del region[index]
return region
def grayImg(img):
# 转化为灰度图
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
gray = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 47,20)
return gray
def detect(img):
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
# 2. 形态学变换的预处理，得到可以查找矩形的图片
dilation = morph_process(gray)
# 3. 查找和筛选文字区域
region = findTextRegion(dilation)
# 4. 用绿线画出这些找到的轮廓
ii = 0
idImgs = []
for box in region:
h = abs(box[0][1] - box[2][1])
w = abs(box[0][0] - box[2][0])
Xs = [i[0] for i in box]
Ys = [i[1] for i in box]
x1 = min(Xs)
y1 = min(Ys)
img2 = img.copy()
if w > 0 and h > 0:
idImg = grayImg(img2[y1:y1 + h, x1:x1 + w])
ii += 1
idImgs.append(idImg)
return idImgs