-
Notifications
You must be signed in to change notification settings - Fork 0
/
opencv_textblocks.Rmd
327 lines (252 loc) · 7.55 KB
/
opencv_textblocks.Rmd
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
---
jupyter:
jupytext:
text_representation:
extension: .Rmd
format_name: rmarkdown
format_version: '1.2'
jupytext_version: 1.11.2
kernelspec:
display_name: Python (py38)
language: python
name: py38
---
```{python}
try:
import cv2 as cv
except:
# !pip install OpenCV-Python
```
```{python}
#https://stackoverflow.com/questions/48768604/how-to-find-a-letter-in-an-image-with-python
```
```{python}
def get_contours(img_path):
### load input image and convert it to grayscale
img = cv.imread(img_path)
imgray = cv.cvtColor(img, cv.COLOR_BGR2GRAY)
height, width, channels = img.shape
#### extract all contours
ret, thresh = cv.threshold(imgray, 127, 255, 0)
contours, hierarchy = cv.findContours(thresh, cv.RETR_TREE, cv.CHAIN_APPROX_SIMPLE)
return contours
```
```{python}
img_path = "Screenshot 20210304115007.png"
img = cv.imread(img_path)
height, width, channels = img.shape
contours = get_contours(img_path)
# debug: draw all contours
img_c = cv.drawContours(img.copy(), contours, -1, (0, 0, 255), 1)
cv.imwrite("all_contours.jpg", img_c)
```
```{python}
#### create one bounding box for every contour found
c_list = []
bb_list = []
for c in contours:
bb = cv.boundingRect(c)
# save all boxes except the one that has the exact dimensions of the image (x, y, width, height)
if (bb[0] == 0 and bb[1] == 0 and bb[2] == img.shape[1] and bb[3] == img.shape[0]):
continue
bb_list.append(bb)
c_list.append(c)
# debug: draw boxes
img_boxes = img.copy()
for c in c_list:
x,y,w,h = cv.boundingRect(c)
cv.rectangle(img_boxes, (x, y), (x+w, y+h), (0, 0, 255), 1)
```
```{python}
from operator import itemgetter, attrgetter
#### sort bounding boxes by the X value: first item is the left-most box
y_list = sorted(bb_list, key=itemgetter(0))
xy_list = sorted(y_list, key=itemgetter(1))
x_lines = [0] * width
y_lines = [0] * height
for bb in xy_list:
x, y, w, h = bb
for i in range(x, x+w):
x_lines[i] = 1
for i in range(y, y+h):
y_lines[i] = 1
pstate = 0
y = 0
y1 = width
for idx, i in enumerate(x_lines):
if i != pstate:
print(idx)
pstate = i
x = idx
x1 = idx
cv.line(img_boxes, (x, y), (x1, y1), (255,0,0), 1)
pstate = 0
x = 0
x1 = width
for idx, i in enumerate(y_lines):
if i != pstate:
pstate = i
y = idx
y1 = idx
cv.line(img_boxes, (x, y), (x1, y1), (0,255,0), 1)
```
```{python}
cv.imwrite("boxes.jpg", img_boxes)
```
```{python}
gutter = 10
# using group rectangles...
rects = bb_list.copy()
#print(len(bb_list))
for bb in bb_list:
x,y,w,h = bb
rects.append((x,y,w,h))
rects.append((x - gutter, y - gutter, w + gutter, h + gutter))
print(len(rects))
rects.extend(bb_list)
cv_rects = cv.groupRectangles(rects, 1)
print(len(cv_rects))
# print image...
img_group = img.copy()
for bb in bb_list:
x,y,w,h = bb
cv.rectangle(img_boxes, (x, y), (x+w, y+h), (0, 0, 255), 1)
for bb in cv_rects[0]:
(x,y,w,h) = bb
cv.rectangle(img_group, (x, y), (x+w, y+h), (0, 255, 0), 1)
cv.imwrite("cv.groups.jpg", img_group)
```
```{python}
class Rectangle(object):
def __init__(self, x, y, w, h):
self._x = x
self._y = y
self._x2 = w + self.x
self._y2 = h + self.y
def check_intersect(self, rect, gutter = 0):
if not rect:
return False
if not isinstance(rect, Rectangle):
x,y,w,d = rect
rect = Rectangle(x,y,w,d)
# If one rectangle is on left side of other
#
# .---. x2, y2
# x,y .___.
#
# this rect is above or right
if (self._y2) < (rect.y - gutter) or (self._x2) < (rect.x - gutter):
return False
# below or left
if (self._x - gutter) > (rect._x) or (self._y- gutter) > (rect._y):
return False
return True
def union(self, rect, gutter=0):
# handle a opencv python tuple
if not isinstance(rect, Rectangle):
x,y,w,d = rect
rect = Rectangle(x,y,w,d)\
x = min(self._x, )
def intersection(self, rect, gutter=0):
# handle a opencv python tuple
if not isinstance(rect, Rectangle):
x,y,w,d = rect
rect = Rectangle(x,y,w,d)
def expand(self, rect, gutter=0):
# handle a opencv python tuple
if not isinstance(rect, Rectangle):
x,y,w,d = rect
rect = Rectangle(x,y,w,d)
if not isinstance(rect, Rectangle):
x,y,w,d = rect
rect = Rectangle(x,y,w,d)
if self.x > rect.x:
self.x = rect.x - gutter
if self.y > rect.y:
self.y = rect.y - gutter
if self.x2 < rect.x2:
self.x2 = rect.x2 + gutter
if self.y2 < rect.y2:
self.y2 = rect.y2 + gutter
def __str__(self):
return "Rectangle<{}, {}, {}, {}>".format(self.x, self.y, self.x2, self.y2)
def __repr__(self):
return "Rectangle<{}, {}, {}, {}>".format(self.x, self.y, self.x2, self.y2)
```
```{python}
def merge_rects(rect_list, gutter = 0):
""" Returns a list of grouped rectangles
https://stackoverflow.com/questions/37847923/combine-overlapping-rectangles-python/53894169
"""
idx = 0
if rect_list is None:
return zones
# move through the array using the idx
while idx < len(rect_list):
# check for overlap in the rest of the list
no_overlap = False
while no_overlap == Flase and len(rect_list) > 1 and idx < len(rect_list):
rect = rect_list[idx]
# get teh remaining rects for match:
tmp_rects = np.delete(rect_list, idx, 0)
```
```{python}
gutter = 10
contour_areas = []
# scan the area and return the bbbox within the aras + gutter size for the document
# add the first bbox to the array, then scan for any values that are in range...
x,y,w,d = xy_list[0]
contour_areas.append(Rectangle(x,y,w,d))
for idx, bb in enumerate(xy_list):
if not bb:
continue
x,y,w,d = bb
rect = Rectangle(x,y,w,d)
# check for existing rectangles
add_rect = True
for g_idx, g_rect in enumerate(contour_areas):
if not g_rect:
continue
if g_rect.intersect(rect, gutter):
# expand the cell
g_rect.expand(rect, gutter)
add_rect = False
break
if add_rect:
#print("+", rect)
contour_areas.append(rect)
# plot the new boxes...
img_group = img.copy()
for bb in contour_areas:
if not bb:
continue
cv.rectangle(img_group, (bb.x, bb.y), (bb.x2, bb.y2), (128, 128, 255), 1)
print(contour_areas)
# merge the boxes till no more can be merged...
merged_areas = []
while len(contour_areas) > 0:
rect = contour_areas.pop()
if not rect:
continue
unique = True
# find all intersecting areas
for g_idx, g_rect in enumerate(contour_areas):
if not g_rect:
continue
if g_rect.intersect(rect, gutter):
rect.expand(g_rect)
print(rect, g_rect)
if unique:
merged_areas.append(rect)
else:
pass
for bb in merged_areas:
if not bb:
continue
cv.rectangle(img_group, (bb.x, bb.y), (bb.x2, bb.y2), (255, 255, 0), 1)
cv.imwrite("groups.jpg", img_group)
```
```{python}
```
```{python}
```