-
Notifications
You must be signed in to change notification settings - Fork 0
/
Invoice_extraction.py
133 lines (107 loc) · 5.51 KB
/
Invoice_extraction.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
import cv2
import numpy as np
from PIL import Image, ImageOps
class InvoiceExtraction:
def __init__(self, model):
self.model = model
def reduce_size(self, path, size=-1, padding=0):
or_image = Image.open(path)
or_image = ImageOps.grayscale(or_image)
if size == -1:
new_img = ImageOps.expand(or_image, padding)
return np.array(new_img).astype('uint8')
size = size - padding
s = or_image.size
if s[0] <= s[1]:
image = or_image.copy().resize((int(s[0] * (size / s[1])), size))
else:
image = or_image.copy().resize((size, int(s[1] * (size / s[0]))))
new_size = image.size
delta_w = size - new_size[0] + padding
delta_h = size - new_size[1] + padding
new_padding = (delta_w // 2, delta_h // 2, delta_w - (delta_w // 2), delta_h - (delta_h // 2))
new_img = ImageOps.expand(image, new_padding)
or_size = or_image.size
delta_w = max(or_size) - or_size[0] + padding * int(max(or_image.size) / (size + padding))
delta_h = max(or_size) - or_size[1] + padding * int(max(or_image.size) / (size + padding))
or_padding = (delta_w // 2, delta_h // 2, delta_w - (delta_w // 2), delta_h - (delta_h // 2))
or_image = ImageOps.expand(or_image, or_padding)
return np.array(new_img)/255, np.array(or_image)/255
def reorder(self, myPoints):
myPoints = myPoints.reshape((4, 2))
myPointsNew = np.zeros((4, 1, 2), dtype=np.int32)
add = myPoints.sum(1)
myPointsNew[0] = myPoints[np.argmin(add)]
myPointsNew[3] = myPoints[np.argmax(add)]
diff = np.diff(myPoints, axis=1)
myPointsNew[1] = myPoints[np.argmin(diff)]
myPointsNew[2] = myPoints[np.argmax(diff)]
return myPointsNew
def drawRectangle(img, biggest, thickness):
cv2.line(img, (biggest[0][0][0], biggest[0][0][1]), (biggest[1][0][0], biggest[1][0][1]), (0, 255, 0),
thickness)
cv2.line(img, (biggest[0][0][0], biggest[0][0][1]), (biggest[2][0][0], biggest[2][0][1]), (0, 255, 0),
thickness)
cv2.line(img, (biggest[3][0][0], biggest[3][0][1]), (biggest[2][0][0], biggest[2][0][1]), (0, 255, 0),
thickness)
cv2.line(img, (biggest[3][0][0], biggest[3][0][1]), (biggest[1][0][0], biggest[1][0][1]), (0, 255, 0),
thickness)
return img
def order_points(self, pts):
rect = np.zeros((4, 2), dtype='float32')
pts = np.array(pts)
s = pts.sum(axis=1)
rect[0] = pts[np.argmin(s)]
rect[2] = pts[np.argmax(s)]
diff = np.diff(pts, axis=1)
rect[1] = pts[np.argmin(diff)]
rect[3] = pts[np.argmax(diff)]
return rect.astype('int').tolist()
def find_dest(self, pts):
(tl, tr, br, bl) = pts
widthA = np.sqrt(((br[0] - bl[0]) ** 2) + ((br[1] - bl[1]) ** 2))
widthB = np.sqrt(((tr[0] - tl[0]) ** 2) + ((tr[1] - tl[1]) ** 2))
maxWidth = max(int(widthA), int(widthB))
heightA = np.sqrt(((tr[0] - br[0]) ** 2) + ((tr[1] - br[1]) ** 2))
heightB = np.sqrt(((tl[0] - bl[0]) ** 2) + ((tl[1] - bl[1]) ** 2))
maxHeight = max(int(heightA), int(heightB))
destination_corners = [[0, 0], [maxWidth, 0], [maxWidth, maxHeight], [0, maxHeight]]
return self.order_points(destination_corners)
def restore_coor(self, contours, ratio):
(tl, tr, br, bl) = contours
# Finding the maximum width.
widthA = np.sqrt(((br[0] - bl[0]) ** 2) + ((br[1] - bl[1]) ** 2))
widthB = np.sqrt(((tr[0] - tl[0]) ** 2) + ((tr[1] - tl[1]) ** 2))
maxWidth = max(int(widthA), int(widthB))
# Finding the maximum height.
heightA = np.sqrt(((tr[0] - br[0]) ** 2) + ((tr[1] - br[1]) ** 2))
heightB = np.sqrt(((tl[0] - bl[0]) ** 2) + ((tl[1] - bl[1]) ** 2))
maxHeight = max(int(heightA), int(heightB))
x = tl[0] * ratio // 2
y = tl[1] * ratio // 2
print(maxWidth, maxHeight)
# Final destination co-ordinates.
destination_corners = [[y, x], [x + maxWidth * ratio, y], [x + maxWidth * ratio, y + maxHeight * ratio],
[x, y + maxHeight * ratio]]
return destination_corners
def extract(self, image_path):
image, or_img = self.reduce_size(image_path, 256, 10)
mask = self.model.predict(image.reshape((1, image.shape[0], image.shape[0], 1))).reshape((256, 256))
canny = cv2.Canny(np.uint8(mask*255), 0, 200)
canny = cv2.dilate(canny, cv2.getStructuringElement(cv2.MORPH_ELLIPSE, (5, 5)))
contours, hierarchy = cv2.findContours(canny, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
page = sorted(contours, key=cv2.contourArea, reverse=True)[:5]
for c in page:
epsilon = 0.02 * cv2.arcLength(c, True)
corners = cv2.approxPolyDP(c, epsilon, True)
if len(corners) == 4:
break
# print(corners, int(or_img.shape[0]/image.shape[0]))
corners = sorted(np.concatenate(corners).tolist())
corners = (np.array(corners) * (or_img.shape[0] / image.shape[0])).tolist()
corners = self.order_points(corners)
destination_corners = self.find_dest(corners)
M = cv2.getPerspectiveTransform(np.float32(corners), np.float32(destination_corners))
final = cv2.warpPerspective(or_img, M, (destination_corners[2][0], destination_corners[2][1]),
flags=cv2.INTER_LINEAR)
return final