-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsudoku_project3.py
139 lines (109 loc) · 4.61 KB
/
sudoku_project3.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
import numpy as np
import cv2
from tensorflow.keras.models import load_model
cap = cv2.VideoCapture(0)
out = 0
model = load_model("computer_mnist_model100.h5")# ("computer_mnist_model.h5") #("mnist_model.h5")
while(1):
ret, im = cap.read()
imgray = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY) # BGR to grayscale
# imgray = cv2.GaussianBlur(imgray, (3, 3), 0)
thresh = cv2.adaptiveThreshold(imgray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,\
cv2.THRESH_BINARY_INV, 35, 12)
# ret, thresh = cv2.threshold(imgray,90, 255, cv2.THRESH_BINARY_INV)
contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) # cv2.CHAIN_APPROX_NONE)
rects = [cv2.boundingRect(ctr) for ctr in contours]
max_area_rect = 0
max_area_ctr = 0
max_area = 0
for ctr in contours:
rect = cv2.boundingRect(ctr)
if(rect[2]*rect[3] > max_area):
max_area = rect[2]*rect[3]
max_area_rect = rect
max_area_ctr = ctr
# cv2.rectangle(im, (max_area_rect[0], max_area_rect[1]), (max_area_rect[0] + max_area_rect[2], max_area_rect[1] + max_area_rect[3]), (0, 0, 255), 1)
if(str(max_area_ctr) == '0'):
continue
epsilon = 0.01 * cv2.arcLength(max_area_ctr, True)
approx = cv2.approxPolyDP(max_area_ctr, epsilon, True)
im = cv2.drawContours(im, [approx], -1, (0, 255, 0), 1)
# im_rect = im[max_area_rect[1]:max_area_rect[1]+max_area_rect[3], max_area_rect[0]:max_area_rect[0]+max_area_rect[2]]
if(approx.shape[0] == 4):
up_left = approx[0][0]
up_right = approx[3][0]
bottom_left = approx[1][0]
bottom_right = approx[2][0]
input_pts = np.float32([up_left,up_right,bottom_left,bottom_right])
height = max(abs(bottom_left[1] - up_left[1]), abs(up_right[1]-bottom_right[1]))
width = max(abs(bottom_left[0] - bottom_right[0]), abs(up_left[0]-up_right[0]))
output_pts = np.float32([[0, 0], [width, 0],[0, height],[width, height]])
# Compute the perspective transform M
M = cv2.getPerspectiveTransform(input_pts, output_pts)
# Apply the perspective transformation to the image
out = cv2.warpPerspective(im, M, (width, height), flags = cv2.INTER_LINEAR)
# cv2.imshow("out", out)
# cv2.imwrite("out.png", out)
thresh_out = cv2.warpPerspective(thresh, M, (width, height), flags = cv2.INTER_LINEAR)
# imgray = cv2.cvtColor(out, cv2.COLOR_BGR2GRAY) # BGR to grayscale
# thresh = cv2.adaptiveThreshold(imgray,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,\
# cv2.THRESH_BINARY_INV,35,12)
edges = cv2.Canny(thresh_out, 1, 1, apertureSize = 3)
# shape = gray.shape
lines = cv2.HoughLines(edges,1,np.pi/180,100)
if(str(lines) == 'None'):
continue
for i in range(len(lines)):
for rho,theta in lines[i]:
a = np.cos(theta)
b = np.sin(theta)
x0 = a*rho
y0 = b*rho
x1 = int(x0 + (1000)*(-b)) # shape[0]
y1 = int(y0 + (1000)*(a)) # shape[1]
x2 = int(x0 - (1000)*(-b))
y2 = int(y0 - (1000)*(a))
if(abs(x1 - x2) > 50 and abs(y1 - y2) > 50):
continue
# cv2.line(img1, (x1,y1),(x2,y2),(255,255,255),3) # 2 points ,color, thikness
cv2.line(thresh_out, (x1,y1),(x2,y2),(0,0,0),3)
im_th = thresh_out
ctrs, hier = cv2.findContours(im_th.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
rects = [cv2.boundingRect(ctr) for ctr in ctrs]
sudoku_area = im_th.shape[0]*im_th.shape[1]//81
digits_array = []
for rect in rects:
if(rect[2]*rect[3] < sudoku_area//15):
continue
cv2.rectangle(out, (rect[0], rect[1]), (rect[0] + rect[2], rect[1] + rect[3]), (0, 255, 0), 1)
from_x = max(0, rect[1] - rect[3]//4)
to_x = min(im_th.shape[0], rect[1] + rect[3] + rect[3]//4)
from_y = max(0, rect[0] - rect[2]//4)
to_y = min(im_th.shape[1], rect[0] + rect[2] + rect[2]//4)
nbr_img = im_th[from_x: to_x, from_y:to_y]/255
nbr_resized_disp = cv2.resize(nbr_img, (28, 28))
nbr_resized = nbr_resized_disp.reshape(-1, 28, 28, 1)
digits_array.append(nbr_resized)
# N = model.predict(nbr_resized).argmax()
# cv2.putText(out, str(N), (rect[0], rect[1]),cv2.FONT_HERSHEY_DUPLEX, 1, (200, 0, 0), 1)
pred_array = model.predict(np.array(digits_array).reshape(-1, 28, 28, 1)).argmax(axis = 1)
N = 0
for rect in rects:
if(rect[2]*rect[3] < sudoku_area//15):
continue
cv2.putText(out, str(pred_array[N]), (rect[0], rect[1]),cv2.FONT_HERSHEY_DUPLEX, 1, (200, 0, 0), 1)
N += 1
# cv2.imshow("out", out)
# cv2.imshow("thresh_out", thresh_out)
# cv2.imshow("thresh", thresh)
if(str(out) == "0"):
continue
cv2.imshow("out", out)
cv2.imshow("title", im)
# cv2.imshow("thresh", thresh)
# print(out.shape)
k = cv2.waitKey(30) & 0xff
if k == 27: # 27 is ascii code for ESC
break
cap.release()
cv2.destroyAllWindows()