-
Notifications
You must be signed in to change notification settings - Fork 0
/
textDetect.py
204 lines (178 loc) · 7.29 KB
/
textDetect.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
## Imports
import cv2
import numpy as np
from imutils.object_detection import non_max_suppression
import math
import os
import pytesseract
from PIL import Image
## LOADING EAST DETECTION Code Source: https://www.pyimagesearch.com/2018/08/20/opencv-text-detection-east-text-detector/
#gets image from path and resizes for EAST detection
#Idea for resizing taken from https://www.pyimagesearch.com/2018/08/20/opencv-text-detection-east-text-detector/
def getImg(path):
aspectRatio = (320,480)
image = cv2.imread(path)
h,w,_ = image.shape
scaledImage = cv2.resize(image, (320,480))
resizeX, resizeY = w/aspectRatio[0], h/aspectRatio[1]
return image, scaledImage, resizeX, resizeY
pathEAST = "frozen_east_text_detection.pb"
#takes in EAST Model path and image
#https://www.pyimagesearch.com/2018/08/20/opencv-text-detection-east-text-detector/
def applyMaps(path, image):
layerNames = ["feature_fusion/Conv_7/Sigmoid", "feature_fusion/concat_3"]
h, w,_ = image.shape
bgrMean = (123.68, 116.78, 103.94)
net = cv2.dnn.readNet(path)
#swapRB swaps R and B so that bgrMean is normal
blob = cv2.dnn.blobFromImage(image, 1.0, (w,h), bgrMean,
swapRB=True, crop=False)
net.setInput(blob)
(scores, geometry) = net.forward(layerNames)
return scores, geometry
# loop over the number of columns
#https://www.pyimagesearch.com/2018/08/20/opencv-text-detection-east-text-detector/
def boundingBox(scores, geometry):
boxes = []
confidences = []
minConfidence = 0.1
rows, cols = scores.shape[2:4]
for y in range(rows):
for x in range(cols):
scoreData = scores[0][0][y][x]
data0 = geometry[0][0][y][x]
data1 = geometry[0][1][y][x]
data2 = geometry[0][2][y][x]
data3 = geometry[0][3][y][x]
angle = geometry[0][4][y][x]
if scoreData > minConfidence:
offsetX, offsetY = x*4, y*4
h, w = data0 + data2, data1 + data3
x1 = int(offsetX + (math.cos(angle)*data1) + (math.sin(angle) * data2))
y1 = int(offsetY - (math.sin(angle)*data1) + (math.cos(angle) * data2))
x0, y0 = int(x1-w), int(y1-h)
boxes.append((x0,y0,x1,y1))
confidences.append(scoreData)
return boxes, confidences
#destructively resizes boxes to original size
#https://www.pyimagesearch.com/2018/08/20/opencv-text-detection-east-text-detector/
def resizeBoxes(boxes, confidences, resizeX, resizeY):
boxes = non_max_suppression(np.array(boxes), probs=confidences)
newBoxes = []
offSet = 5
for (x0,y0,x1,y1) in boxes:
x0 = int(x0 * resizeX) - offSet
y0 = int(y0 * resizeY) - offSet
x1 = int(x1 * resizeX) + offSet
y1 = int(y1 * resizeY) + offSet
newBox = (x0,y0,x1,y1)
newBoxes.append(newBox)
return newBoxes
def drawBoxes(image, boxes):
for (x0,y0,x1,y1) in boxes:
cv2.rectangle(image, (x0, y0), (x1, y1), (0,255,0), 2)
return image
path = "testPosters/poster3.jpg"
pathEAST = "frozen_east_text_detection.pb"
def detectText(path, pathEAST):
image, scaledImage, resizeX, resizeY = getImg(path)
scores, geometry = applyMaps(pathEAST, scaledImage)
boxes, confidences = boundingBox(scores, geometry)
boxes = resizeBoxes(boxes, confidences, resizeX, resizeY)
return boxes, image
# image = drawBoxes(image, boxes)
# cv2.imshow("Detect Text", image)
# cv2.waitKey(0)
#detectText(path, pathEAST)
def detectVideoText(pathEAST):
cap = cv2.VideoCapture(0)
aspectRatio = (320, 480)
frames = 0
while True:
_, frame = cap.read()
if frames % 10 == 0:
#Want to freeze frame and add a loading screen
h,w,_ = frame.shape
scaledFrame = cv2.resize(frame, (320,480))
resizeX, resizeY = w/aspectRatio[0], h/aspectRatio[1]
scores, geometry = applyMaps(pathEAST, scaledFrame)
boxes, confidences = boundingBox(scores, geometry)
newBoxes = resizeBoxes(boxes, confidences, resizeX, resizeY)
frame = drawBoxes(frame, newBoxes)
cv2.imshow("Video Detect Text", frame)
ch = cv2.waitKey(1)
frames += 1
if ch & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
## Research for better methods of segmenting text
#Inspired by https://www.researchgate.net/publication/280105485_Text_Extraction_and_Recognition_from_Posters_for_Movie_Title_Retrieval
def binarization(image):
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
thresh = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 115, 2)
restoredCharacters = mendBrokenCharacters(thresh, image)
dilate = cv2.dilate(thresh, (5,5), iterations = 5)
#erode = cv2.erode(thresh, (5,5), iterations = 5)
return dilate
def mendBrokenCharacters(erode, image):
originalEdge = cv2.Canny(image, 85, 100)
binaryEdge = cv2.Canny(erode, 85, 100)
cv2.imshow("BE",binaryEdge)
cv2.imshow("OE", originalEdge)
initialEdge = cv2.bitwise_and(originalEdge, binaryEdge)
cv2.imshow("preious", initialEdge)
initialEdge = verticleEdgeExtension(initialEdge, originalEdge)
# for row in range(len(originalEdge)):
# newRow = []
# for col in range(len(originalEdge[0])):
# if originalEdge[row][col] > 0 and binaryEdge[row][col] > 0:
# newRow.append((255,255,255))
# else:
# newRow.append((0,0,0))
# initialEdge.append(newRow)
# print(initialEdge)
cv2.imshow("Hi", initialEdge)
pass
def verticleEdgeExtension(edge, oE):
directions = [(-1,0),(-1,-1),(-1,1)]
for x in range(len(edge[0])-1):
for y in range(len(edge)-1):
if edge[y,x] == 255:
if edge[y+1,x-1] == edge[y+1,x] == edge[y+1,x+1] == 0:
neighbors = (oE[y+1,x-1],oE[y+1,x],oE[y+1,x+1])
neighborCount = 3 - neighbors.count(0)
if neighborCount == 1:
for i in range(len(neighbors)):
if neighbors[i] != 0:
edge[y+1, x+i-1] += 255
# elif neighborCount > 1:
# for i in range(len(neighbors)):
# if neighbors[i] != 0:
return edge
def horizontalEdgeExtension(edge, oE):
pass
#detectVideoText(pathEAST)
## Takes in bounding box, spits out text
def saveFiles(boxes, image):
for i in range(len(boxes)):
x0,y0,x1,y1 = boxes[i]
box = image[y0:y1, x0:x1]
betterBox = binarization(box)
fileName = "tmpFile_"+str(i)+".png"
cv2.imwrite("tmp/"+fileName, betterBox)
def getText():
textList = []
for i in range(len(os.listdir("tmp"))-1):
fileName = "tmpFile_" + str(i) + ".png"
text = pytesseract.image_to_string(Image.open("tmp/" + fileName))
textList.append(text)
return textList
def recognizeText(path, pathEAST):
boxes, image = detectText(path, pathEAST)
cv2.imshow("image", image)
saveFiles(boxes, image)
textList = getText()
print(textList)
cv2.waitKey(0)
recognizeText(path, pathEAST)