1

What I want to do is segment the lines and characters of an image's text that looks like this: Sample For line segmentation, I have done the following (based on this answer):

import cv2
import numpy as np
from matplotlib import pyplot as plt

#path
img = cv2.imread("1.png")

#grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

#threshold
th, threshed = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY_INV|cv2.THRESH_OTSU)

#r
pts = cv2.findNonZero(threshed)
ret = cv2.minAreaRect(pts)


(cx,cy), (w,h), ang = ret
if w>h:
    w,h = h,w
    ang += 90

#find  top and bottom line
hist = cv2.reduce(threshed,1, cv2.REDUCE_AVG).reshape(-1)

th = 2
H,W = img.shape[:2]
uppers = [y for y in range(H-1) if hist[y]<=th and hist[y+1]>th]
lowers = [y for y in range(H-1) if hist[y]>th and hist[y+1]<=th]

threshed = cv2.cvtColor(threshed, cv2.COLOR_GRAY2BGR)

# merge lines for detecting both above and below cosonants
temp = uppers[0]


i=0
while i<len(uppers)-1:
    if (uppers[i+1]-temp)<18:
        print(uppers[i+1]-temp)
        lowers.pop(i)
        temp = uppers[i+1]
        uppers.pop(i+1)
    else:
        temp = uppers[i+1]
        i+=1

y1 = []

#lines
for y in uppers:
    cv2.line(threshed, (0,y), (W, y), (255,0,0), 1)
    y1.append(y)
y2 = []
for y in lowers:
    cv2.line(threshed, (0,y), (W, y), (0,255,0), 1)
    y2.append(y)


c=0

while c<len(y1):
    imageCropped = threshed[y1[c]:y2[c]]
    # cv2.imshow('ld', imageCropped)
    # cv2.waitKey()
    cv2.imwrite(str(y1[c])+'.JPG', imageCropped) ##save cropped image
    c=c+1

cv2.imwrite("resultofwholeimage.png", threshed)

The image here shows resulting lines being drawn

However, now I want to segment individual characters, the characters should be segmented with the consonant that's above or below it; for example the first character from the right segmented should look like this: Single character segmentation

Is such a method possible?

liazmehdi
  • 11
  • 2

0 Answers0