pytesser is an open source project of Google, which can be used for text recognition in pictures

Here we will use it for simple verification code identification. The code is as follows:

#!/usr/bin/env python
# -*- coding:utf-8 -*-
__author__ = 'admin'
from pytesser import *

flag = 140
table = []
for i in range(256):
    if i < flag:
        table.append(0)
    else:
        table.append(1)

# Since they are all numbers, if there are letters, use this table for correction, and take the similar letters and values as the mapping
rep = {'O': '0',
       'I': '1',
       'L': '1',
       'Z': '2',
       'S': '8',
       'Q': '0'}


class AuthCode(object):
    def __init__(self, img_name):
        self.img = img_name

    def tran_to_str(self):
        # Open the picture.
        im = Image.open(self.img)
        #   Image to gray
        img_ry = im.convert('L')
        #  Save grayscale image
        img_ry.save('g'+self.img)
        #   Binarization to remove image noise
        out = img_ry.point(table, '1')
        # Save the image after noise removal
        out.save('b'+self.img)
        #   Distinguish
        text = image_to_string(out)
        #   Processing of recognized characters
        text = text.strip()
        text = text.upper()
        for r in rep:
            text = text.replace(r, rep[r])
        #   out.save(text+'.jpg')
        print text
        return text
if __name__ == '__main__':
    i = AuthCode('r2.jpg')
    i.tran_to_str()

Specific process description:

Here RBG is used to convert to HSI color space, and L component is used
Now the gray image looks like this

We set the pixels greater than the threshold value to 1, and the others to 0. To solve this problem, Mr. Zhang made a look-up table and asked the library function to help us with the mapping process.
flag = 140
table = []
for i in range(256):
if i < flag:
table.append(0)
else:
table.append(1)

Now the image looks like this

text = image_to_string(out)

Output result