Gif verification code analysis

GIF Verification Code and Common captcha The difference is that the words on the picture are hidden and visible from time to time , If you view by frame , Every frame is missing text .

GIF Example of verification code picture ：

So now we have the following ideas ：

• Scheme 1 ： Take several screenshots of the picture , Then the array of all pictures is merged and overwritten in one picture , Then we can recognize the complete picture .
• Option two ： Frame the picture multiple times , Then identify each picture individually , The highest frequency of characters in each position is the correct result .

---

Scheme I code ：

from PIL import Image
import numpy as np
import cv2

path = r"C:\Users\lixi\Desktop\p1.gif"
image =Image.open(path)

shapes = []
for i in range(1,4):
    image.seek(i)
    image.save(f'image/{
      i}.png')
    shapes.append(np.array(image))

result = np.subtract(shapes[0], shapes[2])
result2 = np.subtract(shapes[0], shapes[1])
result3 = np.subtract(shapes[1], shapes[2])
cv2.imwrite("image/result.png", shapes[0] + result+result2+result3)

The merged image ：

---

Scheme II code ：（ Code transferred from https://juejin.cn/post/6855483334512869389）

# -*- coding: utf-8 -*-
# @Software: PyCharm
import requests
import time
import json
from PIL import Image
from io import BytesIO
from collections import Counter
def get_max_char(str):
    '''  Get the most frequent character  :param str: :return: '''
    count = Counter(str)
    count_list = list(count.values())
    max_value = max(count_list)
    max_list = []
    for k, v in count.items():
        if v == max_value:
            max_list.append(k)
    return max_list[0]
def recogition(yzm_data):
    '''  Verification code recognition  :param yzm_data: :return: '''
    resp = requests.post('http://127.0.0.1:8080', data=yzm_data)
    return resp.text

def img_to_text(yzmdatas):
    '''  Picture to character  :param length: :return: '''
    yzm1 = ""
    yzm2 = ""
    yzm3 = ""
    yzm4 = ""
    for data in yzmdatas:
        text = recogition(data)
        json_obj = json.loads(text)
        yzm_text = json_obj.get("code","")
        # The length of the verification code in this paper is 4  In the actual test, as long as the length is greater than or equal to 4 All of them can be counted , It does not affect the recognition accuracy 
        if len(yzm_text) == 4:
            l_yzm = list(yzm_text)
            yzm1 = yzm1 + l_yzm[0]
            yzm2 = yzm2 + l_yzm[1]
            yzm3 = yzm3 + l_yzm[2]
            yzm4 = yzm4 + l_yzm[3]
    yzm1 = get_max_char(yzm1)
    yzm2 = get_max_char(yzm2)
    yzm3 = get_max_char(yzm3)
    yzm4 = get_max_char(yzm4)
    return yzm1+yzm2+yzm3+yzm4
def download():
    '''  Download verification code  :return: '''
    # Verification code address 
    url = 'http://credit.customs.gov.cn/ccppserver/verifyCode/creator'
    resp = requests.get(url)
    data = resp.content
    return data
def gif_to_png(length,image):
    ''' gif Frame extraction  :param length: :param image: :return: '''
    try:
        yzm_list = []
        for i in range(1, length):
            image.seek(i)
            stream = BytesIO()
            image.save(stream, 'PNG')
            s = stream.getvalue()
            yzm_list.append(s)
        return yzm_list
    except Exception as e:
        print(e)
    return None
def handle_yzm(length):
    '''  Process verification code  :return: '''
    gif = download()
    start = time.time()
    if gif:
        data = BytesIO(gif)
        image = Image.open(data)
        png_list = gif_to_png(length, image)
        if png_list:
            yzm_text = img_to_text(png_list)
    with open("./Gif_IMG/{}_{}.gif".format(yzm_text, str(time.time())),"wb") as fw:
        fw.write(gif)
    end = time.time()
    print(" Frame extraction length:{}- Spend time ：{}".format(length, end - start))
def run():
    # Frame extraction length ： The specific number of frames can be based on the actual gif Recognition accuracy to adjust .
    # The less frames are extracted, the lower the recognition rate may be , But the time required for identification will be reduced .23 Frame accuracy 98%, Time 1s; 6 Frame accuracy 85%, Time 0.5s about 
    # Find a balance between recognition speed and accuracy 
    length = 10
    # Number of recognized pictures 
    num = 20
    for i in range(num):
        handle_yzm(length)
if __name__ == '__main__':
    run()

Scheme 1 is simple and crude , But it will also combine other interference elements , Therefore, other processing methods are required after the merger .

Scheme 2 code transfer to others , The effect will be better , First identify and then synthesize , But it's not efficient , You need to change it yourself .

The code of both schemes needs to be optimized and adjusted when used , We choose the corresponding scheme according to our own needs .