Python实现新版正方系统滑动验证码识别

Python实现新版正方系统滑动验证码识别算法和方案

步骤一：点击数据分析

点击滑动按钮，将发送一个请求到 /zfcaptchaLogin

请求内容

"type": "verify"
"rtk": "6cfab177-afb2-434e-bacf-06840c12e7af"
"time": "1624611806948"
"mt": "W3sieCI6OTY1LCJ5IjoxNjksInQiOjE2MjQ2MTE4MDY4Njh9LHsieCI6OTY1LCJ5IjoxNjksInQiOjE2MjQ2MTE4MDY5NDh9XQ=="
"instanceId": "zfcaptchaLogin"
"extend": "eyJhcHBOYW1lIjoiTmV0c2NhcGUiLCJ1c2VyQWdlbnQiOiJNb3ppbGxhLzUuMCAoTWFjaW50b3NoOyBJbnRlbCBNYWMgT1MgWCAxMF8xNV83KSBBcHBsZVdlYktpdC81MzcuMzYgKEtIVE1MLCBsaWtlIEdlY2tvKSBDaHJvbWUvOTEuMC40NDcyLjEwNiBTYWZhcmkvNTM3LjM2IiwiYXBwVmVyc2lvbiI6IjUuMCAoTWFjaW50b3NoOyBJbnRlbCBNYWMgT1MgWCAxMF8xNV83KSBBcHBsZVdlYktpdC81MzcuMzYgKEtIVE1MLCBsaWtlIEdlY2tvKSBDaHJvbWUvOTEuMC40NDcyLjEwNiBTYWZhcmkvNTM3LjM2In0="

通过 base64 解密 mt和 extend 得出解密的数值

# mt
[{"x":965,"y":169,"t":1624611806868},{"x":965,"y":169,"t":1624611806948}]
# extend
{"appName":"Netscape","userAgent":"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.106 Safari/537.36","appVersion":"5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.106 Safari/537.36"}

mt 为用户的点击行为，x为X轴上的值，y为Y轴上的值，t为时间戳。通过大量点击分析，发现x值最小值为 950，得出950 为 X轴的起点，y值随机无固定值。

extend 为请求头部内容

步骤二：滑动验证码图像分析，计算滑动距离x值

将图像灰度化，通过getpixel可以获取图像某一点的颜色值， 颜色值越高代表图像越浅，所以寻找纵向连续50个像素点均是 getpixel(x+1, y) > getpixel(x, y)（X轴=x 比 X轴=x+1 颜色浅）

并扫描图像，当x=130、扫描高度=50时，的颜色比x+1时深。

from PIL import Image
import matplotlib.pyplot as plt
import numpy as np

scanf_height= 50 # 扫描的高度
img = Image.open("zfcaptchaLogin.png")

def contrast(imgl, x, y,scanf_height):
   # 黄框颜色值比红框颜色值浅的个数
   count = 0
   for i in range(scanf_height):
       if imgl.getpixel((x+1, y+i)) > imgl.getpixel((x, y+i)):
           count += 1
   # 当 count = scanf_height, 代表黄条区域 整体 红条区域 颜色值浅，则是验证码框位置
   return count

def scanf(img):
   imgx, imgy = img.size
   imgl = img.convert('L') # 图像灰度化
   plt.yticks([])
   plt.xticks([i for i in range(0, imgx, 25)])
   plt.imshow(img)
   plt.pause(0.5)
   for y in range(0, imgy-scanf_height, 10):
       plt.pause(0.01)
       plt.clf()
       plt.yticks([])
       plt.xticks([i for i in range(0, imgx, 25)])
       plt.imshow(imgl, cmap=plt.cm.gray)
       for x in range(1, imgx-1, 1):
           plt.pause(0.0001)
           plt.plot([x-1,x-1], [y, y+scanf_height], color='white')
           plt.plot([x,x], [y, y+scanf_height], color='red')
           plt.plot([x+1,x+1], [y, y+scanf_height], color='yellow')
           count = contrast(imgl, x,y, scanf_height)
           plt.title('count: {}'.format(count) )

print("x,y=[{}, {}], 黄条区域值比红条区域颜色值浅的个数:{}".format(x,y, count))
           if count == scanf_height:
               return

scanf(img)
plt.show()

优化代码计算x，y值

import json
import random
import time
from io import BytesIO

from PIL import Image

class ZfCaptchaRecognit(object):
   def __init__(self, img_path):
       self.img = Image.open(img_path)

def _get_xy(self):
       # 计算 x,y 值
       def _is_dividing_line(img_l, x, y):
           for n in range(50):
               # 寻找纵向连续50个像素点均是 X=x 比 X=x+1 颜色深
               if y + n >= img_l.size[1] or x >= img_l.size[0] - 1:
                   return False
               if img_l.getpixel((x + 1, y + n)) - img_l.getpixel((x, y + n)) < 2:
                   return False
           return True

img_l = self.img.convert("L")
       for x in range(img_l.size[0]):
           for y in range(img_l.size[1]):
               if _is_dividing_line(img_l, x, y):
                   return (x, y)

def show_tag(self):
       # 展示 切分点
       X, Y = self._get_xy()
       img2 = Image.new("RGB", self.img.size, (255, 255, 255))
       for x in range(self.img.size[0]):
           for y in range(self.img.size[1]):
               pix = self.img.getpixel((x, y))
               img2.putpixel((x, y), pix)
               if x == X or y == Y:
                   img2.putpixel((x, y), 225)

img2.save("show_tag.png")
       img2.show()

captcha = ZfCaptchaRecognit("zfcaptchaLogin.png")
captcha.show_tag()

步骤三：生成提交参数

通过 步骤一得出x值最小为950，y值无规律

则提交参数mt的大致格式数据是

[{
   "x":950+ 滑动距离 + 浮动值,  #  浮动值的范围通过分析提交参数得出在10~20内
   "y":random.randint(150, 190),  # 无规律，暂定150到190范围内
   "t":int(time.time() * 1000)},  # 时间戳
...]

获取mt 参数

import json
import random
import time
from io import BytesIO

from PIL import Image

class ZfCaptchaRecognit(object):
   def __init__(self, img_stream):
       obj = BytesIO(img_stream)
       self.img = Image.open(obj)

def _get_xy(self):
       ...

def generate_payload(self):
       base_x = 950
       X, Y = self._get_xy()
       payloads = [{"x": base_x + random.randint(5, 20), "y": random.randint(150, 190), "t": int(time.time() * 1000)}]
       for i in range(random.randint(15, 30)):
           # 在上一个参数基础下浮动
           last_payload = payloads[-1].copy()
           payloads[0]["x"] += random.choice([0] * 8 + [1, -1] * 2 + [2, -2])
           last_payload["t"] += random.randint(1, 20)
           last_payload["y"] += random.choice([0] * 8 + [1, -1] * 2 + [2, -2])
           payloads.append(last_payload)

payloads[-1]["x"] = base_x + random.randint(10, 20) + X
       return json.dumps(payloads)

captcha = ZfCaptchaRecognit("zfcaptchaLogin.png")
captcha. generate_payload()

来源：https://blog.csdn.net/veloi/article/details/115416495