第二十三节：scrapy爬虫识别验证码（二）图片验证码识别

 1 # -*- coding:utf-8 -*-
 2 import tesserocr
 3 from PIL import Image
 4 import requests
 5 
 6 # 通过url链接获取验证码图片，并写入本地文件夹里
 7 def get_image(path,url):
 8     """
 9     :param path: 文件夹路径
10     :param url:  验证码url链接
11     """
12     respon = requests.get(url=url)      # 请求验证码url
13     with open(path,"wb") as file:
14         file.write(respon.content)      # 将验证码写到本地
15 
16 
17 # 由于验证码图片太小，需要对验证码图片放大处理，以便识别
18 def reset_image_size(image_path):
19     """
20     :param image_path: 图片所在的路径
21     :return:
22     """
23     image = Image.open(fp=image_path)   # 打开图片
24     pic_resize = 5                      # 设置图片放大或者缩小倍数
25     (x, y) = image.size                 # 获取图片的大小
26     x_s = int(x * pic_resize)           # 放大5倍(可调)
27     y_s = int(y * pic_resize)           # 放大5倍(可调)
28     out = image.resize((x_s, y_s), Image.ANTIALIAS)     # ANTIALIAS表示高质量图片
29     out.save(image_path)
30 
31 
32 # 读取验证码图片文本
33 def read_image(image_path):
34     """
35     :param image_path: 验证码图片路径
36     :return:
37     """
38     image = Image.open(fp=image_path)       # 打开验证码图片
39     image = image.convert(‘L‘)              # 将验证码图片转换为灰度图(L表示灰度图)
40     threshold = 127                         # 设置灰度图二值化阈值
41     table = []
42     for i in range(256):                    # 像素为256
43         if i < threshold:
44             table.append(0)
45         else:
46             table.append(1)
47     image = image.point(table, ‘1‘)         # 二值化处理后的副本(1表示二值化)
48     image.show()
49     result = tesserocr.image_to_text(image) # 验证码图片转换为文本
50     return result
51 
52 
53 # 验证码识别信息脏数据处理
54 def VerifInfo(result):
55     """
56     :param result: 验证码图片通过初步识别后得到的脏数据
57     :return:
58     """
59     verif_str = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ1234567890"
60     verif_list = []
61     for i in result:
62         if i in verif_str:
63             verif_list.append(i)
64     return "".join(verif_list)
65 
66 
67 
68 if __name__ == ‘__main__‘:
69     img_path = "D:\photo\image"                                 # 文件夹目录
70     img_path = img_path + "\VerificationCode.png"               # 验证码图片所在的目录及名称
71     img_url = "http://my.cnki.net/elibregister/CheckCode.aspx"  # 验证码url
72     get_image(img_path,img_url)                                 # 获取验证码图片
73     reset_image_size(img_path)                                  # 调整验证码图片大小
74     result = read_image(img_path)                               # 读取验证码图片内容
75     verif_info = VerifInfo(result)                              # 验证码内容数据处理
76     verif_len  = len(verif_info)                                # 验证码识别长度
77     if verif_len == 4 and verif_info:
78         print(verif_info)
79     else:
80         pass