进行逻辑验证码码的破解识别

1.验证码/PIL_EXAMPLES/rotate.py

@@ -1,29 +1,31 @@
 # coding:utf-8
 from __future__ import print_function
-from PIL import Image
-import glob, os
+#from PIL import Image
+import Image
+#import glob, os
 
 
 
-size = 128,128
+#size = 128,128
 # 打开图片
-im = Image.open("1.png")
+#im = Image.open("11.png")
 # 旋转45度 并且 显示
 
 # im.rotate(45).show()
 
 # 遍历该目录下的所有PNG文件，打开并且创建略缩图
-for infile in glob.glob("*.png"):
-    file,ext = os.path.splitext(infile)
-#    print file,ext
-    im = Image.open(infile)
-    im.thumbnail(size,Image.ANTIALIAS)
-    #im.save(file+".thumbnail","JPEG")
+# for infile in glob.glob("*.png"):
+#     file,ext = os.path.splitext(infile)
+# #    print file,ext
+#     im = Image.open(infile)
+#     im.thumbnail(size,Image.ANTIALIAS)
+#     #im.save(file+".thumbnail","JPEG")
 
 # 创建新图片，并保存
-new_img = Image.new("RGB",(512,512),"white")
+# 101 31
+#new_img = Image.new("RGB",(101,31),"white")
 #new_img.save("NEW.png")
-im_2 = Image.open("1.1.png")
+# im_2 = Image.open("1.1.png")
 #im_3 = Image.composite(im.copy(),im_2.copy(),'L')
 
 # 将im与im_2 根据透明度进行合并
@@ -37,17 +39,14 @@
 #输出im的最大最小rgb
 #print im_2.getextrema()
 
-im = Image.open('6.png')
-im = im.convert('L')
-im = im.resize((32,32),Image.ANTIALIAS)
-im.save('6_L.png')
-
-#print im.getpixel((0,0))#灰度图没有rgb直接是统一值
-#print im.size
+im = Image.open('11.png')
+#im = im.convert('L')
+#im = im.resize((32,32),Image.ANTIALIAS)
+#im.save('11_l.png')
 for i in xrange(im.size[0]):
     for j in xrange(im.size[1]):
 
-        #print im.getpixel((i,j))
+        print im.getpixel((i,j))
         l = im.getpixel((j,i))
         if( l > 220):
             print ('0',end='')

1.验证码/num/recognise.py

@@ -3,7 +3,7 @@
 import os
 import requests
 from PIL import Image
-import math
+import math,time
 
 def imagesget():
     os.mkdir('images')
@@ -24,7 +24,7 @@ def convert_image(image):
             pix=image.getpixel((x,y))
             if pix<120:#灰度低于120 设置为 0
                 image2.putpixel((x,y),0)
-    image2.save('L.jpeg')#将灰度图存储下来看效果
+    image2.save('L.png')#将灰度图存储下来看效果
     return image2
 
 def cut_image(image):
@@ -86,12 +86,14 @@ def magnitude(self,concordance):
 
     #计算矢量之间的 cos 值
     def relation(self,concordance1, concordance2):
-            
+
         relevance = 0
         topvalue = 0
         for word, count in concordance1.items():
             if word in concordance2:
+                print type(topvalue),topvalue,count,concordance2[word]
                 topvalue += count * concordance2[word]
+                time.sleep(10)
         return topvalue / (self.magnitude(concordance1) * self.magnitude(concordance2))
 
     def recognise(self,image):
@@ -119,7 +121,8 @@ def recognise(self,image):
 if __name__=='__main__':
     imageRecognize=CaptchaRecognize()
     # 设置图片路径
-    image=Image.open('3.jpeg')
+    image=Image.open('3.png')
+    print image.mode
 
     result=imageRecognize.recognise(image)
 

1.验证码/num副本/README.md

@@ -0,0 +1,12 @@
+本部分测试了简单的图像验证码的处理
+
+主要代码为:recognise.py
+
+在main中可以替换为你需要解析的图片路径与名称,
+
+产生L.png是进行了灰度以及二值转换之后的图片在本目录下
+
+还会产生字符切割的效果
+
+用户可以将切割出来的字符人工标记后加入到icon中成为新字符集合.
+

1.验证码/num副本/login.py

@@ -0,0 +1,49 @@
+#coding:utf-8
+
+import requests
+from recognise import *
+from PIL import Image
+import base64
+import getpass
+
+
+def login(username,passwd):
+    session=requests.session()
+    session.get('http://wsxk.hust.edu.cn/login.jsp').text
+    img=session.get('http://wsxk.hust.edu.cn/randomImage.action').content
+    with open('captcha.jpeg','wb') as imgfile:
+        imgfile.write(img)
+    imageRecognize=CaptchaRecognize()
+    image=Image.open('captcha.jpeg')
+    result=imageRecognize.recognise(image)
+    string=''
+    for item in result:
+        string+=item[1]
+    print(string)
+    data={
+    'usertype':"xs",
+    'username':username,
+    'password':passwd,
+    'rand':string,
+    'sm1':"",
+    'ln':"app610.dc.hust.edu.cn"
+    }
+    headers = {
+        'Host':"wsxk.hust.edu.cn",
+        "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8",
+        "Accept-Encoding": "gzip, deflate",
+        "Accept-Language": "en-US,en;q=0.5",
+        "Connection": "keep-alive",
+        'Referer':"http://wsxk.hust.edu.cn/login.jsp",
+        "User-Agent": "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:39.0) Gecko/20100101 Firefox/39.0"}
+    session.post('http://wsxk.hust.edu.cn/hublogin.action',data=data,headers=headers)
+    html=session.get('http://wsxk.hust.edu.cn/select.jsp',headers=headers).text
+    print(html)
+    return session
+
+def main():
+    username=input('username:')
+    passwd=base64.b64encode(getpass.getpass('Passwd:').encode()).decode()
+    login(username,passwd)
+
+main()

1.验证码/num副本/recognise.py

@@ -0,0 +1,144 @@
+#coding:utf-8
+
+import os
+import requests
+from PIL import Image
+import math,time
+
+def convert_image(image):
+    image=image.convert('L')#灰度
+    image2=Image.new('L',image.size,255)
+    for x in range(image.size[0]):
+        for y in range(image.size[1]):
+            pix=image.getpixel((x,y))
+            if pix<120:#灰度低于120 设置为 0
+                image2.putpixel((x,y),0)
+    image2.save('L.png')#将灰度图存储下来看效果
+    return image2
+
+def cut_image(image):
+    ''' 字符切割,根据黑色的连续性,当某一列出现黑色为标志,当黑色消失为结束点'''
+    inletter=False
+    foundletter=False
+    letters=[]
+    start=0
+    end=0
+    for x in range(image.size[0]):
+        for y in range(image.size[1]):
+            pix=image.getpixel((x,y))
+            if(pix==0):
+                inletter=True
+        if foundletter==False and inletter ==True:
+            foundletter=True
+            start=x
+        if foundletter==True and inletter==False:
+            end=x
+            letters.append((start,end))
+            foundletter=False
+        inletter=False
+    images=[]
+    for letter in letters:
+        img=image.crop((letter[0],0,letter[1],image.size[1]))
+        #img.save(str(letter[0])+'.png')#展示切割效果
+        img.save("./cat/"+str(int(time.time()))+'.png')#展示切割效果
+        images.append(img)
+    return images
+
+def buildvector(image):
+    ''' 图片转换成矢量,将二维的图片转为一维'''
+    result={}
+    count=0
+    for i in image.getdata():
+        result[count]=i
+        count+=1
+    #print result
+    return result
+
+
+class CaptchaRecognize:
+    def __init__(self):
+        self.letters=['0','1','2','3','4','5','6','7','8','9']
+        self.loadSet()
+
+    def loadSet(self):
+        self.imgset=[]
+        for letter in self.letters:
+            temp=[]
+            for img in os.listdir('./icon/%s'%(letter)):
+                temp.append(buildvector(Image.open('./icon/%s/%s'%(letter,img))))
+            self.imgset.append({letter:temp})
+
+    #计算矢量大小
+    def magnitude(self,concordance):
+        total = 0
+        for word,count in concordance.items():
+            try:
+                if(type(count) == type(())):
+                    total += count[0] ** 2
+                #print type(total),total,type(count),count ** 2
+                else:
+                    total += count ** 2
+            except Exception,e:
+                print type(total),total,type(count),count
+                print e
+        return math.sqrt(total)
+
+    #计算矢量之间的 cos 值
+    def relation(self,concordance1, concordance2):
+        relevance = 0
+        topvalue = 0
+        for word, count in concordance1.items():
+            if word in concordance2:
+                print type(topvalue),topvalue,count,concordance2[word]
+                time.sleep(1)
+                topvalue += count * concordance2[word][0]
+                #time.sleep(10)
+        return topvalue / (self.magnitude(concordance1) * self.magnitude(concordance2))
+
+    def recognise(self,image):
+        image=convert_image(image)#二值化
+        images=cut_image(image)#字符单独切割出来
+        vectors=[]
+        for img in images:
+            vectors.append(buildvector(img))
+        result=[]
+        for vector in vectors:
+            guess=[]
+            for image in self.imgset:
+                for letter,temp in image.items():
+                    relevance=0
+                    num=0
+                    for img in temp:
+                        relevance+=self.relation(vector,img)
+                        num+=1
+                    relevance=relevance/num
+                    guess.append((relevance,letter))
+            guess.sort(reverse=True)
+            result.append(guess[0])
+        return result
+
+if __name__=='__main__':
+    import os
+    dir="./temp"
+    name_list = []
+    for root,dirs,files in os.walk(dir):
+        for file in files:
+            #name_list.append(file)
+            name = os.path.join(root,file)
+            name_list.append(name)
+
+    print name_list
+    for i in name_list:
+        #name = '11'
+        name = i
+        print name
+
+        imageRecognize=CaptchaRecognize()
+        # 设置图片路径
+        image=Image.open(name)
+        #image=Image.open('./temp/2.png')
+        print image.mode
+        result=imageRecognize.recognise(image)
+        string=[''.join(item[1]) for item in result]
+        print(string)
+        break