›› 2012, Vol. 25 ›› Issue (6): 149-.

• 论文 • 上一篇    下一篇

基于字符特征的文本验证码破解

高原   

  1. (江苏省邳州市机动车综合性能检测站,江苏 邳州 221300)
  • 出版日期:2012-06-15 发布日期:2012-08-23
  • 作者简介:高原(1970—),男,工程师。研究方向:计算机安全。

A Method for Breaking the Text-based CAPTCHA with Character Features

 GAO Yuan   

  1. (Pizhou City Motor Vehicle Test Station,Pizhou 221300,China)
  • Online:2012-06-15 Published:2012-08-23

摘要:

文中使用基于字符特征识别方法对基于文本图像的验证码进行破解研究。以某银行的验证码为例,介绍基于文本图像验证码的识别和破解过程。具体分为验证码预处理、字符提取和字符识别3个阶段。在识别阶段,分别使用颜色填充法、三线条法、全面扫描法的字符特征识别方法,在准确率和速度上都取得了理想的效果,对验证码的识别正确率达78%。文中所提字符特征识别和验证码破解方法,同样适用于其他非粘连的基于文本图像的验证码。

关键词: 验证码, 特征识别, 颜色填充, 三线条法, 全面扫描

Abstract:

This paper introduces a method for breaking the text-based image CAPTCHA based on feature recognition.It also describes the procedures of breaking SPD Bank CAPTCHA in details:preprocessing,extraction,and recognition.In the recognition phase,we propose three feature-based recognition methods for SPD Bank:color-filling,three-lines and full scan.The success rate and the speed of the results are desirable.The experiments show that our attack achieves a success rate of as high as 78%.Both the feature recognition method and the crack method are also applicable to other non-adhesion of the text-based CAPTCHA.

Key words: CAPTCHA;feature recognition;color-filling;three-line;full scan

中图分类号: 

  • TP398.08