网络中舆论战的验证码技术研究

发布时间：2018-05-18 15:02

本文选题：验证码分割 + 复杂网络　；参考：《南京航空航天大学》2014年硕士论文

【摘要】：随着互联网的迅猛发展，网络给人们的生活带来了极大方便，同时网络上的安全问题也日益突出。验证码作为一种区分机器和人类的手段，已广泛应用于网络安全等领域。目前，基于字符的验证码识别研究已经越来越成熟，而验证码分割由于起步比较晚并且验证码中的字符可变性也比较大，使得验证码分割比验证码识别研究更为困难。因此，越来越复杂的验证码分割研究已经成为当前最为棘手的问题，但研究验证码分割技术在增强网络的安全性，防止对网站的恶意攻击等方面有着非常重要的意义。本文主要针对粘连字符验证码进行分析研究，并将研究结果与不同的算法进行对比，使得该研究具有一定的实际应用价值和作用。主要的工作和研究成果如下： 1）针对粘连字符验证码的特点，，提出了基于复杂网络的社区划分分割算法，并将不同的粘连字符个数分割结果进行分析对比。实验结果表明该算法能够有效的分割粘连字符，对于Authorize、京东商城、天涯社区、Windows Live和淘宝网验证码的分割成功率分别能达到98%、95%、71%、55%、33%。同时，随着粘连字符的增多，字符分割成功率随之降低，分割时间也随之增长。 2）针对天涯社区验证码的特点设计出了基于蓄水池的分割算法，并与不同的分割方法进行了对比，实验结果表明该方法的分割成功率能达到92%，并且对于验证码中有字符重叠或者两字符之间像素点相差过大情况的分割非常有效。该方法同时也解决了社区划分分割算法的局限性，有效地缩短了字符分割时间。 3）针对验证码的识别提出了33个特征提取的方法，并用C支持向量分类机对其进行识别，其实验结果表明该方法对天涯社区验证码字符的识别率能够达到93%。最后，介绍了本文从验证码获取、图片预处理到字符分割以及字符识别的系统实现。
[Abstract]:With the rapid development of the Internet, the network has brought great convenience to people's life, and the security problems on the network have become increasingly prominent. As a means of distinguishing machine from human, CAPTCA has been widely used in network security and other fields. At present, the research of character based verification code recognition has become more and more mature, and because of the relatively late start and the character variability in the verification code, it is more difficult to segment the verification code than the research of the verification code recognition. Therefore, more and more complex research on CAPTC-code segmentation has become the most difficult problem at present, but the research of CAPTC-code segmentation technology in enhancing the security of the network, preventing malicious attacks on websites and other aspects has a very important significance. This paper focuses on the analysis and research of the adhesive character verification code, and compares the research results with different algorithms, which makes the research have a certain practical application value and function. The main findings and findings are as follows: 1) according to the characteristics of the adhesive character verification code, a community partition segmentation algorithm based on complex network is proposed, and the results of the different number of adhesion characters are analyzed and compared. The experimental results show that the algorithm can effectively segment the adhesive characters. For Authorize, JingDong Mall, Tianya Community Windows Live and Taobao, the success rate of the segmentation of the verification codes can reach 98%, 95% and 71% and 5555%, respectively. At the same time, with the increase of adhesion characters, the success rate of character segmentation decreases and the segmentation time increases. 2) according to the characteristics of Tianya community verification code, a segmentation algorithm based on cistern is designed and compared with different segmentation methods. The experimental results show that the method can achieve a success rate of 92 and is very effective for the segmentation where there is overlap of characters in the verification code or the pixel difference between two characters is too large. The method also solves the limitation of community partitioning algorithm and effectively shortens the time of character segmentation. 3) A method of 33 feature extraction is proposed for the identification of CAPTC-code, and it is recognized by C support vector classifier. The experimental results show that the recognition rate of this method for the character of verification code in Tianya community can reach 933%. Finally, this paper introduces the implementation of the system from the acquisition of verification code, image preprocessing to character segmentation and character recognition.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.08

【参考文献】