Font Size: a A A

The Research And Study On Picture CAPTCHA

Posted on:2019-03-31Degree:MasterType:Thesis
Country:ChinaCandidate:G G WangFull Text:PDF
GTID:2416330566999460Subject:Applied statistics
Abstract/Summary:PDF Full Text Request
With the development of Internet technology,network information provides great convenience to human beings.However,the threat of network information security also increases.CAPTCHAs as a tool to differentiate between computer and human beings,make a significant contribution to the network information security,which restricting the breeding of online spam and enhancing the defensive power of cyberattacks.This article describes the current method of cracking different types of interference verification code from the point view of captcha character preprocessing,character segmentation and character recognition,and made relevant supplements to captcha character preprocessing,character segmentation.The main contents and achievements of this paper are as follows?Describe the basic flow of the CAPTCHA character and the basic techniques of CAPTCHA character processing.This paper takaes line interference items out of a specific verification code based on Huff transform and Canny edge detection on account of the fact that commonly used CAPTCHA character preconditioning techniques can not effectively handle character verification codes that contain line interference items,compared with the 6% recognition accuracy based on commonly used preprocessing method,the accuracy of the recognition result reaches 71.52%,which will greatly explain the superiority of this method.In the aspect of character segmentation,this paper proposes a character segmentation scheme based on the JiKe website login verification code.In this paper,the inclined area of the CAPTCHA is dealt with by tilting the connected area in view of the fact that the character size of the CAPTCHA is fixed,the characters of the CAPTCHA characters have rotated,tilted,and overlapping transformations,and achieves a better cutting result.For the existing vertical projection method,this method has the disadvantage of not suitable for some characters.For example,characters such as W,V contain troughs themselves,this article set the minimum length of the character is also called the threshold.If the trough spacing is less than the threshold,the next trough will be the projection position.This method can effectively make up for the lack of vertical projection method.In the aspect of character recognition,this paper first finds the center of each character and then uses the horizontal direction as the baseline,divides the character into eight parts at a 45-degree angle which is used as the character identification character,the method greatly improves the recognition efficiency.Although the method proposed in this paper has applicability,the verification code proposed in this paper belongs to a simple verification code,and the verification code recognition technology needs to be further improved.With the constant progress of CAPTCHA,only by mastering all kinds of recognition and verification code technology,can we know each other,the CAPTCHA generation technology also needs to be further improved to improve the network information security.
Keywords/Search Tags:CAPTCHA, recognize, Hough transform, Canny edge detection, Tesseract character recognition, Random forest
PDF Full Text Request
Related items