Text-Based Captcha Using Text-Adhesion and Visual Compensation
Dangquan Zeng · Proceedings of the 2020 4th International Conference on Electronic Information Technology and Computer Engineering · 2020
CAPTCHA (Completely Automated Public Turing Test to tell Computers and Humans Apart) is a test for distinguishing between computers and humans, and has a very wide range of applications on the Internet. Most websites require users to submit CAPTCHAs when registering to log in or submit some form data to improve the security of the site, thus avoiding malicious attacks by automated robots and spammers. In this paper, a text-based CAPTCHA Using text-adhesion and visual compensation is introduced. This CAPTCHA is designed to use character cascading technology and partial defect technology to effectively prevent the machine from using the character cutting and machine learning techniques to verify the CAPTCHA, but human vision can easily separate the characters that are layered together and can fill in some of the missing characters to easily identify the CAPTCHA. In order to test the ability of the CAPTCHA to resist automatic machine identification, four kinds of specialized OCR recognition software and two online OCR recognition websites were used to identify and verify 1000 CAPTCHAs. The test results show that none of the six OCR recognition tools can correctly identify one completely CAPTCHA(correctly recognized), and the probability of not being recognized and misidentified is over 99.5%, which proves that the CAPTCHA is a very high security technology.