The Research And Implementation Of Captcha Recognition System

Internet security issues, such as auto account registration, account signing in, malicious voting, email spamming and password guessing with robot programs, become increasingly prominent as further development of Internet applications all over the world with various resources it provides in our daily life.CAPTCHA is widely adopted by lots of websites to avoid such Internet security problems by distinguishing robot programs from real world human users due to its simplicity, easy to deploy and low cost of data transmission. The nature of CAPTCHA is a picture of randomly generated numbers, letters and other symbols with a few disrupting pixels that can prevent OCR recognition. With CAPTCHA protection, user can't access the protected functions unless he passes the CAPTCHA verification by recognizes the information in CAPTCHA and filling out the CAPTCHA from correctly.This paper introduces a CAPTCHA recognition method based on neural networks and genetic algorithms, which can be used to verify the effectiveness of CAPTCHA in prevention of website security problems, promoting the development of CAPTCHA generation technique. It includes:1. briefing the CAPTCHA recognition technique and its current status of research.2. describing CAPTCHA image acquisition and processing, including graying, binarization, denoising, dividing and normalization.3. describing CAPTCHA image recognition, including characteristic extraction, training and recognition of processed image symbols.4. discussing the prospects of CAPTCHA development and its affect to the Internet security issues.