Analysis of CAPTCHA decoding

Abstract

CAPTCHAs are short for Completely Automated Public Turing test to tell Computers and Humans Apart which is now almost a standard security technology to distinguish whether a client is human or not. The most widely used CAPTCHAs rely on the distortion of text images rendering them unrecognizable to the state of the art of pattern recognition techniques, and these text-based schemes have found widespread applications in commercial or educational websites. A systematic analysis of existing text-based CAPTCHAs is carried out based on distorted characteristics. Also, I design an application that solving CAPTCHA by applying filtering techniques and recognizing by Optical Character Recognition. Moreover, applying a systematic evaluation to current CAPTCHA schemes from popular Hong Kong or international websites, some of them can be vulnerable to attack automatically. Based on these evaluations, I identify a series of suggestions for CAPTCHA designers and computer analyser, and possible future directions for producing more reliable CAPTCHA to distinguish human and computer analysers.