Design of a Variable-length Accented Character-based CAPTCHA System

Abstract

Completely Automated Public Turing test to Tell Humans and Computers Apart (CAPTCHA) is aimed at distinguishing humans from computers by using text, images, audio, or video. Text-based CAPTCHAs, are the most common but face vulnerabilities due to their limited use of latin characters. This study enhanced the design of the NaijaCAPTCHA by generating variable-length accented character-based CAPTCHA codes, utilising latin and accented characters to bolster online transaction security. The design was implemented using Javascript, PHP, HTML, and CSS. The enhanced NaijaCAPTCHA comprised of four modules: CAPTCHA generator, obfuscator, display unit, and database. The generator employed the Gimpy algorithm to create codes of varying lengths (4 to 7 characters) with at least two accented characters. The obfuscator manipulated the code's appearance through color, text distortion, background noise, and skewing. User presentation and authentication occurred in the display unit, with code correctness verified against the obfuscated value. Thirty CAPTCHA categories were generated from the combination of background and text modification. The enhanced NaijaCAPTCHA system presents a formidable challenge to bots, suggesting its potential adoption for government websites and transactional purposes.