A Benchmark for Designing Usable and Secure Text-based Captchas

An automated public Turing test to distinguish between computers and humans known as CAPTCHA is a widely used technique on many websites to protect their online services from malicious users. Two fundamental aspects of captcha considered in various studies in the literature are robustness and usability. A widely accepted standard benchmark, to guide the text-based captcha developers is not yet available. So this paper proposes a benchmark for designing usable-secure text-based captchas based on a community driven evaluation of the usability and security aspects. Based on this benchmark, we develop four new textbased captcha schemes, and conduct two separate experiments to evaluate both the security and usability perspectives of the developed schemes. The result of this evaluation indicates that the proposed benchmark provides a basis for designing usable-secure text-based captchas.