Abstract
A CAPTCHA is a test that can, automatically, tell human and computer programs apart. It is a mechanism widely used nowadays for protecting web applications, interfaces, and services from malicious users and automated spammers. Usability and robustness are two fundamental aspects with CAPTCHA, where the usability aspect is the ease with which humans pass its challenges, while the robustness is the strength of its segmentation-resistance mechanism. The collapsing mechanism, which is removing the space between characters to prevent segmentation, has been shown to be reasonably resistant to known attacks. On the other hand, this mechanism drops considerably the human-solvability of text-based CAPTCHAs. Accordingly, an optimizer has previously been proposed that automatically enhances the usability of a CAPTCHA generation without sacrificing its robustness level. However, this optimizer has not yet been evaluated in terms of improving the usability. This paper, therefore, evaluates the usability of this optimizer by conducting an experimental study. The results of this evaluation showed that a statistically significant enhancement is found in the usability of text-based CAPTCHA generation.