Presentation Information
[A-7-03]Proposal of a Munker CAPTCHA Employing a Text-Input Interface
◎△Murai Koki1, Murakami Yasuyuki1, Doi Motomori1 (1. Osaka Electoro-Communication University)
Keywords:
CAPTCHA,Munker illusion,SUS
In recent years, security threats from BOTs have increased rapidly, making CAPTCHA essential. However, advances in AI allow BOTs to breach conventional CAPTCHAs, while they become harder for humans to recognize.
To solve this trade-off, using human-specific optical illusions has attracted attention. Particularly, CAPTCHAs using the Munker illusion have been proposed. However, conventional methods have a limited number of pre-prepared images, making them vulnerable to brute-force or machine-learning attacks due to fewer problem patterns.
Therefore, this study proposes a novel CAPTCHA method that changes the objects in the illusion image from conventional circles to characters, requiring users to input the character of a specified color. This approach prevents the depletion of patterns and hinders AI recognition while maintaining human usability. Furthermore, we developed a prototype and conducted a usability survey using the System Usability Scale (SUS). This paper reports on the system design and evaluation results.
To solve this trade-off, using human-specific optical illusions has attracted attention. Particularly, CAPTCHAs using the Munker illusion have been proposed. However, conventional methods have a limited number of pre-prepared images, making them vulnerable to brute-force or machine-learning attacks due to fewer problem patterns.
Therefore, this study proposes a novel CAPTCHA method that changes the objects in the illusion image from conventional circles to characters, requiring users to input the character of a specified color. This approach prevents the depletion of patterns and hinders AI recognition while maintaining human usability. Furthermore, we developed a prototype and conducted a usability survey using the System Usability Scale (SUS). This paper reports on the system design and evaluation results.