3D CAPTCHA Generation based on Ternary Tree Bounding Volume Hierarchy

CAPTCHA primarily serves to distinguish between real humans and automated programs, such as bots. However, rapid advancements and growing complexity in artificial intelligence present significant threats to CAPTCHA security. This research seeks to address these concerns by developing a 3D CAPTCHA system based on a ternary bounding volume hierarchy (BVH). The proposed 3D CAPTCHA is designed to be easily recognizable by humans while posing substantial challenges for AI systems. In this study, 3D objects were approximated with bounding volumes, and a video-based Turing test was implemented by introducing noise to the images. Consequently, the AI recognition rate was significantly reduced, effectively overcoming limitations of traditional CAPTCHA systems. This method adds an enhanced layer of security, fortifying the existing CAPTCHA system against vulnerabilities brought on by advancing AI technology.