CAPTCHA System by Differentiating the Awkwardness of Objects

The "Completely Automated Public Turing test to tell Computers and Humans Apart" (CAPTCHA) is a technique that prevents unauthorized access by bots. Most studies of CAPTCHA systems use human cognitive capacities as a countermeasure to facilitate recognition techniques. Differentiating between natural and awkward objects is an approach used to distinguish humans from bots. However, this approach is vulnerable to adversaries who exploit the differences in relative frequency between natural and awkward objects because of the difficulty in collecting natural objects. In this study, we propose a new scheme that does not require the utilization of natural objects, thereby addressing this shortcoming. Our proposed method requires that humans always distinguish awkward objects, which are generated by different parameters. We evaluated our scheme in several experiments.