Image CAPTCHA: Based on human understanding of real world distances

Abstract

We propose a Completely Automated Public Turing Test to Tell Computers and Humans Apart scheme based on a human's understanding of real world objects and their relative distances. Solving this CAPTCHA involves answering about the relative distances of objects present in an image. What is subtle to humans may not be so to computing machines. As an object moves farther away from the point of perception, its size decreases. But equipped with only the information on size, a decision cannot be made on its position from the point of perception. The real world knowledge on the object dimensions is essential since both a big and small object near the perception point will appear big and cannot be differentiated. This ability of humans, aided with experience, is a unique feature and would thus serve to distinguish them from computers. The queries on the CAPTCHA may include - "who stands behind whom", "which is nearer" or "which is larger in real life".