Explaining Human Visual Space Distortion.

A number of experiments have recently been conducted to compare aspects of depth judgment due to stereoscopic and monocular motion perception. In these experiments, it has been shown that from stereo vision humans over-estimate depth (relative to fronto-parallel size) at near fixations and under-estimate it at far fixations, whereas human depth estimates from visual motion are not affected by the fixation point. On the other hand, the orientation of an object in space does not affect depth judgment in stereo vision while it has a strong effect in motion vision, for the class of motions tested. This paper develops a computational geometric model that explains why such distortion might take place. The basic idea is that, both in stereo and motion, we perceive the world from multiple views. Given the rigid transformation between the views and the properties of the image correspondence, the depth of the scene can be obtained. Even a slight error in the rigid transformation parameters causes distortion of the computed depth of the scene. The unified framework introduced here describes this distortion in computational terms, in order to explain a number of recent psychophysical experiments on the perception of depth from motion or stereo.