Properties of Surface Reconstruction Based on Binocular Disparity and Motion Parallax

Binocular disparity and motion parallax are particularly important for our 3D surface perception, and the properties of perceived surfaces are similar when these cues are presented separately. Moreover, when we perceive surfaces on random dot patterns, we can perceive smooth surfaces even though the dot density may be low. That is, we reconstruct a surface by interpolating areas between dots. In this research, we used random dot patterns in which we can perceive 3D surfaces by disparity or parallax. We made a region without dots (a gap) on the surface, and investigated the surface properties perceived. The experimental results were then compared with output of a computational model based on the standard regularization theory. We conclude as follows; (1) The ability for surface reconstruction due to binocular disparity is more effective than that by parallax. (2) The experimental data corresponded well to a model which uses the second order differential of depth data. This indicates the usage of the curvature of the disparity or parallax in the surface perception. (3) The gap width difference between disparity and parallax can be explained by considering the magnitude of the receptive fields in the neural pathways where disparity and parallax are processed.