Depth perception under motion and stereo with implications for 3D TV

Abstract

This paper presents a theoretical model as well as relevant psychophysical studies to understand the impact that errors in the motion and stereo estimates have on shape perception. With the confidence level of the shape estimates recovered using motion or stereo under various scene configurations ascertained, a 3D TV system can adaptively select the resolution level of the 3D information to be presented to the viewers. Depth cues not effective and aspect of shape not perceivable under a particular configuration can be omitted with little or no apparent adverse effect on the quality of the 3D percept.