RECOVERING 3-DIMENSIONAL STRUCTURE-FROM-MOTION WITH SURFACE RECONSTRUCTION

This paper addresses the computational cole that the construction of a complete surface representation may play in the recovery of 3-D struc ture from motion. We first discuss the need to integrate surface recon struction with the structure-from-motion process, both on computationa l and perceptual grounds. We then present a model that combines a feat ure-based structure-from-motion algorithm with a smooth surface interp olation mechanism. This model allows multiple surfaces to be represent ed in a given viewing direction, incorporates constraints on surface s tructure from object boundaries, and segregates image features onto mu ltiple surfaces on the basis of their 2-D image motion. We present the results of computer simulations that relate the qualitative behavior of this model to psychophysical observations. In a companion paper, we discuss further perceptual observations regarding the possible role o f surface reconstruction in the human recovery of 3-D structure from m otion.