A Binocular Stereopsis Algorithm Based on Gaussian Markov Random Field Model

Abstract

In this paper we discuss problems of traditional stereopsis algorithms, and propose a novel stereopsis model. This model detects disparity on edge sites and, at the same time, interpolate between them. The interpolation algorithm is based on Gaussian Markov Random Field Model which makes MRF computation easy. An experiment result shows that this model extracts disparity even on no edge sites.