Boltzmann machines for depth recovery using a MRF model

Abstract

The authors deal with the problem of depth recovery or surface reconstruction from sparse and noisy range data. Based on earlier insights from Markov random field models for such data, a Boltzmann machine is proposed for the parallel computation of the maximum a posteriori (MAP) estimate of the data. A new consensus function is developed to effectively detect discontinuities in highly sparse and noisy images. Interpolation over missing data sites is first done using only local characteristics of the network. Simulation results are also presented. >