Optimization and parallelization of a geographical stereo vision code

This paper describes the sequential and parallel versions of a stereo vision algorithm. In this study, we address the particular stereo vision problem of accurate Digital Elevation Models (DEMs) reconstruction from a pair of images of the SPOT satellite. We start from an existing algorithm made by M.Memier. we optimize it while focusing on the cross-correlation problem based on a statistical operator. And we propose a parallel implementation with special care to load balancing