Methods of complete surface reconstruction through merging of point clouds according to stereo vision data

This article concerns the stack of algorithms which can be applied to the task of 3d reconstruction using stereo vision techniques. Stereo vision data represented as a set of single-valued surfaces of the point clouds that have overlaping areas. They may differ in density and regularity, but the whole reconstructed object model consists of a subset of single-valued surfaces. Thus, the task is to find a method of transformation of these original surfaces by minimizing some functional chracterizing the degree of matching. Considered the methods of representation of the point cloud as two-dimensional discrete functions, suggested variants of the measure of differences between source datasets, as well as methods of mutual localization of surfaces.