Estimation of depth in stereo images using hybrid segmentation approaches

Abstract

Segmentation is the process by which image is divided in to certain parts or regions in such a way that each part and region share a distinct property in common. This paper compare various segmentation approaches applicable to gray scale images. A comprehensive analysis has been done to choose best power and time efficient scheme from available state of art algorithms. SAD approach along with heuristic based segmentation as well as Fuzzy based clustering methods has been used for Depth estimation of stereo images. A case of selective blurring is also introduced that can be applied to remove undesired portion of images.