Stereovision matching through support vector machines

Abstract

This paper presents an approach to the local stereovision matching problem using edge segments as features with four attributes. In this paper we design a Support Vector Machine classifier for solving the stereovision matching problem. We obtain a matching decision function to classify a pair of features as a true or false match. The use of such classifier makes up the main finding of the paper. A comparative analysis among other existing approaches is included to show that this finding can be justified theoretically. From these investigations, we conclude that the performance of the proposed method is appropriate for this task.