An improved matching algorithm for feature points matching

Abstract

Binocular stereo matching is always hot and difficult issues in the machine vision, the advantages and disadvantages of feature stereo matching method is analyzed in the paper, and an improved feature stereo matching method is proposed. First, using the SIFT feature extracting algorithm to extract basic feature points, utilizing these feature points as the starting points, and then finding the similar feature points around them to form the feature point block; second, using the matching algorithm to match, and then analyzing the mathematical morphology, repairing the disparity map to get the final disparity map. The experimental results show that: the mismatching that is brought by feature matching can be solved effectively through the method, and the problems such as the lacks of detail texture in the matching process are solved, a more accurate and clear disparity map can be got.