Robust image matching based on rotation and scale invariant shape context

Abstract

In order to solve the problem of false matches generated from the similar region in image matching, In this paper, an algorithm for a robust image matching method based on the combination of the improved Shape context method and SIFT is proposed. First we extracted the key points and matching them using SIFT. Then we exploited the rotation and scale information from matched SIFT key points, K-means Clustering can estimate the ratio scale and rotation angle accurately and with which we achieve the rotation and scale invariant shape context. Finally The matches is obtained by Shape Context Matching. Experiment show that, the algorithm can improve the accuracy of matching.