A Two-Layer Descriptor and Two-Step Matching for Stereoscopic Images

Abstract

Among all the feature matching algorithms, SURF is famous for its computation efficiency in lots of practical applications. For recently popular stereoscopic videos, sparse feature matching, in which SURF is usually adopted, is the pre-processing of adjusting the 3D effect. However, significant improvement still can be made for SURF in stereo images. In this paper, a new two-layer feature descriptor and the corresponding two-step matching method which is based on the stereo images is proposed. Experimental results show that the proposed algorithm achieves comparable or even better performance of matching at much less time compared against the SURF in stereo images.