A study on wavelet-based edge detection for improved stereo matching

Abstract

Development of stereopsis technology that gives reality is an interesting research area for a long time. Stereopsis technology expresses actuality with effect of two eyes. Correct stereo matching depends on the exact estimation of image disparity that contains depth information. Wavelet transform can be applied to analyze signals in stereo vision, but related research is insufficient. In this paper, we propose a fast and correct stereo image matching method through edge detection that uses multi-resolution wavelet transform. Also, we analyze the effect of noise during the edge detection process, and make a comparison with traditional methods.