실시간 스테레오 동영상 처리를 위한 빠른 디스패리티 추정

Fast disparity estimation for real-time stereo video systems

In stereo vision applications, disparity estimation is often performed to corresponding pixels. Using window-based correlations is a fast and standard approach to the disparity estimation. In this paper, we analyse the behaviour of the correlation-based disparity estimation and improve its performance by combining it with a segmentation scheme. Simulation results demonstrate that the proposed algorithm provides faithful disparity maps.