An asymmetric post-processing for correspondence problem

This paper presents a novel approach that performs post-processing for stereo matching. We improve the performance of stereo matching by performing consistency check and adaptive filtering in an iterative filtering scheme. The consistency check is only done with asymmetric information so that very few additional computational loads are necessary. The information in the valid pixels is propagated into invalid pixels through the adaptive filtering. The proposed post-filtering method can be used in various methods for stereo matching. We demonstrate the validity of the proposed method by applying it to hierarchical belief propagation and semi-global matching. The performance of the post-processing method for hierarchical belief propagation is comparable to state-of-the-art methods in the Middlebury stereo datasets. In order to verify the performance of asymmetric consistency check, we compare it with other reliability estimation methods in the proposed post-processing scheme. Moreover, in order to verify the performance of post-filtering method in noisy environment, the proposed post-filtering method is applied to the stereo images denoised by NL-means algorithm. We find that the iterative filtering scheme reduce an error which may be caused in stereo matching for the denoised images and improve the performance of stereo matching.