A Novel Method for Stereo Matching Using Gabor Feature Image and Confidence Mask

Haixu Liu,Yang Liu,Shuxin OuYang,Chenyu Liu,Xueming Li
DOI: https://doi.org/10.1109/vcip.2013.6706388
2013-01-01
Abstract:In this paper, we present a novel local-based algorithm for stereo matching using Gabor-Feature-Image and Confidence-Mask. Various local-based schemes have been proposed in recent years, most of them mainly use color difference as evaluation criterion when constructing the initial cost volume, however, color channel is highly sensitive to noise, illumination changes, etc. Therefore, we develop a new cost function based on Gabor-Feature-Image for obtaining a more accurate matching cost volume. Furthermore, in order to eliminate the matching ambiguities brought by the winner-takes-all method, an effective disparity refinement strategy using Confidence-Mask is implemented to select and refine the less reliable pixels. The proposed algorithm ranks 23th out of over 150 (global-based and local-based) methods on Middlebury data sets, both quantitative and qualitative evaluation show that it is comparable to state-of-the-art local-based stereo matching algorithms.
What problem does this paper attempt to address?