Stereo Orthogonal Feature-mapping transform (SOFT) for stereo matching

This paper presents a novel stereo matching algorithm that utilizes the disparity variations of each pixel by cost formulation in a cooperative manner to solve several stereo disambiguation problems. A completely novel Stereo Orthogonal Feature-mapping transform (SOFT) has been proposed to compute the local as well as semi-global stereo matching cost that involves the extraction of the color information of the stereo images to make the process illumination invariant and improve the robustness and stability against noise. The computation of the simultaneous left and right disparity helps in removal of occlusion regions as outliers. The stereo matching is achieved only by a constraint based optimization of the cost function relying on the pixel wise matching costs. Our algorithm was verified on the Middlebury and KITTI datasets and the performance is compared to the classical as well as the state-of the art algorithms. The results clearly demonstrate that the proposed approach yields higher accuracy and incisiveness in reconstruction at the cost of lesser computational complexity.