Effective stereo matching method with equicrural triangle census tansform

A new stereo matching method using equicrural triangle census transform is presented in this work. In this method, the equicrural triangle census transform is firstly proposed to improve the robustness of the census transform. Meanwhile, gradient-based matching cost is combined to acquire the final cost volume. In order to improve the robustness of cost aggregation and disparity optimization, a superpixel segmentation method based on the simple linear iterative clustering is employed to guide the cost aggregation. Furthermore, a modified random walk method, named the adaptive random walk is used to optimize the disparity. Finally, a new disparity map post-processing method named wavelet edge joint bilateral filter is proposed to eliminate error regions remain after the cost optimization. The experimental results present that our proposed method significantly presents higher performance of the robustness than the local methods on the Middlebury dataset. Meanwhile, the processing speed of proposed method is faster than the global methods (GC, BP, etc.). In addition, the wider applicability of the proposed method is demonstrated on the KITTI dataset and some typical real-world sequences. Copyright © 2015 Binary Information Press.