Fig. 6From: A learned sparseness and IGMRF-based regularization framework for dense disparity estimation using unsupervised feature learningExperimental results for the Middlebury stereo 2001 and 2003 datasets [2], Venus (first row), Teddy (second row), and Cones (third row). The left image I L and the ground truth disparity map are shown in first and second columns, respectively. The third column shows the initial disparity map used in optimizing the energy function given in Eq. (17). The final disparity and the error maps estimated using the proposed method are shown in the fifth and the sixth columns, respectivelyBack to article page