Fig. 7From: A learned sparseness and IGMRF-based regularization framework for dense disparity estimation using unsupervised feature learningExperimental results for the Middlebury stereo 2014 datasets [2], Adirondack, Motorcycle, Pipes, Playroom, PlaytableP, Recycle, Shelves, Vintage. The left image I L , ground truth and disparity map estimated using the proposed method for each stereo pair are shown in the first, second, and third rows, respectivelyBack to article page