Stable Matching Based on Disparity Components

In this paper, we demonstrate how disparity maps could be improved by applying adaptive windows (varying in their shape). In our approach, windows adapt to high-correlation structures in disparity space, which we call disparity components. The method is straightforward and it is applicable to any stereo algorithm. This helps us in solving standard stereo problems such as: (1) to enforce the inter-line continuity constraint, (2) to improve disparity map accuracy. To show the proposed method properties, we measure the improvement of stereo matching failure due to adaptive windows in a rigorous ground-truth evaluation experiment. We demonstrate that adaptive windows account for 15× improvement in false positive rate and 5× improvement in mismatch rate. All other measured types of error improve only marginally. Our results on standard stereo pairs are also presented.