This paper presents a window-based method for solving the stereo matching problem. Our contribution is to introduce the concept of connectivity for computing adaptive support weights. We show that our new support weight computation strategy leads to improved disparity maps. This is demonstrated on the well-known Middlebury images. According to the Middlebury online competition, the proposed algorithm is the top-performer among all local stereo methods.
A. Hosni, M. Bleyer, M. Gelautz: "Geodesic Support Weights For Local Stereo"; Talk: Junior Scientist Conference 2010, Vienna; 04-07-2010 - 04-09-2010.
Click into the text area and press Ctrl+A/Ctrl+C or ⌘+A/⌘+C to copy the BibTeX into your clipboard… or download the BibTeX.