Enhanced Stereo Matching Technique using Image Gradient for Improved Search Time
Stereo matching algorithms developed from local area based correspondence matching involve extensive search. This paper presents a stereo matching technique for computation of a dense disparity map by trimming down the search time for the local area based correspondence matching. We use constraints such as epipolar line, limiting the disparity, uniqueness and continuity to obtain an initial dense disparity map. We attempt to improvise this map by using color information for matching. A new approach has been discussed which is based on the extension of the continuity constraint for reducing the search time. We use correspondence between rows and gradient of image to compute the disparity. Thus we achieve a good trade off between accuracy and search time.
Keywords: Color stereo matching, Image Gradient, Continuity constraint, Epipolar geometry, Stereo Vision
Download Full-Text