A partial shape matching using wedge wave feature extraction
Kenya Mori, Masakazu OHIRA, Motoko Obata, Kentaro Wada, Kazuo Toraichi · 2002
We introduce wedge wave feature extraction for partial shape matching, which is an essential process for image retrieval and computer vision. Using wedge wave feature extraction and applying extended dynamic programming (DP) matching, our matching method can recognize partial similarity that is location-free, size-flee, orientation-free and noise tolerant. The basic idea of our matching method is classifying the convexity of the shape's contour with a wedge wave, the base of convexity, and detecting the correspondences of convexities between two shapes by extended DP matching, which detects one-to-zero correspondence as well as one-to-one correspondence.