Title :
Segment based stereo matching using shape description and hierarchical maximal clique
Author_Institution :
NTT Human Interface Lab., Kanagawa, Japan
Abstract :
A novel stereo matching method operating on edge segments is presented. In this method shape measure of edge segments is introduced to weigh candidates for matching. Matching determination is performed by a hierarchical clique method that is based on the weight of the candidates. The method is shown by simulation to perform fairly well if the scene produces good edge images; otherwise, the matching result degrades rapidly due to epipolar constraint too strong for the composite segments´ tips
Keywords :
computer vision; computerised pattern recognition; computerised picture processing; candidates; edge images; edge segments; epipolar constraint; hierarchical maximal clique; matching determination; scene; shape description; shape measure; simulation; stereo matching method; weight; Cameras; Explosions; Humans; Image segmentation; Laboratories; Layout; Robots; Shape; Stereo vision; Telephony;
Conference_Titel :
Industrial Applications of Machine Intelligence and Vision, 1989., International Workshop on
Conference_Location :
Tokyo
DOI :
10.1109/MIV.1989.40539