Optimization of Matching-Point-Detection in Stereoscopic Images

Abstract

Several methods of matching-point-detection in stereoscopic images were studied for the purpose of minimizing the detecting error. As the result of the experiments, we have proposed a new matching method called 'Rubber-matching', into which the characteristics of the human visual system have been introduced. Using this method, the matching-error rate has been reduced to about 10% of that of conventional block-matching.