Matching Algorithm of Binocular Dynamic Vision Measurement

Abstract

In order to realize the correspondence of feature points in the binocular dynamic vision measurement system, the two homologous images matching problem in the paper is studied. First of all, a method of strong space resection. The strong space resection optimizes the elements of exterior orientation, improves the precision, so as to improve the calculating precision of epipolar line. Secondly, an algorithm of multiple restriction matching which is for two homologous images matching of feature points has been introduced. This algorithm based on epipolar line constrain, increases the uniqueness constraint and double constraint, combined with the second match with disparity gradient constrain, finally get the correct matching relationship of feature points in two homologous images. The experimental results show that: this method for matching of binocular dynamic vision measurement system, can get 100% of the matching accuracy rate. And it would satisfy the need of binocular dynamic vision measurement system.