The Study Of The Binocular Stereo Matching Technology Of The Computer Vision

Computer vision is a widely and rapidly developing field, binocular stereo matching technology is one of the most important research topics in this field. Achieving faster, more accurate stereo matching is the research direction. This paper studies camera calibration, binocular passive ranging technology and stereo matching algorithm, and focuses on the region-based stereo matching algorithm, build a binocular stereo vision system.In order to solve the problem that uniform support window size and identical reference value will influence matching effects in binocular stereo matching, a novel stereo matching algorithm is proposed. The proposed algorithm can adaptively select the support window size based on image edge, adaptively distribute weights based on geometric distances, and weight based on color distance. Firstly, select the support window size dynamically, using image edge information. Then present a new weight model that calculate weights with the geometrical distances between field points and window center, according to the probability curve characteristics of elements’matching values. Finally, combined with color similarity constraint, obtain dense disparity map using sum of the weighted color distances. Experimental results show that this algorithm is fast and efficient, and can well reduce the matching noise and improve the matching precision in depth discontinuities and low-textured regions.To test the performance of the binocular passive ranging method and stereo matching algorithm, this paper builds a binocular stereo vision experiment. Firstly, adjust PTZ and calibrate the system. Then, do passive ranging with the collected images. Finally, use the proposed algorithm to stereo matching with the collected images, and apply the matching results to achieve distance measurement. Experimental results show that this algorithm can be effectively applied to images collected by the experimental platform with less noise and faster speed, and this system is able to effectively achieve binocular passive ranging function with higher precision and better effect.