Research On Binocular Stereo Vision Scene Matching And Scene Description

Stereo binocular vision technology is an important research direction of machine vision, It can realize the function which is similar to biological eyes in the machine through the technology. Stereo binocular vision technology is a cross-subject, involving image processing、Computer science、pattern recognition and so on. Whether in theory or in the practical application, Stereo binocular vision technology has high academic value and practical value.At present, there are many organizations research on binocular stereo vision technology, such as research institutes、well-known universities both at home and abroad. This paper is based on previous research in related fields, through refering to and analyzing large amounts of academic papers and literature of binocular stereo vision system at home and abroad on, efforts to study the key techniques and methods on camera calibration and stereo matching. Purpose of this paper is to seek algorithms with achieving relatively simple, high matching rate stereo binocular visual scene matching algorithm. This paper mainly works as follows:1、From Marr visual theory, this paper comprehensive overview the background and significance,, research status and development trend of facial expression about binocular vision system. Then, from the technical aspects of binocular vision system, focusing on the camera calibration、feature extraction and stereo matching. And finally presents the experimental simulation results.2、A method of stereo matching based on binocular vision system is proposed in this paper. This method is composed of two parts below, First, the paper adopts RAC camera calibration method rectified input images. The second, in the stereo matching section, we adopt Segment-Based stereo matching algorithm. Segment-Based stereo matching algorithm is the core of this research in the paper, there are four major steps of the proposed algorithm, namely, color segmentation, local matching in pixel domain, the disparity plane estimation and disparity plane assignment. Experimental results demonstrated that satisfactory results can be computed efficiently by the proposed method, it also can provide good foundation for3D reconstruction.