Research On Binocular Vision Stereo Matching Technology

Binocular stereo vision technology is a key technology branch of computer vision research field, and its ultimate goal is to simulate the people’s perception of the eyes by computer. At present, the binocular stereo vision technology has already been widely applied in actual production and people’s life, such as industrial production precision of3 D measurement and 3D scanning, robot navigation, 3D virtual reality and so on. In this paper, the basic theory of binocular stereo vision technology has been carried on the detailed introduction, focuses on the feature extraction, and the algorithm of stereo matching technology.Stereo matching based on feature extraction is a kind of important technology in binocular vision. In addition, the corresponding features matching algorithm is also the difficulty and key. Based on image edge feature extraction, the feature matching algorithm of binocular stereo vision is proposed, and we study the target object stereo matching with image features.The main research object of this paper o is the flame image processing and perception.Based on the characteristics of the flame image, as well as the defects of traditional edge detection algorithm, an improved SUSAN algorithm is proposed for flame image edge detection. This method aims to solve the problem of the SUSAN algorithm: 1) in some special situation, it cannot effectively extract image edge because of the arbitrarythreshold; 2) the too much computation is not suitable for real-time applications. The experimental results show that the edge detection algorithm proposed in this paper can effective achieve the flame image edge detection.Due to consideration of the target and background separation, the edge of the flame extracted by the improved SUSAN algorithm is more complete, clear and accurate reflection of marginal distribution. In this paper, on the basis of the feature matching algorithm by introducing the SSDA algorithm, we use polar and disparity gradient constraints enhance the accuracy of stereo matching algorithm, to reduce the algorithm complexity, and get the matching relation between an image pair. And the stereo matching algorithm is then applied to flame image for stereo matching problem, the experimental results show that it can effectively complete the stereo matching test improve the matching accuracy.