Study On Stereo Matching Algorithm Based On SIFT Feature Descriptor

Vision is an important way to observe and perceive our world for human beings, it is said that about 75 percent of human's information comes from our eyes. Stereo vision is a subject on how to understand and perceive the objective world by machine rather than human beings. It is the method to acquire the 3D geometry information of objects by multiple images (generally two images). On contrast with other methods, the three dimensional reconstruction via stereo vision has low hardware cost and simplicity for implementation, so stereo vision gains ground very fast among the researches as an important branch of computer vision. And after the studies in recent 20 years, stereo vision has been applied successfully in more and more fields such as robots'vision navigation, avigation mapping, military application, medical diagnose and industrial inspection etc. With the rapid development of new technologies and mathematics, the performance and theoretical fundamental of stereo vision has been greatly improved recently.This paper focused on stereo matching algorithm which is the most important and difficult issue in stereo vision and made a deep research. On basis of abundant inland and overseas reference papers, the theory and fundamentals of stereo vision is introduced in detail. In addition, four key components of stereo matching algorithm, including feature space, similarity measure, search space and search strategy, are analyzed in a full and system level. In the light of image primitives used for matching, this paper divided the stereo matching algorithms in common use into three main categories, that is, area-based matching, feature-based matching and phase-based matching. And the basic idea, characteristics, advantages and disadvantages of each kind of algorithm is explained detailedly. Research on state of the art of stereo matching algorithm shown that stereo matching algorithm is greatly image-dependent. In...