Fast Image Matching Algorithm And Distributed Image Matching System

Image matching is an image processing technique used in a wide range. It is also named scene matching guidance being applied in the missile guidance. Limited to the missile's high flying speed, the matching processing should be implemented in a short time after getting its real-time image. So the high real-time quality is needed in the process of missile guidance. In this thesis, a research of fast image matching is carried out with image matching technique, optimized searching theory, wavelet theory in signal system; and a distributed image matching system is built with the database theory and distributed calculating knowledge.On the research of fast image matching algorithm, first, the key factors in image matching are analyzed, and the advantages as well as disadvantages of image matching algorithms are pointed out after the usually-used ones at present are compared, summzrized and classified. Then the ways to reduce the amount of calculation and to increase the speed of calculating are analyzed: reducing the amount of a single time matching calculation and searching positions. A new scheme of fast image matching is proposed on the basis of some popular algorithms' advantages, which adopts an improved algorithm combined with physics and logic hierarchical strategy. As the lower-to-upper form of pyramid, the physics hierarchical strategy is a method which decomposes an image into a group of big-to-small layers using wavelet multiresolution transform. And the logic hierarchical strategy is devided into two steps on the matching method: the first step is to get some coarse matching points with a rapid but rough algorithm; the second step is to get the accurate matching point with a slow but refined one. The rough algorithm used in this thesis is composed of genetic algorithm (GA) and successive similarity detection algorithm (SSDA). In this fast image matching algorithm based on the layering method, the image's wavelet decomposition together with GA reduces searching positions in the course of matching, and SSDA reduces the amount of the single time matching calculation. Finally, it is proved by experiments that this method can improve the speed of image matching distinctly with less loss of gray scale correlation matching accuracy.On the realization of distributed image matching system, first, the background to builtd this system is introduced, then the factors of the database in this system and the structure of"client/server"are analyzed, and the scheme of the projects controlling module is described in detail. At the end of the thesis, this system using PostgreSQL (a database management system) as the base is programmed with the Visual C++. The functions and feasibility of this system are tested by the simulation experiments. And the distributed image matching system has passed the accepted test.