| Image matching is one of the most important technologies in the field of digital image processing. It is very helpful for many industries to have a fast and efficient image matching method. The fields such as robot visual, navigation and positioning and medical image processing will benefit from the image matching technology. The paper research is focused on the detection and matching algorithm of object image with complex deformation.First of all, the paper gives the introduction of the development and history of the image matching technology. Then the paper focuses on the essential theory of the image matching. In this paper, the PSO(Particle Swarm Optimization) algorithm is combined with the image matching technology. In order to optimize the affine target model, one target image matching algorithm based on PSO and gradient maximum points is proposed. With aspect to the search strategy of PSO, the paper redefines the parameters in this strategy, adopting the method of dynamic update strategy. Through the method of dynamic update strategy, we define the max velocity, inertia weight and search space, improving the convergence of PSO algorithm. In the field of multi-target image matching, this paper creatively puts forward MPSO algorithm to accomplish the matching task of multi-target and the theory and related experiment is given. The image features of affine invariance are studied in this paper and the matching method based on ASIFT is introduced. RANSAC is used to enhance the correct matching ratio of the feature points. Finally, this paper put forward the innovation ASIFT algorithm based on PSO algorithm. The detailed theory and experimental results are given in the paper. The result proves the rationality and validity of the proposed algorithm.PSO algorithm is easy to be implemented, and it has a certain intelligence and social behavior. It is an important branch of global optimization algorithm. Image matching technology finds the image transformation model through similarity measure criteria. PSO can smartly and quickly finish the model estimation in the image matching problem.Finally, this paper proposed the improved PSO-ASIFT algorithm according to the theory of ASIFT. Combining the PSO strategy and ASIFT, it enhances the precision of camera position sampling. The original discrete sampling problem can be converted to continuous camera position optimization. It greatly improves the ability of image feature extraction and matching. The experiments show that the proposed algorithm effectively improves the target image matching precision. |
