Algorithm Researches For Point Pattern Matching And Multi-Point Pattern Matching

Point pattern matching is a core problem in the field of image matching,and has been the focus of many scientists.Nowadays,the technology of point pattern matching is more and more widely used in daily life,such as computer vision,security identification,moving target tracking,etc.Although many research methods have been put forward by different scholars,the matching problem is a NP-hard problem,and due to the existence of complex geometric deformation and external interference,it is still a challenging task to solve the problem.In this paper,a variety of point pattern matching algorithms are analyzed,and the point pattern matching algorithm based on shape context is mainly studied.The idea of shape context algorithm is to use the whole spatial distribution of the point set as a method to characterize the point feature.This makes the algorithm greatly affected by the point distribution,and the robustness of the algorithm is poor.Aiming at the problems of the shape context algorithm,this paper presents a new density shape context algorithm(Density Shape Context,DSC).The core idea of density shape context algorithm is to deal with the low density points in the point set,to optimize the selection of point features,and to enhance the robustness of the algorithm.In the density shape context algorithm,this paper introduces a new evaluation criterion named QScore.This standard combines the size of the matching set and the effect of the root mean square deviation of the matching points,and quantifies the matching effect more accurately.On this basis,combined with the idea of maximum matching of bipartite graph,the matching set is further optimized,and the matching effect of the algorithm is improved.Multi-point pattern matching is a study of pattern matching between multiplepoint sets.It has higher computational complexity and difficulty,and less research is conducted at home and abroad,but it is also an urgent problem to be solved in actual production and life,such as the application of the multi-structure comparison of protein binding site.This paper presents an approximate algorithm named Cluster Matching Algorithm to solve the problem of multiple pattern matching.The core idea of cluster matching algorithm is to propose a method to construct undirected graph,and the multi-point pattern matching problem is mapped indirectly to solve the maximum cluster problem in graph theory,which provides a new solution for the multi-point pattern matching problem.In this paper,the simulated data and real data in 3D space are used as test data,and the density shape context algorithm and cluster matching algorithm proposed in this paper are tested.The density shape context algorithm is compared with other point pattern matching algorithms.A large number of experimental results show that the proposed density shape context algorithm has obvious advantages over the traditional point pattern matching algorithm in terms of robustness and correct matching rate.The cluster matching algorithm has better matching effect and lower computational complexity in dealing with the multi-point set pattern matching problem.