Research On Image Feature Extraction And Matching Algorithm

In the process of matching the characteristics of the first match,coupled with the second match,you can further eliminate the wrong match point,greatly improving the matching rate of the algorithmIn the feature extraction,the recognition rate and stability of the SIFT algorithm are superior to those of other algorithms.However,there are many problems,such as large computational workload and complex and some hysteresis.The SIFT algorithm has four steps: first establish the scale space,and then detect the spatial extreme point,and then determine the direction of the movement of the feature point,and finally generate the feature descriptor.After the four steps of statistical operation,the operation time of the whole algorithm is more than 70%,so it is a research direction to generate the operation of the feature descriptor.In order to save the running time,this paper makes a corresponding improvement on the SIFT algorithm,which is in the process of feature extraction,using Harris self correlation matrix to get the regional self mutual change vector construct SIFT feature point detection method for image feature point descriptor this method has the advantages of simple structure,small amount of calculation,so as to improve the algorithm the efficiency.In the process of matching features,the initial matching and the second matching can be used to eliminate the matching points,thus greatly improving the matching rate of the algorithm.Experiments show that the algorithm has improved the algorithm,and the feature extraction and matching have become timely,and the matching accuracy has been greatly improved.