The Research Of Image Matching Algorithm Based On Local Feature

Image matching is a fundamental part of the field of image processing,which is a process of matching images by analyzing the similarity and consistency between the image and the image to be matched in characteristics,grayscale,and structure.Many fields,such as medicine,agriculture,remote sensing,machinery and artificial intelligence,are associated with image matching technology.For image matching,research is mainly conducted from the perspective of both grayscale and features.Among them,the matching method based on image feature has higher accuracy and faster speed than the matching method based on gray scale,and has better adaptability to gray scale change,image deformation and occlusion.Therefore,the matching of feature information is in line with the requirements of current practical applications.It is the focus of scholarly research.In this paper,by studying the image matching principle and the feature-based matching algorithm proposed by predecessors,we propose an image matching algorithm with superior performance in many complex situations such as image geometry transformation,scale transformation,and image blurring.The researchs are as follows:(1)A comprehensive and systematic study of the image matching algorithm is conducted.The matching algorithm is deeply analyzed and compared with its performance.And the research direction of this paper is the image matching technology based on feature.(2)An improved ORB feature extraction algorithm is proposed.The feature extraction method is introduced in detail and compared with experiments,including: Harris,Fast,SIFT,SURF ORB,FAST-SURF and Harris-SURF.And the detection results of each algorithm are verified by experiments.As one of the most widely used feature extraction algorithms,ORB algorithm has the advantages of rapidity and rotation invariance,etc.And the use of SURF algorithm can make up for the disadvantage of ORB algorithm without scale invariance.Therefore,the ORB-SURF feature extraction algorithm(OR-SURF algorithm)is proposed,which makes full use of the fast superiority of the ORB algorithm and the scale invariance of the SURF algorithm,and can greatly improve the matching effect.(3)An image matching algorithm based on OR-SURF is proposed.The improved K-D tree search strategy is adopted to find out the feature highlights of two nearest neighbor points,the preliminary match point is obtained.At last the RANSAC algorithm is adopted to eliminate false matching points and complete image matching.In this paper,the experimental results show that the algorithm can not only improve the performance of the ORB algorithm in terms of scaling,but also effectively improve the matching speed and accuracy when the scale of the image changes.Meanwhile the algorithm has strong robustness in the complex situations.