Research And Application Of ORB Algorithm In Image Matching

As an analytical method to detect the similarity or consistency of two or more images,image matching is widely used in face recognition,object tracking,visual navigation,image Mosaic and retrieval.In the image matching with different transformation degrees,the running time and computational efficiency of the image matching algorithm play a decisive role in the quality of matching,and image matching is an important image processing technology in the field of image Mosaic,which determines the effect of image Mosaic quality.However,the traditional matching algorithms have problems such as slow running time and low extraction accuracy.In order to solve the problems of low matching accuracy and long time in image matching algorithms,this paper proposes an improved image matching algorithm based on the statistical feature of grid motion(RANSAC-GMS),which improves the algorithm's image matching accuracy and speeds up the running time.For image fusion splicing ghost and ghosting and other problems,this article proposed image matching algorithm on image spell the immediate area,puts forward an algorithm combined with improved RANSACGMS best sutures image matching algorithm,the proposed algorithm can eliminate image stitching good fusion of ghosting and crack problem,this algorithm has realized the image mosaicing running time is fast,high efficiency advantages,to ensure the stability of stitching effect.The research work of this paper is as follows:First of all.The development status of image matching are discussed in detail,the angular point extraction,image matching involved feature point matching point description,purification principle and makes a broad overview of the advantages and disadvantages,which focuses on the SIFT algorithm,SURF the image matching algorithm of image matching and the ORB image matching algorithms such as image matching algorithm for image matching matching point in the process of purification,lists the common k-d tree and the BBF method and RANSAC algorithm,an improved algorithm of image matching is proposed in this paper and the application has made the very good theoretical support.Secondly.In view of the slow matching time and low accuracy of SIFT algorithm.This paper proposes an image matching algorithm which combines the statistical feature model of grid motion with the improved random sampling consistency method.The algorithm USES fast rotation invariance feature(ORB)algorithm to prematch the image,and grid motion statistics(GMS)support estimator to distinguish the correct matching points from the wrong matching points.Then,the improved random sampling consistency(RANSAC)algorithm is adopted to screen the feature points through distance similarity between the matching points,and the new data set after screening is rearranged by using the evaluation function,so as to eliminate the false matching points and further improve the image matching accuracy.Finally,the image matching algorithm proposed in this paper is applied to the field of image Mosaic.Aiming at the problems of ghosting and ghosting in the traditional image Mosaic and fusion part,a mesh motion statistical feature model combined with the improved optimal stitching algorithm is proposed in this paper.This algorithm USES the improved gms-ransac image matching algorithm proposed in this paper to extract feature points and complete the image matching process.Then the optimal suture line was used to divide the optimal suture line according to the repeated local area of the image,and the improved weighted average fusion algorithm was used to smooth the fissure transition near the splicing line to complete the image splicing.Compared with ORB algorithm,SIFT algorithm and GMS algorithm,the algorithm in this paper has better performance in computing time and stitching efficiency in image scenes with more image textures and different transformation degrees,such as scaling/rotation,and can meet the requirements of image stitching.