Research On Image Mosaic Technology Based On Improved BRISK Algorithm

Image mosaic technology refers to a technology that seamlessly stitches multiple images with overlapping areas in the same scene into a high-resolution image with a large viewing angle.As a popular research in the field of digital images,Image mosaic technology has attracted much attention in various application fields,with the development of image mosaic technology,it has gradually become an indispensable part of scientific research and daily life.Image mosaic based on feature matching is widely used,and image registration is the most important part of this method.We study and compare the implementation process of SURF(Speed-Up Robust Features,SURF),ORB(Oriented FAST and Rotated BRIEF,ORB)and BRISK(Binary Robust Invariant Scalable Keypoints,BRISK)image registration algorithms.The SURF algorithm is developed on the basis of SIFT(Scale Invariant Feature Transform,SIFT)algorithm,which has strong robustness,but the computational complexity is not real-time.The ORB algorithm is fast in computation but weak in scale transformation.Therefore,this paper focuses on the image mosaic technology based on the improved BRISK algorithm.In the image registration part,the traditional BRISK algorithm has two problems: single matching strategy and low registration accuracy under external interference.The two stages of feature point extraction and feature point matching are improved respectively.In the feature point extraction stage,image processing using an improved bilateral filtering method based on a normalized gray filter function,combined with the AGAST algorithm to detect feature points,suppress noise errors,and remove irrelevant feature points.In the initial stage of feature matching,rough matching of feature points is performed through a two-way Hamming distance,combined with the improved RANSAC algorithm by increasing interior point probability for fine matching.Finally,experiments verify that the improved BRISK algorithm reduces the matching time while improving the registration accuracy.In the image fusion part,in view of the influence factors such as illumination and environment,the overlapped area of the image after splicing is blurred and there are splicing gaps,we propose a fusion method combining the fade-in and fade-out method with the Laplacian pyramid.This method can solve the gap problem of mosaic images,make image transitions smooth and achieve high-resolution fusion.Experimental results show that the improved BRISK algorithm and the improved fade-in and fadeout method are better than the traditional algorithm.