Research And Application Of Image Mosaic Method

Image matching technology has been the key research direction all the time, suchas computer vision, computer graphics and multimedia imaging system and so on. It isable to match and align two pictures which have overlapping area of many pictures, tosplit it into a scene picture, and make pictures in keeping the resolution of the casehave a wider perspective. Therefore, image matching and solve the conflict betweenvision and resolution, has been widely applied to many practical fields. Such as aerialsurvey, satellite remote sensing image processing, virtual reality scene again andmedical image analysis, and other fields, the application prospect is very broad.In this paper, we study and analyze some key technologies in the field of imagematching, including: image feature point extraction, feature matching, image fusion,etc.In this paper, the main job:1) research and implements the several kinds of cornerdetection algorithm of image feature points extraction, and compares their advantagesand disadvantages;2) The traditional SIFT operator are studied, and the process offeature extraction and matching is improved, and the improved SIFT feature extractionand matching process mainly aimed at the pyramid building of SIFT feature extractiontakes up most of the above time of the entire feature extraction, adopting a method ofreasonable class constraint to controlling the number of feature points and reducing thetime consumption. Then combined with Harris corner detection method to SIFTfeature point number control strategy, and thus improve the efficiency of registration;3)Studying the image fusion method, choose the appropriate integration strategy forimage fusion, as far as possible keep the original image and achieve the result ofnatural fusion of information itself.4) In this paper, the improved algorithm is used totourism scenic spot image stitching, the experimental results show that improvedalgorithm is effective.