Font Size: a A A

The Research And Application Of Image Mosaicing Based On SURF Feature Points

Posted on:2015-09-12Degree:MasterType:Thesis
Country:ChinaCandidate:X G WangFull Text:PDF
GTID:2298330467962928Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Image mosaic technology is a technology which makes the number of sheets of overlappingportions of an image (which may be different time, a different perspective or different sensors) up alarge, seamless high-resolution image. This technology originate in humans photography, shootingangle lens in order to solve the limitations of other issues. With the development of computertechnology and digital image processing technology, image mosaic technology has become aresearch focus of computer vision, photogrammetry, image processing and computer graphics, and iswidely used in deep space exploration, remote sensing image processing, medical image analysis,cartography, virtual reality technology and other fields.Image mosaic technology generally includes image acquisition, image preprocessing, imageregistration and image fusion procedure. Among them, the image registration and image fusionsplicing are the most important key technologies of the whole image mosaic technology. Imageregistration methods can be roughly divided into three categories: gray information-based approach,transform domain-based methods, feature-based methods. The purpose of image fusion is to removethe mosaic image distortion and obvious seam, general median filtering methods are fusion method,the weighted average fusion method, wavelet fusion method.This paper studies the image mosaic technique based on SURF feature points. First, the basictheoretical knowledge of image stitching techniques are studied and analyzed, including basicprocesses image stitching technology, image preprocessing, image registration, image fusion andother key aspects; Second, there has a in-depth research to the feature points based imageregistration techniques to classical SIFT feature point detection algorithm as a reference, a detailedanalysis of SURF feature point detection algorithm, and verified by experiments comparing theperformance of the algorithm, the results show that based on SURF feature point matching algorithmis robust, unique and immunity, and have a distinct advantage over SIFT algorithm in computingspeed. In the feature point matching process, using a random BBF (Best-Bin-First) algorithm toimprove search speed of feature points. Finally, several common image fusion methods are analyzed,detailing the weighted fusion method used in this paper, and a detailed comparison with experiment.
Keywords/Search Tags:Image Mosaic, Image registration, Image Fusion, SURF, SIFT, Weighted fusion
PDF Full Text Request
Related items