Font Size: a A A

Research Of Compression And Rendering Based Array Images

Posted on:2013-02-20Degree:MasterType:Thesis
Country:ChinaCandidate:J XuFull Text:PDF
GTID:2248330377959165Subject:System theory
Abstract/Summary:PDF Full Text Request
With the development of information technology, a technology of image-based renderingplays more important role in computer vision. It can not only enhance the speed of reply theimages, but also make the image more similar to the real world. However, in order to makethe rendering result be true to nature, enough scene images should be stored. So it is necessaryto compress the images before rendering.The paper mainly researched array images compression and rendering technology.For array images compression, the paper proposes a new compression method based ondisparity compensating. First, in order to gain the precise parallax between the adjacentimages, the paper proposes the novel corner detection algorithm based on multi-scalecurvature and corner matching algorithm based on rotation invariant. So the range of parallaxcan be gained and the foundation is established for accurate parallax. And then the paperproposes disparity compensating compression algorithm for array images based on accuratelymatching from coarse to fine. This algorithm can gain accurate parallax, eliminate theredundancy between adjacent images greatly and get very high compression ratio between200:1and1000:1depending on reconstruction quality.For array images rendering, the paper proposes two novel rendering algorithms fordifferent positions of the new viewpoints. When the new viewpoint and the cameras are in thesame plane, a image rendering algorithm is proposed based on vector distance and spaceinformation. When the new viewpoint and the cameras are not in the same plane, a imagerendering algorithm is proposed based on geometry information.The paper has done the corresponding simulation experiment for corner detection andmatching, disparity compensates; image compression and image rendering. The experimentresults indicate that the proposed compression algorithm can get a higher compression ratiounder the good reconstructed quality. And the proposed rendering algorithm can realize realtime rendering and get good image effect for the new viewpoint anywhere.
Keywords/Search Tags:Corner detection, Corner matching, Disparity compensating, Image compression, Image rendering
PDF Full Text Request
Related items