Font Size: a A A

Research On Key Technology Of Panoramic Image Mosaic

Posted on:2019-03-09Degree:MasterType:Thesis
Country:ChinaCandidate:S K ZhangFull Text:PDF
GTID:2428330566981224Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Panorama image mosaic has widely used in many research fields of urban planning,virtual reality,navigation and medicine etc.And,it is the hot research in the fields of image processing and computer vision.Its main purpose is to use an ordinary image acquisition device to complete the production of 360-degree panoramic images.This paper deeply analyzes and studies the principle and method of the current technology of panoramic image mosaic.In the two key steps of image matching and image blending,a fast image matching algorithm based on color information and an image blending algorithm based on optimal suture improvement are proposed.The final design a panoramic image mosaic system.The specific research content is as follows:(1)Neglecting the color information,image grayscale is widely used in image matching,which may lead the problem of mismatch in the zone with similar colors.This algorithm calculates color invariant to establish multi-scale space.Then,take SIFT thinking as a lesson,extract color scale invariant feature vector.Finally,using FANN search strategy to realize the precise and efficient match of high dimension feature vector.(2)Aiming at the problems of blending gaps and ghosting,an image blending algorithm based on the improvement of the optimal suture is proposed.Firstly,the pixels in the overlap region of the image to be fused are calculated to search for the best suture.The gradient field is calculated by extending the overlapping area at both ends of the suture.The Poisson editing method is used to eliminate the stitching and ghosting.(3)Finally,combined with the proposed algorithm,efficient and automatic panoramic image mosaic system is researched.The correctness and effectiveness of the system in this paper are verified by experiments.
Keywords/Search Tags:Panoramic stitching, Color model, Scale space, Feature extraction, Suture
PDF Full Text Request
Related items