The Research On Real-time Panorama Stitching Technology Based On Image Corner Feature

Virtual reality technology is widely used in medicine, remote sensing, security,military and other aspects. It is a new performance method that can produce a full range offeelings, of which the most common is the panoramic image technology in visual terms.Panoramic image can provide the Periscopic visual experience, and compared to traditionalimage display technologies, it is more suitable for display comprehensive imageinformation, so that the user can better judge the relationships by showing or analyze eachsituation kinds of information between images. From the way to generate the panoramaview, there are four ways to generate panorama. They are phase-based stitching method,area-based stitching method, cost function-based stitching method and imagefeatures-based stitching method. Image features-based stitching method is the least one, itcompare the two image feature to complete image registration, then compute the projectionmatrix from matching the characteristics, using the matrix stitch two images. Because ofgood effect and efficiency of the algorithm, features-based stitching method has beenwidely used.Compared with the other three stitching method, features-based stitching methodstitching method has a high success rate with stitching algorithm robust and strong, highstitching speed. But if people want the algorithm to achieve the real-time, then it need tomake the original algorithm considerable improvement.The main research contents and results are stated as follow.Reading the virtual reality relevant literature and four kinds of stitching methodsliterature, proposed own solutions and to improve it. Proposed strategies and methodsoptimization, comparing optimization algorithm and original algorithm.Finally, using the algorithm to process3D imaging lidar. Through practicalexperiments to verify the feasibility of this algorithm processing performance on3Dimaging lidar depth data. Proposed a new depth data processing method.