A Study On Image-based Cylinder Panoramic Image Generation Technology

Image-based rendering (IBR) technology is a new hotspot arising in recent years. An important goal of virtual reality is building a living virtual world used computer so that the users are allowed to control the virtual objects interactively. Traditionally, the virtual reality system uses 3D computer graphics to model and tender a virtual environment in real time. However, the mosaic method based on image rendering composes the sequence images of same scene into panorama. It avoids the complicated modeling and rendering and constructs the virtual reality environment by the scene images shot by the camera or vidicon .It also overcomes the disadvantage of 3Dcomputer graphics method .Consequently ,it provides an effective measure for analysis of reality scene .The article introduces the Panorama Generation research present situation, application domain .We researched the emphasis in the image projection, image matching and the image fusion, analyzed and summarized their main method, the performance and the existence question. In this foundation, we implement cylindrical projection algorithm used to generate uniform space to solve the problem of panoramic vision consistency. In order to eliminate image distortion and jagged phenomenon, using the bilinear interpolation algorithm for interpolation; In image matching, the thesis uses based on feature block according to color different and the smallest deviation feature lines matching method combined. By overlapping regions divided into blocks, the Euclidean distance of the each component from color images to calculate color different of each block. Then the boundary of the region as the feature line for image stitching; In image matching, the thesis uses based on feature block according to color different and the smallest deviation feature lines matching method combined. By overlapping regions divided into blocks, the Euclidean distance of the each component from color images to calculate color different of each block. Then the boundary of the region as the feature line for image stitching. It effectively overcomes that the algorithm of based on the smallest deviation feature lines appears false match when the image content is little difference. Experiment shows that the method has a good result In the robustness and accuracy. Also the method accelerated the speed of image mosaic; In image integration, In order to make the image to be smoother in the seam, we adopt the weighted smoothing algorithm with threshold to fuse the two images in the transition. It effectively solves that the conventional weighted smoothing algorithm generate image blurring and ghosting.In addition, the method and experimentation are given in the paper. It proves that we can get ideal panorama using the method in the dissertation.