The Research Of Virtual View Synthesis Based On Layered Depth

With the continuous development of digital multimedia technology,people for video interactivity and sensory experiences demand higher planevisual effect. In the context, stereo video arises at the historic moment. Stereovideo because of its users can experience the characteristics ofthree-dimensional visual perception, in a communication, TV, monitor,education, military and defense etc medical attention, many scholars andscientific research institutions also started the three-dimensional video systemand its related to technology detailed discussion and study. The virtual viewsynthesis is stereo video system is a key technology. It not only to ensure thatthe audience can eventually fluent enjoy stereoscopic feeling, Ensures theaudience can free choice perspective.To base on depth image rendering technology as the foundation, hasdiscussed based on the depth of the image rendering technology depthextraction algorithm. Put forward based on the depth of the image copymethods stratified image. The method used to mean drift clusteringalgorithms will depth chart segmentation processing. Make the objects havethe same depth chart depth data. Then the video information and process afterthe depth information of3D converted image. Make the world coordinatesystem transform to the point in the picture frame image. In the hole in theimage processing to fill, Finally complete virtual view synthesis.Through to the Nagoya Japan university offers kendo, dog the test seriessimulation experiment. The result show that the method solved the originalframe image of certain areas are occluded which lead to3D conversionerrors， and solve the formation of a virtual image can’t meet the subjectiverequirements. Through the PSNR and experiment show that the evaluationSSIM improved algorithm has good result.