Research On Virtual View Rendering Method Based On Depth Image

With the vigorous development of the3D film market, the development of3DTV has also been greatly facilitated.3DTV technology can provide audiences any viewpoint in a certain range, and viewers can get an immersive viewing experience, so3DTV gets a lot of attention. With high definition television (HDTV),3DTV becomes the mainstream trend of television development, and is one of the digital multimedia businesses which have very broad development prospects. Therefore, the development of3DTV technology has become particularly critical.Given the current development status of3DTV system, how to achieve real-time and efficient transmission becomes a research focus, if transport the color images corresponding to left eye and right eye simultaneously, it will greatly increase the transmission bandwidth and system costs, and it is difficult to meet the demand. Therefore, the virtual view rendering technology has become a hot research. In this paper, we give an in-depth research in the virtual view rendering technology, detail the model-based rendering (MBR) and image-based rendering (IBR). MBR is able to obtain better detail information, but it needs complex model, so practicality is poor. IBR generates the virtual view images by the reference view images; it can get better rendering quality while greatly reducing the transmission bandwidth. We mainly research the Depth Image-Based Rendering (DIBR), which is the most important technology in image rendering, and renders virtual view image according to the color image and the depth image of reference view. DIBR can greatly save transmission bandwidth as the depth image takes up less space than the color image.Although DIBR has wide application in3DTV, medical image processing, aviation and other fields, it also has own limitations; the disocclusion is the obvious issues. In this paper, we smooth the depth map to reduce holes, but it will cause problems such as edge distortion; and edge ghosting often appears after the holes filling.In this paper, we give the reasons of edge ghosting and disocclusions in the DIBR; point out the disocclusions are caused by that some occlusion areas in the reference image expose in the depth discontinuity area in the target view. We give a method eliminate the edge ghosting, and it is effective. Using the improved DIBR system, we can get a better result without preprocessing the depth image, thus we can save storage space and improve the computing speed.Experimental results show that, the improved DIBR system can save the transmission bandwidth in the3DTV system, meet the receiving end real-time demand; At the same time, it can better improve the quality of the generated image in the target view. Comparing with the previous methods, the virtual view images obtained by our method achieve good results both in the subjective evaluation and objective evaluation.