Research On Virtual View Generation Algorithm Of3DTV

Three-dimensional television (3DTV) technology can record, transmit and display the three-dimensional color images, and bring the audiences immersive experience when watching TV. Naked-eye3DTV has been a research hotspot in recent years because audiences could experience the feeling of three-dimensional without the aid of stereoscopic glasses. It has huge potential in the entertainment industry, advertising media, remote control, art and many other applications. This paper is focused on virtual view generation algorithm for current3DTV display equipment, explores methods to generate multiple virtual views and analyzes the method of Depth Image-Based Rendering Algorithm (DIBR) deeply. DIBR, which is the most widely used technology in view generation, renders target virtual view images according to the reference color images and the depth images. DIBR could not only render high quality images fast, but also save transmission bandwidth greatly. However, DIBR always results in hole problems where the depth and color values are not known. Current hole-filling methods often cause other problems, such as edge-ghosting and cracks. This paper proposes an algorithm that uses the depth and color images to address the holes. It exploits the assumption of a virtual view between two laterally aligned reference cameras. Here is the major content of the paper:1. The paper analyzes the problems in current blended depth images, and proposes two methods to remove holes. One is morphological closing operations, and the other is depth image layering. The holes in depth images could be removed by those two methods.2. The paper proposes a view generation method which uses the position information provided by the filtered depth maps. A novel blending algorithm is presented that is more efficient than preprocessing depth maps, but obtains similar quality. An image inpainting method and an interpolation method are also presented to fill holes and eliminate edge-ghosting, which additionally use a post-processing technique to improve image quality. The method proposed in this paper can effectively remove holes and edge-ghosting.Experimental quantitative and qualitative results show that the proposed algorithm improves quality dramatically compared with traditional methods.