Multiview Video Presentation and Encoding

The three-dimensional television system is likely to play an important role in the future broadcasting. Currently this research area is mostly related to the computer graphics and animation since the multi-view camera systems and displays have not entered into practical use, hence the capturing of the multi-view sequences are implemented by a computer graphic software. The hybrid video encoding systems like MPEG-1 and MPEG-2 are based on the motion compensated DCT coding scheme. The block-based prediction scheme could be easily used for removing the redundancy between the adjacent views (disparity compensation), hence the conventional video encoding methods could be used for motion and disparity compensation. The multi-view profile of the MPEG-2 video encoding contains both motion and disparity compensation, and the several tools in MPEG-4 support to view an object from any viewpoint. In an MPEG-4 scene the virtual and natural video objects could be mixed since in the MPEG2 multi-view profile uses only binocular representation. Beside the motion and disparity compensation hybrid schemes the Depth Image-based Representation (DIBR) could be also used for multi-view video coding. This toolkit is also the part of the MPEG-4 video encoding tools. This paper organized as follows. In the next section, the prevalent multi-view rendering devices are introduced. In Section 3 we describe the Depth Image-based Representation of the three-dimensional objects. In Section 4 we review the multi-view image and video encoding systems where both the motion and/or disparity compensation techniques and the methods based on Depth Image-based Representation are introduced. In Section 5 we show the developed multi-view encoding system which uses DIBR. In our experiments the reference views are encoded and the others are rendered. The MPEG-4 AVC is used to encode the color information by using motion compensation and the depth information in encoded by using a lossless codec. Finally we give some further research objectives to complete the developed hybrid method.