A 3DTV System Using New Depth Image-based Videos in the MPEG-4 Multimedia Framework

Abstract

In this paper, we present a 3DTV system using a new depth image-based representation (DIBR). After obtaining depth images from multi-view cameras or a depth-range camera, we decompose the depth image into three disjoint layer images and a layer descriptor image. Then, we combine the decomposed images with color images to generate a new representation of dynamic 3D scenes, called as a 3D depth video. The 3D depth video is compressed by a H.264/AVC coder, and streamed to clients over IP networks in the MPEG-4 multimedia framework. The proposed 3DTV system enables not only enjoying a high-quality 3D video in real time, but also experiencing various user-friendly interactions such as free viewpoint changing, composition of computer graphics, and even haptic display.