Depth Video Coding In3D Video

With the development of3D display and interactive multimedia systems,3Dvideo applications, such as3DTV and Free Viewpoint TV, are attracting more andmore academic and industrial research interests. To efficiently represent the3D scene,new data format-multi-view video plus depth has been proposed and widely studied.Multi-view video plus depth is composed of texture videos as well as thecorresponding depth videos from multiple views. The characteristics of the depthvideo are very different from those of the conventional texture video. Depth video iscomposed of large homogeneous areas bounded by sharp edges, which can be definedas piece-wise smooth. In addition, depth video and texture video both represent thesame scene, only from a different perspective. Hence, motion similarity and structuresimilarity can be expected between them. In the conventional video coding, theprediction for the pixels around the edge is generally poor, resulting in large residualerrors at these locations. In the transform-based video coding framework, such largeresidual errors degrade the coding efficiency significantly. Therefore,high-performance depth coding method based on the specific characteristics of thedepth video is highly demanded.This paper first thoroughly studies the motion similarity and structure similaritybetween the depth video and the corresponding texture video. Based on the statistic, anew depth coding framework is proposed, together with some new coding techniques,namely weighted averaging based omnidirectional prediction, depth-texturecooperative clustering based prediction and detection and rectification of themisalignment between depth video and texture video. Experimental results validateour proposed depth coding scheme in terms of subjective and objective quality,compared with the conventional video coding scheme.With respect to the inter prediction for depth coding, a pixel-based interprediction scheme is proposed. This scheme first estimates the pixel-based motionvectors of the texture video, and then maps them for depth coding in view of themotion similarity between the depth video and the texture video. In a sense, thisscheme performs the pixel-based inter prediction for depth coding without sendingany motion vectors. Therefore, it greatly enhances the coding efficiency.