A scalable depth coding with arc breakpoints based synthesis in 3-D video

Depth map represents three-dimensional information and is used for depth image-based rendering to support 3-D video applications. Conventional video coding standards may cause serious coding artifacts along the depth discontinuities, which ultimately affect the synthesized view quality. This paper proposes an efficient technique to compress the depth map by encoding a reduced resolution of depth map and using special upsampling method to reconstruct the original depth map. Unlike previous upsampling schemes with the guidance of corresponding color video, the proposed method upsamples the depth map using the breakpoints which are generated at the decoder side with only a few breakpoints. Instead of the edge representation of the depth map, the breakpoints would substantially take less bits and low errors with the regularity of the contour. Experimental results show that the proposed technique significantly reduces bit rate while achieving a better quality of the synthesized view in terms of subjective and objective measures.