Slice header prediction for depth maps bit reduction

With the 3D Video (3DV) representation of multi-view video plus depth, the 3DV video can be compressed by coding the texture video and depth map independently in current H.264/MVC. This will introduce syntax redundancy between the bitstream of texture video and depth map. To remove the redundancy and improve the coding efficiency of 3D videos, a slice header prediction method is proposed in this paper. Based on the analyses of layered based coding structure and the syntax elements in slice header, the common slice header syntaxes in texture videos and depth maps are jointly coded. Experimental results demonstrate that compared with the existing method in MVC, the proposed method can achieve average and highest bitrate reduction for depth map coding by 2.95% and 7.73%, respectively.

I. Introduction

Nowadays, 3D video (3DV) is emerging as a new media and attracting many interests [1]. Different from the traditional 2D video, the 3DV technique can provide users a real-world scene and deep impression. Therefore, it can be used in many fields, such as 3D television (3DTV) [2], free viewpoint television (FTV) [3], computer games, and visual reality, etc.