Joint Bit Allocation for 3D Video with Nonlinear Depth Distortion—An SSIM-Based Approach

Abstract

Perceptual quality improvement approach for 3D video through bit allocation is presented in this paper. Bit allocation between texture video and depth map plays an important role in deciding quality of synthesized views at the decoder end. To have better visual quality, structural similarity (SSIM) index is used as a distortion metric in rate distortion optimization (RDO) of the 3D video. In this paper, we used the nonlinear relationship of depth distortion with synthesis distortion in computing rate distortion cost resulting in better mode decision. Using the same depth map RDO in bit allocation algorithm, more accurate results are obtained when compared to the linear relation of depth distortion with synthesis distortion.