We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
SSIM-based joint-bit allocation for 3D video coding.
- Authors
Y, Harshalatha; Biswas, Prabir Kumar
- Abstract
The quality of a 3D video display depends on virtual view synthesis process which is affected by the bit allocation criterion. The performance of a bit allocation algorithm is dependent on various encoding parameters like quantization parameter, motion vector, mode selection, and so on. Rate-distortion optimization (RDO) is used to efficiently allocate bits with minimum distortion. In 3D video, rate-distortion (RD) property of synthesized view is used to assign bits between texture video and depth map. Existing literature on bit allocation methods use mean square error (MSE) as distortion metric which is not suitable for measuring perceptual quality. In this paper, we propose structural similarity (SSIM)-based joint bit allocation scheme to enhance visual quality of 3D video. Perceptual quality of a synthesized view depends on texture and depth map quality. Thus, SSIM-based RDO is performed on both texture and depth map where SSIM is used as distortion metric in mode decision and motion estimation. SSIM-based distortion model for synthesized view is determined experimentally. As SSIM cannot be related to quantization step, SSIM-MSE relation is used to convert distortion model in terms of MSE. The Lagrange multiplier method is used to solve the bit allocation problem. The proposed algorithm is implemented using 3DV-ATM as well as HEVC. RD curves show reduction in bitrate with an improvement in SSIM of synthesized view.
- Subjects
3-D video (Three-dimensional imaging); BIT allocation analysis; RATE distortion functions; MEAN square algorithms; MOTION estimation (Signal processing)
- Publication
Multimedia Tools & Applications, 2018, Vol 77, Issue 15, p19051
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-017-5327-0