Auto-Weighted Layer Representation Based View Synthesis Distortion Estimation for 3-D Video Coding

Recently, various view synthesis distortion estimation models have been studied to better serve 3-D video coding. However, they can hardly model the relationship quantitatively among different levels of depth changes, texture degeneration, and (VSD), which is crucial for rate-distortion optimization rate allocation. In this paper, an auto-weighted layer representation based developed. Firstly, sub-VSD (S-VSD) defined according level changes their associated degeneration. After that, a set theoretical derivations demonstrate that VSD be approximately decomposed into S-VSDs multiplied by weights. To obtain efficiently, layer-based method developed, where all pixels with same are represented layer. It enables S-VSD calculation at level. Meanwhile, nonlinear mapping function learnt accurately represent between S-VSDs, automatically providing weights during estimation. learn such function, dataset its built, termed as VSDSet. Experimental results show estimated once available. The proposed outperforms relevant state-of-the-art methods in both accuracy efficiency. VSDSet source code will available https://github.com/jianjin008/.