论文部分内容阅读
在保证虚拟视点绘制质量的前提下,如何降低深度视频的编码复杂度是一个亟需解决的问题。本文提出了一种基于最大可容忍深度失真(MTDD,maximum tolerable depth distortion)模型的低复杂度深度编码算法,将MTDD模型引入到率失真(RD)代价函数,通过对编码单元(CU)进行失真度量判决,如果对当前CU进行编码不会导致可察觉的绘制失真,则失真度量只考虑深度编码失真,否则失真度量同时考虑深度编码失真和视点合成失真(VSD),从而降低编码复杂度。实验结果表明,所提出的算法在不降低虚拟视点绘制质量和不增加编码码率的情况下,能显著降低深度视频的编码复杂度。
How to reduce the coding complexity of deep video is an urgent problem to be solved under the premise of ensuring the virtual viewpoint rendering quality. In this paper, a low complexity depth coding algorithm based on maximum tolerable depth distortion (MTDD) model is proposed. The MTDD model is introduced into the rate-distortion (RD) cost function, and the coding unit (CU) Metric decisions, if the current CU is coded without causing perceptible rendering distortions then the distortion metric takes into account only the depth encoding distortion, otherwise the distortion metric will consider both the depth encoding distortion and the view synthesis distortion (VSD), thereby reducing the encoding complexity. The experimental results show that the proposed algorithm can significantly reduce the coding complexity of the depth video without reducing the rendering quality of the virtual viewpoint and without increasing the coding bit rate.