一种基于多视点视频的低复杂度自适应环路滤波算法

A low complexity adaptive loop filter algorithm for multi-view video coding

自适应环路滤波(ALF)是面向高清的多视点视频编码(MVC)中的一项新的滤波技术,虽然能提高视频压缩的主观质量和客观质量,但是其复杂度过高阻碍了其实时应用。本文提出了一种低复杂度的ALF算法,充分利用了视点间相关性和层间相关性以减少亮度和色度分量的分割次数,并且能自适应跳过大部分帧的块控制(DBC)处理过程。实验结果表明,所提出的算法能在保证视频主观质量和客观质量基本不变的情况下减少61%左右的ALF处理时间。因此,所提出的方法在编码效率和复杂度上做到了很好的折中处理,并且能很好地改善ALF的性能。

Adaptive loop filter （ALF） is a newly introduced technique for high-d efinition multi-view video coding. Its basic idea is to estimate Wiener filter parameters at the encoder and filter the reconstructed image with them to depress coding artifacts.ALF can improve video compression in both objective qu ality and subjective performance by adopting Wiener filtering with the goal of minimizing the mean square error b etween the original frame and the coded frame.However,the high computation complexity prevents it from the real -time application.Therefore,a low complexity algorithm is proposed to speed up the process of ALF.The basic i dea of the method is to exploit inter-view correlation and inter-layer correlation to reduce the cumbersome pr ocess.We not only narrow down the maximum division depth candidates of the chrominance component and luminance com ponent for every frame, but also adaptively terminate the process of deciding block control （DBC） of mos t frames in advance.Simulation results show that the proposed method can achieve the encoding time saving up to about 61% for ALF compared with the HEVC-based test model （i.e.,HTM 6.1） witho ut significant loss of encoding performance.Therefore,our algorithm can achieve a better tradeoff between coding efficiency and the ALF co mplexity,and well improve the performance of ALF.