基于RDWT-子带运动补偿的可扩展视频编码方法

Highly Scalable Video Codec Based on RDWT-Subband Motion Compensation

随着技术的发展,视频编码方法要求的不再仅是压缩效率,还包括码流的可扩展性.本文提出了一种基于RDWT-子带运动补偿的可扩展视频编码方法.首先对输入帧组进行空域DWT分解,在其后集成运动补偿过程的时域分解步骤中,使用冗余离散小波变换(RDWT)分解得到的各相位子带作为候选,选取一个最佳相位子带作为预测,提高运动补偿效率;在编码阶段,提出基于子带扫描顺序的改进3DSPIHT技术,保障码流的扩展特性.实验表明,所提出的方法获得较高编码效率的同时,提供了码流的高度可扩展特性.

With development of technology, video codec demands not only the compression efficiency, but also other important functionalities, such as scalabilities. This paper presents a highly scalable video codec based on RDWT （Redundant Discrete Wavelet Transform）-Subband motion compensation, which first spatially decomposes the input video group of frames, then in motion compensation temporal filtering uses all phase RDWT subbands as candidates and selects one best-matching RDWT subband as references to improve the efficiency of motion compensation. In order to keep the scalabilities of wavelet coefficients the codec also utilizes an amended 3D SPIHT method with new scan order. The experimental results prove that the codec can produce encoded stream with higher PSNR and highly scalable functions.