基于宏块的具有时域和SNR精细可伸缩性的视频编码

提出了一个高效灵活的视频编码方法，称为基于宏块的具有时域和SNR精细可伸缩性的视频编码方法（简称为PFGST视频编码方法），将原有的基于帧的渐进精细可伸缩的视频编码技术扩展为基于宏块的编码技术，即增强层编码中运动补偿和重建时所用的参考信息是基于宏块选择的而不是基于帧、然后将时域可伸缩特性引入到基于宏块的渐进精细可伸缩的视频编码中，从而实现了PFGST编码方案，在时域可伸缩的增强层编码中，根据运动补偿中使用的参考宏块的不同，提出了时域可伸缩增强层宏块编码的两种方法，由于在时域可伸缩的增强层编码中使用高质量的参考宏块不会造成任何误差传播，因此通过选用最佳的参考宏块，PFGST方法的编码效率得到了显著的提高，实验结果显示，同MPEG－4标准中的FGST编码方法相比，基于宏块的PFGST视频编码技术的编码效率提高了2．8dB，并且同样具有FGST的所有特性，即可以根据不同的通道，客户和服务器的需求来分别支持精细的SNR可伸缩特性，时域可伸缩特性和SNR－时域混合可伸缩特性。

Macroblock-Based Progressive Fine Granularity Scalable (PFGS) Video Coding with Flexible Temporal-SNR Scalabilities

A flexible and efficient architecture for scalable video coding is proposed in this paper, namely, the macroblock (MB) based progressive fine granularity scalable video coding with hybrid temporal SNR scalability (PFGST in short). In order to develop a high efficient hybrid temporal-SNR scalable coding scheme, authors first extend the progressive fine granularity scalable (PFGS) coding from the frame level to the macroblock level, in which the references for temporal prediction and reconstruction are selected per macroblock instead of per frame in the enhancement layer coding. Furthermore, temporal scalability is introduced into the MB based PFGS to form the MB based PFGST. Two modes are proposed for the temporal enhancement macroblocks coding corresponding to the different references used in the temporal prediction. Since it would not cause any error propagation even if using the high quality reference in the temporal enhancement macroblock coding, PFGST can achieve high coding efficiency by choosing the most suitable reference for motion compensation. Experimental results show that the MB based PFGST video coding scheme can significantly improve the coding efficiency up to 2.8dB compared with MPEG 4 FGST, meanwhile supporting SNR, temporal, and hybrid SNR temporal scalabilities to satisfy the different requirements from channels, clients and/or servers.