基于H.264平台就多视点视频压缩中多参考帧技术存在计算复杂度高的现状,根据多视点视频序列特性,提出了一种针对多参考B帧的快速多视点视频编码压缩算法。新算法改进了多参考帧列表策略,并结合多视点视频序列的时空域相关性特性,提出了适用于多视点视频编码的参考帧列表调整策略。此外,还针对多种块模式编码引入了块模式预判策略。新算法在保证率失真性能的基础上,有效地降低了计算复杂度,提高了多视点视频编码速度。虽然本文是基于M-Picture编码结构实现了新算法,但它同样适用于其它应用多参考B帧预测技术的多视点视频编码结构中。
Considering high complexity of multiple reference frames in Multi-view Video Compression (MVC), a fast coding algorithm for MVC is presented based on the traits of multi-view video. The propose algorithm improves the design of reference-lists in H.264 standard that adapts to traditional video coding, and adjusts the old reference-lists using the temporal and spatial correlations of multi-view video. Additionally, the strategy of mode pre-decision is also used to reduce the computational complexity. The experimental results show that the proposed method can reduce encoding complexity and speed up encoding process effectively. Though the new algorithm is implemented on M-Picture prediction structure, it can be also used in other multi-view video coding structures with multiple reference B-frames.