论文部分内容阅读
为实现时间可伸缩性,可伸缩视频编码采用分级B帧结构进行编码,但是该结构较大地增加了计算的复杂度。对此,本文深入分析了模式选择的基本原理,提出一种适合于分级B帧结构的快速编码算法,首先根据参考模式进行预测,然后根据宏块的运动和纹理预测可能的编码模式,排除可能性较小的模式从而提高编码速度。实验表明,本文算法在视觉质量和网络带宽影响很小的情况下,能够显著地提高编码速度。
For temporal scalability, scalable video coding uses a hierarchical B-frame structure for encoding, but this structure greatly increases the computational complexity. In this paper, the basic principle of pattern selection is analyzed in depth and a fast coding algorithm suitable for hierarchical B frame structure is proposed. Firstly, the prediction is based on the reference pattern and then the possible coding modes are predicted according to the motion and texture of the macroblock. Smaller mode to improve the encoding speed. Experiments show that the proposed algorithm can significantly improve the coding speed with little impact on visual quality and network bandwidth.