Abstract: Aiming at hierarchical B prediction structure utilized in Scalable Video Coding(SVC) standard, this paper proposes a picture-level Content-Adaptive Quantization Parameter Cascading(CAC) algorithm. The Quantization Parameter(QP) value for each picture is decided based on the hierarchical prediction structure and prediction modes of the macroblocks in the picture. Experimental result demonstrates that compared with the technique used in SVC standard, the algorithm increases rate distortion performance by 0.363 dB.

Key words: Scalable Video Coding(SVC), hierarchical prediction, content-adaptive Quantization Parameter(QP)

摘要： 针对可伸缩视频编码标准中采用的分级B帧预测结构，提出一种图像级的内容自适应量化参数(QP)级联算法。在该算法中，任意一帧图像的最终QP值由分级预测的结构及该图像内所有宏块的运动预测模式决定。实验结果证明，与目前SVC标准采用的技术相比，该算法最多可以提升0.363 dB的编码性能。

关键词: 可伸缩视频编码, 分级预测, 内容自适应量化参数

CLC Number: 

• TN919.81