基于时空域复杂度特征的LCU级码率分配算法

A LCU-Level Bitrate Allocation Algorithm Based on Spatial-Temporal Domain Complexity

提出一种新的最大编码单元(Largest Coding Unit,LCU)级码率分配算法,在保证码率准确性的前提下提升视频质量。首先,选择更符合人眼视觉特性的结构相似性(Structural Similarity,SSIM)图像评价算法来衡量每个LCU的空域复杂度,通过已编码帧的失真度来衡量每个LCU的时域复杂度,基于多元线性回归函数对时空域复杂度特征与实际编码比特数之间的关系进行数学建模。最后,依据时空域复杂度特征因子指导编码帧中每个LCU分配合理的目标码率。实验结果表明,在保证码率精确性的情况下,所提的LCU级码率分配算法具有较高的编码效率,率失真性能平均提高了1.25%。

In this paper,a new Largest Coding Unit(LCU)-level bitrate allocation algorithm is proposed to improve the video quality while ensuring the bitrate accuracy.Firstly,the Structure Similarity Index Measure(SSIM)image evaluation algorithm,which is more conforming to the visual characteristics of the human eye,is chosen to measure the spatial domain complexity of each LCU,and the time domain complexity of each LCU by the distortion of the encoded frames.Then,the relationship between the spatial-temporal domain complexity and coded bits is mathematically modelled based on the multiple linear regression function.Finally,the spatial-temporal domain complexity characteristics factor is used to allocate target bits for each LCU in the coded frame.The experimental results show that the LCU-level bitrate allocation algorithm proposed in this paper has high coding efficiency and the rate distortion performance is improved by 1.25%on average under the guarantee of bitrate accuracy.