【摘要】 针对实时视频通信具有的低时延特性,通过研究速率模型形状参数和图像内容之间存在的关系,提出了一种视频图像自适应速率模型,此模型较其它速率模型能够更好地拟合真实码率输出,使得基于该模型速率控制下的输出速率更加平稳。为了避免TMN8模型中使用宏块方差方法得到宏块量化步长所引入的计算量,在帧层一次计算得到视频图像的基本量化步长,并在宏块层对其进行简单调整,从而降低了系统的处理时延。仿真实验表明,此文提出的基于新模型的速率控制算法较TMN8在输出码率稳定性及系统时延上都具有较好的性能。更多还原

【Abstract】 According to the relationship between the shape parameter of the rate model and the image content, a novel rate control algorithm based on an adaptive rate model in video coding for low-delay communications is presented. Compared with other rate models, the model can fit the real coding rate better, and the output rate is more stability by using the rate control scheme based on the model. In order to avoid the computational complexity introduced by the calculation of macroblock (MB) variance in Test Model Near-Term 8 (TMN8), the basic quantization step of video image coding is calculated in frame level and adjusted in the MB level so as to reduce the system processing delay. The results of experiments show that the proposed rate control algorithm based on the adaptive rate model is better than TMN8 in the aspect of stability of output rate and system delay.更多还原