Abstract: Rate control is one of the most important mechanisms affecting digital video quality in network environments. As for the problem for low-quality and fluctuation of video images in JVT-G012 algorithm, caused by unreasonable choice of I-frame quantization parameter, this paper put forward a selection algorithm of I-frame quantization parameter, based on the R-Q model. Firstly, R-Q model of I-frame was got by curve fitting; secondly, the information of encoded frames was used to update I-frame weight relative to P-frame, and a reasonable target-bit was allocated for I-frame; finally, I-frame quantization parameter was calculated under the target-bit by R-Q model. Simulation results show that compared to the H.264/AVC standard algorithm, the improved algorithm accurately controls the bit-rate while improving the average PSNR of testing sequences, and makes the quality of video images more stable and smooth.

Key words: rate control, JVT-G012, R-Q model, quantization parameter, Peak Signal-to-Noise Ratio (PSNR)

摘要： 在网络环境中，码率控制是影响数字视频质量的重要机制之一。针对提案JVT-G012算法中I帧量化参数选择的不合理造成的视频图像质量(PSNR)低、波动大的问题，提出了一种基于R-Q模型的I帧量化参数选择算法。首先，通过曲线拟合得到I帧的R-Q模型;其次，利用已编码帧的信息更新I帧相对于P帧的权重，为I帧分配合理的目标比特数;最后，在该目标比特下由R-Q模型计算得到I帧量化参数。实验结果表明: 与H.264/AVC原算法相比，改进的算法在准确控制码率的同时，视频图像平均质量得到提高且更加平稳。

关键词: 码率控制, JVT-G012, R-Q模型, 量化参数, 峰值信噪比