一种改进的JVT-H017码率控制算法

摘要

This paper proposes an improved rate control algorithm for the basic-unit layer encoder of JVT-H017. Adopting a new weighted time and space model to predict the Mean Absolute Difference(MAD) and according to the Human Visual System(HVS) characteristic to design a weighted Y component of Peak Signal to Noise Ratio(PSNR), it recalculates the picture complexity of current basic-unit layer with the new MAD and PSNR, which highlights the importance of Y component compared to UV component, and then allocates target bits with above. Experimental results show that compared with the algorithms of JVT-H017, proposed algorithm is closer to the target bit, the average error rate reduces 0.269%. The average of PSNR raises 0.278 dB. The proposed algorithm still ensures the quality of the video after transporting in the 3G wireless channel.%提出一种改进的JVT-H017基本单元层码率控制算法.采用时空加权模型预测平均绝对差值(MAD),根据人类视觉系统特性设计加权Y分量的峰值信噪比(PSNR),联合新的MAD和PSNR估计基本单元层的复杂度,突出Y分量相对UV分量的重要性,并依此分配目标比特.实验结果显示,相比JVT-H017算法,改进算法更接近目标码率,码率误差平均降低0.269％,峰值信噪比平均提高0.278 dB,可保证具有误码的3G无线网络中视频图像的质量.