一种空域可分级的自适应视频码流控制方法

在传统的CBR码流控制中，为防止缓冲器上溢，通常采用跳帧的方法来保持码率恒定。由于跳帧引起解码端运动补偿误差增大，客观上造成跳帧期间图像PSNR降低，图像失真度迅速增加，在主观上造成图像有跳跃感，画面不流畅。针对以上缺陷，该文基于H．263＋的TMN8帧级码流控制算法，提出了一种窄域可分级的自适应码流控制方法。该方法在编码端对原跳帧方法中不编码的帧进行自适应下采样编码，解码端通过上采样恢复。实验结果证明，该文提出的码流控制算法在低带宽条件下，图像质量较跳帧方法有较大提高。在低码率条件下，对于高运动图像该方法的平均PSNR值比采用跳帧的码流控制方法高0．3～0．8dB；比采用增大量化步长的方法高0．1-0．4dB，而对于跳帧阶段PSNR值，该算法比传统的跳帧方法高1-2dB左右。

A New Adaptive Rate Control Algorithm with Spatial Scalability for Video Coding

In the traditional CBR rate control algorithm, the skipping method is adopted to combat the overflow of video buffer, in which the motion compensation error and picture distortion are increased with the number of skipping frame. To deal with the disadvantage of traditional skipping method, a new adaptive rate control algorithm with spatial scalability for video coding is presented in this paper. In the method, the original skipped frame is adaptively sub-sampled at encoder and is up-sampled in decoder. Simulation results show that this scheme can highly improve the quality of reconstructed video and obtain a gain of about0.3-0.8dB and 0.1-0.4dB in PSNR respectively compared with the traditional skipping method and the method of increasing quantization step.