论文部分内容阅读
提出一种新的基于运动估计的数据分割方法,在无线网络上传输视频流时能够有效地降低误差扩散。该方法将宏块作为基本的处理对象,根据两个方面将一个编码帧的所有宏块区分为重要性不同的级别:一方面是宏块对本帧图像质量的影响,另一方面是宏块对下一编码帧可能产生的误差扩散的影响。根据宏块中的像素被下一编码帧参考的次数总合来评估误差扩散的影响。在此基础上,码流被分为三种不同级别的子流,分别执行不同速率的信道编码,以保证在满足带宽约束的条件下,重要的信息和宏块数据能被优先保护。除了与其它方法一样能改善视频质量,本文方法的一个明显优势是平衡当前帧的质量与下一帧的质量。实验结果表明,比较263++和h.264中给出的数据分割方法,在两状态马尔可夫信道模型下,本文方法能获得更稳定的和更好的视频质量,提高PSNR值超过0.3dB。
A new method of data segmentation based on motion estimation is proposed, which can effectively reduce the error diffusion when transmitting video streams on wireless networks. This method uses the macroblock as the basic processing object and divides all the macroblocks of one coded frame into different levels according to two aspects: on the one hand, the effect of the macroblock on the image quality of the frame and on the other hand, The next encoded frame may have the effect of error diffusion. The effect of error diffusion is evaluated according to the sum of the number of times the pixels in the macroblock are referenced by the next coded frame. On this basis, the code stream is divided into three different sub-streams, and different channel coding rates are respectively implemented to ensure that the important information and the macro-block data can be preferentially protected when the bandwidth constraint is satisfied. In addition to improving video quality like any other method, one of the obvious advantages of our approach is to balance the quality of the current frame with the quality of the next frame. Experimental results show that comparing with the data partitioning method given in 263 ++ and h.264, the proposed method can obtain more stable and better video quality under the two-state Markov channel model and improve the PSNR value to exceed 0.3dB .