摘要
H．264标准中亮度分量的9种帧内预测模式的序号是预先设定的，这对于具体的视频序列并不是最优的。为了进一步提高帧内编码的效率，通过对帧内预测编码算法的深入研究，利用相邻宏块间的纹理相关性和预测模式的方向性，提出一种新的帧内预测模式编码算法。通过在和当前编码块纹理相关性最大的区域中动态的统计出各个预测模式的使用概率，然后根据预测模式使用概率的大小来计算当前块的最可能编码模式。实验结果表明：与H．264参考模型JM86相比，该算法可以显著提高当前图像块的最优预测模式和最可能预测模式的匹配概率，使预测模式信息编码所需要的比特数平均减少5％～7％，从而降低了编码后的码率，而峰值信噪比（PSNR）基本保持不变。
There are a total of nine prediction modes for each 4×4 luma block in H.264 video coding standard to achieve coding efficiency, However, the number of each mode is preassigned. Obviously this method is not suit for every video sequences. In order to improve the intra coding efficiency, special research efforts are made on intra prediction coding and an efficient Intra prediction mode coding algorithm is proposed based on regional maximum probability criterion. Experimental results show that the proposed scheme is able to achieve more most_probable_modes, with the reduction of the bit rate for prediction mode coding is between 5%～7%, and the negligible peak signal-noise ratio （PSNR） loss compared with the original H.264 reference software.
出处
《电路与系统学报》
CSCD
北大核心
2008年第5期53-57,52,共6页
Journal of Circuits and Systems
