摘要
为了降低高效视频编码(HEVC)的复杂度,提出了一种基于纹理方向和空域相关性的帧内快速编码算法。一方面,利用相邻编码单元(CU,coding unit)的最佳分割尺寸确定相关性的权重,并根据已编码CU的率失真代价值定义了CU分割尺寸的相关性因子,通过比较该因子提前终止CU分割;另一方面,利用Sobel算子求取预测单元(PU,prediction unit)子块的纹理方向,通过判定其纹理方向显著性确定相应的候选模式集,然后根据PU大小对所得的预测模式修正处理,最后遍历这些候选模式选取最优模式。实验结果表明:本文算法相对于原始HEVC编码方法,在全I帧模式下编码时间平均减少36.84%,BDBR(Bjntegaard delta bit rate)上升约0.81%,BDPSNR(Bjntegaard delta peak signal-to-noise rate)降低了0.047dB。
In order to reduce the encoding computational complexity of the emerging high efficiency video coding (HEVC) ,a fast intra coding algorithm with spatial correlation whose development is based on the texture has been proposed. On one hand,a weighting factor is proposed through the effective selection of the optimal size of adjacent coding unit (CU). Then,a correlation factor is defined based on rate distortion cost and this factor is able to terminate the division of the CU ahead of time. On the other hand,Sobei operator is utilized to get the texture direction of the sub-block of prediction unit (PU). By judging the conspicuousness of the PU texture direction,a corresponding candidate mode Set is selected. Furthe more, further refinement which means amending the prediction mode according to the PU size is done. And finally,all of the coding modes in the mode set are traversed and the mode selection is finished. Experimental results show that the proposed method reduces encoding time of 36.84% on average, with a BD-hit rate increase of 0. 81% and a BD-PSNR 10ss of 0. 047 dB compared with the original HEVC encoding scheme. Consequently,the proposed algorithm can reduce the encoding computational complexity efficiently,while the rate-distortion performance remains unchanged practically.
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2014年第9期1777-1784,共8页
Journal of Optoelectronics·Laser
基金
国家自然科学基金(61171163
61271270
61111140392
U1301257)资助项目