摘要
新一代国家自主音视频编码标准AVS2已制定完成,相对于上一代标准其具有更高的编码效率,但同时编码复杂度大幅度增加。为了降低编码器的编码复杂度以达到实时应用性能,针对编码单元的预测单元划分进行深入研究,提出了一种帧间预测模式快速判决方法,该方法利用编码单元时域与空域的信息再结合编码单元的纹理信息来预测其划分方式,从而在帧间编码过程中避免遍历所有划分模式。实验表明,该算法可以使编码时间平均下降38%,并且对编码效率影响很小。
The new generation of the national independent audio and video coding standard AVS2 has been finalized. It has higher coding efficiency relatives to the previous generation, hut it also increases a substantial the encoding complexity. In order to reduce AVS2 encoding complexity to achieve real-time application performance, this paper makes in-depth research on prediction unit of encoding unit and proposes a fast decision algorithm of inter prediction mode. The algorithm utilizes spatio-temporal correlation and the texture feature of coding unit to forecast coding unit' s partition mode, which can avoid iterating over all partition modes. Experiments show that the algorithm can reduce encoding time by 38% on average and the algorithm has only a little impact on encoding efficiency.
出处
《电视技术》
北大核心
2016年第11期1-6,共6页
Video Engineering
基金
上海市自然科学基金项目(14ZR1415200)
上海市"科技创新行动计划"项目(14511105602)