基于提前终止编码单元划分的快速帧内预测算法

FAST INTRA PREDICTION ALGORITHM BASED ON EARLY TERMINATION OF CODING UNIT

为了降低高效视频编码HEVC(High Efficiency Video Coding)帧内预测过程的计算复杂度,提出一种提前终止编码单元(CU)划分的快速帧内预测算法。首先,针对HEVC编码结构,提出一种纹理复杂度测度的方法(即像素亮度方差值)。然后,分析不同纹理特性对帧内预测编码单元层次结构的影响,提出一种基于像素亮度方差值统计的提前终止编码单元的方法。对不同层次结构的编码单元进行像素亮度方差值统计,同时结合平均像素代价值统计,设置合理的阈值,自适应提前终止编码单元的划分,从而降低了帧内预测的编码复杂度。实验结果表明,在保证信噪比和比特率不变的情况下,该算法与HEVC参考软件HM10.0相比,编码时间平均节省32%。

This paper presents a fast intra prediction algorithm which is based on early termination of coding unit（ CU） partition in order to reduce the computational complexity of intra prediction process in HEVC. First,aiming at HEVC coding structure we put forward a texture complexity measurement method（ i. e. the pixel luminance variance value）. Then we analysed the influences of different texture features on CU hierarchical structure of intra prediction,and gave a new method for early termination of CU based on pixel luminance variance value statistics. By counting the luminance variance value of coding units with different hierarchical structure and in combination with counting the average pixel cost value,we set the reasonable threshold to adaptively terminate the CU partition early,so that reduced the coding complexity of intra prediction. Experimental results showed that comparing with the reference software HM10. 0 of HEVC,the algorithm reduced the encoding time by 32% in average with negligible changes in PSNR and bit rate.