针对新一代视频压缩编码标准HEVC计算复杂度较高的特点,利用视频序列间时域上的相关性,提出了一种基于灰度差值的编码单元快速划分策略,该策略根据当前编码块与参考块之间的灰度差值进行运动条件判决,在进行编码之前提前确定当前编码单元的编码深度信息,减少帧间预测编码的次数,从而有效地降低了编码端的计算复杂度。实验结果表明,该算法在编码效率和峰值信噪比(PSNR)损失都很小的情况下,和HM标准中的帧间预测算法相比,平均降低了50.18%的编码时间。
With respect to the eomputatinnal complexity of the High Efficiency Video Coding(HEVC) , a rapid partitioning strategy of coding trail (CU) which is based on the gray value difference algorithm by using the strong temporal correlation in video sequences is proposed. The method is to determine the depth of the current coding unit before coding process, according to the differences of gray level between the current coding block and referencing block, which can reduce the number of times of inter - frame prediction and lower the computation complexity of the encoding. As a result, in the cases that both coding efficiency and PSNR are tiny, the coding lime is decreased by 50. 18% when enmpared with the inter prediction algorithm recommended in HM standard.