论文部分内容阅读
根据视频序列时域相关性强的特性,提出了一种基于时域的HEVC快速帧内预测算法。在帧间预测中,通过在参考帧中运动搜索找到当前预测单元(PU)的帧间最佳匹配块,当搜索当前PU的帧内预测模式时,把帧间最匹配参考块遍历帧内模式时的最佳帧内模式,当作当前PU进行帧内预测时的最佳帧内模式,以达到减少帧内预测模式数量,降低计算复杂度的目的。实验结果表明,在码率和图像质量基本不变的情况下,提出的算法将HEVC帧内预测时间平均减少了33.79%。
According to the strong temporal correlation of video sequences, a fast HEVC intra prediction algorithm based on time domain is proposed. In inter prediction, finding the inter-optimally matching block of the current prediction unit (PU) by motion search in a reference frame traverses the intra-frame best match reference block when searching for the intra-prediction mode of the current PU , The optimal intra mode is regarded as the optimal intra mode when the current PU performs intra prediction, so as to reduce the number of intra prediction modes and reduce the computational complexity. The experimental results show that the proposed algorithm reduces the HEVC intra prediction time by an average of 33.79% with the same code rate and image quality.