Abstract: Based on the analysis of the next-generation High Efficiency Video Coding（HEVC） intra prediction Planar and DC mode algorithms, a high-efficiency Very Large Scale Integration Circuit（VLSI） design scheme is proposed, for solving the problem of large resource occupation and long processing delay. Aiming at Planar mode, a framework for predicting block multiplexing based on recombination and merging algorithm is proposed. For DC mode, a basic block separation of dcValue calculation and filtering is proposed, and the architecture of different blocks is reused. The experimental results show that the proposed architecture reduces the average processing delay by 21% and the resource consumption by 14.7% and 7% respectively compared with the other two architectures of the same type. The average processing delay of the DC mode is reduced by 55%. At the same time, resource consumption is reduced by 22% and 15%, which can meet the real-time encoding requirements of 1 920×1 080@30 f/s video sequences.

Key words: high efficiency video coding, intra prediction, Planar mode, DC mode, very large scale integration circuit

摘要： 在对新一代高效视频编码（High Efficiency Video Coding，HEVC）帧内预测Planar和DC模式算法分析的基础上，分别提出了高效的超大规模集成电路（Very Large Scale Integration Circuit，VLSI）设计方案，旨在解决处理延时较长、资源占用较大的问题。针对Planar模式，提出一种在重组、合并算法的基础上，预测块复用的架构；针对DC模式，提出一种dcValue计算和滤波的基本块分离、各自复用不同块的架构。实验结果表明：所提架构与其他两种同类型架构相比，Planar模式实现平均处理延时减少了21%，资源消耗分别减少了14.7%和7%；DC模式实现平均处理延时减少了55%，同时资源消耗减少了22%和15%，能够满足1?920×1?080@30?f/s视频序列实时编码的需求。

关键词: 高效视频编码, 帧内预测, Planar模式, DC模式, 超大规模集成电路