基于高效率视频帧内编码的快速编码单元和预测模式决策算法

针对高效率视频编码采用四叉树分割和35种模式预测等技术来提高视频编码质量，但引起算法复杂度剧增这一问题，本文利用当前编码单元及其相邻单元空间相关性对当前编码单元分割模式进行预测，从而快速决策编码单元的分割模式.此外，通过跳过若干低概率预测模式的率失真最优化处理过程，实现了一种快速的预测模式决策算法.测试结果显示，提出的算法可以平均减少53%的编码时间，并保持BDBR仅为1.17%的良好编码质量. 

Fast CU Depth Decision and Direction Mode Decision Algorithm for HEVC Intra Coding

High efficiency video coding is the latest video coding standard. It adopts a quad-tree structure and 35 direction modes to improve coding performance, however, is at the expense of high coding complexity. To speed up the encoder, a fast CU depth decision method was proposed in this paper. Firstly, a spatial correlation between the current CU and its neighboring CUs was utilized to predict the depth for the current CU, accordingly a quick decision could be made for the division mode of CU. And then, the direction mode decision algorithm was implemented by a rate-distortion optimization process, skipping some unimportant direction modes. Test results show that the proposed algorithm can achieve 53% coding time saving on average, and can keep a better coding quality, while BDBR is only 1.17%.