Fast Intra Mode Decision for High Efficiency Video Coding (HEVC).docx

Fast Intra Mode Decision for High Efficiency Video Coding (HEVC)高效快速帧内模式决策视频编码（HEVC）Hao Zhang, Member, IEEE, and Zhan Ma, Member, IEEE张昊， IEEE会员，和马站，IEEE会员Abstract—The latest High Efficiency Video Coding (HEVC)摘要----必威体育精装版的高效视频编码（HEVC）standard only requires 50% bit-rate of the H.264/AVC at标准仅要求H.264/AVC比特率在50%the same perceptual quality, but with a significant encoder相同的感知质量，但有一个显着的编码器complexity increase. Hence, it is necessary and inevitable to复杂性的增加。因此，它是必要的和不可避免的develop fast HEVC encoding algorithms for its potential mar-快速发展的HEVC编码算法，其潜在的损坏—ket adoption. In this paper, we propose a fast intra modeKET收养。在本文中，我们提出了一个快速的帧内模式decision for the HEVC encoder. The overall fast intra mode为HEVC编码器的决定。整体快速帧内模式decision algorithm consists of both micro- and macro-level决策算法包括微观和宏观两个层次schemes. At the micro-level, we propose the Hadamard cost-方案。在微观层面上，我们提出了Hadamard成本—based progressive rough mode search (pRMS) to selectively基于渐进粗糙模式有哪些信誉好的足球投注网站（PRMS）选择性check the potential modes instead of traversing all candidates检查潜在的模式，而不是遍历所有候选人(i.e., up to 35 in HEVC). Fewer effective candidates will（即，高达35的HEVC）。较少有效的候选人将be chosen by the pRMS for the subsequent rate-distortion可选择由PRMS为随后的率失真optimized quantization (RDOQ) to derive the rate-distortion优化量化（rdoq）导出的率失真(R-D) optimal mode. An early RDOQ skip method is also（R-D）优化模式。早期的rdoq跳的方法是introduced to further the complexity reduction. At the macro-介绍了进一步的复杂性减少。在宏—level, we introduce the early coding unit (CU) split termination水平，我们介绍了早期的编码单元（铜）拆分终止if the estimated R-D cost [through aggregated R-D costs of如果估计RD成本[通过聚集RD成本(partial) sub-CUs] is already larger than the R-D cost of the（部分）子因为]已经比的率失真成本current CU. On average, the proposed fast intra mode decision目前的铜。平均而言，拟议的快速帧内模式决定provides about 2.5× speedup (without any platform or source提供约2.5的加速比（没有任何平台或来源code level optimization) with just a 1.0% Bjontegaard delta代码级优化）仅约1%的三角洲rate (BD-rate) increase using the HEVC common test condition.率（BD率）增加使用HEVC通用试验条件。Moreover, our proposed solution also demonstrat