Abstract In combination with Internet-of-things (IoT) technology, realistic media will continue to evolve with more smart and intelligent media services via interactions between users and displays. Therefore, it is important to efficiently compress smart contents on low-power and low-complexity hand-held displays. High efficiency video coding (HEVC) is the latest video coding technology that can achieve excellent coding performance, but at the cost of tremendously increased complexity due to the use of newly adopted video coding tools. Most previous fast encoding methods were mainly investigated in terms of spatio-temporal correlation between the current coding unit (CU) and neighboring CUs. As HEVC has a recursive quad-tree structure, the correlation between different CU depths can be higher than the spatio-temporal correlation. In this study, a fast CU mode decision method is proposed for estimating the mode of a current CU to be encoded either as intra or inter mode prior to encoding a current CU. The method uses the information of the best upper CU mode, where an upper CU indicates a larger block than a current CU in the quad-tree structure. The experimental results show that the encoding process of the proposed method is faster than that of a previous method with an unnoticeable coding loss.

中文翻译：

---

一种基于上层编码单元模式继承的快速编码单元模式决策方法，用于智能内容的低复杂度压缩

摘要 结合物联网（IoT）技术，现实媒体将通过用户和显示器之间的交互，以更加智能和智能的媒体服务不断发展。因此，在低功耗、低复杂度的手持显示器上高效压缩智能内容非常重要。高效视频编码 (HEVC) 是最新的视频编码技术，可以实现出色的编码性能，但代价是由于使用了新采用的视频编码工具，复杂性大大增加。大多数以前的快速编码方法主要是根据当前编码单元（CU）和相邻 CU 之间的时空相关性来研究的。由于 HEVC 具有递归四叉树结构，因此不同 CU 深度之间的相关性可以高于时空相关性。在这项研究中，提出了一种快速 CU 模式决策方法，用于在对当前 CU 进行编码之前估计要被编码为帧内或帧间模式的当前 CU 的模式。该方法使用最佳上CU模式的信息，其中上CU表示四叉树结构中比当前CU更大的块。实验结果表明，所提出的方法的编码过程比以前的方法快，编码损失不明显。