Bagged Tree and ResNet-Based Joint End-to-End Fast CTU Partition Decision Algorithm for Video Intra Coding
Bagged Tree and ResNet-Based Joint End-to-End Fast CTU Partition Decision Algorithm for Video Intra Coding
复制标题
DOI：
10.3390/electronics11081264
复制
发表时间：
2022-04-01
期刊：
影响因子：
2.9
通讯作者：
Ling, Nam
中科院分区：
文献类型：
--
作者：
Li, Yixiao;Li, Lixiang;Ling, Nam
诸如高效视频编码(HEVC)、多功能视频编码(VVC)和AOMedia Video 2(AV2)等视频编码标准通过遍历编码单元(CU)分区的所有可能组合并选择具有最小编码成本的组合来实现最佳编码性能。由于HEVC是使用最广泛的编码标准之一，因此仍有必要进一步减少HEVC的编码时间。在HEVC中，寻找最佳性能的过程是大部分编码复杂度的来源。为了降低HEVC中编码块划分的复杂度，提出了一种新的端到端快速算法来辅助帧内编码中编码树单元(CTU)的划分结构决策。在该方法中，采用了一种新颖的两阶段策略来解决CTU的分区结构决策问题。在第一阶段，使用袋装树模型来预测CTU的分裂。在第二阶段，首次将32×32大小CU的划分问题建模为一个17个输出的分类任务，从而可以通过一次预测来解决该问题。为了达到较高的预测精度，采用了34层的残差网络(ResNet)。该快速CTU划分算法通过端到端的预测过程生成CTU的划分四叉树结构，摒弃了传统的在不同深度层次上进行多决策的方案。此外，本文使用了多个数据集，为获得较高的预测精度奠定了基础。实验结果表明，与原HM16.7编码器相比，该算法的编码时间平均减少了60.29%，而Bjontegaard Delta Rate(BD-Rate)丢失率仅为2.03%，优于目前大多数快速CU帧内划分方法的结果。
Video coding standards, such as high-efficiency video coding (HEVC), versatile video coding (VVC), and AOMedia video 2 (AV2), achieve an optimal encoding performance by traversing all possible combinations of coding unit (CU) partition and selecting the combination with the minimum coding cost. It is still necessary to further reduce the encoding time of HEVC, because HEVC is one of the most widely used coding standards. In HEVC, the process of searching for the best performance is the source of most of the encoding complexity. To reduce the complexity of the coding block partition in HEVC, a new end-to-end fast algorithm is presented to aid the partition structure decisions of the coding tree unit (CTU) in intra coding. In the proposed method, the partition structure decision problem of a CTU is solved by a novel two-stage strategy. In the first stage, a bagged tree model is employed to predict the splitting of a CTU. In the second stage, the partition problem of a 32 x 32-sized CU is modeled as a 17-output classification task for the first time, so that it can be solved by a single prediction. To achieve a high prediction accuracy, a residual network (ResNet) with 34 layers is employed. Jointly using bagged tree and ResNet, the proposed fast CTU partition algorithm is able to generate the partition quad-tree structure of a CTU through an end-to-end prediction process, which abandons the traditional scheme of making multiple decisions at various depth levels. In addition, several datasets are used in this paper to lay the foundation for high prediction accuracy. Compared with the original HM16.7 encoder, the experimental results show that the proposed algorithm can reduce the encoding time by 60.29% on average, while the Bjontegaard delta rate (BD-rate) loss is as low as 2.03%, which outperforms the results of most of the state-of-the-art approaches in the field of fast intra CU partition.