Reducing Complexity on Coding Unit Partitioning in Video Coding: A Review

Abstract

In this article, we present a survey on the low complexity video coding on a coding unit (CU) partitioning with the aim for researchers to understand the foundation of video coding and fast CU partition algorithms. Firstly, we introduce video coding technologies by explaining the trending standards and reference models. They are High Efficiency Video Coding (HEVC), Joint Exploration Test Model (JEM), and VVC, which introduce novel quadtree (QT), quadtree plus binary tree (QTBT), quadtree plus multi-type tree (QTMT) block partitioning with expensive computation complexity, respectively. Secondly, we present a comprehensive explanation of the time-consuming CU partitioning, especially for researchers who are not familiar with CU partitioning. The newer the video coding standard, the more flexible partition structures and the higher the computational complexity. Then, we provide a deep and comprehensive survey of recent and state-of-the-art researches. Finally, we include a discussion section about the advantages and disadvantage of heuristic based and learning based approaches for the readers to explore quickly the performance of the existing algorithms and their limitations. To our knowledge, it is the first comprehensive survey to provide sufficient information about fast CU partitioning on HEVC, JEM, and VVC.