Context-Based Ternary Tree Decision Method in Versatile Video Coding for Fast Intra Coding

Sang-Hyo Park,Je-Won Kang

doi:10.1109/access.2019.2956196

Abstract

This paper presents a fast encoding method for versatile video coding (VVC) using an early determination scheme that skips redundant multi-type tree (MTT) pruning. MTT consists of binary and ternary trees (TTs) with traditional quadtrees and has recently attracted considerable VVC research interest due to coding efficiency beyond HEVC. However, the additional trees, particularly TTs, significantly increased VVC complexity, which has rarely been studied previously. Therefore, we identified TT characteristics in the VVC encoding context, and hence propose a simple early decision technique that can effectively reduce TT complexity. Experiments over various test sequences show that the proposed method substantially decreased the total encoding time for a VVC test model 4.0 (VTM 4.0) with only slight coding loss under the all-intra configuration. In particular, the proposed method reduced TT encoding time by up to 66% compared with VTM 4.0.

Highlights

Versatile video coding (VVC) is a new standardization project established by the Joint Video Experts Team (JVET) of ISO/IEC JTC 1/SC 29/WG 11 and ITU-T SG16 WP3 to improve compression efficiency compared with HEVC
We proposed a method to efficiently skip ternary trees (TTs), exploiting RD costs from the previously encoded coding units (CUs) data based on probabilistic approach
Experimental results confirmed that the proposed method significantly reduced VTM 4.0 encoding time while sustaining reasonable coding efficiency

Summary

INTRODUCTION

Versatile video coding (VVC) is a new standardization project established by the Joint Video Experts Team (JVET) of ISO/IEC JTC 1/SC 29/WG 11 and ITU-T SG16 WP3 to improve compression efficiency compared with HEVC. This paper proposes a context-based TT decision (C-TTD) method to significantly reduce TT computational complexity in VVC intra-coding. By skipping redundant TT split tests, the proposed method can improve the entire VVC encoding complexity. B. EXISTING TREE DECISION METHODS FOR FAST VIDEO ENCODING Various methods have been proposed to reduce complexity to make a recursive QT structure for a square block in HEVC. A pruning-based early CU decision was proposed to judge whether to split a block according to the best prediction mode [4]. A fast encoding method based on decision tree was proposed, targeting partitioning and prediction mode complexity [21]. In the random-access configuration, this approach showed 0.61% coding loss with 30% total encoding time saving These learning-based approach still required complex and intensive preprocessing to extract texture information for every CU, greatly increasing total encoder complexity. The proposed method can substantially decrease both BT and TT encoding complexity by pruning TT split processes

EXPERIMENTAL RESULT

Findings

CONCLUSION