Close
Go To Macao Polytechnic University

2020/2021

More probability estimators for CABAC in versatile video coding

2020 IEEE 5th International Conference on Signal and Image Processing (ICSIP 2020)*, IEEE, Virtual, Nanjing, China, 2020: 366-370

Author(s)Sio-Kei Im,
Ka-Hou Chan
Summary

In the next-generation Versatile Video Coding (VVC), its Context-based Adaptive Binary Arithmetic Coding will have evolved to use linear quantization and multi-probability estimators for prediction, instead of using the finite state table and lookup table in the current standard. This paper highlights the key techniques that involve more probability estimators to enable VVC to potentially achieve a linear quantization representation of the probability state, and a multiplication-free operation can be achieved through a series of bit operations. This article focuses on key techniques involving more probability estimators, so that VVC can linearly achieve a quantized representation of probability prediction and describes the scalability potential for higher accuracy. The BD-RATE gain of the proposed approach is up to 4.0% for all-intra mode and 2.0% for inter mode, the improvement showing that gain is provided in the encoding, and the proposed method can be used in the next generation VVC standard.


* Also listed in Scopus.

Top Top