A Hardware-Friendly Fast VVC Test Zone Search Algorithm Using Machine Learning
Ramiro Viana, Fernando Sagrilo, Rafael Ferreira, Marta Breunig Loose, Marcelo Porto, Guilherme Corrêa, Luciano Volcan Agostini · 2024
The Versatile Video Coding (VVC) currently stands as the state-of-the-art in video coding efficiency, but with the cost of substantial computational effort and, consequently, high energy consumption. This elevated energy usage poses a critical challenge, particularly for applications designed for battery-powered devices. A key component of the VVC is the Test Zone Search (TZS) algorithm, known for its demanding computational requirements. This paper addresses this issue by introducing a machine learning based solution for the TZS algorithm. Notably, this solution is hardware-friendly, leveraging Decision Trees that are straightforward to implement in hardware. The proposed solution achieved an average 86.98 % reduction in TZS encoding time with a mere 0.45% impact on BD-BR.