Low-complexity AV1 intra prediction algorithm☆

doi:10.1016/j.jvcir.2025.104464

Low-complexity AV1 intra prediction algorithm☆

2025-07-01

0

PRE

AI

DOI：10.1016/j.jvcir.2025.104464

Original

Original request for help

Save

Abstract

En 中文

As a new-generation video coding standard, Alliance for Open Media Video 1 (AV1) introduces flexible and diverse block partition types to improve coding efficiency, but also increases coding complexity. To address this issue, we propose a low-complexity AV1 intra prediction algorithm using Long-edge Sparse Sampling (LSS) and Chroma Migrating from Luma (CML) for efficiently encoding video sequences. First, we develop an LSS method by selecting key reference pixels based on block partition condition to reduce computational complexity. Second, we exploit a CML algorithm which combines the angle mode of the luma component and the spatial correlations of chroma components to derive more accurate linear model parameters between the luma and chroma components. Experimental results show that LSS avoids division operations, reducing 93% addition operations. Combined with CML, our approach saves 4.97% time and enhances coding performance compared to standard AV1, particularly improving chroma component quality.

Keywords:

Video coding

Intra prediction

Direct current

Chroma from Luma

Low-complexity AV1 intra prediction algorithm☆

Abstract

Journal

Organization