Predicting Chroma from Luma in AV1

TL;DR

This paper introduces a new chroma prediction method called CfL in AV1, which improves accuracy and reduces decoder complexity, leading to significant bitrate savings in still images and videos.

Contribution

The paper presents the integration of CfL prediction into AV1, demonstrating its advantages over previous methods in accuracy and complexity reduction.

Findings

Reduces BD-rate by 5% for images

Reduces BD-rate by 2% for videos

Produces more accurate chroma predictions

Abstract

Chroma from luma (CfL) prediction is a new and promising chroma-only intra predictor that models chroma pixels as a linear function of the coincident reconstructed luma pixels. In this paper, we present the CfL predictor adopted in Alliance Video 1 (AV1), a royalty-free video codec developed by the Alliance for Open Media (AOM). The proposed CfL distinguishes itself from prior art not only by reducing decoder complexity, but also by producing more accurate predictions. On average, CfL reduces the BD-rate, when measured with CIEDE2000, by 5% for still images and 2% for video sequences.