Learning Optimal Linear Block Transform by Rate Distortion Minimization

TL;DR

This paper introduces a novel linear block transform called RDLT, optimized specifically for rate-distortion minimization, outperforming traditional DCT and existing data-driven transforms in image and video compression.

Contribution

The paper presents the RDLT, a data-driven, linear transform optimized for rate-distortion, designed for easy integration into existing standards, unlike nonlinear VAE-based methods.

Findings

RDLT significantly outperforms DCT and other data-driven transforms.

Integration of RDLT into VVC-like frameworks yields substantial quality improvements.

The method is practical and compatible with current compression standards.

Abstract

Linear block transform coding remains a fundamental component of image and video compression. Although the Discrete Cosine Transform (DCT) is widely employed in all current compression standards, its sub-optimality has sparked ongoing research into discovering more efficient alternative transforms even for fields where it represents a consolidated tool. In this paper, we introduce a novel linear block transform called the Rate Distortion Learned Transform (RDLT), a data-driven transform specifically designed to minimize the rate-distortion (RD) cost when approximating residual blocks. Our approach builds on the latest end-to-end learned compression frameworks, adopting back-propagation and stochastic gradient descent for optimization. However, unlike the nonlinear transforms used in variational autoencoder (VAE)-based methods, the goal is to create a simpler yet optimal linear block transform, ensuring practical integration into existing image and video compression standards. Differently from existing data-driven methods that design transforms based on sample covariance matrices, such as the Karhunen-Lo\`eve Transform (KLT), the proposed RDLT is directly optimized from an RD perspective. Experimental results show that this transform significantly outperforms the DCT or other existing data-driven transforms. Additionally, it is shown that when simulating the integration of our RDLT into a VVC-like image compression framework, the proposed transform brings substantial improvements. All the code used in our experiments has been made publicly available at [1].