Accelerating block-level rate control for learned image compression

TL;DR

This paper introduces a block-level rate control method for learned image compression that significantly speeds up encoding while maintaining high accuracy, leveraging a novel D-λ model and inter-block correlations.

Contribution

It proposes a new block-wise R-D prediction algorithm and a D-λ model tailored for LIC, enabling fast and accurate rate control at the block level.

Findings

Achieves up to 100x speed-up in rate control

Maintains over 98% accuracy in bitrate approximation

Improves overall compression performance

Abstract

Despite the unprecedented compression efficiency achieved by deep learned image compression (LIC), existing methods usually approximate the desired bitrate by adjusting a single quality factor for a given input image, which may compromise the rate control results. Considering the Rate-Distortion (R - D) characteristics of different spatial content, this work introduces the block-level rate control based on a novel D - {\lambda} model specific for LIC. Furthermore, we try to exploit the inter-block correlations and propose a block-wise R - D prediction algorithm which greatly speeds up block-level rate control while still guaranteeing high accuracy. Experimental results show that the proposed rate control achieves up to 100 times, speed-up with more than 98% accuracy. Our approach provides an optimal bit allocation for each block and therefore improves the overall compression performance, which offers great potential for block-level LIC.