Staggered Quantizers for Perfect Perceptual Quality: A Connection between Quantizers with Common Randomness and Without

TL;DR

This paper explores the connection between randomized and non-randomized quantizers within the rate-distortion-perception framework, introducing staggered quantizers to improve neural compression quality.

Contribution

It provides a novel interpretation of dithered quantization's advantage and proposes a new RDP coding method using staggered quantizers based on this insight.

Findings

Dithered quantization offers specific benefits in RDP settings.

A conceptual link between randomized and non-randomized quantizers is established.

A new RDP coding procedure using staggered quantizers is introduced.

Abstract

The rate-distortion-perception (RDP) framework has attracted significant recent attention due to its application in neural compression. It is important to understand the underlying mechanism connecting procedures with common randomness and those without. Different from previous efforts, we study this problem from a quantizer design perspective. By analyzing an idealized setting, we provide an interpretation of the advantage of dithered quantization in the RDP setting, which further allows us to make a conceptual connection between randomized (dithered) quantizers and quantizers without common randomness. This new understanding leads to a new procedure for RDP coding based on staggered quantizers.