Learning Optimal Lattice Vector Quantizers for End-to-end Neural Image Compression

TL;DR

This paper introduces a novel learning method for optimizing lattice vector quantizers tailored to the distribution of neural image compression latent features, significantly enhancing rate-distortion performance.

Contribution

It proposes a rate-distortion optimal lattice vector quantization (OLVQ) method that adapts to latent feature distributions, improving compression efficiency over traditional LVQ.

Findings

Enhanced rate-distortion performance in neural image compression.

Better fit of LVQ structures to latent feature distributions.

Retains computational efficiency similar to scalar quantization.

Abstract

It is customary to deploy uniform scalar quantization in the end-to-end optimized Neural image compression methods, instead of more powerful vector quantization, due to the high complexity of the latter. Lattice vector quantization (LVQ), on the other hand, presents a compelling alternative, which can exploit inter-feature dependencies more effectively while keeping computational efficiency almost the same as scalar quantization. However, traditional LVQ structures are designed/optimized for uniform source distributions, hence nonadaptive and suboptimal for real source distributions of latent code space for Neural image compression tasks. In this paper, we propose a novel learning method to overcome this weakness by designing the rate-distortion optimal lattice vector quantization (OLVQ) codebooks with respect to the sample statistics of the latent features to be compressed. By being able to better fit the LVQ structures to any given latent sample distribution, the proposed OLVQ method improves the rate-distortion performances of the existing quantization schemes in neural image compression significantly, while retaining the amenability of uniform scalar quantization.