Generalized Gaussian Model for Learned Image Compression

TL;DR

This paper introduces a generalized Gaussian model for learned image compression that balances modeling flexibility and complexity, leading to improved performance over traditional Gaussian models.

Contribution

The paper extends the Gaussian model to the generalized Gaussian family with an additional shape parameter and proposes training methods to reduce train-test mismatch.

Findings

Outperforms Gaussian and Gaussian mixture models in compression tasks

Enhanced training methods improve model robustness and accuracy

Achieves better compression performance with comparable complexity

Abstract

In learned image compression, probabilistic models play an essential role in characterizing the distribution of latent variables. The Gaussian model with mean and scale parameters has been widely used for its simplicity and effectiveness. Probabilistic models with more parameters, such as the Gaussian mixture models, can fit the distribution of latent variables more precisely, but the corresponding complexity is higher. To balance the compression performance and complexity, we extend the Gaussian model to the generalized Gaussian family for more flexible latent distribution modeling, introducing only one additional shape parameter beta than the Gaussian model. To enhance the performance of the generalized Gaussian model by alleviating the train-test mismatch, we propose improved training methods, including beta-dependent lower bounds for scale parameters and gradient rectification. Our proposed generalized Gaussian model, coupled with the improved training methods, is demonstrated to outperform the Gaussian and Gaussian mixture models on a variety of learned image compression networks.