Regularization via deep generative models: an analysis point of view

TL;DR

This paper introduces an analysis-based regularization method using deep generative models for inverse imaging problems, demonstrating improved performance and robustness over synthesis-based approaches across tasks like inpainting, deblurring, and super-resolution.

Contribution

It proposes a novel analysis formulation for regularization with deep generative models, directly optimizing images and penalizing latent vectors, unlike previous synthesis-based methods.

Findings

Improves performance in inpainting, deblurring, and super-resolution tasks.

Shows increased robustness to initialization.

Achieves better results compared to synthesis-based approaches.

Abstract

This paper proposes a new way of regularizing an inverse problem in imaging (e.g., deblurring or inpainting) by means of a deep generative neural network. Compared to end-to-end models, such approaches seem particularly interesting since the same network can be used for many different problems and experimental conditions, as soon as the generative model is suited to the data. Previous works proposed to use a synthesis framework, where the estimation is performed on the latent vector, the solution being obtained afterwards via the decoder. Instead, we propose an analysis formulation where we directly optimize the image itself and penalize the latent vector. We illustrate the interest of such a formulation by running experiments of inpainting, deblurring and super-resolution. In many cases our technique achieves a clear improvement of the performance and seems to be more robust, in particular with respect to initialization.