Zero-shot inverse imaging with auto-tuned hyperparameters and untrained generative priors

Inverse imaging aims to reconstruct underlying images from noisy and corrupted observations. Conventional model-based methods heavily rely on prior informatioand hyperparameter tuning, while recent deep learning-based methods inherently require large training datasets. This study presents a robust zero-shot inverse imaging approach based on untrained generative priors. Our method combines constrained model-based reconstruction with an untrained generative model as a prior, enabling simultaneous network training and target image estimation through an iterative process. The key contributions of this study include: (1) proposing a new method for simultaneous image reconstruction and prior learning, thereby eliminating the need for training datasets; (2) introducing an automated hyperparameter tuning for both image reconstruction and network training; and (3) proposing an ensemble of networks with automated outlier removal, ensuring enhanced inverse imaging robust to training failure. Our method is evaluated across various inverse imaging tasks, consistently outperforming existing methods without requiring large training datasets and hyperparameter tuning.