Joint recovery of an image and its point spread function using a physics-based model with natural image and kernel priors. Supports multiple PSF types, MAP optimization, and scalable experiment sweeps with logging.
pytorch unsupervised-learning inverse-problems computational-imaging image-restoration image-deblurring blind-deconvolution diffusion-models psf-estimation image-priors
-
Updated
Dec 21, 2025 - Python