2022/03/18 by Pakshal Bohra, Bohra, Pakshal, Thanh-an Pham +5 · 1 citation
Computer Science · Earth and Planetary Sciences · Physics and Astronomy · #Advanced X-ray Imaging Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Generative Adversarial Networks and Image Synthesis #Seismic Imaging and Inversion Techniques #Signal Processing (eess.SP) #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2203.10078
openalex publication_date 2022/03/18 · openalex created_date 2022/06/13 · openalex updated_date 2026/07/28
Most modern imaging systems incorporate a computational pipeline to infer the\nimage of interest from acquired measurements. The Bayesian approach to solve\nsuch ill-posed inverse problems involves the characterization of the posterior\ndistribution of the image. It depends on the model of the imaging system and on\nprior knowledge on the image of interest. In this work, we present a Bayesian\nreconstruction framework for nonlinear imaging models where we specify the\nprior knowledge on the image through a deep generative model. We develop a\ntractable posterior-sampling scheme based on the Metropolis-adjusted Langevin\nalgorithm for the class of nonlinear inverse problems where the forward model\nhas a neural-network-like structure. This class includes most practical imaging\nmodalities. We introduce the notion of augmented deep generative priors in\norder to suitably handle the recovery of quantitative images.We illustrate the\nadvantages of our framework by applying it to two nonlinear imaging\nmodalities-phase retrieval and optical diffraction tomography.\n