Abstract:Recently, various methods have been proposed to solve Image Restoration (IR) tasks using a pre-trained diffusion model leading to state-of-the-art performance. However, most of these methods assume that the degradation operator in the IR task is completely known. Furthermore, a common characteristic among these approaches is that they alter the diffusion sampling process in order to satisfy the consistency with the degraded input image. This choice has recently been shown to be sub-optimal and to cause the restored image to deviate from the data manifold. To address these issues, we propose Blind Image Restoration via fast Diffusion inversion (BIRD) a blind IR method that jointly optimizes for the degradation model parameters and the restored image. To ensure that the restored images lie onto the data manifold, we propose a novel sampling technique on a pre-trained diffusion model. A key idea in our method is not to modify the reverse sampling, i.e., not to alter all the intermediate latents, once an initial noise is sampled. This is ultimately equivalent to casting the IR task as an optimization problem in the space of the input noise. Moreover, to mitigate the computational cost associated with inverting a fully unrolled diffusion model, we leverage the inherent capability of these models to skip ahead in the forward diffusion process using large time steps. We experimentally validate BIRD on several image restoration tasks and show that it achieves state of the art performance on all of them. Our code is available at <a class="link-external link-https" href="https://github.com/hamadichihaoui/BIRD" rel="external noopener nofollow">this https URL</a>.

Efficient blind image restoration using discrete periodic radon transform

Efficient Blind Image Restoration Based on 1-D Generalized Cross Validation

Blind Deconvolution Using an Improved L0 Sparse Representation

Restoration method of TDI remote sensing image based on optimization of numerical fidelity term

Blind restoration of compressed degraded image based on block effect suppression

Iterative Blind Image Restoration With Ringing Reduction

Fast non-iterative blind restoration of hyperspectral images with PSFs

A Robust Blind Deblurring Method for Motion Images Based on Edge Information

Restoration of the Image Blurred by Motion Based on High-speed CCD Motion Detection

A Blind Restoration Method for Remote Sensing Images

A Blind Restoration Method of Three Dimensional Positron Emission Tomography Image

Reconstruction of Single Image from Multiple Blurry Measured Images

A Method for Image Restoration with Iterative Blind Deconvolution in Frequency Domain

Blind Image Restoration Algorithm in Bayesian Framework

Blind Image Restoration via Fast Diffusion Inversion

A new approach of blind image restoration

Threshold based Approach for Image Blind Deconvolution

Blind Image Deblurring with FFT-ReLU Sparsity Prior

Image Restoration for Ring-Array Photoacoustic Tomography System Based on Blind Spatially Rotational Deconvolution

A new efficient alternating method for image restoration and texture extraction

Improved estimation of motion blur parameters for restoration from a single image