Taming Generative Diffusion Prior for Universal Blind Image Restoration
This work addresses the limitation of previous blind image restoration methods that require assumptions about degradation types, offering a more versatile solution for real-world applications with complex or multiple degradations.
The paper tackles the problem of blind image restoration by proposing BIR-D, a method that uses an optimizable convolutional kernel to simulate degradation models and dynamically updates parameters during diffusion steps, achieving superior results on real-world and synthetic datasets compared to existing unsupervised methods.
Diffusion models have been widely utilized for image restoration. However, previous blind image restoration methods still need to assume the type of degradation model while leaving the parameters to be optimized, limiting their real-world applications. Therefore, we aim to tame generative diffusion prior for universal blind image restoration dubbed BIR-D, which utilizes an optimizable convolutional kernel to simulate the degradation model and dynamically update the parameters of the kernel in the diffusion steps, enabling it to achieve blind image restoration results even in various complex situations. Besides, based on mathematical reasoning, we have provided an empirical formula for the chosen of adaptive guidance scale, eliminating the need for a grid search for the optimal parameter. Experimentally, Our BIR-D has demonstrated superior practicality and versatility than off-the-shelf unsupervised methods across various tasks both on real-world and synthetic datasets, qualitatively and quantitatively. BIR-D is able to fulfill multi-guidance blind image restoration. Moreover, BIR-D can also restore images that undergo multiple and complicated degradations, demonstrating the practical applications.