Flow-based Kernel Prior with Application to Blind Super-Resolution
This work addresses kernel estimation for blind super-resolution, an incremental improvement over methods like Double-DIP and KernelGAN by better exploiting the anisotropic Gaussian kernel assumption.
The paper tackled the problem of kernel estimation in blind image super-resolution by proposing a normalizing flow-based kernel prior (FKP) that learns an invertible mapping between anisotropic Gaussian kernel distributions and a latent space, resulting in significant improvements in kernel estimation accuracy with fewer parameters, runtime, and memory usage, leading to state-of-the-art blind SR results.
Kernel estimation is generally one of the key problems for blind image super-resolution (SR). Recently, Double-DIP proposes to model the kernel via a network architecture prior, while KernelGAN employs the deep linear network and several regularization losses to constrain the kernel space. However, they fail to fully exploit the general SR kernel assumption that anisotropic Gaussian kernels are sufficient for image SR. To address this issue, this paper proposes a normalizing flow-based kernel prior (FKP) for kernel modeling. By learning an invertible mapping between the anisotropic Gaussian kernel distribution and a tractable latent distribution, FKP can be easily used to replace the kernel modeling modules of Double-DIP and KernelGAN. Specifically, FKP optimizes the kernel in the latent space rather than the network parameter space, which allows it to generate reasonable kernel initialization, traverse the learned kernel manifold and improve the optimization stability. Extensive experiments on synthetic and real-world images demonstrate that the proposed FKP can significantly improve the kernel estimation accuracy with less parameters, runtime and memory usage, leading to state-of-the-art blind SR results.