Image restoration improved with faster hybrid scheduling and projection method
HP-PnP-Flow: Hybrid Time Scheduling and Projection for Plug-and-Play Image Restoration via Flow Matching
Computer Vision and Pattern Recognition
Summary
Restoring clear images from blurry or damaged ones is a challenging problem. The paper presents a new way to improve image restoration by combining two existing techniques: plug-and-play methods and flow matching. The authors’ approach uses a smart schedule and an extra projection step to make the process more efficient and reliable. Their method works better and faster on tasks like fixing blurred images or filling in missing parts.
What this means in practice
- •For photo editing software developers: Reduce the number of algorithm iterations needed to enhance or repair images, improving speed and quality in editing tools.
- •For surveillance system engineers: Enable more reliable restoration of low-quality or corrupted video frames, aiding clearer image recovery for analysis.
Authors
Yong Li, Xiaojie Wang, Xiaoyan Zhang
Abstract
Image restoration is a fundamental inverse problem that seeks to recover high-quality images from degraded observations. Plug-and-Play (PnP) methods provide a flexible framework for integrating pretrained denoisers into iterative reconstruction, whereas flow matching offers an efficient generative modeling approach based on continuous-time ODE dynamics. Very recently, Martin et al.\ (ICLR, 2025) developed a PnP-Flow method, which integrates the PnP framework with flow matching through a time-dependent denoiser constructed from a pretrained flow matching model.By incorporating a different time schedule and an additional pixel-space projection operator, this paper proposes a novel efficient PnP-Flow algorithm (HP-PnP-Flow) that enjoys rigorous convergence guarantees under mild assumptions, without imposing restrictive conditions on iterates required by the aforementioned work. Numerical experiments on CelebA and AFHQ-Cat demonstrate that the newly proposed HP-PnP-Flow algorithm significantly outperforms PnP-Flow on several restoration tasks such as denoising, deblurring, super-resolution, random inpainting, and box inpainting. Notably, our algorithm achieves comparable or better reconstruction quality with substantially fewer iterations, reducing the iteration count from $500$ to $100$ for deblurring and super-resolution on AFHQ-Cat.