Spectral-Structured Diffusion for Single-Image Rain Removal
This work addresses the challenge of removing rain streaks from single images, which is important for applications like autonomous driving and surveillance, but it is incremental as it builds on existing diffusion-based approaches with specific modifications.
The paper tackled the problem of single-image rain removal by introducing SpectralDiff, a spectral-structured diffusion-based framework that incorporates structured spectral perturbations to suppress multi-directional rain components, achieving competitive performance with improved model compactness and inference efficiency compared to existing diffusion-based methods.
Rain streaks manifest as directional and frequency-concentrated structures that overlap across multiple scales, making single-image rain removal particularly challenging. While diffusion-based restoration models provide a powerful framework for progressive denoising, standard spatial-domain diffusion does not explicitly account for such structured spectral characteristics. We introduce SpectralDiff, a spectral-structured diffusion-based framework tailored for single-image rain removal. Rather than redefining the diffusion formulation, our method incorporates structured spectral perturbations to guide the progressive suppression of multi-directional rain components. To support this design, we further propose a full-product U-Net architecture that leverages the convolution theorem to replace convolution operations with element-wise product layers, improving computational efficiency while preserving modeling capacity. Extensive experiments on synthetic and real-world benchmarks demonstrate that SpectralDiff achieves competitive rain removal performance with improved model compactness and favorable inference efficiency compared to existing diffusion-based approaches.