Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Spectral-Structured Diffusion for Single-Image Rain Removal

Yucheng Xing, Xin Wang

TL;DR

SpectralDiff is introduced, a spectral-structured diffusion-based framework tailored for single-image rain removal that achieves competitive rain removal performance with improved model compactness and favorable inference efficiency compared to existing diffusion-based approaches.

Abstract

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.

Spectral-Structured Diffusion for Single-Image Rain Removal

TL;DR

SpectralDiff is introduced, a spectral-structured diffusion-based framework tailored for single-image rain removal that achieves competitive rain removal performance with improved model compactness and favorable inference efficiency compared to existing diffusion-based approaches.

Abstract

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.
Paper Structure (20 sections, 22 equations, 4 figures, 3 tables, 2 algorithms)

This paper contains 20 sections, 22 equations, 4 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Works
  3. Single Image Rain Removal
  4. Spectral Domain Convolution
  5. Methodology
  6. Layered Characteristic of Rain Streaks
  7. Iterative Rain Removal via Spectral Diffusion
  8. Radial Mask $\mathbf{M}_{d, r}$.
  9. Angular Mask $\mathbf{M}_{d, \theta}$.
  10. Accelerating with Full-Product U-Net
  11. Experiments
  12. Datasets
  13. Overall Performance
  14. Ablation Study
  15. Spatial vs. Spectral Diffusion.
  16. ...and 5 more sections

Figures (4)

  • Figure 1: Illustration of the rain layer adding process. \ref{['fig:rain_process_a1']} is the original image $\textbf{B}$, \ref{['fig:rain_process_b1']} - \ref{['fig:rain_process_d1']} are example rain-streak layers $\textbf{R}_{d}$$(d \in [1, 3])$, and \ref{['fig:rain_process_f1']} is the final rainy image $\textbf{O}$.
  • Figure 2: Illustration of forward / reverse diffusion process.
  • Figure 3: Illustration of rain layers in the frequency domain.
  • Figure 4: Structure of our Product U-Net and Product Res-Block.