Swin-Diff: a single defocus image deblurring network based on diffusion model

Single Image Defocus Deblurring (SIDD) remains challenging due to spatially varying blur kernels, particularly in processing high-resolution images where traditional methods often struggle with artifact generation, detail preservation, and computational efficiency. This paper presents Swin-Diff, a n...

Full description

Saved in:
Bibliographic Details
Published in:Complex & intelligent systems Vol. 11; no. 3; pp. 170 - 13
Main Authors: Liang, Hanyan, Chai, Shuyao, Zhao, Xixuan, Kan, Jiangming
Format: Journal Article
Language:English
Published: Cham Springer International Publishing 01.03.2025
Springer Nature B.V
Springer
Subjects:
Blurring
Complexity
Computational Intelligence
Data Structures and Information Theory
Diffusion model
Diffusion models
Engineering
Image resolution
Image restoration
Original Article
Regression models
Single image defocus deblurring
Single image defocus deblurring
Diffusion model
Image restoration
ISSN:2199-4536, 2198-6053
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Single Image Defocus Deblurring (SIDD) remains challenging due to spatially varying blur kernels, particularly in processing high-resolution images where traditional methods often struggle with artifact generation, detail preservation, and computational efficiency. This paper presents Swin-Diff, a novel architecture integrating diffusion models with Transformer-based networks for robust defocus deblurring. Our approach employs a two-stage training strategy where a diffusion model generates prior information in a compact latent space, which is then hierarchically fused with intermediate features to guide the regression model. The architecture incorporates a dual-dimensional self-attention mechanism operating across channel and spatial domains, enhancing long-range modeling capabilities while maintaining linear computational complexity. Extensive experiments on three public datasets (DPDD, RealDOF, and RTF) demonstrate Swin-Diff’s superior performance, achieving average improvements of 1.37% in PSNR, 3.6% in SSIM, 2.3% in MAE, and 25.2% in LPIPS metrics compared to state-of-the-art methods. Our results validate the effectiveness of combining diffusion models with hierarchical attention mechanisms for high-quality defocus blur removal.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2199-4536
2198-6053
DOI:10.1007/s40747-025-01789-w