A Single Image Deraining Algorithm Based on Convolutional Neural Networks and Swin Transformer
DU Yao
XUE Tao
LI Meng
Abstract：Aiming at the issue of blurriness and detail loss in outdoor images captured during rainy days,this paper proposes a single-image deraining algorithm that leverages convolutional neural networks(CNN)and Transformer technology.Initially,a fea-ture extraction module that combines the local modeling strengths of CNNs with the global context capturing ability of the Swin Trans-former is developed.Subsequently,incorporating contrastive constraint loss,a comprehensive loss function is proposed to optimize network training.When compared to prevailing deraining algorithms,the proposed solution not only removes rain streaks more effec-tively,but also retains image details more efficiently.
Keywords：single image derainingconvolutional neural networkTransformercontrastive learningattention mechanism
Publication Date：2025-12-20
Online Publishing Date：2026-03-23(First online date of this platform, not the publication date of the document)
Pages：6( 30-35 )