Sequential Dual Attention Network for Rain Streak Removal in a Single Image

Area of Science:

• Computer Vision
• Image Processing
• Deep Learning

Background:

• Adverse weather conditions like rain, haze, and snow degrade image and video quality.
• Degraded visual quality significantly impacts the performance of computer vision applications.

Purpose of the Study:

• To propose a novel deep learning framework for single image deraining.
• To develop a method that effectively removes rain streaks while preserving image details.

Main Methods:

• A novel framework named SSDRNet (Sequential dual attention-based Single image DeRaining deep Network) is proposed.
• A two-stage learning strategy is implemented to capture the distribution of rain streaks.
• The network incorporates Residual Dense Blocks (RDBs), Sequential Dual Attention Blocks (SDABs), and Multi-scale Feature Aggregation Modules (MAMs).

Main Results:

• The proposed two-stage strategy effectively learns fine details of rain streaks.
• SSDRNet successfully removes rain streaks from single images.
• Extensive experiments demonstrate superior performance over state-of-the-art methods in both qualitative and quantitative metrics.

Conclusions:

• SSDRNet offers a highly effective solution for single image deraining.
• The framework achieves state-of-the-art performance, outperforming existing methods.
• The code and trained model are publicly available for further research and application.