Single Image De-Raining via Improved Generative Adversarial Nets.

Yi Ren,Mengzhen Nie,Chuankun Li,Shichao Li

doi:10.3390/s20061591

Abstract

Capturing images under rainy days degrades image visual quality and affects analysis tasks, such as object detection and classification. Therefore, image de-raining has attracted a lot of attention in recent years. In this paper, an improved generative adversarial network for single image de-raining is proposed. According to the principles of divide-and-conquer, we divide an image de-raining task into rain locating, rain removing, and detail refining sub-tasks. A multi-stream DenseNet, termed as Rain Estimation Network, is proposed to estimate the rain location map. A Generative Adversarial Network is proposed to remove the rain streaks. A Refinement Network is proposed to refine the details. These three models accomplish rain locating, rain removing, and detail refining sub-tasks, respectively. Experiments on two synthetic datasets and real world images demonstrate that the proposed method outperforms state-of-the-art de-raining studies in both objective and subjective measurements.

Highlights

Rainy images have poor visual qualities, and heavily affect analysis related tasks, such as detection, classification, recognition, and tracking
The experimental results on two synthetic datasets and real-world images show that the proposed method outperforms state-of-the-art de-raining methods
We proposed an Improved Generative Adversarial Network for Single Image De-raining (IGAN-SID)

Summary

Introduction

Rainy images have poor visual qualities, and heavily affect analysis related tasks, such as detection, classification, recognition, and tracking. Developing algorithms for automatically removing rain streaks is essential. Many studies have been proposed to solve the image de-raining problem. The studies can be categorized as prior-based methods, convolutional neural network (CNN)-based methods and Generative. The prior-based methods achieve various degrees of success [1,2,3,4]. The prior information is usually insufficient to cover all shapes, densities or directions of rain streaks. Prior-based methods may only work in a part of rain conditions

Methods

Results

Conclusion