A Fast and Efficient Network for Single Image Deraining

Rain streaks will degrade the visibility of images. To tackle this problem, we propose a novel Adaptive Dilated Network (ADN) to remove rain streaks from a single image while using less parameters and running faster than previous methods. Specifically, an Adaptive Dilated Block (ADB) is constructed as the sub-module of ADN. In ADB, we apply a shared dilated block to extract multi-scale features. Then a dilated selection block is added to leverage the importance of features in different scales. All the multi-scale features are fused together to obtain features with rich rain details. To further model the inter-dependencies of the fused features, a feature selection block is employed in ADB to assign different weights to each feature. Moreover, all the hierarchical features extracted by each ADB are concatenated together and fed into a rainy map generator to estimate rain layer. Experimental results demonstrate that the proposed method is superior to the state-of-the-art methods on performances and running time while using less parameters. The source code is available at https://github.com/nnUyi/ADN.

1. INTRODUCTION

Single image deraining is a classical problem in low-level computer vision which aims to remove rain streaks from a rainy image. Traditional methods, such as Gaussian mixture model [1], kernel based method [2], [3], sparse coding [4]–[6], low rank approximation [7],[8], representation learning [9], and dictionary learning [10],[11] have been employed in deraining task. Though these methods can improve the quality of rainy images to some extent, they fail to learn the distribution of rain streaks with different scales, shapes, and directions. Thus relatively low performances are obtained.