Progressive Dilation Dense Residual Fusion Network for Single-Image Deraining

Abstract: Deraining is very important for many applications in computer vision, and it is a challenging problem due to its ill-posed nature, especially for single image deraining. In order to remove rain streaks more thoroughly, as well as retain more details, a progressive dilation dense residual fusion network (PDDRFN) is proposed. More precisely, the PDDRFN is designed in a cascade manner with multiple fusion blocks. The fusion block consists of a dilation dense residual block (DDRB) and a dense residual feature fusion block (DRFFB), where DDRB is created for feature extraction and DRFFB is mainly designed for feature fusion operation. Meanwhile, detail compensation memory mechanism (DCMM) is leveraged between each of two cascade modules to retain more background details. Extensive experiments on synthetic benchmark datasets and real-world rainy images demonstrate that the proposed method can achieve better results compared with previous state-of-the-art methods, in terms of rain streaks removal and background details preservation. Furthermore, PDDRFN also shows its superiority in image denoising task.