DPAFNet:Dual Path Attention Fusion Network for Single Image Deraining

Image is an important source of information in modern society, so image quality is particularly important. However, bad weather dramatically affects the quality of images, thus hindering the expected acquisition of information, thereby affecting the normal operation of imaging equipment. Therefore, the restoration of degraded images is essential and has attracted extensive attention from researchers in recent years. Before data-driven methods show their superiority in the filed of computer vision, model-driven methods are primarily used in single image deraining, including filter-based and prior-based methods in more detail. In recent years, deep learning methods have shone in computer vision and surpassed traditional model-based methods in performance, such as object detection[1], semantic segmentation[2], image classification[3], person reidentification[4].