With the development of autonomous driving technology, the problem of how to recover the poor quality images captured by cameras under bad weather has attracted much attention. Traditional single-image restoration methods have been difficult to meet the practical needs, so multi-weather image restoration has become a research hotspot. In this paper, we propose a novel network architecture DNSWNet, which adopts a hybrid model of convolutional neural network (CNN) and Swin transformer, which can effectively extract the residual information of an image and obtain high-quality restored images by the difference between the original clean image and the residual information. We conducted experiments on several rain, snow, and fog datasets, and the results show that DNSWNet can significantly improve the recovery effect compared with existing mainstream denoising algorithms.