Single image rain removal method based on attention model

The invention belongs to the technical field of image processing, and provides a single image rain removal method based on an attention model, comprising the following steps: S1, constructing a neural network model; s2, designing a loss function; s3, training a neural network model by using the public data to obtain model parameters of the neural network model; and S4, importing the model parameters trained in the step S3 into the neural network model, inputting a rain image, and outputting to obtain a rain-removed image. The technical problem to be solved by the invention is to provide a single image rain removal method based on an attention model, a plug-and-play channel-space attention module is designed, the channel attention module ignores the space information, the space attention module ignores the channel information, the advantages of the channel attention module and the space attention module are combined and applied to a rain removal network, and the rain removal efficiency is improved. And a bette