Network model for optimizing degraded image and image optimization method

The invention provides a network model for optimizing a degraded image and an image optimization method, and belongs to the technical field of big data, the network model comprises a first type of neural network model and a second type of neural network model, the first type of neural network model is used for carrying out feature extraction on a target image, and the second type of neural network model is used for carrying out feature extraction on the target image; the first neural network comprises a down-sampling module and an up-sampling module, and the down-sampling module comprises a plurality of continuous convolutional layers and pooling layers; the second type neural network model comprises processing units at multiple moments, the processing unit at each moment is composed of a gate structure, and the gate structures are used for continuously transmitting data in the sequence and discarding determined invalid data. According to the scheme, the degraded image is optimized and repaired through the mu