IMAGE PROCESSING METHOD AND APPARATUS, AND DEVICE AND MEDIUM

Disclosed in the present invention are an image processing method and apparatus, and a device and a medium. In the present invention, target images to be processed are acquired; convolution processing is performed on inputted target images on the basis of a convolutional layer of a pre-trained hybrid attention super-resolution network model, so as to obtain shallow feature maps; and by means of a set number of attention residual layers which are connected in series in the model, perception processing is performed on the shallow feature maps sequentially by means of the attention residual layers. A deeper network is constructed by the set number of attention residual layers that are connected in series, such that the model pays more attention to detail regions in an image, and the deeper network can accurately extract deep feature maps of detail features. Pixel values of pixel points in corresponding rows and columns in the deep feature maps and the shallow feature maps are added together, and are then inputte