Single-image super-resolution reconstruction method and system based on channel attention

The invention provides a single-image super-resolution reconstruction method and system based on channel attention, and relates to the technical field of image reconstruction. The method comprises the following steps: S1, utilizing a shallow feature extraction module to map each pixel of an original input image to a high-order feature space to extract shallow features of the original input image; s2, utilizing a multi-scale feature extraction structure in the multi-scale channel attention learning module to extract deep features from shallow features of the original input image; s3, performing channel attention learning on the deep features by using a channel attention learning branch in the multi-scale channel attention learning module to obtain a nonlinear transformation relation from a low-resolution image to a high-resolution image; and step S4, based on the nonlinear transformation relationship, using a reconstruction module to perform image reconstruction in a mode of changing the number of channels of