Image super-resolution reconstruction system and method based on cross-scale attention network

Abstract

The invention provides an image super-resolution reconstruction system and method based on a cross-scale attention network, including a shallow feature extraction module, a multi-scale residual module and a reconstruction module; the shallow feature extraction module is used to extract low-resolution images Shallow features; the multi-scale residual module includes several cascaded high-order cross-scale attention groups, a feature fusion module and a global residual adder; the high-order cross-scale attention group and feature fusion module are cascaded At the same time, each high-order cross-scale attention group takes the input features and output features of its previous high-order cross-scale attention group as input. The constructed image super-resolution reconstruction system learns the channel correlation of multi-scale features across scales and within scales, and adjusts the attention weight according to the correlation, realizing the adaptive adjustment of multi-scale features, thereby improving the network's discrimination and analysis of features Learning ability, which in turn improves the quality of image super-resolution reconstruction.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to an image super-resolution reconstruction system and method based on a cross-scale attention network. Background technique [0002] As a post-processing method, image super-resolution reconstruction technology can enhance the resolution of images without increasing hardware costs, and has become a research hotspot in the fields of image processing and computer vision in recent years. Traditional super-resolution reconstruction methods are mainly divided into three categories: interpolation-based methods, reconstruction-based methods, and learning-based methods. Learning-based methods have gained more attention from academia and industry because they are significantly superior to the other two types of methods in terms of computing speed and reconstruction quality. Many current super-resolution reconstruction schemes utilize multi-scale features to improve the qua...

AI Technical Summary

Problems solved by technology

However, these schemes do not consider the cross-scale and intra-scale channel correlation of multi-scale features, and the quality of image reconstruction still needs to be improved.

Images