Dual Learning-Based Graph Neural Network for Remote Sensing Image Super-Resolution

High-resolution (HR) remote sensing imagery plays a critical role in remote sensing image interpretation, and single image super-resolution (SISR) reconstruction technology is becoming increasingly valuable and significant. The state-of-the-art deep-learning-based SISR methods have demonstrated remarkable advantages while reconstructing complex texture details still remains a big challenge. Besides, as a typical ill-posed inverse problem, how to determine the optimal solution is another important topic. To address these problems, in this work, a dual learning-based graph neural network (DLGNN) is proposed, in which the graph neural network (GNN) is utilized to consider the self-similarity patches in remote sensing imagery by aggregating cross-scale neighboring feature patches, and dual learning strategy is adopted to refine the reconstruction results by constraining the mapping process in terms of the loss function, transferring the typical ill-posed problem to a well-posed one. Abundant experiments on 3K VEHICLE_SR datasets and Massachusetts Roads demonstrate the validity and outstanding performance for remote sensing image super-resolution (SR) tasks compared with other state-of-the-art SR construction methods. Code is available at https://github.com/CUG-RS/DLGNN

I. Introduction

The rapid development of modern aerospace technology has advanced the wide use of remote sensing images in remote sensing applications. Nonetheless, for some highly accurate remote sensing image interpretation assignments, such as military target detection [1], fine-grained classification [2], object tracking [3], and detailed environment monitoring [4], the spatial resolution of the optical sensors cannot meet the accuracy requirements, and becomes the limitation in image interpretation. Even though satellites with the most advanced technology can provide a resolution within a square meter, the high costs of this acquisition technology hinders its application, then single image super-resolution (SISR) technology has been widely studied in depth to improve the spatial resolution of remote sensing imagery [5], [6]. SISR algorithm aims to generate high-resolution (HR) images from a low-resolution (LR) image input by yielding more detailed information and improving the quality of images [7]. Due to the wide use of SISR, there is increasing interest in HR images, especially in the remote sensing field.