Unsupervised Remoting Sensing Super-Resolution via Migration Image Prior

Recently, satellites with high temporal resolution have fostered wide attention in various practical applications. Due to limitations of bandwidth and hardware cost, however, the spatial resolution of such satellites is considerably low, largely limiting their potentials in scenarios that require spatially explicit information. To improve image resolution, numerous approaches based on training low-high resolution pairs have been proposed to address the super-resolution (SR) task. De-spite their success, however, low/high spatial resolution pairs are usually difficult to obtain in satellites with a high temporal resolution, making such approaches in SR impractical to use. In this paper, we proposed a new unsupervised learning framework, called "MIP", which achieves SR tasks without low/high resolution image pairs. First, random noise maps are fed into a designed generative adversarial network (GAN) for reconstruction. Then, the proposed method converts the reference image to latent space as the migration image prior. Finally, we update the input noise via an implicit method, and further transfer the texture and structured information from the reference image. Extensive experimental results on the Draper dataset show that MIP achieves significant improvements over state-of-the-art methods both quantitatively and qualitatively. The proposed MIP is open-sourced at https://github.com/jiaming-wang/MIP.

1. INTRODUCTION

Recently, remote sensing satellites, which are especially appropriate for uninterrupted observing targets, have drawn widespread concerns in various practical applications. Continuously monitoring moving targets by high temporal resolution satellites, can expand the application range than satellites with a static image, such as, the Jilin-1 and Zhuhai-1 OVS-1 A/B video satellites. It is common knowledge that the spatial resolution and spectral resolution are always a pair of contradictory for the optical remote sensor. Additionally, due to bandwidth and hardware cost limitations, the spatial resolution of high temporal resolution satellite images is decreased that cannot meet the demand of high precision applications. Therefore, improving the spatial resolution of satellite images with large compression ratios, has become an urgent issue in remote sensing applications.