Single Space Object Image Denoising and Super Resolution Reconstructing based on Unpaired images
Errui Chen, Xubin Feng · 2022 IEEE 10th Joint International Information Technology and Artificial Intelligence Conference (ITAIC) · 2022
High quality space target image is of great significance for space attack and defense because space exploration missions are becoming more and more important. But high quality images of space object are difficult to obtain due to the large number of cosmic rays in the space environment, as well as the limitations of optical lenses, detectors and transmission links on satellites. Image denoising and super-resolution reconstruction are the most economical and effective methods to solve this problem. This paper presents an unpaired denoising and super-resolution reconstruction method for optical remote sensing images which could obtain the images has higher quality than dataset itself. In order to further improve the quality of optical remote sensing images, high quality natural images are added into the training set, and unpaired image data sets (natural images and optical remote sensing images belong to different fields and cannot correspond one to one) are adopted to complete the training of the whole network by using the idea of unsupervised learning. Through the verification tests of three optical remote sensing image data sets, it can be seen that the method in this paper has reconstructed high quality optical remote sensing images with higher resolution than the dataset itself.