Multi-frame super-resolution of remote sensing images using attention-based GAN models

Multi-frame super-resolution (MFSR) of remote sensing (RS) imageries becomes a critical research topic with the launch of new satellites having video capturing capability and the advancement of artificial intelligence techniques. In this study, an attention-based Generative Adversarial Network (GAN) algorithm is proposed for the multi-frame remote sensing image super-resolution (MRSISR). Firstly, we introduced an attention module to the generator and designed a space-based net that worked on every single frame for better temporal information extraction. Secondly, we proposed a novel attention module for better spatial and spectral information extraction. Thirdly, we applied an attention-based discriminator for the discriminative ability improvement of the discriminator. We implemented several experiments with the state-of-the-art models and the proposed approach using SpaceNet7 and Jilin-1 datasets. We quantitatively and qualitatively compared the results of different multi-frame super-resolution models.