Super-resolution Reconstruction of Airborne Remote Sensing Images based on Multi-scale Fusion
Fengguo Chu, Liu Hu, Ziyu Wang, Zhiyuan Cao · 2022 3rd International Conference on Big Data, Artificial Intelligence and Internet of Things Engineering (ICBAIE) · 2022
To extract more detailed features of airborne remote sensing images to obtain more information, super-resolution reconstruction is performed on them. However, the existing super-resolution reconstruction algorithms of airborne remote sensing images have poor feature extraction capabilities, and smooth image edges, and are difficult to restore high-frequency information effectively. In this paper, the residual features of different residual modules are densely connected to form a dense group (DG), which combines different residual features to reduce the redundancy of features and ensure the effective transmission of high-frequency residual features. Further, the residual features of DG are densely connected to realize the reuse of information, and combined with multi-scale fusion, a two-branch lightweight multi-scale fusion super-resolution reconstruction network is proposed. The experimental results show that the algorithm has good performance and is lightweight, and can obtain a better reconstruction effect.