Int J Performability Eng ›› 2019, Vol. 15 ›› Issue (7): 1783-1791.

### Remote Sensing Image Super-Resolution Reconstruction based on Generative Adversarial Network

Aili Wanga,*, Ying Wanga, Xiaoying Songa, and Yuji Iwahorib

1. a Higher Education Key Lab for Measuring and Control Technology and Instrumentations of Heilongjiang Harbin University of Science and Technology, Harbin, 150001, China
b Computer Science, Chubu University, Aichi, 487-8501, Japan
• Submitted on  ;
• Contact: * E-mail address: aili925@hrbust.edu.cn
• About author:Aili Wang received her B.S. degree from the Department of Electronic and Communication Engineering at Harbin Institute of Technology in 2002 and her M.S. degree and Ph.D. from the Department of Information and Communication Engineering at Harbin Institute of Technology in 2004 and 2008, respectively. She joined Harbin University of Science and Technology as an assistant in 2004 and became an associate professor in 2010. She was a visiting professor at Chubu University in 2014. Her research interests include image super-resolution, image fusion, and object tracking.Ying Wang is a master's student at Harbin University of Science and Technology. His research interests include image classification.Xiaoying Song is a master's student at Harbin University of Science and Technology. His research interests include image super-resolution.Yuji Iwahori received his B.S. degree from Nagoya Institute of Technology in 1983 and his M.S. degree and Ph.D. from Tokyo Institute of Technology in 1985 and 1988, respectively. He joined Nagoya Institute of Technology in 1988 and became a professor there in 2002. He has been a professor at Chubu University since 2004. Meanwhile, he has also collaborated with UBC since 1991, IIT Guwahati since 2010, and Chulalongkorn University since 2014.

Abstract: The super-resolution reconstruction algorithm based on generative adversarial network (GAN) can generate realistic texture in the super-resolution process of a single remote sensing image. In order to further improve the visual quality of the reconstructed image, this paper will improve the generation network, discrimination network, and perceptual loss of the generated confrontation network. Firstly, the batch normalization layer is removed and dense connections are used in the residual blocks, which effectively improves the performance of the generated network. Then, we use the relative discriminant network to learn more detailed texture. Finally, we obtain the perception loss before the activation function to maintain the consistency of brightness. In addition, transfer learning is used to solve the problem of insufficient remote sensing data. The experimental results show that the proposed algorithm has superiority in the super-resolution reconstruction of remote sensing images and can obtain better subjective visual effects.