MRA-IDN: A Lightweight Super-Resolution Framework of Remote Sensing Images Based on Multiscale Residual Attention Fusion Mechanism
Open Access
- 27 March 2024
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 17 (19391404) , 7781-7800
Abstract
The emergence of deep learning technology has significantly improved the performance of super-resolution algorithms for a single remote sensing image; however, the number of deep learning model parameters is large, which limits its real-time deployment. In addition, the reconstructed image quality still needs improvement. To deal with the aforementioned problems, a lightweight multi-scale residual attention information distillation network (MRA-IDN) is proposed in this paper. It achieves high quality and fast super-resolution processing of remote sensing images. First, a high and low frequency separation reconstruction strategy is adopted that enables the network to improve the reconstructed details of high frequency components while keeping the number of model parameters low. Second, a novel multi-scale residual attention information distillation group (MRA-IDG) is designed as the key component to further extract richer regional features with different perceptual fields and high frequency information while reducing the number of network parameters. This is achieved by combining a multi-scale residual information distillation (MSR-ID) block that consists of multiple residual convolutional sub-blocks (RCBs) and a high frequency channel aware (HFCA) attention block. Last, the experimental results show that, compared with existing mainstream methods such as the MHAN, the number of model parameters can be reduced by 75%, and the edge details of the reconstructed images are richer and more complete. The corresponding PSNR and SSIM can reach 31.59 dB and 0.824, respectively, under the condition of a ×4 magnification factor, and 27.39 dB and 0.668, respectively, under the condition of a ×8 magnification factor.Keywords
Funding Information
- Key Area R&D Program of Guangdong Province (2018B030338001)
- Basic and applied basic research project of Guangzhou basic research program (202201010595)
- Guangdong Education Department
- Guangdong University of Technology (220413548)
This publication has 72 references indexed in Scilit:
- From degrade to upgrade: Learning a self-supervised degradation guided adaptive network for blind remote sensing image super-resolutionInformation Fusion, 2023
- Single Image Super-Resolution via a Holistic Attention Network2020
- Efficient Image Super-Resolution Using Pixel Attention2020
- Deep Learning Based Single Image Super-Resolution: A Survey2018
- Image Super-Resolution Using Very Deep Residual Channel Attention Networks2018
- Sketch-based manga retrieval using manga109 datasetMultimedia Tools and Applications, 2016
- Accelerating the Super-Resolution Convolutional Neural Network2016
- Identity Mappings in Deep Residual Networks2016
- Learning a Deep Convolutional Network for Image Super-Resolution2014
- On Single Image Scale-Up Using Sparse-Representations2012