Abstract
Hyperspectral Image (HSI) can continuously cover tens or even hundreds of spectral segments for each spatial pixel. Limited by the cost and commercialization requirements of remote sensing satellites, HSIs often lose a lot of information due to insufficient image spatial resolution. For the high-dimensional nature of HSIs and the correlation between the spectra, the existing Super-Resolution (SR) methods for HSIs have the problems of excessive parameter amount and insufficient information complementarity between the spectra. This paper proposes a Multi-Scale Feature Mapping Network (MSFMNet) based on the cascaded residual learning to adaptively learn the prior information of HSIs. MSFMNet simplifies each part of the network into a few simple yet effective network modules. To learn the spatial-spectral characteristics among different spectral segments, a multi-scale feature generation and fusion Multi-Scale Feature Mapping Block (MSFMB) based on wavelet transform and spatial attention mechanism is designed in MSFMNet to learn the spectral features between different spectral segments. To effectively improve the multiplexing rate of multi-level spectral features, a Multi-Level Feature Fusion Block (MLFFB) is designed to fuse the multi-level spectral features. In the image reconstruction stage, an optimized sub-pixel convolution module is used for the up-sampling of different spectral segments. Through a large number of verifications on the three general hyperspectral datasets, the superiority of this method compared with the existing hyperspectral SR methods is proved. In subjective and objective experiments, its experimental performance is better than its competitors.
Highlights
The development of aerospace technology and remote sensing technology has promoted the application of hyperspectral image (HSI) based on remote sensing satellite [1,2,3,4], such as land monitoring [5,6], urban planning [7], road network layout [8], agricultural yield estimation [9] and disaster prevention and control [10]
To qualitatively measure the proposed Multi-Scale Feature Mapping Network (MSFMNet), four evaluation methods are employed to verify the effectiveness of the algorithm, including Peak Signal-to-Noise Ratio (PSNR), Mean Peak Signal-to-Noise Ratio (MPSNR), Structural Similarity (SSIM), and Spectral Angle Mapping (SAM)
To evaluate the performance of the MSFMNet method as comprehensively as possible, several popular Super Resolution (SR) methods are used for comparison, including four commonly used algorithms, namely Bicubic, VDSR, EDSR, ERCSR, and MCNet
Summary
The development of aerospace technology and remote sensing technology has promoted the application of hyperspectral image (HSI) based on remote sensing satellite [1,2,3,4], such as land monitoring [5,6], urban planning [7], road network layout [8], agricultural yield estimation [9] and disaster prevention and control [10]. Through supervised learning of the feature mapping relationship between the pairs of high-resolution (HR) images and low-resolution (LR) images, the SR method based on the CNN network can efficiently reconstruct high-quality images with rich details. Focusing on the problem of fully mining the prior knowledge of remote sensing images, and based on the SR method of classic CNN model, a variety of simple and effective CNN modules with different functions is designed from image feature extraction, feature nonlinear mapping, and image reconstruction. The multi-scale structure module based on wavelet transform and attention mechanism is designed, which can efficiently learn the prior knowledge of global and local features with a small number of model parameters. The network model based on multi-scale structure can obtain more similar image feature information from low-resolution remote sensing image blocks. A large number of experiments are carried out based on benchmark datasets, and the experimental results signify that the proposed method is superior to the existing methods
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
