Hyun-Ho Kim's Paper Accepted for IEEE TGRS Journal
Hyun-Ho Kim's paper has been accepted for IEEE TGRS Journal. Congratulation~
Title: FBS-PS: Fully Band-Separable PAN-Sharpening Considering the Physical Characteristics of Electro-Optical Sensors
Authors: Hyun-Ho Kim and Munchurl Kim
Abstract:
Electro-optical (EO) satellites are primarily used for reconnaissance, national defense, and cartography. However, most high-resolution (HR) EO satellites obtain images at a lower resolution (LR) in the multispectral (MS) band compared to the panchromatic (PAN) band due to technical limitations. Deep learning-based PAN-sharpening (PS) methods have been continuously developed to address the growing demand for MS images with the same ground sample distance as PAN images. The improvements in deep learning-based PS methods have focused on enhancing the network structure and often overlooked the physical characteristics of satellite EO sensors, leading to artifacts such as noise and color distortions in the PS images. Thus, we propose a fully band-separable PAN-sharpening method (FBSPS) that processes MS images separately, effectively considering the physical properties of the corresponding EO sensors in the acquisition when generating PS images. This helps prevent unrelated information from being mixed among MS images and enables accurate feature extractions. Therefore, the generated PS images have less noise and color distortions than previous PS methods that typically fuse MS images from front-end layers. Additionally, we design a novel training method and loss function to handle the problem of misregistered MS and PAN images. The proposed FBS-PS achieves a PS performance with 0.21∼1.02 dB and 0.10∼1.23 dB higher peak signal-to-noise ratio (PSNR) values for the WorldView-III and KOMPSAT-3A datasets, respectively, compared to the state-of-the-art (SOTA) methods. Furthermore, the proposed FBS-PS is relatively lightweight, having 31.3∼99.2%
fewer floating point operations per second and 8.6∼95.2% fewer parameters than those in SOTA methods.
Index Terms—PAN-sharpening, deep learning, electro-optical sensor, remote sensing
Comentarios