Pub Date : 2022-01-01DOI: 10.1109/LGRS.2021.3121094
J. Dong, Wufan Zhao, Shuai Wang
The existing methods of building change detection (CD) using remote sensing (RS) images are still deficient in handling scale variation and class imbalance problems, indicating a decrease in the robustness of small-object detection and pseudo-change information. Thus, a novel building CD framework called the multiscale context aggregation network (MSCANet) is proposed. The high-resolution network is integrated into the feature extracting stage to maintain high-resolution representations throughout the whole process. Then, multiscale context information is aggregated using a scale-aware feature pyramid module (FPM). Recognition performance can be improved from discriminant feature representation learning by using a channel–spatial attention module. Furthermore, a class-balanced loss is proposed to reduce the impact of class imbalance in long-tail datasets. Experimental results from using the LEVIR-CD and SZTAKI AirChange benchmark datasets prove the superiority of the MSCANet over the other baseline methods, with improved maximum F1 scores of 5.28 and 8.47, respectively.
{"title":"Multiscale Context Aggregation Network for Building Change Detection Using High Resolution Remote Sensing Images","authors":"J. Dong, Wufan Zhao, Shuai Wang","doi":"10.1109/LGRS.2021.3121094","DOIUrl":"https://doi.org/10.1109/LGRS.2021.3121094","url":null,"abstract":"The existing methods of building change detection (CD) using remote sensing (RS) images are still deficient in handling scale variation and class imbalance problems, indicating a decrease in the robustness of small-object detection and pseudo-change information. Thus, a novel building CD framework called the multiscale context aggregation network (MSCANet) is proposed. The high-resolution network is integrated into the feature extracting stage to maintain high-resolution representations throughout the whole process. Then, multiscale context information is aggregated using a scale-aware feature pyramid module (FPM). Recognition performance can be improved from discriminant feature representation learning by using a channel–spatial attention module. Furthermore, a class-balanced loss is proposed to reduce the impact of class imbalance in long-tail datasets. Experimental results from using the LEVIR-CD and SZTAKI AirChange benchmark datasets prove the superiority of the MSCANet over the other baseline methods, with improved maximum F1 scores of 5.28 and 8.47, respectively.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62482892","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-01-01DOI: 10.1109/lgrs.2021.3135310
Yi Kong, Xuesong Wang, Yuhu Cheng, Yangchi Chen, C. L. P. Chen
A hyperspectral image (HSI) classification method named graph domain adversarial network with dual-weighted pseudo-label loss (GDAN-DWPL) is proposed in this letter. First, in order to extract more discriminative features, GDAN is applied to the transfer task of HSI. Then, a more reliable spectral–spatial graph is constructed by comprehensively utilizing the abundant spectral features and spatial contextual information. Finally, due to the misalignment of probability distribution on class-level caused by inaccurate pseudo-labels of target domain, a dual-weighted pseudo-label loss is proposed from the perspective of spatiality and confidence. By assigning larger weights to more reliable pixels and eliminating pixels with false pseudo-labels, the negative impact on learning process of prediction model can be reduced. Experimental results on four real HSI datasets show the superiority of GDAN-DWPL.
{"title":"Graph Domain Adversarial Network With Dual-Weighted Pseudo-Label Loss for Hyperspectral Image Classification","authors":"Yi Kong, Xuesong Wang, Yuhu Cheng, Yangchi Chen, C. L. P. Chen","doi":"10.1109/lgrs.2021.3135310","DOIUrl":"https://doi.org/10.1109/lgrs.2021.3135310","url":null,"abstract":"A hyperspectral image (HSI) classification method named graph domain adversarial network with dual-weighted pseudo-label loss (GDAN-DWPL) is proposed in this letter. First, in order to extract more discriminative features, GDAN is applied to the transfer task of HSI. Then, a more reliable spectral–spatial graph is constructed by comprehensively utilizing the abundant spectral features and spatial contextual information. Finally, due to the misalignment of probability distribution on class-level caused by inaccurate pseudo-labels of target domain, a dual-weighted pseudo-label loss is proposed from the perspective of spatiality and confidence. By assigning larger weights to more reliable pixels and eliminating pixels with false pseudo-labels, the negative impact on learning process of prediction model can be reduced. Experimental results on four real HSI datasets show the superiority of GDAN-DWPL.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62485540","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-01-01DOI: 10.1109/lgrs.2021.3140097
Hafize Hasar, U. Hasar, Y. Kaya, T. Oztas, M. Y. Canbolat, Nevzat Aslan, M. Ertugrul, O. Ramahi
A new de-embedding line–line method has been proposed for accurate complex relative permittivity ($varepsilon _{r}$ ) determination of soil samples loaded into an EIA 1-5/8” coaxial transmission line measurement system. The method has three main features. First, it bypasses the requirement of calibration of this system by using only two identical coaxial lines with different lengths. Second, it does not need any numerical technique for $varepsilon _{r}$ determination. Third, it does not require knowledge of electromagnetic properties and thickness information of the bead used for supporting soil samples. The method is next validated by simulations performed using a full 3-D electromagnetic simulation program (CST Microwave Studio) and by $varepsilon _{r}$ measurement of a polyethylene (PE) material. Finally, $varepsilon _{r}$ values of three air-dried and water-saturated soil samples having 90% or more sand content with different electrical conductivities (ECs) and gathered from different areas of the city Gaziantep in Turkey, were measured.
{"title":"Broadband Soil Permittivity Measurements Using a Novel De-Embedding Line–Line Method","authors":"Hafize Hasar, U. Hasar, Y. Kaya, T. Oztas, M. Y. Canbolat, Nevzat Aslan, M. Ertugrul, O. Ramahi","doi":"10.1109/lgrs.2021.3140097","DOIUrl":"https://doi.org/10.1109/lgrs.2021.3140097","url":null,"abstract":"A new de-embedding line–line method has been proposed for accurate complex relative permittivity (<inline-formula> <tex-math notation=\"LaTeX\">$varepsilon _{r}$ </tex-math></inline-formula>) determination of soil samples loaded into an EIA 1-5/8” coaxial transmission line measurement system. The method has three main features. First, it bypasses the requirement of calibration of this system by using only two identical coaxial lines with different lengths. Second, it does not need any numerical technique for <inline-formula> <tex-math notation=\"LaTeX\">$varepsilon _{r}$ </tex-math></inline-formula> determination. Third, it does not require knowledge of electromagnetic properties and thickness information of the bead used for supporting soil samples. The method is next validated by simulations performed using a full 3-D electromagnetic simulation program (CST Microwave Studio) and by <inline-formula> <tex-math notation=\"LaTeX\">$varepsilon _{r}$ </tex-math></inline-formula> measurement of a polyethylene (PE) material. Finally, <inline-formula> <tex-math notation=\"LaTeX\">$varepsilon _{r}$ </tex-math></inline-formula> values of three air-dried and water-saturated soil samples having 90% or more sand content with different electrical conductivities (ECs) and gathered from different areas of the city Gaziantep in Turkey, were measured.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62486020","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Inverse synthetic aperture radar (ISAR) is an effective detection method for targets. However, for the maneuvering targets, the Doppler frequency induced by an arbitrary scatterer on the target is time-varying, which will cause defocus on ISAR images and bring difficulties for the further recognition process. It is hard for traditional methods to well refocus all positions on the target well. In recent years, generative adversarial networks (GANs) achieve great success in image translation. However, the current refocusing models ignore the information of high-order terms containing in the relationship between real and imaginary parts of the data. To this end, an end-to-end refocusing network, named complex-valued pix2pixHD (CVPHD), is proposed to learn the mapping from defocus to focus, which utilizes complex-valued (CV) ISAR images as an input. A CV instance normalization layer is applied to mine the deep relationship between the complex parts by calculating the covariance of them and accelerate the training. Subsequently, an innovative adaptively weighted loss function is put forward to improve the overall refocusing effect. Finally, the proposed CVPHD is tested with the simulated and real dataset, and both can get well-refocused results. The results of comparative experiments show that the refocusing error can be reduced if extending the pix2pixHD network to the CV domain and the performance of CVPHD surpasses other autofocus methods in refocusing effects. The code and dataset have been available online (https://github.com/yhx-hit/CVPHD).
{"title":"High-Resolution Refocusing for Defocused ISAR Images by Complex-Valued Pix2pixHD Network","authors":"Haoxuan Yuan, Hongbo Li, Yun Zhang, Yong Wang, Zitao Liu, Chenxi Wei, Chengxin Yao","doi":"10.1109/LGRS.2022.3210036","DOIUrl":"https://doi.org/10.1109/LGRS.2022.3210036","url":null,"abstract":"Inverse synthetic aperture radar (ISAR) is an effective detection method for targets. However, for the maneuvering targets, the Doppler frequency induced by an arbitrary scatterer on the target is time-varying, which will cause defocus on ISAR images and bring difficulties for the further recognition process. It is hard for traditional methods to well refocus all positions on the target well. In recent years, generative adversarial networks (GANs) achieve great success in image translation. However, the current refocusing models ignore the information of high-order terms containing in the relationship between real and imaginary parts of the data. To this end, an end-to-end refocusing network, named complex-valued pix2pixHD (CVPHD), is proposed to learn the mapping from defocus to focus, which utilizes complex-valued (CV) ISAR images as an input. A CV instance normalization layer is applied to mine the deep relationship between the complex parts by calculating the covariance of them and accelerate the training. Subsequently, an innovative adaptively weighted loss function is put forward to improve the overall refocusing effect. Finally, the proposed CVPHD is tested with the simulated and real dataset, and both can get well-refocused results. The results of comparative experiments show that the refocusing error can be reduced if extending the pix2pixHD network to the CV domain and the performance of CVPHD surpasses other autofocus methods in refocusing effects. The code and dataset have been available online (https://github.com/yhx-hit/CVPHD).","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62496447","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-01-01DOI: 10.23977/geors.2022.050106
Chuanyue Yang
: The area of arid and semi-arid areas in the world is increasing; in order to solve the issues related to the shallow groundwater enrichment assessment of the arid semi-arid areas, take the typical arid and semi-arid area as the research area of Wuwei Citizen Qin County, Gansu, through remote sensing, GF-6, CBERS04 and DEM are used as data sources to use layer analysis to build an evaluation model for hierarchical enrichment results. It has obtained the laws of shallow groundwater distribution in the research zone in the past five years and the next five years. The trend of water level distribution in the past five years is generally consistent, showing from the southwest to the northeast gradually decreases, there are multiple groundwater funnels, and the shallow groundwater content will remain stable and will increase slightly in the next five years. The results of this study evaluate the development trend of shallow groundwater in Wuwei citizens in Gansu; it provides a scientific basis for future shallow groundwater management.
{"title":"Research on Shallow Groundwater Enrichment Assessment Based on RS and GIS Arid and Semi-Arid Areas","authors":"Chuanyue Yang","doi":"10.23977/geors.2022.050106","DOIUrl":"https://doi.org/10.23977/geors.2022.050106","url":null,"abstract":": The area of arid and semi-arid areas in the world is increasing; in order to solve the issues related to the shallow groundwater enrichment assessment of the arid semi-arid areas, take the typical arid and semi-arid area as the research area of Wuwei Citizen Qin County, Gansu, through remote sensing, GF-6, CBERS04 and DEM are used as data sources to use layer analysis to build an evaluation model for hierarchical enrichment results. It has obtained the laws of shallow groundwater distribution in the research zone in the past five years and the next five years. The trend of water level distribution in the past five years is generally consistent, showing from the southwest to the northeast gradually decreases, there are multiple groundwater funnels, and the shallow groundwater content will remain stable and will increase slightly in the next five years. The results of this study evaluate the development trend of shallow groundwater in Wuwei citizens in Gansu; it provides a scientific basis for future shallow groundwater management.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"29 1","pages":""},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75838844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-01-01DOI: 10.1109/LGRS.2021.3064334
G. Wen, A. Marshak
A new technique to retrieve precipitable water vapor (PWV) amount in the clear-cloud transition zone using ground-based zenith spectral radiance is developed. The method uses zenith radiances at the water vapor band at 720 nm and at the adjacent nonadsorbing band at 750 nm. Radiative transfer calculations show that the relative difference in zenith radiance between the two bands depends on PWV and the variations in cloud optical depth introduce a small change in the relative difference which is independent of PWV amount. This allows us to retrieve PWV variations in the clear-cloud transition zone for clouds over dark ocean surface. We applied this method to a Cu cloud case with zenith radiance observations by Shortwave Array Spectroradiometer-Zenith (SASZe) during the Marine ARM GPCI Investigation of Clouds (MAGIC) field campaign. We found that there is about 10% change in PWV amount from the known-cloudy region to known-clear sky in the cloud edges.
{"title":"Precipitable Water Vapor Variation in the Clear-Cloud Transition Zone From the ARM Shortwave Spectrometer","authors":"G. Wen, A. Marshak","doi":"10.1109/LGRS.2021.3064334","DOIUrl":"https://doi.org/10.1109/LGRS.2021.3064334","url":null,"abstract":"A new technique to retrieve precipitable water vapor (PWV) amount in the clear-cloud transition zone using ground-based zenith spectral radiance is developed. The method uses zenith radiances at the water vapor band at 720 nm and at the adjacent nonadsorbing band at 750 nm. Radiative transfer calculations show that the relative difference in zenith radiance between the two bands depends on PWV and the variations in cloud optical depth introduce a small change in the relative difference which is independent of PWV amount. This allows us to retrieve PWV variations in the clear-cloud transition zone for clouds over dark ocean surface. We applied this method to a Cu cloud case with zenith radiance observations by Shortwave Array Spectroradiometer-Zenith (SASZe) during the Marine ARM GPCI Investigation of Clouds (MAGIC) field campaign. We found that there is about 10% change in PWV amount from the known-cloudy region to known-clear sky in the cloud edges.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://sci-hub-pdf.com/10.1109/LGRS.2021.3064334","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62477271","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-01-01DOI: 10.1109/lgrs.2022.3166209
Fangyuan Shi, Zhiqiang Li, M. Zhang, Jinxing Li
Differences in the motion of different parts of a target cause the echo signal to contain specific Doppler modulation information, i.e., the micro-Doppler (m-D) effect. This phenomenon provides an effective way to detect targets in marine environments. In this study, based on the establishment of the micromotion model of a rotating surveillance radar and analysis of the m-D frequency, the geometrical optics and physical optics (GO-PO) method and the time-frequency analysis technique are used to obtain the radar cross section (RCS) and m-D signature of a ship with a shipborne radar at different observation angles. The ship, as the main component of the echo, is associated with the main energy. Finding the optimum angle to observe the shipborne radar is of great importance. The results show that the m-D signatures of the shipborne radar are not clear when the elevation angle is greater than 60° but are clear when the elevation angle is less than 55°. Moreover, some motion parameters can be extracted from the m-D signature, such as the period of the ship micromotion. The rotation speed of the shipborne radar can be obtained and is consistent with the set speed. This can help identify and track the key parts of a ship with local motion.
{"title":"Analysis and Simulation of the Micro-Doppler Signature of a Ship With a Rotating Shipborne Radar at Different Observation Angles","authors":"Fangyuan Shi, Zhiqiang Li, M. Zhang, Jinxing Li","doi":"10.1109/lgrs.2022.3166209","DOIUrl":"https://doi.org/10.1109/lgrs.2022.3166209","url":null,"abstract":"Differences in the motion of different parts of a target cause the echo signal to contain specific Doppler modulation information, i.e., the micro-Doppler (m-D) effect. This phenomenon provides an effective way to detect targets in marine environments. In this study, based on the establishment of the micromotion model of a rotating surveillance radar and analysis of the m-D frequency, the geometrical optics and physical optics (GO-PO) method and the time-frequency analysis technique are used to obtain the radar cross section (RCS) and m-D signature of a ship with a shipborne radar at different observation angles. The ship, as the main component of the echo, is associated with the main energy. Finding the optimum angle to observe the shipborne radar is of great importance. The results show that the m-D signatures of the shipborne radar are not clear when the elevation angle is greater than 60° but are clear when the elevation angle is less than 55°. Moreover, some motion parameters can be extracted from the m-D signature, such as the period of the ship micromotion. The rotation speed of the shipborne radar can be obtained and is consistent with the set speed. This can help identify and track the key parts of a ship with local motion.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62489340","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-01-01DOI: 10.1109/lgrs.2022.3194702
Chi Zhang, Zegang Ding, Han Li, Tianyi Zhang
Distributed multichannel synthetic aperture radar (MC-SAR) is a system in which transmitting or receiving arrays are distributed on multiple platforms or at different locations on one platform. The along-track component of the baseline makes distributed SAR promising in high-resolution wide-swath (HRWS) imaging such as azimuth MC-SAR. However, the additional channel mismatch introduced by the cross-track baseline (CTB) is considered for the distributed SAR. When the azimuth beam is wide, the azimuth-variant channel mismatch caused by the CTB must be compensated before SAR imaging. First, an improved azimuth signal reconstruction algorithm for distributed wide-beam SAR is proposed in this article. The azimuth variance of the channel mismatch is considered in a reconstruction filter to further suppress the ambiguity, and the computational consumption is decreased by approximately decomposing the mismatch matrix. Second, the ambiguity suppression performance of the proposed method is analyzed quantitatively. Finally, a simulation and real data processing are provided to demonstrate the effectiveness of the proposed method.
{"title":"An Improved Azimuth Signal Reconstruction Algorithm for Wide-Beam Distributed SAR","authors":"Chi Zhang, Zegang Ding, Han Li, Tianyi Zhang","doi":"10.1109/lgrs.2022.3194702","DOIUrl":"https://doi.org/10.1109/lgrs.2022.3194702","url":null,"abstract":"Distributed multichannel synthetic aperture radar (MC-SAR) is a system in which transmitting or receiving arrays are distributed on multiple platforms or at different locations on one platform. The along-track component of the baseline makes distributed SAR promising in high-resolution wide-swath (HRWS) imaging such as azimuth MC-SAR. However, the additional channel mismatch introduced by the cross-track baseline (CTB) is considered for the distributed SAR. When the azimuth beam is wide, the azimuth-variant channel mismatch caused by the CTB must be compensated before SAR imaging. First, an improved azimuth signal reconstruction algorithm for distributed wide-beam SAR is proposed in this article. The azimuth variance of the channel mismatch is considered in a reconstruction filter to further suppress the ambiguity, and the computational consumption is decreased by approximately decomposing the mismatch matrix. Second, the ambiguity suppression performance of the proposed method is analyzed quantitatively. Finally, a simulation and real data processing are provided to demonstrate the effectiveness of the proposed method.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62494140","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2022-01-01DOI: 10.1109/lgrs.2021.3130625
Hongqi Zhang, Xudong Sun, Yuan Zhu, Fengqiang Xu, Xianping Fu
Band selection (BS) methods based on deep learning have achieved significant development. However, most existing band selection methods commonly utilize a fully connected neural network (FCN) or convolutional neural network (CNN) to explore the correlation among bands and rarely combine the two styles of the network to select bands. Moreover, almost all the methods employ the form of the combination of $L_{1}$ norm and Sigmoid to constitute attention model, which may lead to losing some informative band feature. To tackle these troubles, this letter proposes a novel band selection network using FCN and CNN, termed as global-local spectral weight network based on attention (GLSWA), in which the band features of each pixel is mined using the network of two types, and designing an attention-based scoring module (ASM) and a convolutional reconstruction module (CRM), respectively, so that each attention of band is adjusted by simultaneous considering the entire band features and successive one. Experimental results on three real hyperspectral image (HSI) datasets show that the proposed method achieves satisfactory accuracy than some state-of-the-art algorithms.
{"title":"A Global-Local Spectral Weight Network Based on Attention for Hyperspectral Band Selection","authors":"Hongqi Zhang, Xudong Sun, Yuan Zhu, Fengqiang Xu, Xianping Fu","doi":"10.1109/lgrs.2021.3130625","DOIUrl":"https://doi.org/10.1109/lgrs.2021.3130625","url":null,"abstract":"Band selection (BS) methods based on deep learning have achieved significant development. However, most existing band selection methods commonly utilize a fully connected neural network (FCN) or convolutional neural network (CNN) to explore the correlation among bands and rarely combine the two styles of the network to select bands. Moreover, almost all the methods employ the form of the combination of $L_{1}$ norm and Sigmoid to constitute attention model, which may lead to losing some informative band feature. To tackle these troubles, this letter proposes a novel band selection network using FCN and CNN, termed as global-local spectral weight network based on attention (GLSWA), in which the band features of each pixel is mined using the network of two types, and designing an attention-based scoring module (ASM) and a convolutional reconstruction module (CRM), respectively, so that each attention of band is adjusted by simultaneous considering the entire band features and successive one. Experimental results on three real hyperspectral image (HSI) datasets show that the proposed method achieves satisfactory accuracy than some state-of-the-art algorithms.","PeriodicalId":13046,"journal":{"name":"IEEE Geoscience and Remote Sensing Letters","volume":"19 1","pages":"1-5"},"PeriodicalIF":4.8,"publicationDate":"2022-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"62484094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：