Pub Date : 2024-10-03DOI: 10.1109/JSTARS.2024.3472751
Yijian Duan;Liwen Meng;Yanmei Meng;Jihong Zhu;Jiacheng Zhang;Jinlai Zhang;Xin Liu
Given the inherent limitations of camera-only and LiDAR-only methods in performing semantic segmentation tasks in large-scale complex environments, multimodal information fusion for semantic segmentation has become a focal point of contemporary research. However, significant modal disparities often result in existing fusion-based methods struggling with low segmentation accuracy and limited efficiency in large-scale complex environments. To address these challenges,we propose a semantic segmentation network with camera–LiDAR cross-attention fusion based on fast neighbor feature aggregation (MFSA-Net), which is better suited for large-scale semantic segmentation in complex environments. Initially, we propose a dual-distance attention feature aggregation module based on rapid 3-D nearest neighbor search. This module employs a sliding window method in point cloud perspective projections for swift proximity search, and efficiently combines feature distance and Euclidean distance information to learn more distinctive local features. This improves segmentation accuracy while ensuring computational efficiency. Furthermore, we propose a cross-attention fusion two-stream network based on residual, which allows for more effective integration of camera information into the LiDAR data stream, enhancing both accuracy and robustness. Extensive experimental results on the large-scale point cloud datasets SemanticKITTI and Nuscenes demonstrate that our proposed algorithm outperforms similar algorithms in semantic segmentation performance in large-scale complex environments.
{"title":"MFSA-Net: Semantic Segmentation With Camera-LiDAR Cross-Attention Fusion Based on Fast Neighbor Feature Aggregation","authors":"Yijian Duan;Liwen Meng;Yanmei Meng;Jihong Zhu;Jiacheng Zhang;Jinlai Zhang;Xin Liu","doi":"10.1109/JSTARS.2024.3472751","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3472751","url":null,"abstract":"Given the inherent limitations of camera-only and LiDAR-only methods in performing semantic segmentation tasks in large-scale complex environments, multimodal information fusion for semantic segmentation has become a focal point of contemporary research. However, significant modal disparities often result in existing fusion-based methods struggling with low segmentation accuracy and limited efficiency in large-scale complex environments. To address these challenges,we propose a semantic segmentation network with camera–LiDAR cross-attention fusion based on fast neighbor feature aggregation (MFSA-Net), which is better suited for large-scale semantic segmentation in complex environments. Initially, we propose a dual-distance attention feature aggregation module based on rapid 3-D nearest neighbor search. This module employs a sliding window method in point cloud perspective projections for swift proximity search, and efficiently combines feature distance and Euclidean distance information to learn more distinctive local features. This improves segmentation accuracy while ensuring computational efficiency. Furthermore, we propose a cross-attention fusion two-stream network based on residual, which allows for more effective integration of camera information into the LiDAR data stream, enhancing both accuracy and robustness. Extensive experimental results on the large-scale point cloud datasets SemanticKITTI and Nuscenes demonstrate that our proposed algorithm outperforms similar algorithms in semantic segmentation performance in large-scale complex environments.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"19627-19639"},"PeriodicalIF":4.7,"publicationDate":"2024-10-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10704067","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142595926","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-02DOI: 10.1109/JSTARS.2024.3472296
Qinfeng Zhu;Yuan Fang;Yuanzhi Cai;Cheng Chen;Lei Fan
Deep learning methods, especially convolutional neural networks (CNNs) and vision transformers (ViTs), are frequently employed to perform semantic segmentation of high-resolution remotely sensed images. However, CNNs are constrained by their restricted receptive fields, while ViTs face challenges due to their quadratic complexity. Recently, the Mamba model, featuring linear complexity and a global receptive field, has gained extensive attention for vision tasks. In such tasks, images need to be serialized to form sequences compatible with the Mamba model. Numerous research efforts have explored scanning strategies to serialize images, aiming to enhance the Mamba model's understanding of images. However, the effectiveness of these scanning strategies remains uncertain. In this research, we conduct a comprehensive experimental investigation on the impact of mainstream scanning directions and their combinations on semantic segmentation of remotely sensed images. Through extensive experiments on the LoveDA, ISPRS Potsdam, ISPRS Vaihingen, and UAVid datasets, we demonstrate that no single scanning strategy outperforms others, regardless of their complexity or the number of scanning directions involved. A simple, single scanning direction is deemed sufficient for semantic segmentation of high-resolution remotely sensed images. Relevant directions for future research are also recommended.
{"title":"Rethinking Scanning Strategies With Vision Mamba in Semantic Segmentation of Remote Sensing Imagery: An Experimental Study","authors":"Qinfeng Zhu;Yuan Fang;Yuanzhi Cai;Cheng Chen;Lei Fan","doi":"10.1109/JSTARS.2024.3472296","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3472296","url":null,"abstract":"Deep learning methods, especially convolutional neural networks (CNNs) and vision transformers (ViTs), are frequently employed to perform semantic segmentation of high-resolution remotely sensed images. However, CNNs are constrained by their restricted receptive fields, while ViTs face challenges due to their quadratic complexity. Recently, the Mamba model, featuring linear complexity and a global receptive field, has gained extensive attention for vision tasks. In such tasks, images need to be serialized to form sequences compatible with the Mamba model. Numerous research efforts have explored scanning strategies to serialize images, aiming to enhance the Mamba model's understanding of images. However, the effectiveness of these scanning strategies remains uncertain. In this research, we conduct a comprehensive experimental investigation on the impact of mainstream scanning directions and their combinations on semantic segmentation of remotely sensed images. Through extensive experiments on the LoveDA, ISPRS Potsdam, ISPRS Vaihingen, and UAVid datasets, we demonstrate that no single scanning strategy outperforms others, regardless of their complexity or the number of scanning directions involved. A simple, single scanning direction is deemed sufficient for semantic segmentation of high-resolution remotely sensed images. Relevant directions for future research are also recommended.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"18223-18234"},"PeriodicalIF":4.7,"publicationDate":"2024-10-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10703181","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142452721","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This study creatively invented a new type of turnbuckle adjustable corner reflector (CR), which greatly enhances the flexible adjustment ability of CR in both vertical and horizontal directions through a unique positive and negative screw structure design, significantly improving the convenience of on-site deployment. Based on the performance of dihedral CR and trihedral CR installed in the South-to-North Water Diversion Channel using back-to-back design on TerraSAR-X and Sentinel-1A images, the performance of different structures of CR in complex environments, especially under heavy precipitation conditions, was deeply analyzed. The experimental results show that the trihedral CR can still maintain stable monitoring efficiency when encountering extreme weather conditions with precipitation exceeding 10 mm. The monitoring effect of traditional dihedral CR drops sharply and is almost ineffective in such environments. At the same time, the combination of theoretical radar cross section (RCS) and measured RCS values confirms the decisive impact of CR geometry and deployment strategy on improving monitoring stability and accuracy. Further precise comparison between CR-InSAR monitoring results and the second-order leveling measurement results shows that the system's average error is controlled within the range of 2–3 mm using trihedral CR. Compared with the results of dihedral CR and InSAR without CR, a significant improvement in accuracy has been achieved. This study provides strong theoretical support and practical guidance for the optimization design and practical application of CR systems, and has important scientific value and application prospects.
{"title":"Refinement Analysis of Real Dihedral and Trihedral CR-InSAR Based on TerraSAR-X and Sentinel-1A Images","authors":"Hui Liu;Bochen Zhou;Changwei Miao;Shihuan Li;Lei Xu;Ke Zheng;Geshuang Li;Shiji Yang;Mengyuan Zhu","doi":"10.1109/JSTARS.2024.3472220","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3472220","url":null,"abstract":"This study creatively invented a new type of turnbuckle adjustable corner reflector (CR), which greatly enhances the flexible adjustment ability of CR in both vertical and horizontal directions through a unique positive and negative screw structure design, significantly improving the convenience of on-site deployment. Based on the performance of dihedral CR and trihedral CR installed in the South-to-North Water Diversion Channel using back-to-back design on TerraSAR-X and Sentinel-1A images, the performance of different structures of CR in complex environments, especially under heavy precipitation conditions, was deeply analyzed. The experimental results show that the trihedral CR can still maintain stable monitoring efficiency when encountering extreme weather conditions with precipitation exceeding 10 mm. The monitoring effect of traditional dihedral CR drops sharply and is almost ineffective in such environments. At the same time, the combination of theoretical radar cross section (RCS) and measured RCS values confirms the decisive impact of CR geometry and deployment strategy on improving monitoring stability and accuracy. Further precise comparison between CR-InSAR monitoring results and the second-order leveling measurement results shows that the system's average error is controlled within the range of 2–3 mm using trihedral CR. Compared with the results of dihedral CR and InSAR without CR, a significant improvement in accuracy has been achieved. This study provides strong theoretical support and practical guidance for the optimization design and practical application of CR systems, and has important scientific value and application prospects.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"18739-18750"},"PeriodicalIF":4.7,"publicationDate":"2024-10-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10702605","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142517954","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The global navigation satellite system reflectometry (GNSS-R) technology has shown significant potential in retrieving snow depth using signal-to-noise ratio (SNR) data. However, compared to traditional in situ snow depth measurement techniques, we have observed that the accuracy and performance of GNSS-R can be significantly impacted under certain conditions, particularly when the elevation angle increases. This is due to the attenuation of the multipath effect, which is particularly evident during snow-free periods and under low-snow conditions where snow depths are below 50 cm. To address these limitations, we propose a snow depth inversion method that integrates SNR signals with the support vector regression algorithm, utilizing SNR sequences as feature inputs. We conducted studies at stations P351 and P030, covering elevation angles ranging from 5° to 20°, 5° to 25°, and 5° to 30°. The experimental results show that the root-mean-square error at both the stations decreased by 50% or more compared to traditional methods, demonstrating an improvement in inversion accuracy across different elevation angles. More importantly, the inversion accuracy of our method does not significantly lag behind that at lower elevation angles, indicating its excellent performance under challenging conditions. These findings highlight the contribution of our method in enhancing the accuracy of snow depth retrieval and its potential to drive further advancements in the field of GNSS-R snow depth inversion.
{"title":"GNSS-R Snow Depth Inversion Study Based on SNR-SVR","authors":"Yuan Hu;Jingxin Wang;Wei Liu;Xintai Yuan;Jens Wickert","doi":"10.1109/JSTARS.2024.3470508","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3470508","url":null,"abstract":"The global navigation satellite system reflectometry (GNSS-R) technology has shown significant potential in retrieving snow depth using signal-to-noise ratio (SNR) data. However, compared to traditional in situ snow depth measurement techniques, we have observed that the accuracy and performance of GNSS-R can be significantly impacted under certain conditions, particularly when the elevation angle increases. This is due to the attenuation of the multipath effect, which is particularly evident during snow-free periods and under low-snow conditions where snow depths are below 50 cm. To address these limitations, we propose a snow depth inversion method that integrates SNR signals with the support vector regression algorithm, utilizing SNR sequences as feature inputs. We conducted studies at stations P351 and P030, covering elevation angles ranging from 5° to 20°, 5° to 25°, and 5° to 30°. The experimental results show that the root-mean-square error at both the stations decreased by 50% or more compared to traditional methods, demonstrating an improvement in inversion accuracy across different elevation angles. More importantly, the inversion accuracy of our method does not significantly lag behind that at lower elevation angles, indicating its excellent performance under challenging conditions. These findings highlight the contribution of our method in enhancing the accuracy of snow depth retrieval and its potential to drive further advancements in the field of GNSS-R snow depth inversion.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"18025-18037"},"PeriodicalIF":4.7,"publicationDate":"2024-10-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10703113","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142524244","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-01DOI: 10.1109/JSTARS.2024.3472021
Nguyen Thi Thu Ha;Pham Quang Vinh;Nguyen Thien Phuong Thao;Pham Ha Linh;Michael Parsons;Nguyen Van Manh
The effective monitoring of eutrophication in inland water bodies is crucial for environmental management and pollution prevention. This study conducts a comprehensive analysis of in situ hyperspectral reflectance data (400–900 nm) and the trophic state index (TSI) obtained from 365 points across ten lakes and reservoirs in Northern Vietnam to propose a trophic classification based on water reflectance spectra features and a TSI estimation model for diagnosis and assessment of lake trophic status. By analyzing the quantity of reflectance peaks and their heights, our study identifies three distinct water reflectance spectra classes corresponding to three trophic levels: mesotrophic to lightly eutrophic, highly eutrophic, and hypertrophic. This classification enables the quick identification of trophic levels directly at the in situ radiometric measurement sites. Our study demonstrates that a logarithmic function of the band ratio, ${{mathbf{R}}_{mathbf{rs}}}( {715} )/{{mathbf{R}}_{mathbf{rs}}}( {560} )$ , is robust for estimating TSI ( ${{{bm{R}}}^2}$ = 0.85 and 0.94; root-mean-square error = 5.0 and 3.7 in calibration and validation, respectively), particularly in algal-dominated waters. These findings represent a practical application of hyperspectral remote sensing for effective eutrophication management. They also highlight the potential use of multispectral optical imagery for monitoring eutrophication in tropical regions.
{"title":"A Method for Assessing the Lake Trophic Status Using Hyperspectral Reflectance (400–900 nm) Measured Above Water","authors":"Nguyen Thi Thu Ha;Pham Quang Vinh;Nguyen Thien Phuong Thao;Pham Ha Linh;Michael Parsons;Nguyen Van Manh","doi":"10.1109/JSTARS.2024.3472021","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3472021","url":null,"abstract":"The effective monitoring of eutrophication in inland water bodies is crucial for environmental management and pollution prevention. This study conducts a comprehensive analysis of in situ hyperspectral reflectance data (400–900 nm) and the trophic state index (TSI) obtained from 365 points across ten lakes and reservoirs in Northern Vietnam to propose a trophic classification based on water reflectance spectra features and a TSI estimation model for diagnosis and assessment of lake trophic status. By analyzing the quantity of reflectance peaks and their heights, our study identifies three distinct water reflectance spectra classes corresponding to three trophic levels: mesotrophic to lightly eutrophic, highly eutrophic, and hypertrophic. This classification enables the quick identification of trophic levels directly at the in situ radiometric measurement sites. Our study demonstrates that a logarithmic function of the band ratio, \u0000<inline-formula><tex-math>${{mathbf{R}}_{mathbf{rs}}}( {715} )/{{mathbf{R}}_{mathbf{rs}}}( {560} )$</tex-math></inline-formula>\u0000, is robust for estimating TSI (\u0000<inline-formula><tex-math>${{{bm{R}}}^2}$</tex-math></inline-formula>\u0000 = 0.85 and 0.94; root-mean-square error = 5.0 and 3.7 in calibration and validation, respectively), particularly in algal-dominated waters. These findings represent a practical application of hyperspectral remote sensing for effective eutrophication management. They also highlight the potential use of multispectral optical imagery for monitoring eutrophication in tropical regions.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"17890-17902"},"PeriodicalIF":4.7,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10702456","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142524189","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Accurately extracting offshore floating raft aquaculture (FRA) areas from remotely sensed images is the key to rationally managing aquaculture resources. Currently, deep learning-based methods perform well in FRA area extraction tasks but are limited by the shortcomings of single-modality remote sensing data, which affect their extraction accuracies. To solve this problem, we constructed a multimodal dataset called CHN-YS3-FRA for FRA area extraction using heterogeneous Sentinel-1/-2 remote sensing image data and proposed a new multiscale synthetic aperture radar (SAR) guidance network (MSARG-Net) for performing FRA area extraction on multimodal remote sensing images. In this network, we designed a global-local Poolformer block to model the local and global relationships of FRA areas to more comprehensively learn the semantic features of these areas. In addition, we designed a multiscale SAR-guided attention block to efficiently fuse the semantic information acquired from different modalities. The experimental results obtained on the CHN-YS3-FRA dataset show that MSARG-Net could robustly extract offshore FRA regions with F1 scores, intersection-over-union and kappa coefficient values of 91.46%, 84.26%, and 89.69%, respectively. Compared with the latest remote sensing-based semantic segmentation methods, MSARG-Net has achieved significant quantitative and qualitative improvements and has significant potential for mapping and monitoring large-scale offshore FRA areas.
从遥感图像中准确提取近海浮筏水产养殖（FRA）区域是合理管理水产养殖资源的关键。目前，基于深度学习的方法在浮筏养殖区域提取任务中表现良好，但受限于单一模态遥感数据的缺点，影响了其提取精度。为了解决这个问题，我们利用异构的 Sentinel-1/-2 遥感图像数据构建了一个名为 CHN-YS3-FRA 的多模态数据集，用于 FRA 区域提取，并提出了一个新的多尺度合成孔径雷达（SAR）制导网络（MSARG-Net），用于在多模态遥感图像上执行 FRA 区域提取。在该网络中，我们设计了一个全局-局部 Poolformer 模块，以模拟 FRA 区域的局部和全局关系，从而更全面地学习这些区域的语义特征。此外，我们还设计了一个多尺度合成孔径雷达引导的注意力模块，以有效融合从不同模态获取的语义信息。在 CHN-YS3-FRA 数据集上获得的实验结果表明，MSARG-Net 可以稳健地提取离岸 FRA 区域，其 F1 分数、交集-重合度和 kappa 系数值分别为 91.46%、84.26% 和 89.69%。与最新的基于遥感的语义分割方法相比，MSARG-Net 在定量和定性方面都有显著提高，在大尺度近海 FRA 区域的测绘和监测方面具有巨大潜力。
{"title":"MSARG-Net: A Multimodal Offshore Floating Raft Aquaculture Area Extraction Network for Remote Sensing Images Based on Multiscale SAR Guidance","authors":"Haomiao Yu;Fangxiong Wang;Yingzi Hou;Junfu Wang;Jianfeng Zhu;Jianke Guo","doi":"10.1109/JSTARS.2024.3471925","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3471925","url":null,"abstract":"Accurately extracting offshore floating raft aquaculture (FRA) areas from remotely sensed images is the key to rationally managing aquaculture resources. Currently, deep learning-based methods perform well in FRA area extraction tasks but are limited by the shortcomings of single-modality remote sensing data, which affect their extraction accuracies. To solve this problem, we constructed a multimodal dataset called CHN-YS3-FRA for FRA area extraction using heterogeneous Sentinel-1/-2 remote sensing image data and proposed a new multiscale synthetic aperture radar (SAR) guidance network (MSARG-Net) for performing FRA area extraction on multimodal remote sensing images. In this network, we designed a global-local Poolformer block to model the local and global relationships of FRA areas to more comprehensively learn the semantic features of these areas. In addition, we designed a multiscale SAR-guided attention block to efficiently fuse the semantic information acquired from different modalities. The experimental results obtained on the CHN-YS3-FRA dataset show that MSARG-Net could robustly extract offshore FRA regions with F1 scores, intersection-over-union and kappa coefficient values of 91.46%, 84.26%, and 89.69%, respectively. Compared with the latest remote sensing-based semantic segmentation methods, MSARG-Net has achieved significant quantitative and qualitative improvements and has significant potential for mapping and monitoring large-scale offshore FRA areas.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"18319-18334"},"PeriodicalIF":4.7,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10700976","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142452701","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-01DOI: 10.1109/JSTARS.2024.3472041
Huinan Guo;Nengshuang Zhang;Jing Zhang;Wuxia Zhang;Congying Sun
Infrared small target (IST) detection involves identifying objects that occupy fewer than 81 pixels in a 256 × 256 image. Because the target is small and lacks texture, structure, and shape information on its surface, this task is highly challenging. CNN-based methods can extract rich features of the target. However, overly deep network structures may increase the risk of losing small targets. In addition, pixel-level positional deviations can also reduce the detection accuracy of IST. To address these challenges, we propose the location-guided dense nested attention network for IST detection. The proposed network consists of a pixel attention guided feature extraction module (PAG-FEM), a channel attention guided feature fusion module (CAG-FFM), and a detection module. First, the PAG-FEM utilizes the DNIM dense nested blocks from the DNANet as the backbone, integrating both channel and pixel attention mechanisms. This method focuses on the semantic and positional information of the targets, yielding semantic features that emphasize the positions of small targets. Second, the CAG-FFM employs upsampling and convolution operations to align the feature sizes, while utilizing the channel attention mechanism to obtain effective channel information. Then, these features are fused through stacking, addition, and averaging operations to obtain more discriminative features. Finally, the detection module uses eight-connected neighborhood clustering method to obtain the centroid coordinates of the targets for subsequent detection evaluation. Three datasets are utilized to verify our method, and experimental results show that our method performs better than other advanced methods.
红外小目标（IST）检测包括识别在 256 × 256 图像中占不到 81 个像素的物体。由于目标很小，且表面缺乏纹理、结构和形状信息，因此这项任务极具挑战性。基于 CNN 的方法可以提取目标的丰富特征。但是，过深的网络结构可能会增加丢失小目标的风险。此外，像素级的位置偏差也会降低 IST 的检测精度。为了应对这些挑战，我们提出了用于 IST 检测的位置引导密集嵌套注意力网络。该网络由像素注意力引导的特征提取模块（PAG-FEM）、通道注意力引导的特征融合模块（CAG-FFM）和检测模块组成。首先，PAG-FEM 利用 DNANet 中的 DNIM 密集嵌套块作为骨干，整合了通道和像素注意机制。这种方法关注目标的语义和位置信息，产生强调小目标位置的语义特征。其次，CAG-FFM 采用上采样和卷积操作来调整特征大小，同时利用信道注意机制来获取有效的信道信息。然后，这些特征通过堆叠、加法和平均运算进行融合，以获得更具辨别力的特征。最后，检测模块使用八连邻域聚类方法获取目标的中心点坐标，用于后续的检测评估。我们利用三个数据集来验证我们的方法，实验结果表明，我们的方法比其他先进方法的性能更好。
{"title":"Location-Guided Dense Nested Attention Network for Infrared Small Target Detection","authors":"Huinan Guo;Nengshuang Zhang;Jing Zhang;Wuxia Zhang;Congying Sun","doi":"10.1109/JSTARS.2024.3472041","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3472041","url":null,"abstract":"Infrared small target (IST) detection involves identifying objects that occupy fewer than 81 pixels in a 256 × 256 image. Because the target is small and lacks texture, structure, and shape information on its surface, this task is highly challenging. CNN-based methods can extract rich features of the target. However, overly deep network structures may increase the risk of losing small targets. In addition, pixel-level positional deviations can also reduce the detection accuracy of IST. To address these challenges, we propose the location-guided dense nested attention network for IST detection. The proposed network consists of a pixel attention guided feature extraction module (PAG-FEM), a channel attention guided feature fusion module (CAG-FFM), and a detection module. First, the PAG-FEM utilizes the DNIM dense nested blocks from the DNANet as the backbone, integrating both channel and pixel attention mechanisms. This method focuses on the semantic and positional information of the targets, yielding semantic features that emphasize the positions of small targets. Second, the CAG-FFM employs upsampling and convolution operations to align the feature sizes, while utilizing the channel attention mechanism to obtain effective channel information. Then, these features are fused through stacking, addition, and averaging operations to obtain more discriminative features. Finally, the detection module uses eight-connected neighborhood clustering method to obtain the centroid coordinates of the targets for subsequent detection evaluation. Three datasets are utilized to verify our method, and experimental results show that our method performs better than other advanced methods.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"18535-18548"},"PeriodicalIF":4.7,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10702466","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142517959","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-01DOI: 10.1109/JSTARS.2024.3471638
Jian Cui;Jiahang Liu;Yue Ni;Jinjin Wang;Manchun Li
Multiple modal data fusion can provide valuable and diverse information for remote sensing image segmentation. However, the existing fusion methods often lead to feature loss during the fusion of various modal data, and the complementarity among multimodal features is insufficient. To address these problems, we propose a multimodal gated segmentation network for remote sensing images based on the frequency decomposition. Complementary information from multimodal features is extracted by establishing a long-distance correlation between the low-frequency components of different modal data. In addition, high-frequency detailed features of different modal data are preserved by residual connection. The adaptive gated fusion method is then used to control the information flow between the complementary information and each modality feature map, enabling adaptive fusion between multimodal features. These operations can effectively improve the adaptability of the proposed method in various scenarios and data changes. Extensive experiments demonstrate that the proposed method has good effectiveness, robustness, and generalization and achieved state-of-the-art performance in several remote sensing image semantic segmentation tasks.
{"title":"FDGSNet: A Multimodal Gated Segmentation Network for Remote Sensing Image Based on Frequency Decomposition","authors":"Jian Cui;Jiahang Liu;Yue Ni;Jinjin Wang;Manchun Li","doi":"10.1109/JSTARS.2024.3471638","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3471638","url":null,"abstract":"Multiple modal data fusion can provide valuable and diverse information for remote sensing image segmentation. However, the existing fusion methods often lead to feature loss during the fusion of various modal data, and the complementarity among multimodal features is insufficient. To address these problems, we propose a multimodal gated segmentation network for remote sensing images based on the frequency decomposition. Complementary information from multimodal features is extracted by establishing a long-distance correlation between the low-frequency components of different modal data. In addition, high-frequency detailed features of different modal data are preserved by residual connection. The adaptive gated fusion method is then used to control the information flow between the complementary information and each modality feature map, enabling adaptive fusion between multimodal features. These operations can effectively improve the adaptability of the proposed method in various scenarios and data changes. Extensive experiments demonstrate that the proposed method has good effectiveness, robustness, and generalization and achieved state-of-the-art performance in several remote sensing image semantic segmentation tasks.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"19756-19770"},"PeriodicalIF":4.7,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10700993","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142595924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-01DOI: 10.1109/JSTARS.2024.3472096
Yunmei Ma;Lei Zhao;Erxue Chen;Zengyuan Li;Yaxiong Fan;Kunpeng Xu;Han Wang
Accurate estimation of forest aboveground biomass (AGB) is crucial for research on terrestrial carbon cycling and global climate change. In this study, we introduce an improved approach for estimating forest AGB combining P-band and X-band interferometric synthetic aperture radar (InSAR) data. Forest AGB was estimated by combining unbiased forest height and volume backscatter intensity. For forest height, a multilayer model and subaperture decomposition technology were used to remove the penetration bias of the X-band and reduce the effects of forest scatterers on the extraction of a pure understory terrain phase based on P-band, respectively. For volume backscatter intensity, a ground cancellation algorithm based on P-band InSAR was used to eliminate ground scattering contributions unrelated to forest AGB. The proposed method was validated using airborne P-band InSAR data and spaceborne X-band InSAR data gathered over the study area on the Saihanba Forest Farm in Hebei, China. The unbiased forest height and volume backscatter intensity had stronger correlations with forest AGB than estimates derived from unimproved features. The proposed method returned high-precision estimates of forest AGB with an accuracy of 83.73%, an improvement of 8.80% over an estimate derived from unoptimized features. Additionally, AGB estimates combined with forest height and backscatter intensity were greater than those based on a single feature, with the contribution of the former is greater than that of the latter.
准确估算森林地上生物量（AGB）对陆地碳循环和全球气候变化研究至关重要。在本研究中，我们介绍了一种结合 P 波段和 X 波段干涉合成孔径雷达 (InSAR) 数据估算森林 AGB 的改进方法。通过结合无偏的森林高度和体积反向散射强度来估算森林 AGB。在森林高度方面，使用了多层模型和子孔径分解技术，以消除 X 波段的穿透偏差，并减少森林散射体对基于 P 波段提取纯林下地形相位的影响。在体积反向散射强度方面，使用了基于 P 波段 InSAR 的地面消除算法，以消除与森林 AGB 无关的地面散射贡献。利用在中国河北塞罕坝林场研究区域上空采集的机载 P 波段 InSAR 数据和机载 X 波段 InSAR 数据，对所提出的方法进行了验证。与未改进的地貌特征相比，无偏的森林高度和体积反向散射强度与森林AGB的相关性更强。所提出的方法可获得高精度的森林 AGB 估计值，准确率达 83.73%，比未优化地物得出的估计值提高了 8.80%。此外，结合森林高度和反向散射强度得出的 AGB 估计值高于基于单一特征得出的估计值，前者的贡献大于后者。
{"title":"Collaborative Estimation of Aboveground Forest Biomass Using P-Band and X-Band Interferometric Synthetic Aperture Radar Based on Feature Optimization","authors":"Yunmei Ma;Lei Zhao;Erxue Chen;Zengyuan Li;Yaxiong Fan;Kunpeng Xu;Han Wang","doi":"10.1109/JSTARS.2024.3472096","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3472096","url":null,"abstract":"Accurate estimation of forest aboveground biomass (AGB) is crucial for research on terrestrial carbon cycling and global climate change. In this study, we introduce an improved approach for estimating forest AGB combining P-band and X-band interferometric synthetic aperture radar (InSAR) data. Forest AGB was estimated by combining unbiased forest height and volume backscatter intensity. For forest height, a multilayer model and subaperture decomposition technology were used to remove the penetration bias of the X-band and reduce the effects of forest scatterers on the extraction of a pure understory terrain phase based on P-band, respectively. For volume backscatter intensity, a ground cancellation algorithm based on P-band InSAR was used to eliminate ground scattering contributions unrelated to forest AGB. The proposed method was validated using airborne P-band InSAR data and spaceborne X-band InSAR data gathered over the study area on the Saihanba Forest Farm in Hebei, China. The unbiased forest height and volume backscatter intensity had stronger correlations with forest AGB than estimates derived from unimproved features. The proposed method returned high-precision estimates of forest AGB with an accuracy of 83.73%, an improvement of 8.80% over an estimate derived from unoptimized features. Additionally, AGB estimates combined with forest height and backscatter intensity were greater than those based on a single feature, with the contribution of the former is greater than that of the latter.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"17876-17889"},"PeriodicalIF":4.7,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10702464","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142524184","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-01DOI: 10.1109/JSTARS.2024.3472045
Feng Jing;Meng Jiang;Ramesh P. Singh
When earthquakes occur in high-mountain areas during the winter season, the epicentral region is often covered by a snow layer, which can be either thin or thick. The presence of snow and/or ice layers affects the detection of thermal anomalies associated with seismic signals. Taking into account the penetration capabilities of microwaves, microwave brightness temperature data were analyzed by using the index of microwave radiation anomaly to study the response of the epicentral region associated with two recent strong earthquakes in Central Asia, which occurred in snow-covered mountainous areas. Increased microwave radiation was observed within one week prior to the earthquakes. By conducting a comparative analysis of different frequencies and a comprehensive examination of meteorological parameters, we distinguished anomalies caused by tectonic activity from those induced by atmospheric water vapor. A robustness analysis from the periods of seismic tranquility and seismic disturbance has been conducted to validate our results. Our findings suggest that regions with less snow cover or shallow snow depth may exhibit high sensitivity to seismic microwave radiation anomalies in high-altitude mountainous areas during the cold season, which can be detected through passive microwave remote sensing. Combined with a further analysis from microwave polarization difference index and distribution of regional lithology, we proposed that the theory of positive holes may be the dominant mechanism for enhanced microwave radiation.
{"title":"Detection of Seismic Microwave Radiation Anomalies in Snow-Covered Mountainous Terrain: Insights From Two Recent Earthquakes in the Pamir–Tien Shan Region","authors":"Feng Jing;Meng Jiang;Ramesh P. Singh","doi":"10.1109/JSTARS.2024.3472045","DOIUrl":"https://doi.org/10.1109/JSTARS.2024.3472045","url":null,"abstract":"When earthquakes occur in high-mountain areas during the winter season, the epicentral region is often covered by a snow layer, which can be either thin or thick. The presence of snow and/or ice layers affects the detection of thermal anomalies associated with seismic signals. Taking into account the penetration capabilities of microwaves, microwave brightness temperature data were analyzed by using the index of microwave radiation anomaly to study the response of the epicentral region associated with two recent strong earthquakes in Central Asia, which occurred in snow-covered mountainous areas. Increased microwave radiation was observed within one week prior to the earthquakes. By conducting a comparative analysis of different frequencies and a comprehensive examination of meteorological parameters, we distinguished anomalies caused by tectonic activity from those induced by atmospheric water vapor. A robustness analysis from the periods of seismic tranquility and seismic disturbance has been conducted to validate our results. Our findings suggest that regions with less snow cover or shallow snow depth may exhibit high sensitivity to seismic microwave radiation anomalies in high-altitude mountainous areas during the cold season, which can be detected through passive microwave remote sensing. Combined with a further analysis from microwave polarization difference index and distribution of regional lithology, we proposed that the theory of positive holes may be the dominant mechanism for enhanced microwave radiation.","PeriodicalId":13116,"journal":{"name":"IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing","volume":"17 ","pages":"18156-18166"},"PeriodicalIF":4.7,"publicationDate":"2024-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10702460","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142452754","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","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：