Pub Date : 2025-02-17DOI: 10.1109/LGRS.2025.3542874
Duo Jia;Cangjiao Wang;Yanchen Bo
The vertical leaf area index (LAI) is crucial for assessing photosynthetic and carbon sequestration dynamics and atmospheric interaction within terrestrial ecosystems. The global ecosystem dynamics investigation (GEDI), the first full-waveform lidar for monitoring global forest structure, has generated a vertical plant area index (VPAI) product at 5 m intervals. This study conducts a comprehensive assessment of the accuracy of GEDI’s vertical plant area index (PAI) across temperate forests in the conterminous United States and analyzes the impact of sensor parameters on the accuracy of the VPAI to provide insights for the optimal application of GEDI’s capabilities. The results show that GEDI can offer reliable layered PAI for heights exceeding 10 m. Substantial inaccuracies across various forest types were observed in layers of 5–10 m, with the worst accuracy observed in mixed forests. The impact of GEDI sensor parameters varies across different layers of PAI with GEDI’s Power beam being more accurate than its Coverage beam in layered PAI below 20 m; night observations are more accurate but also less available than day observations. A significant negative correlation between the signal-to-noise ratio (SNR) and errors of layered PAI exists below 15 m. Prioritizing the use of the Power beam and weighting the GEDI footprint based on the SNR for layers within 15 m, are recommended ways to improve the accuracy of the subsequent layered PAI mapping.
{"title":"Evaluation of GEDI for Estimating the Vertical Distribution of PAI in Temperate Forests: A Case Study of the Conterminous United States","authors":"Duo Jia;Cangjiao Wang;Yanchen Bo","doi":"10.1109/LGRS.2025.3542874","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3542874","url":null,"abstract":"The vertical leaf area index (LAI) is crucial for assessing photosynthetic and carbon sequestration dynamics and atmospheric interaction within terrestrial ecosystems. The global ecosystem dynamics investigation (GEDI), the first full-waveform lidar for monitoring global forest structure, has generated a vertical plant area index (VPAI) product at 5 m intervals. This study conducts a comprehensive assessment of the accuracy of GEDI’s vertical plant area index (PAI) across temperate forests in the conterminous United States and analyzes the impact of sensor parameters on the accuracy of the VPAI to provide insights for the optimal application of GEDI’s capabilities. The results show that GEDI can offer reliable layered PAI for heights exceeding 10 m. Substantial inaccuracies across various forest types were observed in layers of 5–10 m, with the worst accuracy observed in mixed forests. The impact of GEDI sensor parameters varies across different layers of PAI with GEDI’s Power beam being more accurate than its Coverage beam in layered PAI below 20 m; night observations are more accurate but also less available than day observations. A significant negative correlation between the signal-to-noise ratio (SNR) and errors of layered PAI exists below 15 m. Prioritizing the use of the Power beam and weighting the GEDI footprint based on the SNR for layers within 15 m, are recommended ways to improve the accuracy of the subsequent layered PAI mapping.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143570818","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-17DOI: 10.1109/LGRS.2025.3542586
Xiaojun Dai;Weiyi Huang;Ming Xi;Yaqi Zhang;Deying Ma;Daguo Wang
The high-consequence area (HCA) is crucial for the safety management and operation of oil and gas pipelines. However, traditional models that rely on manual field investigations are costly, inefficient, and risky. Deep learning (DL)-based instance segmentation (IS) has the potential to enable automatic HCA identification. Unfortunately, the existing studies lack methods specifically designed to identify HCAs from remote sensing (RS) images. This letter proposes an IS network (HCA-Net) with spatial relation enhancement and mask decoupling refinement for HCA recognition is proposed. The proposed method first develops a spatial relation enhancement module (SREM) that queries the similarity of features at different spatial locations to represent spatial relations, further enhancing these features to promote completeness. Moreover, a unique decoupled mask refinement head (DMRH) is designed to refine the mask by decoupling boundary features from body features and optimally integrating them into the final features. Experiments on the constructed gas pipeline aerial dataset (GPAD) show that our method outperforms eight state-of-the-art (SOTA) methods. Compared to the baseline model mask R-CNN, HCA-Net improves the mAP of masks and the mIoU of HCA by 3.9% and 6.9%, respectively.
{"title":"HCA-Net: An Instance Segmentation Network for High-Consequence Areas Identification From Remote Sensing Images","authors":"Xiaojun Dai;Weiyi Huang;Ming Xi;Yaqi Zhang;Deying Ma;Daguo Wang","doi":"10.1109/LGRS.2025.3542586","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3542586","url":null,"abstract":"The high-consequence area (HCA) is crucial for the safety management and operation of oil and gas pipelines. However, traditional models that rely on manual field investigations are costly, inefficient, and risky. Deep learning (DL)-based instance segmentation (IS) has the potential to enable automatic HCA identification. Unfortunately, the existing studies lack methods specifically designed to identify HCAs from remote sensing (RS) images. This letter proposes an IS network (HCA-Net) with spatial relation enhancement and mask decoupling refinement for HCA recognition is proposed. The proposed method first develops a spatial relation enhancement module (SREM) that queries the similarity of features at different spatial locations to represent spatial relations, further enhancing these features to promote completeness. Moreover, a unique decoupled mask refinement head (DMRH) is designed to refine the mask by decoupling boundary features from body features and optimally integrating them into the final features. Experiments on the constructed gas pipeline aerial dataset (GPAD) show that our method outperforms eight state-of-the-art (SOTA) methods. Compared to the baseline model mask R-CNN, HCA-Net improves the mAP of masks and the mIoU of HCA by 3.9% and 6.9%, respectively.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143496602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Deep metric learning (DML) has shown promising results in few-shot hyperspectral image (HSI) classification. The core idea of DML is to learn a generalized metric space, in which pixels from unseen classes can be effectively classified with only a few labeled samples. However, the existing DML methods mainly adopt traditional Euclidean distance to achieve the feature metric, which ignores the category uncertainty of spatial-spectral features in mixed and edge pixels. To address this issue, we fully exploit fuzzy logic theory and propose a deep fuzzy metric learning (DFML) method for few-shot HSI classification. First, we design a novel hybrid CNN-transformer spatial-spectral feature extraction network to fully capture the spatial-spectral features of HSI pixels. Then, a fuzzy set representation method based on Gaussian membership function for spatial-spectral features is proposed, which describes the inherent fuzziness of the spatial-spectral features. Finally, to perform the fuzzy similarity measure between the fuzzy sets of query samples and prototypes, we construct a spatial-spectral fuzzy metric space, in which HSI pixels with category uncertainty in their features can be better classified under the condition of small-scale labeled samples. Extensive experimental results on three public HSI datasets demonstrate that the proposed DFML method outperforms the state-of-the-art few-shot HSI classification methods.
{"title":"Few-Shot Hyperspectral Image Classification With Deep Fuzzy Metric Learning","authors":"Haojin Tang;Chao Zhang;Dong Tang;Xin Lin;Xiaofei Yang;Weixin Xie","doi":"10.1109/LGRS.2025.3542571","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3542571","url":null,"abstract":"Deep metric learning (DML) has shown promising results in few-shot hyperspectral image (HSI) classification. The core idea of DML is to learn a generalized metric space, in which pixels from unseen classes can be effectively classified with only a few labeled samples. However, the existing DML methods mainly adopt traditional Euclidean distance to achieve the feature metric, which ignores the category uncertainty of spatial-spectral features in mixed and edge pixels. To address this issue, we fully exploit fuzzy logic theory and propose a deep fuzzy metric learning (DFML) method for few-shot HSI classification. First, we design a novel hybrid CNN-transformer spatial-spectral feature extraction network to fully capture the spatial-spectral features of HSI pixels. Then, a fuzzy set representation method based on Gaussian membership function for spatial-spectral features is proposed, which describes the inherent fuzziness of the spatial-spectral features. Finally, to perform the fuzzy similarity measure between the fuzzy sets of query samples and prototypes, we construct a spatial-spectral fuzzy metric space, in which HSI pixels with category uncertainty in their features can be better classified under the condition of small-scale labeled samples. Extensive experimental results on three public HSI datasets demonstrate that the proposed DFML method outperforms the state-of-the-art few-shot HSI classification methods.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143535511","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-17DOI: 10.1109/LGRS.2025.3542797
Ming Jin;Jiashuo Zhang;Lifei Jiang;Jieying He;Jiakai He;Jing Xu;Yunan Han;Ming Bai
In this work, the brightness temperature (TB) transfer from the microwave calibration target (MCT) to the feeder antenna in the near-field region is investigated, which is the fundamental physical process in microwave radiometer calibration. As in this scenario, the MCT and antenna cannot be separately considered as points like in far fields, it is interesting and important to study the TB characteristics of the target in cases of feeder antennas at different positions and with different aperture sizes. Recently, as reciprocity in the near field has been established, this fundamental issue can be now investigated. The study starts at a high frequency of 89 GHz when the free-space lambda is much smaller than the unit period of MCT; then, the distributions of the local TB contribution rate are calculated to understand the possible TB transfer variation. It is found that the key factor for the TB variation is the illumination area upon the array-type MCT, and as the footprint is sufficiently large to cover several pyramid units, the TB can be stable versus relative position.
{"title":"On the Transferred Brightness Temperature Variation Versus Feeder Antenna Position in a Fundamental Calibration Link","authors":"Ming Jin;Jiashuo Zhang;Lifei Jiang;Jieying He;Jiakai He;Jing Xu;Yunan Han;Ming Bai","doi":"10.1109/LGRS.2025.3542797","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3542797","url":null,"abstract":"In this work, the brightness temperature (TB) transfer from the microwave calibration target (MCT) to the feeder antenna in the near-field region is investigated, which is the fundamental physical process in microwave radiometer calibration. As in this scenario, the MCT and antenna cannot be separately considered as points like in far fields, it is interesting and important to study the TB characteristics of the target in cases of feeder antennas at different positions and with different aperture sizes. Recently, as reciprocity in the near field has been established, this fundamental issue can be now investigated. The study starts at a high frequency of 89 GHz when the free-space lambda is much smaller than the unit period of MCT; then, the distributions of the local TB contribution rate are calculated to understand the possible TB transfer variation. It is found that the key factor for the TB variation is the illumination area upon the array-type MCT, and as the footprint is sufficiently large to cover several pyramid units, the TB can be stable versus relative position.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143553455","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1109/LGRS.2025.3542065
Chenyin Ding;Qianwen Cheng;Yukun Lin;Jia Yu;Shiqiang Du;Bo Du
The change detection of high-resolution images plays an important role in practical applications. However, most existing studies use local or global attention mechanisms alone to filter and screen for changing features. In this study, we proposed a triple attention multiscale fusion network (TAMFNet) that can effectively utilize both global and local attention mechanisms, thereby improving the ability to detect the location of change areas and fully outline the change areas. First, we employed a fully convolutional network to extract features from dual temporal images at different scales. Second, three complementary attention mechanisms, namely, the spatial attention mechanism (SAM), channel attention mechanism (CAM), and the multihead self-attention (MSA) module, were integrated to extract and fuse global and local features. Finally, to address semantic and scale differences, we utilized the cross scale fusion (CSF) module, pyramid pooling module (PPM), and pyramid receptive field (PRF) module to aggregate features from adjacent scales for comprehensive feature transmission. To demonstrate the effectiveness of our method, we tested it on the LEVIR-CD and WHU-CD datasets. The results showed that our model achieved intersection over union (IOU) scores of 80.17% and 77.23% on the datasets, outperforming comparative models. Ablation experiments on the LEVIR-CD dataset confirmed the positive impact of each intermediate module in TAMFNet, with an overall 2.35% increase in IOU score.
{"title":"Remote Sensing Images Change Detection Using Triple Attention Mechanism to Aggregate Global and Local Features","authors":"Chenyin Ding;Qianwen Cheng;Yukun Lin;Jia Yu;Shiqiang Du;Bo Du","doi":"10.1109/LGRS.2025.3542065","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3542065","url":null,"abstract":"The change detection of high-resolution images plays an important role in practical applications. However, most existing studies use local or global attention mechanisms alone to filter and screen for changing features. In this study, we proposed a triple attention multiscale fusion network (TAMFNet) that can effectively utilize both global and local attention mechanisms, thereby improving the ability to detect the location of change areas and fully outline the change areas. First, we employed a fully convolutional network to extract features from dual temporal images at different scales. Second, three complementary attention mechanisms, namely, the spatial attention mechanism (SAM), channel attention mechanism (CAM), and the multihead self-attention (MSA) module, were integrated to extract and fuse global and local features. Finally, to address semantic and scale differences, we utilized the cross scale fusion (CSF) module, pyramid pooling module (PPM), and pyramid receptive field (PRF) module to aggregate features from adjacent scales for comprehensive feature transmission. To demonstrate the effectiveness of our method, we tested it on the LEVIR-CD and WHU-CD datasets. The results showed that our model achieved intersection over union (IOU) scores of 80.17% and 77.23% on the datasets, outperforming comparative models. Ablation experiments on the LEVIR-CD dataset confirmed the positive impact of each intermediate module in TAMFNet, with an overall 2.35% increase in IOU score.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143553533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1109/LGRS.2025.3542121
Mengxin Guo;Siyuan Chen;Zuoshi Liu;Weihong Wang
In geophysics, structure-oriented filtering (SOF) has been extensively utilized as an effective algorithm to eliminate random noise. This type of filtering is mainly restricted by the local dip of the data and smoothed along the structural direction. The algorithm assumes the strata of a single local dip, and even if the seismic data contain strata and faults, the SOF remains smoothed with a single dip angle. This condition leads to the smoothing of faults or strata in the wrong direction, which results in the effective removal of reflected signals. Given this, we herein propose a multidip SOF (Mdip-SOF) algorithm for complex structure data, including faults. This investigation employs a guided filtering theory to pre-extract structural features of faults. Guided by this attribute, we construct an edge-preserving filter, where faults may exist, aiming to preserve faults and other structures. Other positions with a single dip are also adopted for conventional SOF. A poststack data denoising test is also conducted, and the results obtained reveal that the proposed multidip constrained SOF algorithm outperforms the existing SOF algorithm in terms of fault handling.
{"title":"Application of Multidip Structure-Oriented Filtering to Poststack Seismic Data","authors":"Mengxin Guo;Siyuan Chen;Zuoshi Liu;Weihong Wang","doi":"10.1109/LGRS.2025.3542121","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3542121","url":null,"abstract":"In geophysics, structure-oriented filtering (SOF) has been extensively utilized as an effective algorithm to eliminate random noise. This type of filtering is mainly restricted by the local dip of the data and smoothed along the structural direction. The algorithm assumes the strata of a single local dip, and even if the seismic data contain strata and faults, the SOF remains smoothed with a single dip angle. This condition leads to the smoothing of faults or strata in the wrong direction, which results in the effective removal of reflected signals. Given this, we herein propose a multidip SOF (Mdip-SOF) algorithm for complex structure data, including faults. This investigation employs a guided filtering theory to pre-extract structural features of faults. Guided by this attribute, we construct an edge-preserving filter, where faults may exist, aiming to preserve faults and other structures. Other positions with a single dip are also adopted for conventional SOF. A poststack data denoising test is also conducted, and the results obtained reveal that the proposed multidip constrained SOF algorithm outperforms the existing SOF algorithm in terms of fault handling.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143496532","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-14DOI: 10.1109/LGRS.2025.3542169
Yaxian Lei;Xiaochong Tong;Chunping Qiu;Haoshuai Song;Congzhou Guo;He Li
Recent advances in stable diffusion models have revolutionized text-to-image generation. However, these models struggle with spatial relationship comprehension in remote sensing (RS) scenarios, limiting their ability to generate spatially accurate imagery. We present a novel framework for generating RS images from spatial relationship descriptions with precise spatial control. Our approach introduces a two-stage pipeline: first, a spatial relationship semantic structuring model converts formalized spatial relationship descriptions into controlled layouts, and second, an enhanced diffusion model incorporates positional prompts and a layout attention mechanism to generate the final image. The positional prompts explicitly encode spatial information, while the layout attention mechanism enables focused region learning. Comprehensive experiments demonstrate that our method achieves superior performance compared with state-of-the-art approaches in both spatial accuracy and image quality.
{"title":"Spatial-Aware Remote Sensing Image Generation From Spatial Relationship Descriptions","authors":"Yaxian Lei;Xiaochong Tong;Chunping Qiu;Haoshuai Song;Congzhou Guo;He Li","doi":"10.1109/LGRS.2025.3542169","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3542169","url":null,"abstract":"Recent advances in stable diffusion models have revolutionized text-to-image generation. However, these models struggle with spatial relationship comprehension in remote sensing (RS) scenarios, limiting their ability to generate spatially accurate imagery. We present a novel framework for generating RS images from spatial relationship descriptions with precise spatial control. Our approach introduces a two-stage pipeline: first, a spatial relationship semantic structuring model converts formalized spatial relationship descriptions into controlled layouts, and second, an enhanced diffusion model incorporates positional prompts and a layout attention mechanism to generate the final image. The positional prompts explicitly encode spatial information, while the layout attention mechanism enables focused region learning. Comprehensive experiments demonstrate that our method achieves superior performance compared with state-of-the-art approaches in both spatial accuracy and image quality.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143570683","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-13DOI: 10.1109/LGRS.2025.3541895
Jie Li;Zhanhong Liu;Shujun Liu;Huajun Wang
The utilization of both optical and synthetic aperture radar (SAR) images for joint semantic segmentation enhances the accuracy of land use classification. Recent advancements in multimodal fusion models, particularly those using self-attention mechanisms and convolutional neural networks (CNNs), have yielded significant results. However, self-attention has quadratic computational complexity, and CNN has insufficient local-global contextual modeling power. Recently, 2-D-selective-scan (SS2D) has emerged as a promising approach. It excels in modeling long-range dependencies while maintaining linear computational complexity. Based on SS2D, we propose a joint semantic segmentation network for optical and SAR images, called MBSSNet. Specifically, we introduce SS2D and design a cross-modal fusion module (CMFM) to fuse multimodal features from dual branches layer by layer, thereby enhancing the consistency of fused feature representations. In addition, during the decoding phase, we integrate contextual information from multiscale fusion features, thereby enhancing the spatial and semantic information of the fused features. Our experimental results show that our method outperforms the state-of-the-art (SOTA), and overall accuracy (OA), mean intersection over union (mIoU), and Kappa outperform other SOTA methods by 1.7%, 3.1%, and 2.2%, respectively.
{"title":"MBSSNet: A Mamba-Based Joint Semantic Segmentation Network for Optical and SAR Images","authors":"Jie Li;Zhanhong Liu;Shujun Liu;Huajun Wang","doi":"10.1109/LGRS.2025.3541895","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3541895","url":null,"abstract":"The utilization of both optical and synthetic aperture radar (SAR) images for joint semantic segmentation enhances the accuracy of land use classification. Recent advancements in multimodal fusion models, particularly those using self-attention mechanisms and convolutional neural networks (CNNs), have yielded significant results. However, self-attention has quadratic computational complexity, and CNN has insufficient local-global contextual modeling power. Recently, 2-D-selective-scan (SS2D) has emerged as a promising approach. It excels in modeling long-range dependencies while maintaining linear computational complexity. Based on SS2D, we propose a joint semantic segmentation network for optical and SAR images, called MBSSNet. Specifically, we introduce SS2D and design a cross-modal fusion module (CMFM) to fuse multimodal features from dual branches layer by layer, thereby enhancing the consistency of fused feature representations. In addition, during the decoding phase, we integrate contextual information from multiscale fusion features, thereby enhancing the spatial and semantic information of the fused features. Our experimental results show that our method outperforms the state-of-the-art (SOTA), and overall accuracy (OA), mean intersection over union (mIoU), and Kappa outperform other SOTA methods by 1.7%, 3.1%, and 2.2%, respectively.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143570562","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-02-13DOI: 10.1109/LGRS.2025.3541342
Chu Jian;Jun Li;Zhengwei Xu;Xiaolin Tian;Zhipeng Cheng;Jiayue Deng;Mujing Lan;Yue Sun
The deep exploration potential of basic-ultrabasic rock masses associated with Panzhihua-type vanadium-titanium magnetite (VTM) deposits are closely tied to the occurrence of deep-seated rock bodies. In this study, we utilized newly acquired 1:50000 scale aeromagnetic data from the Panxi region to perform a 3-D magnetization inversion using an improved regularized focusing conjugate gradient approach to achieve high-resolution 3-D magnetic imaging of basic-ultrabasic rock masses within the “Panzhihua-type” VTM concentration zone at depths reaching 10 km. The inversion results reveal that the 3-D magnetic anomalies of strong magnetic sources correspond with the distribution of the NS fault zones in the study area. However, these anomalies are predominantly located within narrow zones between the fault zones rather than directly along the fault lines. It also suggests that during the Late Huashan period, two rift regions might have developed in the Panxi area: the Anninghe Rift and the Panzhihua Rift. The deep and large faults within these confined rift valleys likely controlled the eruption and intrusion of mantle-derived magma, facilitating the emplacement of basic-ultrabasic strong magnetic rock masses along these zones. Additionally, the local shear structures within the paleo-rift zones may have provided ample space and a relatively stable environment conducive to the formation of VTM deposits.
{"title":"Magnetic Characteristics of Deep-Seated “Panzhihua-Type” Vanadium-Titanium Magnetite Based on 3-D Aeromagnetic Inversion","authors":"Chu Jian;Jun Li;Zhengwei Xu;Xiaolin Tian;Zhipeng Cheng;Jiayue Deng;Mujing Lan;Yue Sun","doi":"10.1109/LGRS.2025.3541342","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3541342","url":null,"abstract":"The deep exploration potential of basic-ultrabasic rock masses associated with Panzhihua-type vanadium-titanium magnetite (VTM) deposits are closely tied to the occurrence of deep-seated rock bodies. In this study, we utilized newly acquired 1:50000 scale aeromagnetic data from the Panxi region to perform a 3-D magnetization inversion using an improved regularized focusing conjugate gradient approach to achieve high-resolution 3-D magnetic imaging of basic-ultrabasic rock masses within the “Panzhihua-type” VTM concentration zone at depths reaching 10 km. The inversion results reveal that the 3-D magnetic anomalies of strong magnetic sources correspond with the distribution of the NS fault zones in the study area. However, these anomalies are predominantly located within narrow zones between the fault zones rather than directly along the fault lines. It also suggests that during the Late Huashan period, two rift regions might have developed in the Panxi area: the Anninghe Rift and the Panzhihua Rift. The deep and large faults within these confined rift valleys likely controlled the eruption and intrusion of mantle-derived magma, facilitating the emplacement of basic-ultrabasic strong magnetic rock masses along these zones. Additionally, the local shear structures within the paleo-rift zones may have provided ample space and a relatively stable environment conducive to the formation of VTM deposits.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143455230","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In recent years, the detection for tiny objects in remote sensing images has become a hot research topic. Tiny objects contain a limited number of pixels and are easily confused with the background, which leads to low detection accuracy. To the end, this letter proposes a tiny object detection method based on adaptive downsampling and scale enhanced detection head (SEDH) to improve the accuracy of detection without increasing the model parameters. First, the dynamic feature extraction module (DFEM) is proposed. The module can obtain the context information of tiny objects. Second, the adaptive downsampling module (ADM) is designed to capture local details of tiny objects. Finally, the scale enhanced detection head is constructed which improves the sensitivity to tiny objects, while reducing the number of parameters of the model. To verify the effectiveness of the proposed method, a series of experiments are conducted on the challenging AI-TOD dataset. The experimental results demonstrate that the proposed method effectively trade-offs the relationship between detection accuracy and the number of model parameters.
{"title":"Adaptive Downsampling and Scale Enhanced Detection Head for Tiny Object Detection in Remote Sensing Image","authors":"Yunzuo Zhang;Ting Liu;Jiawen Zhen;Yaoxing Kang;Yu Cheng","doi":"10.1109/LGRS.2025.3532983","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3532983","url":null,"abstract":"In recent years, the detection for tiny objects in remote sensing images has become a hot research topic. Tiny objects contain a limited number of pixels and are easily confused with the background, which leads to low detection accuracy. To the end, this letter proposes a tiny object detection method based on adaptive downsampling and scale enhanced detection head (SEDH) to improve the accuracy of detection without increasing the model parameters. First, the dynamic feature extraction module (DFEM) is proposed. The module can obtain the context information of tiny objects. Second, the adaptive downsampling module (ADM) is designed to capture local details of tiny objects. Finally, the scale enhanced detection head is constructed which improves the sensitivity to tiny objects, while reducing the number of parameters of the model. To verify the effectiveness of the proposed method, a series of experiments are conducted on the challenging AI-TOD dataset. The experimental results demonstrate that the proposed method effectively trade-offs the relationship between detection accuracy and the number of model parameters.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"22 ","pages":"1-5"},"PeriodicalIF":0.0,"publicationDate":"2025-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143430564","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","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：