IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society最新文献

{"title":"Multiscale Classifier Ensemble Knowledge Distillation for Remote Sensing Image Classification","authors":"Pan He;Xiao Wang;Hang Pu","doi":"10.1109/LGRS.2026.3658005","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3658005","url":null,"abstract":"To address the problems of significant remote sensing image ground object scale differences and traditional classification models struggling to balance accuracy and lightweight performance, this letter proposes a multiscale classifier ensemble knowledge distillation (MCEKD) method for remote sensing image classification. Based on deep neural networks (DNNs), the method constructs multiscale classifiers as teacher models and transfers multiscale ground object discrimination knowledge to corresponding student models through ensemble soft labels. Experimental results on the UC Merced dataset, aerial image dataset (AID), and NWPU-RESISC45 dataset demonstrate that when compressing the parameter quantity of mainstream models such as GCSANet model, Vision Transformer (ViT) model, and adaptive feature interaction network (AFIMNet) model, the performance of student models remains close to that of teacher models. The effectiveness of this method has also been demonstrated through experiments on the PatternNet, EuroSAT, RSD46-WHU, and remote sensing object detection (RSOD) datasets. Results of visualization experiments verify that the proposed method can better focus on important features.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1-5"},"PeriodicalIF":4.4,"publicationDate":"2026-01-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146223583","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}

{"title":"FFO-DETR: Using Feature Filter to Improve the Performance of Oriented Detection Transformer","authors":"Weiyi Zhang;Jun Wang;Xiaofeng Xu;Qimeng Shi","doi":"10.1109/LGRS.2026.3657158","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3657158","url":null,"abstract":"In recent years, oriented object detection has garnered considerable interest due to its wide range of applications and the impressive performance of recent models. Due to the high resolution of remote sensing images, object detection models often require significant computational resources and lengthy inference times. To tackle this issue, this letter introduces a sparse attention module into the oriented object detector, enhancing efficiency by selectively computing attention on foreground features. We have optimized key components to better adapt the pipeline for remote sensing images and oriented object detection. With these improvements, our model achieves notable performance gains of 2% AP50 on DIOR-R and 1.2% AP50 on DOTA-v1.0 compared to the baseline ARS-DETR, along with a 10.9% increase in FPS and a 24.2% decrease in FLOPs.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1-5"},"PeriodicalIF":4.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146175881","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}

{"title":"Acoustic-to-Seismic Signal Attenuation and Aircraft Trajectory Tracking","authors":"Hongbin Lu;Jie Shao;Yibo Wang","doi":"10.1109/LGRS.2026.3652414","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3652414","url":null,"abstract":"Airborne sound sources generate strong pressure disturbances that couple with the ground to produce acoustic-to-seismic signals, leaving a continuous seismic footprint detectable by sensors. This study simulates the seismic response of moving aircraft using the finite difference method (FDM) and analyzes the relationship between signal amplitude and the distance to the flight path. By integrating the wavefront diffusion equation with the distance formula, we derive a mathematical link between maximum amplitude and trajectory parameters. Based on this relation, an inversion algorithm using linear array data is developed for trajectory tracking. Field data from Beijing Capital International Airport validate the feasibility of the method. Results demonstrate that the proposed approach enables reliable large-scale and long-term aircraft monitoring, complementing traditional radar and acoustic technologies for aviation surveillance.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1-5"},"PeriodicalIF":4.4,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146026393","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}

{"title":"RARM-YOLO: Remote Sensing Small-Target Detection Model Enhanced by Dual-Branch Region-Aware Refinement Module","authors":"Weiyong Tang;Xiao Yang;Haihe Zhou;Yingli Liu","doi":"10.1109/LGRS.2026.3653619","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3653619","url":null,"abstract":"Remote sensing image object detection is characterized by dim features of small objects and complex backgrounds. In most networks, simply enhancing small object features may disrupt global consistency and affect detection results. This letter designs a region-aware refinement module (RARM) to locate the enhanced target semantic representation and suppress background noise interference and a dual detection branch to fuse underlying details and shallow semantic features at the neck of the feature pyramid, enhancing the model’s ability to focus on very small targets. Experimental results show that the improved model achieves an mAP50% of 76.0% on the Vehicle Detection in AI (VEDAI) dataset, which is 10.6% higher than the original YOLOv8s. The mAP<inline-formula> <tex-math>${}_{mathrm {50-95}}$ </tex-math></inline-formula>% is 46.5%, which is 7% higher. The precision and recall rates are improved by 7.5% and 12.2%, respectively. The generalization performance was verified on VisDrone2019 and SODA-A remote sensing datasets, with mAP50% of 47.1% and 73.8%, respectively, a model size of only 29.5 MB, balancing lightweight and high detection performance. This method provides a technical approach involving the cooperative optimization of multiple modules for target detection in complex remote sensing scenes, offering significant application value.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1-5"},"PeriodicalIF":4.4,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081995","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}

{"title":"A Practical Landsat-9 Land Surface Temperature Sharpen Method Combining Top-of-Atmosphere Multispectral and Panchromatic Reflectance","authors":"Jian Hui;Jie Liu;Xue Liu;Jian Zhu;Yanhong Duan;Xin Ye","doi":"10.1109/LGRS.2026.3652424","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3652424","url":null,"abstract":"The thermal infrared (TIR) band of the Landsat-9 satellite has a spatial resolution of 100 m, which is coarser than the optical multispectral bands of the 30-m resolution and the panchromatic band of the 15-m resolution. Existing image fusion algorithm studies primarily focus on multispectral and panchromatic bands. For the TIR band with the lowest spatial resolution, although Landsat-9 land surface temperature (LST) products are resampled to 30 m using the cubic algorithm to match the multispectral bands, the physical mechanism of sharpening methods remains unclear, and they fail to fully leverage the higher spatial resolution advantage provided by the panchromatic band. This study proposes a practical LST sharpening method (TOA-MsPS) by integrating the correlation between top-of-atmosphere TIR brightness temperature (BT) and the fused reflectance of multispectral and panchromatic bands. The TOA-MsPS method comprises three steps: panchromatic and multispectral band fusion, correlation modeling of reflectance and BT, and LST end-to-end retrieval. Compared to existing methods, the TOA-MsPS method sharpens the spatial resolution of both TIR band BT images to 15 m without relying on external parameters, simultaneously deriving the LST data. All input data for the method consists of remote sensing observations at the TOA, reducing external parameter uncertainties and error propagation inherent in the preprocessing steps of current LST retrieval algorithms, such as atmospheric correction, resampling, and emissivity estimation. Qualitative visual interpretation based on visual inspection indicates that TOA-MsPS-derived LST images exhibit significantly enhanced detail and reasonable local spatial distribution. Quantitative validations using ground site measurements demonstrate that the sharpened LST images achieve comparable accuracy to Landsat-9 LST products while substantially improving spatial resolution, with root mean square errors of 2.45 K (LST product) and 2.35 K (LST sharpen), respectively. Furthermore, the TOA-MsPS method can be directly applied to remote sensing data from multiple other remote sensing data sources, further enhancing the precision of long-term land surface thermal radiance monitoring.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1-5"},"PeriodicalIF":4.4,"publicationDate":"2026-01-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081999","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}

{"title":"PWIoU-YOLO: A Position-Wise IoU-Optimized Detector With Multiscale Context Fusion for Aerial Oriented Objects","authors":"Shihao Chen;Junwei Yu;Hao Yang;Xinmo Zhao;Chunhua Zhu","doi":"10.1109/LGRS.2026.3677568","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3677568","url":null,"abstract":"Accurate oriented object detection in aerial imagery is crucial for remote sensing applications, yet remains challenging due to arbitrary orientations, extreme aspect ratios, and dense distributions of objects. Existing rotated detectors often suffer from unstable training gradients and inadequate geometric representation in intersection-over-union (IoU) computation. This study proposes PWIoU-YOLO, a novel detection framework that introduces position-wise IoU (PWIoU) as the core optimization strategy. Specifically, we develop: 1) a contextual spatial pyramid pooling fast (CSPPF) module that enhances multiscale feature representation while suppressing background interference; 2) a multiscale channel attention mechanism (MSCAM) that adaptively fuses features across scales; and 3) the PWIoU loss, which combines the smallest oriented enclosing rectangle, probabilistic IoU, and dynamic gradient modulation to stabilize training and improve regression accuracy. Extensive experiments on DOTA-v1.0 demonstrate that our method achieves 81.2% mAP, outperforming state-of-the-art approaches by 2.5% while maintaining computational efficiency.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1-5"},"PeriodicalIF":4.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147737030","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}

{"title":"An Intelligent Prestack Seismic Inversion Framework Using the Propagator Matrix Method","authors":"Shenghuang Li;Sanyi Yuan;Lu Qin;Shangxu Wang","doi":"10.1109/LGRS.2026.3682542","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3682542","url":null,"abstract":"Simultaneous inversion of P-wave velocity, S-wave velocity, and density remains a major challenge in quantitative seismic interpretation. Conventional inversion techniques based on the Zoeppritz equations and their approximations often fail to adequately capture the complexity of wave propagation in layered media, especially in the presence of interbed multiples and mode conversions. These limitations become particularly evident in waveform-matching-based inversion methods. To address this issue, we propose an intelligent inversion framework constrained by the propagator matrix (PM) method, in which a differentiable wave-propagation operator is embedded into a data-driven bidirectional gated recurrent unit (Bi-GRU) network. Compared with approaches that characterize only single-interface reflection/transmission behavior, the introduced PM module provides a more complete 1-D wave-propagation description, thereby enhancing well-to-seismic tie quality. The PM module is then used to constrain both the network’s learning over the entire reservoir interval and the prediction process. Both synthetic and field experiments demonstrate that, compared with a purely data-driven inversion and a Zoeppritz-constrained model-driven approach, the proposed method achieves improved accuracy and stability. Integrating the PM method with deep learning (DL) provides an effective strategy for physics-constrained, high-fidelity elastic-parameter inversion.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1-5"},"PeriodicalIF":4.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147828782","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}

{"title":"Learning Distributional Differences in High-Dimensional Seismic Data Using Modified Variational Autoencoders","authors":"Tiash Ghosh;Aurobinda Routray;Mamata Jenamani;Sanjai Kumar Singh","doi":"10.1109/LGRS.2026.3694346","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3694346","url":null,"abstract":"Seismic data plays a critical role in several geophysical tasks. However, the statistical properties of seismic datasets vary significantly across geographical regions, leading to distributional shifts that undermine the reliability of traditional algorithms. In this letter, we propose a deep generative framework based on variational autoencoders to learn the underlying probability distributions of high-dimensional seismic data. By modeling the underlying latent representations from geologically distinct regions, we demonstrate how the proposed method can effectively capture distributional shifts. Such differences in the underlying data distributions often affect the performance of traditional seismic interpretation algorithms. Visual comparison and multivariate statistical tests of the latent space confirm the distributional differences between seismic data of different regions. Hence, we highlight the potential of variational autoencoders (VAEs) as seismic data descriptors, offering a probabilistic approach to model the variability across complex high-dimensional seismic datasets.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"7504305-7504305"},"PeriodicalIF":4.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148236083","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}

{"title":"MSCTFormer: A High-Resolution Remote Sensing Image Water Body Fine Extraction Network Based on a Hybrid CNN–Transformer Architecture","authors":"Bo Song;Zhiyong Zhang;Zhili Chen;Qingyang Wang;Bo Li","doi":"10.1109/LGRS.2026.3693304","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3693304","url":null,"abstract":"Water resources form a vital foundation for human survival and development, and their precise monitoring is crucial for addressing global climate change. Water body extraction based on remote sensing imagery serves as a primary technical approach. However, existing methods still face challenges such as blurred boundaries and significant scale variations when processing high-resolution imagery, resulting in insufficient segmentation accuracy. To address this, this letter proposes a residual network model integrating multiscale contextual attention, called MSCTFormer. This model employs ResNet50 as its encoder, while its decoder core incorporates three innovative modules: 1) enhanced multiscale context attention (EMSCA) module, which integrates multiscale local feature extraction with cross-window global attention (CWGA) to effectively capture multiscale water body characteristics; 2) enhanced feedforward network (EFFN) module, which enhances local feature representation capabilities through deep separable convolutions and residual connections; 3) dual-path feature fusion module (D-FFM), which achieves complementary fusion of multilevel encoder features and decoder features through parallel spatial and passage paths, thereby optimizing boundary information. We evaluated the proposed model on two public water body datasets. Experimental results demonstrate that MSCTFormer achieves performance improvements of 0.94%–3.30% in IoU and 1.06%–2.91% in <inline-formula> <tex-math>$F1$ </tex-math></inline-formula> score, providing an effective solution for water body segmentation in high-resolution remote sensing imagery.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"1503005-1503005"},"PeriodicalIF":4.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148488417","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}

{"title":"Attention-Guided Hybrid Learning for Tree Species Classification in Mining Restoration Areas Using UAV Hyperspectral and LiDAR Data","authors":"Guo Wang;Cheng Wang;Yuncheng Deng;Jinliang Wang","doi":"10.1109/LGRS.2026.3692141","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3692141","url":null,"abstract":"Accurate tree species classification in open-pit mining regions is crucial for assessing ecological restoration effectiveness. However, it remains challenging due to extreme surface reflectance, dust contamination, and spectral whitening effects prevalent. To address these challenges, this letter proposes an attention-guided misclassification-aware hybrid learning (AG-MAHL) framework for tree species classification. Unlike existing fusion methods that simply concatenate features or ignore hard samples, our framework introduces: 1) an attention-guided fusion strategy to align spectral–structural features under high-reflectance conditions and 2) a misclassification-aware learning mechanism that dynamically reweights difficult samples. Experiments conducted at the Haikou phosphogypsum tailings reservoir in Kunming, China, demonstrate that the proposed AG-MAHL framework achieves an overall accuracy (OA) of 0.8571, a weighted <inline-formula> <tex-math>$F1$ </tex-math></inline-formula>-score of 0.8423, and a balanced accuracy of 0.8402. This performance surpasses that of conventional machine learning methods and representative deep learning baselines. The results indicate that the proposed attention-guided hyperspectral–LiDAR fusion, coupled with misclassification-aware learning, offers an effective solution for tree species classification in spectrally challenging open-pit mining restoration environments.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"2503205-2503205"},"PeriodicalIF":4.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148488668","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}