IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society最新文献_第3页

A Wildfire Monitoring Method Based on Daily Nighttime Light Data: A Case Study of the Los Angeles Wildfire 基于夜间光照数据的野火监测方法——以洛杉矶野火为例

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-02-27 DOI: 10.1109/LGRS.2026.3669070

Yuan Yuan;Congxiao Wang;Wei Xu;Lefeng Zhang;Jianbin Zhu;Yue Tu;Bailang Yu

{"title":"A Wildfire Monitoring Method Based on Daily Nighttime Light Data: A Case Study of the Los Angeles Wildfire","authors":"Yuan Yuan;Congxiao Wang;Wei Xu;Lefeng Zhang;Jianbin Zhu;Yue Tu;Bailang Yu","doi":"10.1109/LGRS.2026.3669070","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3669070","url":null,"abstract":"Wildfires are among the most destructive natural disasters, inflicting severe and irreversible damage to ecosystems and human societies. However, most existing research lacked sufficient spatial coverage and spatiotemporal continuity to provide near-real-time assessments of wildfire impacts, including their extent and severity. Leveraging globally accessible, temporally continuous nighttime light (NTL) data from National Aeronautics and Space Administration (NASA)’s Black Marble VNP46A1 product, this study developed a threshold-based method to delineate wildfire-affected areas and quantify impact severity from daily observations, providing a rapid and scalable solution for near-real-time wildfire assessment across diverse regions and events. The method was applied to the Los Angeles (LAs) wildfire of January 2025, which resulted in infrastructure losses exceeding U.S. <inline-formula> <tex-math>${$}$ </tex-math></inline-formula>50 billion. The results showed that the proposed approach effectively captured wildfire-affected regions—particularly in the Palisades, Eaton, and Hurst areas—covering approximately 307.75 km2. The sharp increase in NTL intensity from ~30 to 4750 nW/cm2/sr revealed the extreme burning severity of the event. Furthermore, approximately 30 252 buildings and 281 240 individuals were estimated to be directly affected, with impacts classified into five severity levels to characterize variations in damage intensity across the affected areas. Overall, the proposed method demonstrated strong potential for large-scale wildfire monitoring and rapid impact assessment, offering near–real-time applicability and globally accessible data support.","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":"147557593","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}

引用次数: 0

Deep Learning Inversion of Gravity Data With Multiple Physical Constraints 多重物理约束下重力数据的深度学习反演

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-03-11 DOI: 10.1109/LGRS.2026.3673191

Haoyuan He;Tonglin Li;Rongzhe Zhang;Yong Li;Guanwen Gu;Zhihe Xu

{"title":"Deep Learning Inversion of Gravity Data With Multiple Physical Constraints","authors":"Haoyuan He;Tonglin Li;Rongzhe Zhang;Yong Li;Guanwen Gu;Zhihe Xu","doi":"10.1109/LGRS.2026.3673191","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3673191","url":null,"abstract":"Gravity inversion is an important method for revealing underground density distribution. Traditional deterministic inversions face challenges such as easily becoming trapped in local minimum solutions and sensitivity to initial models, while conventional data-driven deep learning inversions (DD-DLIs) also exhibit limitations including dependence on datasets and lack of physical constraints. In this letter, we propose a multiphysical constrained deep learning inversion (MPC-DLI) method for gravity data based on the U-Net. Within an iterative descent inversion framework, the method employs a trained neural network to predict model updates for each iteration. Constraints about forward modeling laws and model smoothness are incorporated into the loss function during the training phase, serving respectively to promote the fitting of observation data and enhance stability. In the prediction phase, a focusing constraint is built into the objective function to yield model results with clear boundaries and pronounced density variations. Synthetic and field data examples are used to demonstrate the effectiveness of the new method.","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":"147736974","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}

引用次数: 0

Efficient L1–TV1 Inversion of the Vertical Gravity Gradient With Application to the Vinton Salt Dome 垂直重力梯度的高效L1-TV1反演及其在Vinton盐丘上的应用

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-04-27 DOI: 10.1109/LGRS.2026.3687542

Jiaxiang Peng;Bo Chen;Shengbo Liu;Jinsong Du

{"title":"Efficient L1–TV1 Inversion of the Vertical Gravity Gradient With Application to the Vinton Salt Dome","authors":"Jiaxiang Peng;Bo Chen;Shengbo Liu;Jinsong Du","doi":"10.1109/LGRS.2026.3687542","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3687542","url":null,"abstract":"Achieving high depth resolution while accurately recovering sharp structural boundaries presents a significant challenge in gravity gradient inversion. This study addresses this issue by employing an L1–TV1 regularization framework for the interpretation of vertical gravity-gradient data, which combines the L1 norm of the model with the L1 norm of the model gradients (TV1). To balance model sparsity and structural smoothness and to improve depth resolution, we introduce four weighting factors. The resulting objective function is solved efficiently using the alternating direction method of multipliers (ADMM). Model experiments demonstrate that this method successfully delineates causative bodies while preserving sharp structural boundaries. The method is then applied to airborne gravity-gradient data from the Vinton Dome in southern Louisiana, USA. The inverted density model reveals a dense rock cap at depths of 300–700 m, exhibiting a density contrast of approximately 0.7 g/cm3. Beneath this cap, a large salt dome is imaged, centered at depths of 2300–4900 m, with a density contrast of about −0.2 g/cm3. The recovered density structure provides a more realistic representation of the deep salt body, characterized by geologically plausible geometry and clear boundaries that closely align with known geological constraints. These results highlight the practical utility of the proposed method for detailed subsurface imaging of complex salt structures.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"7503405-7503405"},"PeriodicalIF":4.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147828772","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}

引用次数: 0

Seismic Impedance Inversion With Adaptive Group Sparse Regularization 基于自适应群稀疏正则化的地震阻抗反演

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-04-02 DOI: 10.1109/LGRS.2026.3680028

Qing Xiang;Ronghuo Dai

{"title":"Seismic Impedance Inversion With Adaptive Group Sparse Regularization","authors":"Qing Xiang;Ronghuo Dai","doi":"10.1109/LGRS.2026.3680028","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3680028","url":null,"abstract":"Seismic impedance inversion, as a core technology in geophysical exploration, has evolved into a pivotal method for the quantitative prediction of subsurface rock elastic properties within exploration geophysics. Its accuracy and resolution directly impact the precision of hydrocarbon reservoir prediction and characterization. Seismic inversion often faces an ill-posedness challenge. Sparse regularization, such as L1-norm, is frequently used to deal with this issue. But the noise interference is still mightiness when dealing with complex geological structures. It is because of the inherent feature of sparse regularization, which tends to suppress minor geological features. To compensate for the limitations of conventional sparse regularization, this letter proposes another regularization, i.e., adaptive group sparse (AGS) regularization. It is the generalization of the existing group sparse (GS) regularization and can adaptively find the optimal results without predefined group partition. Based on the theory of GS regularization, it can overcome the limitations of sparse regularization. We use adaptive group sparsity to regularize the seismic impedance inversion. To effectively solve the objective function, the variable splitting alternating iteration has been used. Through a synthetic data test, the feasibility of the proposed method has been verified. And then, through actual data tests, it has been verified that the advantages in resolution enhancement in complex geological structures are compared to the inversion method only with sparse regularization.","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":"147737029","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}

引用次数: 0

Multitrace Bayesian Orthogonal Matching Pursuit Incorporating Multiple Prior Knowledge for Seismic Resolution Enhancement 基于多先验知识的多道贝叶斯正交匹配追踪提高地震分辨率

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-04-27 DOI: 10.1109/LGRS.2026.3687721

Kun Li;Wei Ge;Xingyao Yin;Danping Cao;Shuangshuang Zhou

{"title":"Multitrace Bayesian Orthogonal Matching Pursuit Incorporating Multiple Prior Knowledge for Seismic Resolution Enhancement","authors":"Kun Li;Wei Ge;Xingyao Yin;Danping Cao;Shuangshuang Zhou","doi":"10.1109/LGRS.2026.3687721","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3687721","url":null,"abstract":"Broadening the seismic frequency band and enhancing resolution play important roles in reservoir quantitative characterization. Seismic sparse representation (SR) provides a vital means of enhancing seismic resolution. Within the framework of SR and Bayesian inference, we propose a multitrace Bayesian orthogonal matching pursuit (OMP) method incorporating multiple prior knowledge for seismic resolution enhancement. Based on the maximum a posteriori probability solution, we first construct an objective function for SR that couples the low-frequency impedance prior, seismic amplitude-envelope prior, and support-position prior. Incorporating the matching pursuit (MP) framework, we propose a multitrace Bayesian OMP algorithm for 3-D seismic data. Based on the SR results, a broadband Ricker wavelet frequency and phase optimization method using particle swarm optimization (PSO) is developed to achieve high-resolution seismic processing constrained by well-logging data. The numerical results show that the approach effectively improves the matching accuracy of the positions and amplitudes of seismic reflection coefficients. Applications to field 3-D seismic data volume and stratigraphic slice analysis demonstrate the practicality and validity of the proposed approach.","PeriodicalId":91017,"journal":{"name":"IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society","volume":"23 ","pages":"7503305-7503305"},"PeriodicalIF":4.4,"publicationDate":"2026-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147828936","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}

引用次数: 0

A Method for Maritime Weak Moving Target Localization Leveraging GNSS-Reflected Baseband Signals and DP-TBD 利用gnss反射基带信号和DP-TBD的海上弱运动目标定位方法

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-03-09 DOI: 10.1109/LGRS.2026.3672312

Zhikun Zhang;Bofeng Guo;Yang Nan;Yulin Han;Xiang Wu

{"title":"A Method for Maritime Weak Moving Target Localization Leveraging GNSS-Reflected Baseband Signals and DP-TBD","authors":"Zhikun Zhang;Bofeng Guo;Yang Nan;Yulin Han;Xiang Wu","doi":"10.1109/LGRS.2026.3672312","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3672312","url":null,"abstract":"To address the limited power budget of the global navigation satellite system (GNSS)-based passive radar and to enable high-frequency localization of maritime moving targets, this letter proposes a weak maritime target localization method based on GNSS baseband signals and a dynamic programming track-before-detect (DP-TBD) framework. First, GNSS baseband signal processing is performed to construct the bistatic range–slow-time (BRST) map of the target over the observation interval. Subsequently, the DP-TBD algorithm is applied to extract the target motion trajectory from the BRST map, yielding a per-second bistatic range history. Finally, under multisatellite illumination, the target position is estimated at a 1-Hz rate through bistatic range intersection, enabling high-frequency localization of maritime moving targets. Experimental validations were conducted in two scenarios involving vessels with different lengths and ranges. Comparisons with automatic identification system (AIS) reference data demonstrate that the proposed method achieves detection ranges of approximately 160 and 1100 m, with corresponding root-mean-square errors (RMSEs) of 13.15 and 41.52 m.","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":"147557464","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}

引用次数: 0

An Improved Dual-Antenna GNSS-IR Method for BDS-Based High-Elevation Soil Moisture Retrieval: Initial Results and Evaluation 一种改进的双天线GNSS-IR方法用于基于bds的高海拔土壤水分检索：初步结果与评价

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-03-13 DOI: 10.1109/LGRS.2026.3673971

Nazi Wang;Yunqiao He;Zhenlong Fang;Lili Jing;Fan Gao;Yang Liu;Xiao Zhang;Tianhe Xu

{"title":"An Improved Dual-Antenna GNSS-IR Method for BDS-Based High-Elevation Soil Moisture Retrieval: Initial Results and Evaluation","authors":"Nazi Wang;Yunqiao He;Zhenlong Fang;Lili Jing;Fan Gao;Yang Liu;Xiao Zhang;Tianhe Xu","doi":"10.1109/LGRS.2026.3673971","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3673971","url":null,"abstract":"GNSS Interferometric Reflectometry (GNSS-IR) has been proven effective for soil moisture monitoring by leveraging the interference effects between direct and reflected signals at low elevation angles. However, at high elevation angles, the polarization of reflected signals shifts to left-handed circular polarization (LHCP), which cannot be captured by right-handed circular polarization (RHCP) antennas—thus rendering conventional GNSS-IR technology ineffective. To develop more robust solutions for harnessing high-elevation GNSS data, this study proposes an improved GNSS-IR configuration that integrates two geodetic GNSS receivers, one upward-looking RHCP antenna, one downward-looking LHCP antenna, and two power couplers for soil moisture retrieval. This optimized setup enables the acquisition and utilization of high-elevation GNSS signals for GNSS-IR-based soil moisture estimation. To validate the proposed method, an observational campaign was conducted at a farmland site. Analyses of the amplitude and phase measurements derived from BDS reflected signals indicate that soil moisture solutions retrieved from high-elevation signals achieve accuracy comparable to those obtained from low-elevation signals. Specifically, the maximum correlation coefficient between retrievals and in situ data reaches 0.95, and the mean root mean square error (RMSE) across all frequencies is no greater than 0.02 m3/m3. Overall, this method maximizes the utilization efficiency of received GNSS signals and provides a more comprehensive detection range compared with conventional GNSS-IR method.","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":"147737004","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}

引用次数: 0

Wave-Equation Diffraction Imaging Using Recursive Beamforming and Envelope Multiplication 采用递推波束形成和包络乘法的波方程衍射成像

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2026-01-01 Epub Date: 2026-04-10 DOI: 10.1109/LGRS.2026.3682671

Lu Liu;Young Seo Kim;Hussain Salim

{"title":"Wave-Equation Diffraction Imaging Using Recursive Beamforming and Envelope Multiplication","authors":"Lu Liu;Young Seo Kim;Hussain Salim","doi":"10.1109/LGRS.2026.3682671","DOIUrl":"https://doi.org/10.1109/LGRS.2026.3682671","url":null,"abstract":"Diffraction imaging provides a direct means of detecting local heterogeneities such as faults, fracture zones, and erosional surfaces, which are critical for seismic interpretation and unconventional reservoir characterization. In this study, we present a wave-equation diffraction imaging method that integrates a recursive beamforming strategy with an envelope-multiplication imaging condition. The proposed approach first uses a fast recursive beamforming algorithm to construct directional beams from the extrapolated receiver-side wavefields. It then employs a recursive Hilbert autocorrelator to generate source-side beams. Combining these source- and receiver-side beams yields dip-dependent images at selected positive and negative dip angles. Subsequently, the envelope-multiplication imaging condition is used to suppress specular reflections, correct phase distortions, and enhance diffraction responses. Numerical experiments on both synthetic and field datasets demonstrate that the method produces accurate and high-resolution diffraction images, while avoiding the explicit construction and I/O of full dip-angle gathers and thereby reducing storage and data-I/O requirements.","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":"147737138","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}

引用次数: 0

Incorporating Stratal Dip to Constrain the Integration Range of Marchenko Imaging 结合地层倾角约束马尔琴科成像积分范围

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2025-12-25 DOI: 10.1109/LGRS.2025.3648652

Xiaochun Chen;Tianjing Shen;Dong Zhang;Yukai Wo;Haoyang Ou;Xuri Huang

{"title":"Incorporating Stratal Dip to Constrain the Integration Range of Marchenko Imaging","authors":"Xiaochun Chen;Tianjing Shen;Dong Zhang;Yukai Wo;Haoyang Ou;Xuri Huang","doi":"10.1109/LGRS.2025.3648652","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3648652","url":null,"abstract":"Marchenko imaging provides a powerful framework for reconstructing amplitude-preserved subsurface images while suppressing migration artifacts caused by internal multiples. This capability is achieved by applying an energy-compensated cross correlation imaging condition to the up- and down-going Green’s functions retrieved through the Marchenko scheme. However, the performance of this approach strongly depends on the choice of the integration range: an excessively large range lowers resolution in shallow areas, whereas an overly small range can degrade the imaging quality of steeply dipping structures. To address this limitation, we propose an adaptive strategy that improves image quality by optimizing the integration range for each imaging point. The integration range is characterized by two parameters—the location of the integration center and the integration radius. Specifically, the radius is determined from the travel time difference between the up- and down-going Green’s functions together with the depth of the imaging point, while the center location is constrained by stratal dip. The resulting optimal integration range, defined by these two parameters, is then applied to both the retrieval of Marchenko Green’s functions and the subsequent imaging. The proposed method was first tested on synthetic data, where it was shown to outperform fixed integration ranges (either too large or too small) by simultaneously enhancing shallow resolution and preserving the fidelity of steeply dipping structures. It was then further applied to field data from western China, which confirmed the feasibility of the approach.","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":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145929486","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}

引用次数: 0

Weakly Supervised Semantic Segmentation of Remote Sensing Scenes With Cross-Image Class Token Constraints 基于跨图像类令牌约束的遥感场景弱监督语义分割

IF 4.4

IEEE geoscience and remote sensing letters : a publication of the IEEE Geoscience and Remote Sensing Society Pub Date : 2025-12-18 DOI: 10.1109/LGRS.2025.3645679

Pengcheng Guo;Zhen Wang;Junhuan Peng;Yuebin Wang;Guodong Liu;Yasong Mi;Dengxiang Wu;Jie Huang

{"title":"Weakly Supervised Semantic Segmentation of Remote Sensing Scenes With Cross-Image Class Token Constraints","authors":"Pengcheng Guo;Zhen Wang;Junhuan Peng;Yuebin Wang;Guodong Liu;Yasong Mi;Dengxiang Wu;Jie Huang","doi":"10.1109/LGRS.2025.3645679","DOIUrl":"https://doi.org/10.1109/LGRS.2025.3645679","url":null,"abstract":"Weakly supervised semantic segmentation (WSSS) based on image-level labels significantly reduces the labeling burden. However, current mainstream approaches optimize solely using single-image information, neglecting the rich semantic correlation among images and struggling to dynamically suppress interfering information. When confronted with complex backgrounds and multicategory remote sensing (RS) images, intraclass consistency and interclass discrimination pose significant challenges. To address these challenges, this letter proposes the cross-image class token constraints network (CICTC-Net). CICTC-Net establishes semantic correlations across multicategory RS images and implements two modules for targeted optimization. Specifically, the cross-image token enhancement (CITE) module constructs intraclass token relationship graphs and applies cross-image consistency constraints to enhance semantic consistency among objects of the same category. The class-patch interaction refinement (CPIR) module constructs a directed graph of class-patch relationships and employs a neighborhood selection mechanism to refine class tokens, thereby enhancing interclass discriminability. Experiments on two RS datasets demonstrate that this approach significantly outperforms existing state-of-the-art solutions.","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":"2025-12-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145830879","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}

引用次数: 0