ISPRS Journal of Photogrammetry and Remote Sensing最新文献

{"title":"Saliency supervised masked autoencoder pretrained salient location mining network for remote sensing image salient object detection","authors":"Yuxiang Fu ,&nbsp;Wei Fang ,&nbsp;Victor S. Sheng","doi":"10.1016/j.isprsjprs.2025.03.025","DOIUrl":"10.1016/j.isprsjprs.2025.03.025","url":null,"abstract":"<div><div>Remote sensing image salient object detection (RSI-SOD), as an emerging topic in computer vision, has significant applications across various sectors, such as urban planning, environmental monitoring and disaster management, etc. In recent years, RSI-SOD has seen significant advancements, largely due to advanced representation learning methods and better architectures, such as convolutional neural networks and vision transformers. While current methods predominantly rely on supervised learning, there is potential for enhancement through self-supervised learning approaches, like masked autoencoder. However, we observed that the conventional use of masked autoencoder for pretraining encoders through masked image reconstruction yields subpar results in the context of RSI-SOD. To this end, we propose a novel approach: saliency supervised masked autoencoder (SSMAE) and a corresponding salient location mining network (SLMNet), which is pretrained by SSMAE for the task of RSI-SOD. SSMAE first uses masked autoencoder to reconstruct the masked image, and then employs SLMNet to predict saliency map from the reconstructed image, where saliency supervision is adopted to enable SLMNet to learn robust saliency prior knowledge. SLMNet has three major components: encoder, salient location mining module (SLMM) and the decoder. Specifically, SLMM employs residual multi-level fusion structure to mine the locations of salient objects from multi-scale features produced by the encoder. Later, the decoder fuses the multi-level features from SLMM and encoder to generate the prediction results. Comprehensive experiments on three public datasets demonstrate that our proposed method surpasses the state-of-the-art methods. Code is available at: <span><span>https://github.com/Voruarn/SLMNet</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 222-234"},"PeriodicalIF":10.6,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820639","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Evaluating saliency scores in point clouds of natural environments by learning surface anomalies","authors":"Reuma Arav ,&nbsp;Dennis Wittich ,&nbsp;Franz Rottensteiner","doi":"10.1016/j.isprsjprs.2025.03.022","DOIUrl":"10.1016/j.isprsjprs.2025.03.022","url":null,"abstract":"<div><div>In recent years, three-dimensional point clouds are used increasingly to document natural environments. Each dataset contains a diverse set of objects, at varying shapes and sizes, distributed throughout the data and intricately intertwined with the topography. Therefore, regions of interest are difficult to find and consequent analyses become a challenge. Inspired from visual perception principles, we propose to differentiate objects of interest from the cluttered environment by evaluating how much they stand out from their surroundings, i.e., their geometric salience. Previous saliency detection approaches suggested mostly handcrafted attributes for the task. However, such methods fail when the data are too noisy or have high levels of texture. Here we propose a learning-based mechanism that accommodates noise and textured surfaces. We assume that within the natural environment any change from the prevalent surface would suggest a salient object. Thus, we first learn the underlying surface and then search for anomalies within it. Initially, a deep neural network is trained to reconstruct the surface. Regions where the reconstructed part deviates significantly from the original point cloud yield a substantial reconstruction error, signifying an anomaly, i.e., saliency. We demonstrate the effectiveness of the proposed approach by searching for salient features in various natural scenarios, which were acquired by different acquisition platforms. We show the strong correlation between the reconstruction error and salient objects. To promote benchmarking and reproducibility, the code used in this work can be found on <span><span>https://github.com/rarav/salient_anomaly/releases/tag/v1.0.0</span><svg><path></path></svg></span> while the datasets are published on doi: <span><span>10.48436/mps0m-c9n43</span><svg><path></path></svg></span> and <span><span>10.48436/fh0am-at738</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 235-250"},"PeriodicalIF":10.6,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820636","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Super-resolution supporting individual tree detection and canopy stratification using half-meter aerial data","authors":"Zhu Mao,&nbsp;Omid Abdi,&nbsp;Jori Uusitalo,&nbsp;Ville Laamanen,&nbsp;Veli-Pekka Kivinen","doi":"10.1016/j.isprsjprs.2025.04.005","DOIUrl":"10.1016/j.isprsjprs.2025.04.005","url":null,"abstract":"<div><div>Individual Tree Detection (ITD) can automatically recognize single trees and generate large-scale individual tree maps, providing essential insights for tree-by-tree forest management. However, ITD using remote sensing data becomes increasingly challenging as data quality and spatial resolution decrease. This paper proposes a Super-Resolution (SR)-based ITD method to predict individual tree locations, delineate crowns, and classify species from half-meter multi-modal aerial data. Based on the predicted ITD results, this study designs a tree-level forest canopy stratification method to further understand the forest structure. Specifically, we first fuse multi-modal data to represent tree features, including spectral RGB and NIR orthophoto images with a spatial resolution of 50 cm, and Canopy Height Model (CHM). The CHM is derived from the laser scanning data with a point density of approximately <span><math><mrow><mn>5</mn><msup><mrow><mtext>points/m</mtext></mrow><mrow><mn>2</mn></mrow></msup></mrow></math></span>. Second, we integrate an SR module into the DCNN-based ITD model and employ multi-scale feature fusion (PANFPN) to address the challenges posed by limited data resolution. The SR module improves the spatial resolution of the data, while PANFPN integrates high- and low-level features to retain spatial details. Consequently, these components help mitigate the loss of tree features during the downsampling or pooling operation in DCNN models, preserving finer details. Third, we estimate tree height using Local Maxima (LM) filtering and derive crown size from ITD results to stratify the forest into three canopy classes: dominant (codominant), intermediate, and suppressed. The study sites are located in the Parkano region of Southwest Finland and encompass a diversity of tree species and forest stand types. Experiments demonstrate that the proposed SR module and PANFPN improve ITD performance, achieving an <span><math><mrow><mi>m</mi><mi>A</mi><mi>P</mi></mrow></math></span> of 69.2% for the predicted bounding boxes and an <span><math><mrow><mi>m</mi><mi>A</mi><mi>P</mi></mrow></math></span> of 64.3% for the segmented boundary masks. Our method is applicable to large-scale ITD and tree-level canopy stratification using half-meter multi-modal aerial data. The code is available at <span><span>https://github.com/zmaomia/SR-Supporting-ITD</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 251-271"},"PeriodicalIF":10.6,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820637","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"RSGPT: A remote sensing vision language model and benchmark","authors":"Yuan Hu ,&nbsp;Jianlong Yuan ,&nbsp;Congcong Wen ,&nbsp;Xiaonan Lu ,&nbsp;Yu Liu ,&nbsp;Xiang Li","doi":"10.1016/j.isprsjprs.2025.03.028","DOIUrl":"10.1016/j.isprsjprs.2025.03.028","url":null,"abstract":"<div><div>The emergence of large-scale Large Language Models (LLMs), with GPT-4 as a prominent example, has significantly propelled the rapid advancement of Artificial General Intelligence (AGI) and sparked the revolution of Artificial Intelligence 2.0. In the realm of remote sensing, there is a growing interest in developing large vision language models (VLMs) specifically tailored for data analysis in this domain. However, current research predominantly revolves around visual recognition tasks, lacking comprehensive, high-quality image–text datasets that are aligned and suitable for training large VLMs, which poses significant challenges to effectively training such models for remote sensing applications. In computer vision, recent research has demonstrated that fine-tuning large vision language models on small-scale, high-quality datasets can yield impressive performance in visual and language understanding. These results are comparable to state-of-the-art VLMs trained from scratch on massive amounts of data, such as GPT-4. Inspired by this captivating idea, in this work, we build a high-quality Remote Sensing Image Captioning dataset (<strong>RSICap</strong>) that facilitates the development of large VLMs in the remote sensing field. Unlike previous remote sensing datasets that either employ model-generated captions or short descriptions, RSICap comprises 2,585 human-annotated captions with rich and high-quality information. This dataset offers detailed descriptions for each image, encompassing scene descriptions (e.g., residential area, airport, or farmland) as well as object information (e.g., color, shape, quantity, absolute position, etc.). To facilitate the evaluation of VLMs in the field of remote sensing, we also provide a benchmark evaluation dataset called <strong>RSIEval</strong>. This dataset consists of human-annotated captions and visual question–answer pairs, allowing for a comprehensive assessment of VLMs in the context of remote sensing. We are actively engaged in expanding the scale of these two datasets to cover a broader spectrum of remote sensing image understanding tasks, further enhancing their utility and applicability. Our dataset and codes will be released at <span><span>https://github.com/Lavender105/RSGPT</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 272-286"},"PeriodicalIF":10.6,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"TKSF-KAN: Transformer-enhanced oat yield modeling and transferability across major oat-producing regions in China using UAV multisource data","authors":"Pengpeng Zhang ,&nbsp;Bing Lu ,&nbsp;Jiali Shang ,&nbsp;Changwei Tan ,&nbsp;Qihan Xu ,&nbsp;Lei Shi ,&nbsp;Shujian Jin ,&nbsp;Xingyu Wang ,&nbsp;Yunfei Jiang ,&nbsp;Yadong Yang ,&nbsp;Huadong Zang ,&nbsp;Junyong Ge ,&nbsp;Zhaohai Zeng","doi":"10.1016/j.isprsjprs.2025.04.004","DOIUrl":"10.1016/j.isprsjprs.2025.04.004","url":null,"abstract":"<div><div>Accurate and efficient estimation of crop yield is crucial for enhancing crop variety testing, optimizing cultivation practices, and supporting effective crop management to ensure sustainable production. However, remote sensing-based yield models often face limitations due to geographical variability and diverse cultivation techniques, affecting both their model accuracy and transferability. This study utilized multiple features, including vegetation indices (VI), color indices (CI), texture features (TF), structural indices (SI), and canopy thermal information (TIR), obtained from RGB, multispectral, and thermal infrared sensors of unmanned aerial vehicles (UAV), to create six scenarios for oat yield estimation across major oat-producing regions in China. We developed a novel deep learning-based architecture, TKSF-KAN, which combines Transformer and Kolmogorov–Arnold Network (KAN) to fuse multimodal data across key growth stages, and benchmarked its performance against stacking ensemble learning (SEL) and standalone Transformer model. While SEL demonstrated the highest accuracy in single-modal scenarios, TKSF-KAN outperformed SEL in multimodal settings (R² = 0.76–0.81). Particularly, TKSF-KAN, with integrated VI, CI, TF, SI, and TIR inputs improved R² by 53.77 % compared with single-modal data sources. By combining Adaptive Batch Normalization (AdaBN) with fine-tuning strategies, transfer performance improved as the proportion of fine-tuned data increased, reaching a peak R² of 0.83 at one study site. In contrast, transferability was more influenced by cultivation practices at another site, with a maximum R² of 0.78. This study presents an innovative framework that seamlessly integrates agricultural practices with remote sensing and transfer learning methodologies, offering a more robust and scalable solution for yield prediction and advancing precision of agricultural management.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 166-186"},"PeriodicalIF":10.6,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820633","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"DeepCropClustering: A deep unsupervised clustering approach by adopting nearest and farthest neighbors for crop mapping","authors":"Hengbin Wang ,&nbsp;Yuanyuan Zhao ,&nbsp;Shaoming Li ,&nbsp;Zhe Liu ,&nbsp;Xiaodong Zhang","doi":"10.1016/j.isprsjprs.2025.04.007","DOIUrl":"10.1016/j.isprsjprs.2025.04.007","url":null,"abstract":"<div><div>Existing crop type maps usually rely on extensive ground truth, limiting the potential applicability in regions without any crop labels. Unsupervised clustering offers a promising approach for crop mapping in regions lacking labeled crop samples. However, due to the high-dimensional complexity and pronounced temporal dependencies of crop time series, existing unsupervised clustering methods are inadequate for effectively capturing deep semantic representations. In this study, we developed a novel deep unsupervised clustering approach, named DeepCropClustering (DCC), for crop mapping without any crop label information. This approach includes a generating cluster feature space component to acquire the semantically meaning features via contractive learning and a learnable deep clustering component for unsupervised clustering using the nearest-farthest neighbor information derived from feature spaces sorting. DCC achieved an average Overall Accuracy (OA) of 70.9% across the four sites, surpassing the average OA of K-means and GMM by 7.0% and 8.6% respectively. Evaluation results of the cluster feature space indicated that the generated feature space contained reliable far-neighbor and near-neighbor samples, providing highly discriminative feature representations. By monitoring the clustering confidence during each training iteration, we found that clustering reliability increased progressively throughout the learning process, gradually converging to appropriate clusters. DCC does not require any crop labels during the clustering process, offering a new option for crop mapping in regions without crop labels and has the potential to become a new method for large-scale crop mapping.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 187-201"},"PeriodicalIF":10.6,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820634","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"SfM on-the-fly: A robust near real-time SfM for spatiotemporally disordered high-resolution imagery from multiple agents","authors":"Zongqian Zhan ,&nbsp;Yifei Yu ,&nbsp;Rui Xia ,&nbsp;Wentian Gan ,&nbsp;Hong Xie ,&nbsp;Giulio Perda ,&nbsp;Luca Morelli ,&nbsp;Fabio Remondino ,&nbsp;Xin Wang","doi":"10.1016/j.isprsjprs.2025.04.002","DOIUrl":"10.1016/j.isprsjprs.2025.04.002","url":null,"abstract":"<div><div>In the last twenty years, Structure from Motion (SfM) has been a constant research hotspot in the fields of photogrammetry, computer vision, robotics etc., whereas real-time performance has only recently emerged as a topic of growing interest. This work builds upon the original on-the-fly SfM (Zhan et al., 2024) and presents an updated version (v2) with three new advancements to get better SfM reconstruction results during image capturing: (i) near real-time image matching is further boosted by employing the Hierarchical Navigable Small World (HNSW) graphs, and more true positive overlapping image candidates can be faster identified; (ii) a self-adaptive weighting strategy is proposed for robust hierarchical local bundle adjustment to improve the SfM results; (iii) multiple agents are included for supporting collaborative SfM and seamlessly merge multiple 3D reconstructions into a complete 3D scene in presence of commonly registered images. Various comprehensive experiments demonstrate that the proposed SfM method (named on-the-fly SfMv2) can generate more complete and robust 3D reconstructions in a time-efficient way. Code is available at <span><span>http://yifeiyu225.github.io/on-the-flySfMv2.github.io/</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 202-221"},"PeriodicalIF":10.6,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143820635","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GACraterNet: A collaborative geometry-attribute domain network for enhanced detection of Martian impact craters","authors":"Fan Hu ,&nbsp;Dong Chen ,&nbsp;Jiaming Na ,&nbsp;Zhen Cao ,&nbsp;Zhenxin Zhang ,&nbsp;Liqiang Zhang ,&nbsp;Zhizhong Kang","doi":"10.1016/j.isprsjprs.2025.03.023","DOIUrl":"10.1016/j.isprsjprs.2025.03.023","url":null,"abstract":"<div><div>Accurately understanding the local and global distribution, categories and morphological parameters of impact craters on Mars, including variations across the southern highlands, northern lowlands, equatorial region and polar zones, is crucial for revealing the geological history and environmental changes. To this end, this paper proposes a multi-task deep learning framework, GACraterNet(Geometric and Attribute Domain-based Crater Detection Network), which addresses both impact crater detection and attribute extraction while facilitating mutual enhancement between these tasks. GACraterNet comprises two main components: the geometric-domain module and the attribute-domain module. The geometric-domain module features a detection network named M<span><math><msup><mrow><mtext>S</mtext></mrow><mrow><mn>2</mn></mrow></msup></math></span>CraterNet, which encodes crater features from two data sources: digital elevation models (DEMs) and digital orthophoto maps (DOMs) using a dual backbone network. This module incorporates a multi-source, multi scale feature fusion module (M<span><math><msup><mrow><mtext>S</mtext></mrow><mrow><mn>2</mn></mrow></msup></math></span>-FFM) to integrate the features, enabling the detection of craters larger than 1 km in diameter. The attribute-domain module is designed to perform three tasks: segmentation, classification and extraction of morphological parameters. First, Segment Anything Model (SAM) is utilized for unsupervised semantic segmentation on terrain maps within the bounding boxes predicted by M<span><math><msup><mrow><mtext>S</mtext></mrow><mrow><mn>2</mn></mrow></msup></math></span>CraterNet. This step enhances the extraction of crater foregrounds and optimizes the positioning and sizing of the bounding boxes. The resultant crater foregrounds are then input to the Swin Transformer network, which categorizes craters into four types: bowl-shaped, flat floor, central peak and central pit. Finally, radial profiles of each crater type are analyzed to extract their 2.5D morphological parameters, followed by a comparative analysis of the morphological differences among the various categories. Validation results on the HRSC Mars remote sensing dataset indicate that M<span><math><msup><mrow><mtext>S</mtext></mrow><mrow><mn>2</mn></mrow></msup></math></span>CraterNet achieved a mean Average Precision (mAP50) of 79.4%, with precision and recall rates of 78.4% and 73.3%, respectively. These results significantly outperform the detection results obtained from a single data source. Furthermore, Swin Transformer attained an overall classification accuracy of 83.9% for the craters, with specific classification F1-score for bowl-shaped, central peak, central pit and flat floor craters reaching 91.5%, 83.4%, 35.3% and 71.9%, respectively. The source code of our GACraterNet is available at <span><span>https://github.com/shincccc/GACraterNet</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 133-154"},"PeriodicalIF":10.6,"publicationDate":"2025-04-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143807841","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GAT-LSTM: A feature point management network with graph attention for feature-based visual SLAM in dynamic environments","authors":"Xuan Wang ,&nbsp;Yuan Zhuang ,&nbsp;Xiaoxiang Cao ,&nbsp;Jianzhu Huai ,&nbsp;Zhenghua Zhang ,&nbsp;Zhenqi Zheng ,&nbsp;Naser El-Sheimy","doi":"10.1016/j.isprsjprs.2025.03.011","DOIUrl":"10.1016/j.isprsjprs.2025.03.011","url":null,"abstract":"<div><div>Visual simultaneous localization and mapping (vSLAM) is crucial in various applications, ranging from robotics to augmented reality. However, dynamic environments cause difficulty to vSLAM, which often relies on extracted feature points (FPs). Effectively managing FPs in dynamic environments poses a significant challenge. To address this challenge, we propose an innovative solution that leverages a graph attention network (GAT) integrated into a long- short-term memory (LSTM) network, enabling the system to prioritize attention on these stable FPs. The GAT component extracts spatial structural information from individual image FPs, which are graph nodes, thereby modeling the local relationship of each FP. Meanwhile, the LSTM module facilitates the local association’s consistent temporal feature analysis. Our approach effectively blends local relationship modeling with global consistency analysis, presenting the first application of GAT-LSTM to tackle the complexities introduced by dynamic and error-tracking FPs. Additionally, we introduce a backpropagating epipolar geometry solver to address this non-back propagatable optimization module in a deep neural network. Moreover, monocular vSLAM cannot directly measure distances and typically depends on reference objects or motion information. Depth estimation is complex and error-prone due to texture deficiency and motion blur. Thus, we present a dense depth estimation approach to mitigate the challenges associated with depth estimation by leveraging the selected stable FPs and a depth estimation network. We validated the GAT-LSTM network within a purely Visual Odometry (VO) framework and a Visual-Inertial Odometer (VIO) using the KITTI, VIODE, and in-house datasets. These experiments demonstrated that the exclusion of dynamic and error-tracking FPs using GAT-LSTM significantly enhances odometry accuracy and robustness. Compared to existing methods, the root-mean-square error of absolute pose error decreased by 4.52%–76.86% in VO and by 9.09%–96.94% in VIO. Our practice offers valuable insights and potential applications for more robust and accurate vSLAM and other related fields, highlighting the benefits of integrating GAT and LSTM networks.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 75-93"},"PeriodicalIF":10.6,"publicationDate":"2025-04-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143808150","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advancements in satellite-based methane point source monitoring: A systematic review","authors":"Fariba Mohammadimanesh ,&nbsp;Masoud Mahdianpari ,&nbsp;Ali Radman ,&nbsp;Daniel Varon ,&nbsp;Mohammadali Hemati ,&nbsp;Mohammad Marjani","doi":"10.1016/j.isprsjprs.2025.03.020","DOIUrl":"10.1016/j.isprsjprs.2025.03.020","url":null,"abstract":"<div><div>Methane (CH4) stands out as the second-largest contributor to the global warming since pre-industrial era. Anthropogenic methane emissions (e.g., oil and gas, waste management, and coal mining) are the major sources of methane release and in the meantime provide an excellent opportunity for emission reduction. Regarding this, observations from satellite remote sensing paly pivotal role in methane detection and quantification, further enhancing the temporal and spatial extent of methane research. A key feature of satellite data is the presence of Shortwave Infrared (SWIR) methane absorption bands, which are essential for identifying methane plumes from space. Detection, monitoring, and characterization are the main components of satellite-based methane studies. In general, quantifying methane emissions from regional and point sources presents two major challenges in the literature and the past few years have witnessed a tremendous success in terms of research, development, and operationalization of new techniques for methane studies, particularly, in the latter domain. As such, there is a need for a systematic analysis and review of existing literature to identify current trends, techniques, earth observation data for methane point source monitoring from space. Accordingly, this study systematically reviews 77 studies and highlights the critical roles of satellite data in detecting methane point source emissions. The literature identifies oil and gas sector as a dominant source of reported methane emission, particularly from countries like Turkmenistan, the United States, and Algeria. The review also categorizes instruments with methane detection capabilities into three main types, namely hyperspectral, multispectral, and SWIR spectrometers, each offering their unique advantages and addressing the limitations of other source of data. The main processing steps for methane point source emission monitoring from space identified in the literature are methane column retrieval, masking/detection, and source rate quantification. The literature reveals while conventional techniques are still widely used for methane detection and quantification, AI-based models are emerging as useful tools in different stages of methane research and significantly address the limitations of conventional techniques. The general characteristics of methane point source studies reveal a diverse array of applications across both atmospheric science and remote sensing fields. The rising number of publications in these areas, especially in high-impact journals, underscores the importance and relevance of methane monitoring. These studies bridge the gap between atmospheric observations and remote sensing technologies, contributing to a more integrated understanding of methane emissions on a global scale.</div></div>","PeriodicalId":50269,"journal":{"name":"ISPRS Journal of Photogrammetry and Remote Sensing","volume":"224 ","pages":"Pages 94-112"},"PeriodicalIF":10.6,"publicationDate":"2025-04-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143808095","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}