Table of Content

25 February 2026, Volume 0 Issue 2

Previous Issue   

High-precision extraction of water bodies in mountainous areas from multi-polarization SAR images via deep learning that integrates multi-scale radar backscatter characteristics

YUAN Zhifang, LI Junxiao, KANG Qian

2026, 0(2):  1-6.  doi:10.13474/j.cnki.11-2246.2026.0201

Asbtract ( )   HTML ( )   PDF (4764KB) ( )  

References | Related Articles | Metrics

To address the challenges of frequent water misclassification and difficulty in preserving water boundary details caused by mountain shadows,dense vegetation,and SAR coherent speckle noise in complex terrains,this study proposes a novel deep learning-based water extraction method for SAR imagery that integrates multi-scale radar backscattering features through multi-polarization SAR data.The developed approach employs multi-resolution residual convolution and multi-scale feature identity mapping to effectively characterize the rich multi-scale terrain features in complex mountainous environments using multi-polarization SAR imagery,thereby enhancing both accuracy and completeness in water body extraction.Experimental validation was conducted using Sentinel-1 dual-polarization data covering a reservoir in Shanxi Province,with comparative analysis against classical SVM,MRF algorithm,and U-Net model through qualitative and quantitative assessments.The experimental results showed that 92.14% F1,91.39% Kappa coefficient,85.43% IoU,and 98.62% OA.This indicates that the method proposed in this paper has the best comprehensive performance in water extraction and can effectively maintain the boundary details of water extraction.

Application of satellite-derived bathymetry in reservoir storage monitoring

WANG Qingqing, ZHANG Zhongliu, LU Gang

2026, 0(2):  7-11.  doi:10.13474/j.cnki.11-2246.2026.0202

Asbtract ( )   PDF (3766KB) ( )  

References | Related Articles | Metrics

To address the issues of high cost and long cycles associated with traditional reservoir storage monitoring methods,this study explores an efficient and low-cost dynamic monitoring technology,providing a novel solution for medium and small reservoir management.Taking Jurong reservoir as the study case,we integrated Sentinel-2 data with in-situ measurements and developed a satellite-derived bathymetry model using the random forest algorithm.By combining water level data,underwater topography was reconstructed to generate an updated storage capacity curve.The results demonstrate good accuracy in water depth inversion (R²=0.90, RMSE=0.43 m),with the derived storage capacity showing strong consistency with official data trends.A linear relationship between surface area and storage increment was established.Compared to conventional methods,this approach significantly shortens the update cycle of storage capacity curves while reducing costs.Therefore,satellite-derived bathymetry method proves to be a reliable method for dynamic storage monitoring,offering advantages in speed,cost-effectiveness,and operational convenience.This technique is particularly suitable for routine management of medium and small reservoirs,providing water authorities with a practical and efficient technical solution.

A comparison study on land subsidence monitoring in the Xiaolangdi Reservoir area based on SBAS and DS-SBAS

LI Penghao, LI Aiguo, ZHANG Shiyu

2026, 0(2):  12-17.  doi:10.13474/j.cnki.11-2246.2026.0203

Asbtract ( )   HTML ( )   PDF (5491KB) ( )  

References | Related Articles | Metrics

Xiaolangdi Reservoir is the core of water and sediment regulation in the Yellow River.The subsidence monitoring of the reservoir is crucial to ensuring its long-term and effective operation.To accurately obtain the settlement information of Xiaolangdi Reservoir,conventional SBAS (small baseline subset)technology and DS-SBAS (distributed scatterer small baseline subset)technology are respectively used for time-series interferometric processing of 60 Sentinel-1A satellite images from 2018 to 2024,and their results are compared.The results show that the correlation coefficient of settlement rates at monitoring points with the same longitude and latitude obtained by SBAS and DS-SBAS is 0.96,indicating a high consistency between the two monitoring results.The number of high-coherence points extracted by DS-SBAS is approximately 54% higher than that by SBAS,and its monitoring effect is more significant in the settlement rate range of 0～5 mm/a.Therefore,DS-SBAS can be used as an effective method for settlement monitoring in the reservoir area.

Construction and application of water-land integrated DEM based on multi-source data fusion of lake wetlands: a case study of Sanchahe National Wetland Park

WEI Linghui, LI Bao, CHEN Liangsong, GAO Yang, SUN Wei

2026, 0(2):  18-23,45.  doi:10.13474/j.cnki.11-2246.2026.0204

Asbtract ( )   PDF (3096KB) ( )  

References | Related Articles | Metrics

The terrain environment of lake wetlands is complex,and a single measurement method is difficult to meet the requirements of high-precision terrain data acquisition and high-precision three-dimensional expression.This paper proposes a multi-source collaborative measurement method,which uses unmanned boats for single-beam depth sounding,unmanned aerial vehicles (UAVs)for LiDAR height measurement,and combines GNSS and depth rods for supplementary measurement in small areas to achieve efficient,high-precision,and high-density data collection of water and land terrain.Based on the vector boundary of the water area,the constrained Delaunay triangulation algorithm is adopted to seamlessly integrate and smoothly transition the water and underwater terrain data,constructing a high-precision water-land integrated digital elevation model (DEM),and then establishing a surface water storage calculation model.The research results show that the collaborative application of multiple measurement technologies can significantly improve the data collection efficiency and accuracy of the water-land integrated DEM.This method can provide reliable technical support for water resources investigation,management,and sustainable utilization.

Dynamics and driving mechanisms of groundwater storage in the Tarim River basin

NIU Fangpeng, LU Zhenlin, CAO Biao, KOU Jing, MA Xueyan

2026, 0(2):  24-30.  doi:10.13474/j.cnki.11-2246.2026.0205

Asbtract ( )   HTML ( )   PDF (10283KB) ( )  

References | Related Articles | Metrics

This study addresses the issues of water scarcity and groundwater over-exploitation in the Tarim River basin,an arid region in northwestern China.By integrating remote sensing and ground-based observation data,we developed an accurate estimation model for groundwater storage in the region using water storage component decomposition techniques,aiming to provide a scientific basis for sustainable regional water resource management.Multi-source data,including terrestrial water storage (TWS)and surface water,were consolidated into a spatiotemporally consistent dataset through preprocessing and spatial resampling.A systematic analysis was conducted to examine the long-term trends,seasonal variations,and spatial distribution of groundwater storage from 2002 to 2024.The results reveal a significant declining trend in groundwater storage across the Tarim River basin,with an average annual reduction rate of -8.6 mm/a(p<0.01).The total cumulative loss over the study period exceeded 198 mm,equivalent to approximately 234 billion m³ of groundwater depletion for the entire basin.Spatially,the groundwater decline was most severe in the Kashgar Oasis,Korla Irrigation District,and Taitema Lake area,where the average annual loss rates generally exceeded -10 mm/a.This study provides critical data support for dynamic groundwater monitoring and rational water resource allocation in arid regions,while also offering a reference for future related research.

Geometry correction accuracy improvement of satellite images guided by control images

WANG Ruixuan, GUO Haoyu, ZHANG Yongjun, ZHOU Bilian, WANG Guangshuai, WAN Yi

2026, 0(2):  31-37.  doi:10.13474/j.cnki.11-2246.2026.0206

Asbtract ( )   PDF (4173KB) ( )  

References | Related Articles | Metrics

To address the limitations of geometric accuracy in high-resolution satellite image production,which relies on orthophoto control sources and suffers from low efficiency due to manual ground control point selection,an efficient and automated geometric accuracy improvement method is urgently needed.This paper proposes an automated geometric control framework based on control image blocks,which integrates adaptive image segmentation and epipolar geometric constraints.An automatic control point matching and hierarchical bundle adjustment workflow is developed to robustly consolidate multi-source control information and enhance geometric constraints.Experiments demonstrate that the proposed method achieves control point matching accuracy comparable to manual operations across multiple typical regions,with matching errors consistently maintained within a reasonable range and stable performance across cross-resolution scenarios.The method offers strong scalability of control data and significantly improves the automation efficiency and geometric accuracy of high-resolution satellite image production,showing promising potential for engineering applications.

Atmospheric water vapor retrieval over China's Taiwan Island using combined IGS/Sentinel-1/ERA5 multi-source data

ZHOU Lei, LI Dewei, LI Xiaohong, DI Guishuan

2026, 0(2):  38-45.  doi:10.13474/j.cnki.11-2246.2026.0207

Asbtract ( )   HTML ( )   PDF (7794KB) ( )  

References | Related Articles | Metrics

Aiming at solving the uncertain problems result from the various differences in the atmospheric water vapor inversion process between GNSS and InSAR techniques,this paper proposes a high-precision,spatially continuous atmospheric water vapor inversion method that integrates multi-source data from IGS,Sentinel-1,ERA5.Firstly,taking China's Taiwan Island as the study area,tropospheric delays are extracted from interferograms generated by Sentinel-1A data of 72 scenes in 2023.Secondly,InSAR dry delays are verified and corrected by IGS station dry delays.Then,the geometric differences between GNSS and InSAR observations are overcome by mapping functions.And then,spatial continuous water vapor conversion coefficients are calculated by ERA5 meteorological reanalysis data.Finally,the accuracy of InSAR PWV inversion is verified by sounding station and IGS station PWV data.The results indicate that compared with PWV measurements from sounding stations and IGS stations,the average deviation of InSAR PWV is less than 0.71 mm,and the root mean square error is less than 1.63 mm.By incorporating GNSS-based quality control in the calculation of dry delay and water vapor conversion coefficients,InSAR enables high-precision,spatially continuous,and finely detailed atmospheric water vapor retrieval,thereby establishing a foundation for monitoring atmospheric water vapor distribution in mountainous regions with complex terrain.

Monitoring and influencing factors analysis of land subsidence along the Beijing-Xiong'an intercity railway using InSAR technology

LIN Yang, YU Bing, TIAN Xin, ZHANG Guanjun, LIU Cheng, GAN Jun, LI Guangyu

2026, 0(2):  46-53.  doi:10.13474/j.cnki.11-2246.2026.0208

Asbtract ( )   PDF (17688KB) ( )  

References | Related Articles | Metrics

To address the limited quantitative research and insufficient temporal coverage of surface subsidence and its driving factors along the Beijing-Xiong'an intercity railway.The SBAS-InSAR technique,Moran's I index,and multi-scale geographically weighted regression (MGWR)model were employed to monitor and analyze surface deformation from November 2022 to November 2024.Within a 4 km buffer zone on both sides of the railway,the maximum average annual vertical deformation rate reached -127 mm/a.Subsidence was mainly concentrated west of Xiong'an Station to Bazhou North Station,showing significant spatial clustering.In order of contributions size,driving factors are groundwater level fluctuation、distance to faults、distance to rivers、surface roughness、distance to roads.Cumulative deformation was negatively correlated with groundwater level fluctuation,indicating that groundwater extraction aggravates subsidence.In the section west of Xiong'an to Bazhou North,subsidence increased with distance to faults,suggesting that the Niudong Fault may cause the observed differential settlement.These findings provide a scientific reference for railway maintenance and groundwater resource management.

Identification of rural homestead utilization status by integrating multi-source high-resolution remote sensing imagery

XIE Jianing, LIU Zhenbo, YANG Yuting

2026, 0(2):  54-59,67.  doi:10.13474/j.cnki.11-2246.2026.0209

Asbtract ( )   PDF (10148KB) ( )  

References | Related Articles | Metrics

To furnish data-driven decision support for rural settlement spatial restructuring,revitalization of underutilized homesteads,and precision land governance through accurate identification and classification of rural homestead utilization states.A rural homestead utilization identification framework driven by multi-source high-resolution remote sensing data is proposed,integrating deep learning and machine learning techniques.The findings indicate that:①The overall accuracy of homestead recognition based on GF imagery and Google Earth imagery exceeds 84%;②The XGBoost model demonstrates superior performance in identifying inhabited homesteads,achieving a precision of 94.6%,while the random forest (RF)model exhibits the best performance in recognizing idle homesteads,with a precision of 77.8%;③According to comprehensive evaluations using ROC and PR curves,the RF algorithm outperforms the others,with the green looking ratio derived from Google Earth imagery contributing 12.7%to feature importance.This study substantiates that fusing multi-source remote sensing and machine learning technologies constitutes an effective approach for homestead utilization mapping,thereby providing a robust technical foundation for advancing land resource intensification and sustainable rural land management.

A robust method for identifying the subsidence basin boundary caused by mining based on SBAS-InSAR and level data fusion

CHEN Yuanfei, WANG Lei

2026, 0(2):  60-67.  doi:10.13474/j.cnki.11-2246.2026.0210

Asbtract ( )   HTML ( )   PDF (5647KB) ( )  

References | Related Articles | Metrics

InSAR monitoring technology struggles to obtain large deformation data in areas of mining subsidence,and the boundary of subsidence is easily influenced by interference,affecting the accuracy of boundary delineation.To address this,a random sample consensus (RANSAC)algorithm is introduced,and a robust fusion method for mining subsidence data is proposed,which couples the Boltzmann function (BRAN).By randomly sampling SBAS-InSAR pixel points and using the number of inliers as an evaluation criterion,we repeatedly invert the model until a robust parameter solution with the maximum number of inliers is obtained.Finally,the predicted subsidence calculated by Boltzmann model and its robust parameter are used to replace outliers pixels and decoherent data,completing the robust fusion of subsidence data.Simulation experiments and a case study from the Zhujidong coal mine in Huainan show that the BRAN method can effectively identify and eliminate the interference of outliers pixel points,with the relative root mean square error of the fused subsidence values being 6.4%.The obtained subsidence basin boundary aligns with the characteristics of mining subsidence.The research findings provide new insights for robust fusion of multi-source data in mining areas and hold certain application value for monitoring and early warning of subsidence disasters in these regions.

Vehicle object detection approach in drone imagery based on improved YOLOv8s

TENG Min, ZHANG Bo, XU Jiawei, LIN Cong, SHEN Yu, CHU Zhengwei

2026, 0(2):  68-73,80.  doi:10.13474/j.cnki.11-2246.2026.0211

Asbtract ( )   PDF (2770KB) ( )  

References | Related Articles | Metrics

Accurate and real-time vehicle detection and tracking provide crucial data support for traffic flow estimation and intelligent traffic management.Drone imagery has emerged as a vital data source for vehicle detection tasks.To address the weak ability of existing YOLO models to detect small objects within complex scenarios and the scarcity of open-source datasets for drone vehicle detection,this paper proposes the YOLOv8s-VOD model specifically designed for vehicle detection tasks,and introduces the open-source dataset NJVOD.This method constructs C2f-PTB and BiFPN-GLSA modules to achieve collaborative extraction of global-local featuresand effective fusion of multi-scale semantic and edge information,thereby improving detection accuracywhile reducing network complexity.Experimental results show that YOLOv8s-VOD achieves the highest detection accuracy with minimal parameters,outperforming existing methods by 2.4～12.2 percentage poin on the VEDAI dataset and 4.1～5.3 percentage point on the NJVOD dataset;The C2f-PTB and BiFPN-GLSA modules proposed in this work both effectively improve small-object detection accuracy.Additionally,the newly created NJVOD dataset offers crucial support for related research.

DEM generation and coastline extraction analysis utilizing airborne LiDAR point cloud data

LI Yingying, LI Xiaona, SUN Liang, ZHANG Baojin, YAN Xing, JI Xue, ZHANG Peixuan, SUN Yumei

2026, 0(2):  74-80.  doi:10.13474/j.cnki.11-2246.2026.0212

Asbtract ( )   PDF (6311KB) ( )  

References | Related Articles | Metrics

The changes in coastline form and location have a profound impact on marine ecology,resource development and coastal economy.Taking the Illinois Beach park beach in the United States as the research object,its on-board point cloud data from 2018 to 2024 is processed to optimize data quality in the paper.In response to the sparse problem of ocean point clouds,Poisson reconstruction method is used to supplement it to improve the integrity and accuracy of DEM.Combined with the DEM contour thick extraction and supervision classification methods,the coastline is initially fitted,and the on-board image is accurately adjusted to ensure that the coastline data is accurate and reliable.The precisely extracted information such as the length and torsion rate of the coastline is compared and analyzed.The study find that the length and torsional rate of the coastline in this area were basically stable and showed a slight upward trend,indicating that there was seawater erosion; but the artificial intervention project launched in spring 2023 effectively slowed down coastal erosion and enhanced the stability of the coastal zone.

Optimization model for GNSS vertical deformation prediction constrained by environmental factor characteristics

ZHANG Yongqi, XIAO Haiping

2026, 0(2):  81-86,125.  doi:10.13474/j.cnki.11-2246.2026.0213

Asbtract ( )   PDF (3272KB) ( )  

References | Related Articles | Metrics

In the traditional time series analysis process,the impact of environmental factors on prediction performance is often overlooked,which limits the model's prediction accuracy.Considering the influence of environmental factors on the vertical deformation of GNSS stations,a BiLSTM model optimized by the VMD and GOOSE algorithms and constrained by environmental factor characteristics is developed.Through experimental comparison,data from 14 GNSS stations between 2006 and 2019 are selected,and prediction accuracy is evaluated using the WQE index.The results show that the average WQE of the VMD-GOOSE-Ef-BiLSTM model is reduced by 84.2%,66.2%,and 50.2%compared to the BiLSTM,VMD-BiLSTM,and VMD-Ef-BiLSTM models,respectively.This demonstrates that the proposed model has higher prediction accuracy and stronger robustness.The experimental results indicate that the proposed model exhibits significant adaptability and accuracy in GNSS vertical deformation prediction,providing effective technical support for surface deformation monitoring.

Factor graph optimization method for mobile phone GNSS positioning in complex environments and performance analysis

ZHOU Zijian, XU Xiaolei, BIN Yuancen, FENG Boqing, YANG Meihao, LIU Weijia

2026, 0(2):  87-91.  doi:10.13474/j.cnki.11-2246.2026.0214

Asbtract ( )   PDF (3710KB) ( )  

References | Related Articles | Metrics

Aiming at the problem of insufficient positioning accuracy of smart phone GNSS caused by multi-path effect,signal occlusion and hardware noise in complex urban scenarios.This paper proposes a positioning method based on factor graph optimization (FGO).By jointly using pseudorange and doppler observations,sliding window technology is adopted to fuse multi-epoch data,so as to optimize state parameters such as position,clock error and inter-system bias.Experiments using Huawei Mate20 (static)and P40 (dynamic)mobile phone data show that compared with the extended Kalman filter (EKF),FGO significantly improves the positioning accuracy.In the static experiment,the errors in the east,north and elevation directions are reduced by 21.2%,9.0% and 15.2%,respectively.In the dynamic experiment,the elevation error is reduced by 37.7%.When the sliding window size is 30,the balance between accuracy and real-time performance can be achieved.The fusion of multi-system data further suppresses the positioning divergence,which can provide an effective solution for high-precision mobile positioning in complex scenarios.

A fusion positioning algorithm of LiDAR and MEMS_IMU

SHEN Wei, LENG Jiaxin, WANG Xiaodan, FENG Qibin, WANG Zicheng, QIAN Enze

2026, 0(2):  92-96.  doi:10.13474/j.cnki.11-2246.2026.0215

Asbtract ( )   PDF (1745KB) ( )  

References | Related Articles | Metrics

This study addresses the challenge of achieving accurate positioning for survey vessels in GNSS-denied environments such as urban waterways,under bridges,and beneath high-pile wharfs.A novel positioning algorithm is proposed that fuses LiDAR with a low-cost MEMS-IMU,aiming to provide relatively high precision under constraints of low computational cost and inexpensive hardware.The algorithm corrects motion distortion in LiDAR point clouds using MEMS-IMU data.It then employs the NDT for point cloud matching to establish a laser odometry function,with KD-tree enhancement for accelerated point cloud search.Finally,the laser odometry is integrated with the MEMS-IMU-derived odometry within a particle filter for fused positioning in these obscured aquatic scenarios.In indoor simulated experiments,the proposed algorithm achieves an average final-position error of 0.20 m,representing a twofold improvement in accuracy over the standalone NDT algorithm and a tenfold improvement over the ICP algorithm.Outdoor experiments conduct under a bridge further demonstrated that the mapping accuracy of the proposed method is significantly superior to both ICP and NDT algorithms.The proposed method achieves relatively high positioning accuracy (better than 0.2 m in both indoor and outdoor tests)at a low cost,demonstrating good potential for real-time positioning of survey vessels or vehicles in GNSS-deprived environments.

Integrating point-line features and geomagnetic constraints in visual-inertial SLAM

WANG Yaohui, ZHANG Zuhao, CHEN Guoliang, WANG Teng

2026, 0(2):  97-103.  doi:10.13474/j.cnki.11-2246.2026.0216

Asbtract ( )   PDF (4153KB) ( )  

References | Related Articles | Metrics

To address the problems of severe localization drift and high false positive/negative rates in loop closure detection under complex conditions in traditional visual-inertial simultaneous localization and mapping (VI-SLAM)systems,we propose a SLAM method that integrates point-line feature extraction and geomagnetic optimization.On one hand,the efficient line feature extraction algorithm Fast-EDLines is introduced into the existing visual-inertial odometry (VIO),accelerated using the AVX2 instruction set,along with a strategy of merging long line segments and eliminating short ones.On the other hand,in the loop closure detection process,magnetometer data from a 9-axis IMU is fused to apply geomagnetic constraints.Combined with a keyframe temporary buffer strategy,the visual matching threshold is dynamically adjusted to reduce false detections and missed detections.The proposed algorithm is tested on the public dataset VECtor Benchmark,and results show that the localization accuracy is improved by 7.0 times and 2.9 times compared to VINS-Mono and PL-VINS,respectively.Effectively improve the positioning accuracy and robustness of SLAM algorithm in complex environments.

Research on the dynamic deformation patterns monitoring and visualization of super high-rise buildings integrating GB-RAR and multi-sensor technologies

ZHANG Guojian, YUE Chengzhao, SANG Wengang, JIN Fengxiang, LIU Shengzhen, YANG Jun, JI Xiang, MIN Fanwen

2026, 0(2):  104-110.  doi:10.13474/j.cnki.11-2246.2026.0217

Asbtract ( )   PDF (6646KB) ( )  

References | Related Articles | Metrics

During long-term operation,factors such as prolonged loading and material aging gradually reduce the stability and bearing capacity of super-tall structures,making them prone to safety incidents.Therefore,it is essential to conduct real-time deformation monitoring of skyscrapers to understand their health status.This study takes the Shandong International Financial Center and the Yunding duilding as research objects.It integrates technologies such as ground-based real aperture radar(GB-RAR),dual-axis tilt sensors,and GNSS technology to monitor the dynamic deformation characteristics of the super-tall structures from multiple dimensions.Furthermore,the study achieves the indoor and outdoor integrated modeling of the Shandong International Financial Center based on handheld and station-based LiDAR technology,establishing a skyscraper deformation monitoring and early warning visualization platform.The results indicate that the GB-RAR measurement accuracy reaches 0.016 5 mm,which satisfies the precision requirements for deformation monitoring.The deformation pattern of the Shandong International Financial Center exhibits a spatio-temporal characteristic of sway and vibration synergistic evolution.The main direction of the sway movement is from the southeast to the northwest,with a sway period of approximately 20 d.The sway amplitude at a height of 400 m is about 180 mm,and the average sway velocity is approximately 1.5 mm/h.The structural transfer floor,located at 330 m,balances the eccentric axial force of the super-tall structure.This floor exhibits a sway amplitude of approximately 551 mm and an average sway velocity of about 5.19 mm/h.This transfer floor is a key factor causing the overall spatial state of the skyscraper to switch back and forth among S-shaped,hyperbolic,and diagonal configurations,thereby presenting non-linear elastic deformation characteristics.

Enhanced visualization method for spatio-temporal evolution process of landslide disasters

FU Lin, ZHANG Junjie, XIE Mingxuan, XIONG Junnan, XIAO Kunhong, WANG Ke

2026, 0(2):  111-117.  doi:10.13474/j.cnki.11-2246.2026.0218

Asbtract ( )   PDF (10007KB) ( )  

References | Related Articles | Metrics

The existing numerical simulation research methods focus on the study and expression of the mechanism of landslide sliding process,lack the display of information before and after disasters,and are difficult to present the complex spatio-temporal process of disaster scenes.The virtual geographical environment can reproduce the spatio-temporal process of disasters,realize the visual expression and sharing of disaster information,and is of great significance to the modernization development of disaster emergency management.To this end,this paper proposes a visualization method for landslide disasters with enhanced spatio-temporal evolution,achieving the integrated modeling of landslide numerical results and multi-source geographical entities.It enhances the expression of the spatio-temporal evolution process of landslides through multi-level semantic constraints and multi-dimensional visual variables.Develop a prototype system to conduct experiments and analyses,display the geographical scenarios of disasters in the front-end system,and fully realize the visual enhancement and visualization expression of the processes before,during and after disasters.Studies show that this method can effectively integrate the numerical results of landslides with virtual geographical scenes,providing more intuitive and information-rich visualization support for the spatio-temporal evolution process of disasters.

Location selection of railway drainage facilities in mountainous areas based on fine terrain and DBSCAN-PCA algorithm

CHEN Liuting, ZHANG Xuanyu, DONG Xiujun, LIU Guiwei, SUN Qihao, DENG Bo

2026, 0(2):  118-125.  doi:10.13474/j.cnki.11-2246.2026.0219

Asbtract ( )   PDF (18520KB) ( )  

References | Related Articles | Metrics

Railway water damage disasters are the most frequent,widespread,and harmful events on railway lines.They often cause the longest interruptions to train services.In mountainous areas,due to complex terrain and variable climate,water damage disasters occur frequently,which have a serious impact on the safe transportation of railways.Moreover,traditional manual location selection is time-consuming,laborious and difficult to carry out macroscopic optimization and allocation of regions.To solve the problem of intelligent location selection for railway drainage facilities in mountainous areas,by extracting three-dimensional refined terrain features and based on the DBSCAN density clustering algorithm and principal component analysis (PCA),an intelligent optimization allocation algorithm for drainage facility selection of “spatial clustering-dimension reduction combined” is proposed.Taking a railway in a mountainous area of Hunan province as an example,the location selection of railway drainage facilities is predicted.The results show that the prediction results match the actual culvert locations with an accuracy of 92%.The design of drainage ditches covers high-risk areas.The optimized drainage system can reduce the probability of water accumulation on the roadbed and significantly improve the railway's resistance to water damage.The research results veriy the applicability of the drainage facility location selection method based on this allocation algorithm under complex terrain conditions,providing a useful reference for the location design and optimization of drainage facilities in mountainous railway areas.

A drone precision landing framework using large language models

CHEN Lijun, CHEN Qing

2026, 0(2):  126-130.  doi:10.13474/j.cnki.11-2246.2026.0220

Asbtract ( )   PDF (1151KB) ( )  

References | Related Articles | Metrics

Traditional landing methods often fall short in response to the limited semantic sensing capabilities of UAVs in dynamic,unstructured environments and their reliance on fixed,context-insensitive safety factors.To address these limitations,a hybrid framework,LLM_Land,is proposed that combines a large language model (LLM)with model predictive control (MPC),starting from a visual language encoder (VLE)(e.g.,BLIP),which converts real-time images into succinct textual scene descriptions,which is used by a retrieval-augmented generation (RAG)-equipped lightweight LLM (e.g.,Qwen 2.5 1.5B or LLaMA 3.2 1B)to categorise scene elements and infer context-aware safety buffers,e.g.,3 m for pedestrians and 5 m for vehicles,and the resulting semantic flags and unsafe zones are subsequently fed into the MPC module,enabling real-time trajectory replanning for collision avoidance whilst maintaining a high level of landing accuracy.The proposed framework is validated in the ROS-Gazebo simulator,which consistently outperformed the conventional vision-based MPC baseline,and the results showed a significant reduction in near-miss accidents due to dynamic obstacles,while maintaining accurate landings in a cluttered environment.

Research on spacing thresholds for image control point distribution in digital aerial photography based on aerial triangulation accuracy

YAO Na, SU Penghao, LI Dezhong, WANG Yanping, XUE Yongjun, CONG Xiaoou

2026, 0(2):  131-136.  doi:10.13474/j.cnki.11-2246.2026.0221

Asbtract ( )   PDF (1714KB) ( )  

References | Related Articles | Metrics

Sparse distribution of image control points is critical for enhancing the economic efficiency of large-scale digital aerial photography.Based on the requirement for constructing a 1∶2000 high-precision real-scene 3D monitoring baseplate within the natural resources survey and monitoring system of the Inner Mongolia Autonomous Region,this paper designs 11 sets of comparative experiments with varying combinations of along-track baseline span and cross-track flight-line span to quantitatively verify the impact of spacing configurations on aerial triangulation accuracy.Results indicate that residual errors of orientation points remain minimally affected by span variations and checkpoint accuracy decreases with increasing spans,where plane accuracy shows higher sensitivity to along-track spans,while elevation accuracy responds more significantly to cross-track spans.Accordingly,span thresholds for image control point distribution balancing accuracy and efficiency in flat terrains are proposed,providing an engineering decision-making basis for precision-economy equilibrium in digital aerial photography.

Semantic segmentation of UAV orthophoto images based on improved U-Net3+ model

JIANG Lei, LIANG Cong, ZHAO Xu, WANG Peng, YAN Wenkai, YANG Hongding, WU Jizhong

2026, 0(2):  137-143.  doi:10.13474/j.cnki.11-2246.2026.0222

Asbtract ( )   PDF (2443KB) ( )  

References | Related Articles | Metrics

To address the limitations of insufficient feature abstraction and cross-scale feature redundancy in semantic segmentation of unmanned aerial vehicle (UAV)orthophoto images using the U-Net3+ model,this study proposes an improved U-Net3+ architecture.The improvement incorporates ResNet50,a deep convolutional neural network based on residual network,as the backbone for feature extraction.Simultaneously,the convolutional block attention module (CBAM)is integrated as a lightweight attention mechanism.Experimental results demonstrate that the proposed U-Net3+ model delivers significant improvements in segmentation performance,achieving an 8.3% increase in overall accuracy,2.6% in mean intersection over union,and 1.9% in F1-score compared to the original U-Net3+ model.The proposed model consistently outperforms established benchmarks,including FCN,U-Net,U-Net++,and the DeepLab series,across all evaluation metrics,demonstrating superior feature discrimination and segmentation accuracy in representative scene types.Moreover,the integration of either ResNet50 or CBAM alone results in moderate gains,their combined implementation leads to a notable synergistic effect,yielding the most effective results in segmentation tasks.The improved U-Net3+ model has significantly improved the segmentation accuracy,providing an effective technical solution for semantic segmentation of UAV orthophoto maps.

Fine-grained crop classification algorithm based on global and local feature fusion

ZHANG Huan, HUANG Qiuying, LIU Sheng, WANG Qi, LI Kai

2026, 0(2):  144-148.  doi:10.13474/j.cnki.11-2246.2026.0223

Asbtract ( )   PDF (2193KB) ( )  

References | Related Articles | Metrics

In the national land use change survey,the manual visual interpretation of field-photographed verification photos is inefficient.Identifying crop types to determine land use types is a key approach to solve this problem.However,crop type identification faces challenges such as large intra-class differences and small inter-class differences.This paper proposes a fine-grained crop classification algorithm based on the fusion of global and local features.Global features are extracted from the original images,and local features are obtained by using a detector to acquire target regions.After feature compression and fusion,the features are input into a classifier.The exponential moving average method is used to update network parameters,and the additive angular cross entropy loss function is selected to complete model training.Experiments on real-world data from the national land use change survey show that this algorithm outperforms comparative algorithms in terms of precision,effectively improving the accuracy of fine-grained crop classification and providing efficient intelligent interpretation technical support for related businesses such as the national land use change survey.

Characterization of surface subsidence in Hefei city based on the new domestic LT-1 SAR satellite

WANG Yingchun, LI Jinchao, WANG Chengzhi, LI Shuiping, TAO Tingye

2026, 0(2):  149-155,173.  doi:10.13474/j.cnki.11-2246.2026.0224

Asbtract ( )   PDF (5334KB) ( )  

References | Related Articles | Metrics

The LuTan-1 (LT-1)satellite is China's first L-band differential interferometry SAR satellite,notable for its high spatial resolution,short revisit cycle,and full polarization capabilities.The current research on the application of SAR images acquired by this novel sensor is insufficient.The city of Hefei is chosen as the focal point for this study,where in three high-resolution radar images from LT-1 spanning July 2023 to May 2024 are utilized.To derive the surface subsidence deformation image of Hefei city,we employ the GACOS atmospheric correction model and delaunay minimum cost flow phase unwrapping method.The findings indicate that the ground deformation in Hefei city is generally smooth,with cumulative settlements in most areas ranging from -200～20 mm.However,certain local areas exhibit noticeable deformation trends.Five regions with significant deformation have been selected for a more detailed analysis,and their results have been compared with concurrent Sentinel observations,thereby validating the reliability of the LT-1 monitoring data.This study demonstrates that LT-1 imagery is capable of detecting finer deformation signals and providing higher-precision subsidence information,which can serve as vital support for urban safety and maintenance monitoring.

3D point cloud semantic segmentation for high-voltage transmission lines

GAO Shuhan, ZHOU Chao, RONG Mengqi, LIU Yangdong, LIU Hui, SHEN Hao, JIA Ran, LIU Chuanbin, ZHANG Yang, LIU Rong, SHEN Shuhan

2026, 0(2):  156-160.  doi:10.13474/j.cnki.11-2246.2026.0225

Asbtract ( )   PDF (2003KB) ( )  

References | Related Articles | Metrics

Due to the phenomenon of sample imbalance among target categories in high-voltage transmission scenarios,the intelligent inspection system for high-voltage transmission lines usually has low recognition accuracy for few-sample targets.To address this issue,this paper proposes a 3D point cloud semantic segmentation method that integrates category awareness.Firstly,an adaptive dynamic sampling strategy is introduced,which optimizes the distribution of point cloud data through density-aware region division to improve data balance.Secondly,a class-aware contextual feature enhancement module is designed,which dynamically fuses points features using class embedding information to strengthen the model's discriminative capability.Finally,a weighted loss function is constructed to alleviate learning bias caused by long-tailed data distributions.Experimental results on real-world point cloud datasets of high-voltage transmission lines demonstrate that the proposed method not only improves overall segmentation accuracy but also achieves superior recognition performance for minority classes.This study provides effective technical support for intelligent recognition of complex structural targets in power line inspection and shows promising potential for practical engineering applications.

Partial discharge detection in 220 kV high-voltage equipment using UAV-based ultraviolet imaging with multi-scale feature fusion

JI Shuolei, HUANG Hengying, LI Yucheng, CHEN Hailin

2026, 0(2):  161-167.  doi:10.13474/j.cnki.11-2246.2026.0226

Asbtract ( )   PDF (5372KB) ( )  

References | Related Articles | Metrics

To address the challenges of weak signal interference and feature extraction in ultraviolet (UV)imaging detection of partial discharges (PD)in 220 kV substation high-voltage equipment,this paper proposes a detection method that partial discharge detection of unmanned aerial vehicle ultraviolet imaging based on multi-scale feature fusion network(MSFF-Net).Firstly,an improved Retinex algorithm and adaptive wavelet threshold denoising are applied to enhance UV images,improving the signal-to-noise ratio in weak discharge regions.Then,a multi-scale feature fusion network is constructed,which combines parallel atrous convolution modules with an attention mechanism to extract and dynamically weight key discharge features under different receptive fields.Experimental results on real-world 220 kV substation datasets demonstrate that the proposed method significantly improves the sensitivity to weak PD signals and successfully identifies over 90% of early-stage discharge defects.The proposed approach effectively enhances the detection and localization capabilities for UV imaging-based PD signals,greatly improving the accuracy and reliability of substation equipment condition monitoring.

Ecological restoration monitoring of abandoned quarry mine based on UAV

WANG Jianbin, CHEN Qiuji, CUI Chuangyi, NIU Baoqi

2026, 0(2):  168-173.  doi:10.13474/j.cnki.11-2246.2026.0227

Asbtract ( )   PDF (3703KB) ( )  

References | Related Articles | Metrics

Establishing a monitoring and evaluation method of mine ecological restoration based on UAV remote sensing images to reveal the spatial distribution characteristics and influencing factors of mine vegetation restoration and providing basis for ecosystem management.An abandoned quarry in Weibei dry belt was taken as the study area,Remote sensing images of the study area after restoration were obtained by unmanned aerial vehicles,combined with the high-resolution images before treatment,the vegetation cover of the study area was divided by green leaf index (GLI)and fractal dimension of GLI was calculated to evaluate the ecological environment changes before and after restoration.Combined with geographic detectors,the influencing factors affecting ecological restoration effect were identified,and corresponding ecological restoration measures and monitoring suggestions were put forward.After treatment,GLI value and vegetation coverage increased significantly,and vegetation types showed diversity.Slope is the key factor restricting vegetation restoration in the study area,and physical and chemical characteristics of reconstructed soil also have great influence on regional ecological restoration.The research results provide basic data for mine ecosystem regulation and enrich the evaluation techniques and methods of ecological restoration effect.

Research on multi-dimensional intelligent site selection and route planning application based on AI+GIS

DAI Zhimin, WANG Haiyan, ZHANG Xing, TANG Hao, TANG Ming, ZHONG Yong, WU Baoyou

2026, 0(2):  174-179,186.  doi:10.13474/j.cnki.11-2246.2026.0228

Asbtract ( )   PDF (2887KB) ( )  

References | Related Articles | Metrics

To address issues such as the reliance on manual experience,low data integration efficiency,and lagging multi-dimensional evaluation in traditional power grid site selection and route planning,this paper proposes and studies a multi-dimensional intelligent site selection and route planning application system based on AI+GIS.With GIS technology as the spatial data support carrier,the system integrates a remote sensing interpretation module to realize the automatic extraction and high-precision analysis of geographical information such as terrain,vegetation,and buildings.It also integrates artificial intelligence algorithms to construct an evaluation model covering dimensions including environmental impact,project cost,and power grid security,enabling intelligent calculation and optimized ranking of site selection and route planning schemes.Verified through application in actual power grid projects,the platform can replace over 70% of repetitive work in traditional manual surveys,shorten the site selection and route planning cycle by 40%,and increase both the scheme compliance rate and the accuracy of economic evaluation to over 90%.The research shows that the in-depth integration of AI and GIS can effectively break through the limitations of traditional methods,provide scientific and efficient technical support for power grid planning,site selection,and route planning,and is of great significance for promoting the intelligent transformation of power grid planning.

Surface sinking prediction using coupled machine learning models

ZHU Weigang, JIN Shengbo, JIANG Shaohua

2026, 0(2):  180-186.  doi:10.13474/j.cnki.11-2246.2026.0229

Asbtract ( )   PDF (5028KB) ( )  

References | Related Articles | Metrics

Surface settlement prediction plays a crucial role in ensuring infrastructure safety and managing construction risks.However,single predictive models often suffer from limited generalization capability,noticeable systematic bias,and poor responsiveness to outliers in surface settlement prediction.This study proposes a coupled machine learning approach for surface settlement prediction.Using settlement monitoring data from a metro station project in Changchun as the engineering background,three machine learning models—CatBoost,GBDT,and AdaBoost—are integrated to predict surface settlement.The results indicate that the coupled model effectively eliminates systematic deviation and suppresses abnormal peak fluctuations.The coefficient of determination (R²)reachs 0.97,while the mean absolute error (MAE)and root mean square error (RMSE)decrease to 0.007 3 and 0.009 6,respectively,demonstrating superior predictive accuracy and stronger generalization capability.The proposed coupled model provides reliable technical support for long-term deformation monitoring and risk early warning in metro projects.It also offers a methodological reference for enhancing the accuracy of surface settlement prediction and ensuring construction safety,with potential applicability to other geotechnical engineering scenarios.