Table of Content

25 August 2026, Volume 0 Issue 8

Previous Issue   

The positioning performance improvement method of BDS-3 PPP-B2b sliding window smoothing

Chen Junnan, Liu Ling, Zheng Yongfeng, Wang Shangqi

2026, 0(8):  1-7.  doi:10.13474/j.cnki.11-2246.2026.0801

Asbtract ( )   PDF (2527KB) ( )  

References | Related Articles | Metrics

[Purposes] The BeiDou-3 Navigation Satellite System (BDS-3)broadcasts PPP-B2b signals via three geostationary earth orbit (GEO)satellites.This effectively overcomes the reliance of conventional precise point positioning (PPP)technology on ground-based data links.This paper aims to systematically evaluate the performance of the PPP-B2b correction signals broadcast by the BDS-3 GEO satellites in precise point positioning and address the real-world issue of clock offset jump in PPP-B2b corrections. [Methods] Observation data from five MGEX stations in and around China during DOY075—080,2024 were selected for analyzing the real-time positioning accuracy of PPP-B2b.An anomaly detection and correction algorithm based on a sliding window was proposed to mitigate the clock offset jumps,and positioning results before and after processing were compared. [Findings] Experiments show that in static mode,the average RMS of PPP-B2b positioning in the N,E,and U directions are 1.0,1.7,and 2.3 cm,respectively.After applying the sliding-window algorithm,the accuracy in the height direction improved by 15%～28% on average.Specifically,the U-direction RMS decreased from 0.023 m to 0.020 m at station URUM and from 0.029 5 m to 0.023 m at station WUH2,effectively suppressing the fluctuations caused by clock offset jumps. [Conclusions] The results demonstrate that PPP-B2b can achieve centimeter-level real-time static positioning,and the proposed algorithm significantly enhances vertical accuracy,providing technical support for its application in fields such as geological hazard monitoring and high-precision surveying and mapping．

PPP-WAR positioning performance analysis based on MADOCA satellite-based augmentation service

Jiao Hui, Shi Shangfeng, Zheng Kai

2026, 0(8):  8-13,20.  doi:10.13474/j.cnki.11-2246.2026.0802

Asbtract ( )   HTML ( )   PDF (3312KB) ( )  

References | Related Articles | Metrics

[Purposes] Real-time，high-precision GNSS positioning is critical for applications such as autonomous driving and unmanned aerial vehicle navigation.Precise point positioning with WL ambiguity resolution (PPP-WAR)offers rapid decimeter-level accuracy but is often limited by network communication，hindering offline use. [Methods] Therefore，this paper proposes a PPP-WAR algorithm based on factor graph optimization and MADOCA-PPP service，validated through simulated kinematic tests and vehicle-borne experiments in urban environments. [Findings] Results show that using the WL UPD products for ambiguity fixation，the method achieved positioning accuracies of 0.54，0.58，and 1.06 m in the E，N and U directions，representing improvements of 28.6%，3.6%，and 25.3% over the float solution，with convergence time shortened by approximately 30%.However，the WL UPD derived from OSB provided by MADOCA showed lower temporal stability，with an standard deviation of approximately 0.02 to 0.03 cycles，leading to a slight degradation in PPP-WAR performance. [Conclusions] Overall，both approaches achieved accuracy comparable to solutions based on WHU real-time precise products．

Distributed parallel solution of large-scale GNSS network based on GAMIT double-difference model

Wang Jianwei, Feng Zaimei, Zhao Hui, Jiang Guangwei, Tian Jie, Ma Runxia

2026, 0(8):  14-20.  doi:10.13474/j.cnki.11-2246.2026.0803

Asbtract ( )   PDF (1779KB) ( )  

References | Related Articles | Metrics

[Purposes] In response to the low efficiency of traditional serial processing in large-scale GNSS networks and the difficulty of meeting demands merely by upgrading single-node configurations,seeking rapid data processing has become a current research hotspot both domestically and internationally. [Methods] Based on the GAMIT software,a parallel computing engine was constructed on the basis of the spatio-temporal integrated two-layer data parallel algorithm； a distributed computing engine was built by using technologies such as remote procedure call,cluster computing,and message passing interface. Both were deeply integrated to propose a spatio-temporal integrated three-layer data parallel architecture for multi-core parallel and multi-node parallel in large-scale GNSS networks. [Findings] In the test environment,the maximum speedup ratio of this scheme reached 83.53,and the baseline solution cycle was significantly shortened from about 2.8 months in the traditional serial mode to about 1 d. [Conclusions] This scheme fully integrates the high performance of shared memory systems and the high scalability of distributed systems,and has significant advantages in terms of technological advancement and practical engineering value,providing strong technical support for the efficient processing of massive GNSS data．

UAV positioning method based on LiDAR-IMU-optical flow tightcoupling in GNSS-denied tunnel

Wang Yicun, Xue Kechong, Ye Liangliang, Chen Diangan, Yu Feiyan

2026, 0(8):  21-27.  doi:10.13474/j.cnki.11-2246.2026.0804

Asbtract ( )   PDF (3088KB) ( )  

References | Related Articles | Metrics

[Purposes] A LiDAR-IMU-optical flow tightly coupling positioning method is proposed for UAV positioning in GNSS-denied tunnel with low illuminance and weak textures. [Methods] Optical flow velocity factors are introduced into a LiDAR-IMU factor graph together with tunnel-oriented feature extraction,FPGA-based hardware synchronization,and joint extrinsic calibration. [Findings] Experiments show that in a 400 m simulated tunnel,the mean ATE is 0.127 m,which is 41.2% and 28.5% lower than LIO-SAM and FAST-LIO2,respectively.In a 2.4 km field round-trip test,the loop translation error is 0.83 m.The average processing time on an NVIDIA Jetson Xavier NX is 51.9 ms per frame. [Conclusions] The method preserves online localization potential when LiDAR registration degenerates in long straight tunnel sections,but performance is still limited by extremely weak textures and the lack of loop-closure constraints．

Fast and accurate estimation method for the installation angle of on-board inertial measurement unit

Chen Gang, Zhang Haijun, Ling Chuangwei, Zhao Yingxiong, Wang Bo, Yang Rendong, Zheng Ke, Ding Lei, Wang Yong

2026, 0(8):  28-34.  doi:10.13474/j.cnki.11-2246.2026.0805

Asbtract ( )   PDF (2246KB) ( )  

References | Related Articles | Metrics

[Purposes] To aim at the calibration of the inertial measurement unit (IMU)mounting angles for vehicle-mounted systems. [Methods] This paper proposes a rapid estimation method for vehicular IMU mounting angles.The method leverages the kinematic characteristics of the vehicle during straight-line motion,enabling the direct analytical calculation of the mounting angles using the velocity in the IMU body frame,without requiring prior trajectory computation. [Findings] Experimental results demonstrate that,for a tactical-grade IMU,the accuracies for the heading and pitch mounting angles are 0.039° and 0.019°,respectively. [Conclusions] Theoretical analysis and simulation experiments demonstrate that the method supports online,real-time,fast,and accurate estimation of the mounting angles,making it suitable for vehicular navigation systems that require frequent IMU mounting angle calibration or employ real-time kinematic constraints.

Semantic segmentation method for remote sensing image through global fusion of mask prior and regional semantics

Rong Huijuan, Zhai Liang, Liu Zhendong, Chen Xinxiang, Fu Yu, Sun Yunchuan, Li Min, He Xiaohui

2026, 0(8):  35-43.  doi:10.13474/j.cnki.11-2246.2026.0806

Asbtract ( )   HTML ( )   PDF (5439KB) ( )  

References | Related Articles | Metrics

[Purposes] To address blurred object boundaries,fragmented small targets,and regional semantic inconsistencies commonly encountered in high-resolution remote sensing image segmentation,this study introduces a fusion method that incorporates boundary priors with region-level semantic consistency. [Methods] The proposed approach employs class-agnostic geometric masks generated by SAM(segment anything model)as boundary priors and aligns them with pixel-level semantic probabilities derived from UNetFormer on a unified raster grid.At the connected-component level,a region-semantic feature cost matrix is constructed,and the region-to-class assignment is formulated as a binary linear integer programming problem solved under global semantic consistency constraints. [Findings] Experiments on three remote sensing datasets,including UAVid,show that the proposed method significantly outperforms single-model baselines in mean intersection-over-union (mIoU)and mean F1 (mF1),with notable gains in boundary completeness and semantic consistency for classes such as buildings and vehicles. [Conclusions] The method is decoupled from backbone architectures and requires no retraining,enabling its integration as an independent semantic fusion module within existing segmentation pipelines.It offers an effective solution for enhancing fine-grained semantic interpretation in high-resolution remote sensing imagery．

Remote sensing inversion of COD _Mn in Guangzhou water bodies based on Sentinel-2 imagery

Zhao Tongtong, Deng Ruru

2026, 0(8):  44-50.  doi:10.13474/j.cnki.11-2246.2026.0807

Asbtract ( )   HTML ( )   PDF (6297KB) ( )  

References | Related Articles | Metrics

[Purposes] COD_Mn is an important indicator for assessing the level of organic pollution in water bodies,and its accurate retrieval is crucial for quality monitoring. [Methods] Based on water radiative transfer theory,a multi-component physical inversion model was developed by considering the interactions among optically active substances,including COD_Mn,chlorophyll-a,and suspended solids.Sentinel-2 imagery was used to retrieve COD_Mn concentrations on a pixel-by-pixel basis for surface waters in Guangzhou,and geographic weighted regression (GWR)was applied to analyze the influencing factors. [Findings] The retrieved COD_Mn values showed good agreement with in situ measurements,with a coefficient of determination (R²) of 0.82,a root mean square error (RMSE)of 0.71,and a mean absolute percentage error (MAPE)of 27.30%, indicating satisfactory accuracy.COD_Mn concentrations in Guangzhou were mainly concentrated in the range of 0～2 mg/L,corresponding to Class I surface water quality,suggesting overall good water conditions.Industrial wastewater significantly affected water quality in Nansha,Panyu,and Zengcheng,agricultural activities played a prominent role in Nansha and Zengcheng,and tourism had a stronger impact in Panyu. [Conclusions] The radiative transfer-based multi-component physical inversion approach can accurately characterize the spatial distribution of COD_Mn,providing technical support for refined water quality.

Supraglacial lakes dynamic changes monitoring in the Baltoro Glacier based on time-series Sentinel-2 data

Sun Yongling, Liu Xiao, Pan Leran, Chen Yizhe, Zhang Zhimin

2026, 0(8):  51-58.  doi:10.13474/j.cnki.11-2246.2026.0808

Asbtract ( )   HTML ( )   PDF (6509KB) ( )  

References | Related Articles | Metrics

[Purposes] As a key component of the cryosphere，the dynamic changes of supraglacial lakes are closely related to the stability of glacial systems.In recent years，affected by global warming，glaciers around the world have generally shown a trend of retreat and thinning，and the problem of negative mass balance has continued to worsen.This has directly led to dramatic fluctuations in the number and area of supraglacial lakes.Therefore，accurate monitoring of the dynamic changes of supraglacial lakes holds important scientific significance for revealing the evolution laws of regional climate and environment. [Methods] Based on Sentinel-2 images from 2016 to 2023，this study adopts the random forest algorithm to extract supraglacial lakes on the Baltoro Glacier in the Karakoram region and analyzes the temporal variation characteristics. [Findings] The results show that the number of supraglacial lakes on the Baltoro Glacier shows an overall upward trend，while their total area exhibits a downward trend.In addition，the analysis of the stability of supraglacial lakes shows that the number of dried-up supraglacial lakes has continued to rise，while the number of newly formed and stable supraglacial lakes first increased and then decreased.Among the small，medium，and large supraglacial lakes classified by area，the proportion of their quantities followed form largest to smallest in order of small、medium、large，while the amplitude of their area fluctuations followed form largest to smallest in order large、medium、small. [Conclusions] The findings of this study provide crucial foundational data support for subsequent in-depth analysis of glacial change mechanisms and the evolutionary laws of the climate and environment in the Karakoram region.

An automatic tidal flat extraction algorithm for the Bohai Bay incorporating irregular hexagonal grid

Zhang Xue, Zhang Shuyuan, Zhang Zhijie, He Junliang, Song Kunlun, Qi Qing, Zhang Miaomiao, Zhang Xiao

2026, 0(8):  59-66.  doi:10.13474/j.cnki.11-2246.2026.0809

Asbtract ( )   HTML ( )   PDF (16689KB) ( )  

References | Related Articles | Metrics

[Purposes] To improve the accuracy of tidal flat extraction in complex environments，this study proposes a tidal flat extraction algorithm that incorporates an irregular hexagonal grid (IHG). [Methods] Firstly，the algorithm uses a time-series remote sensing image dataset to generate extreme tidal level images based on the modified normalized difference water index (mNDWI)via the maximum spectral index composite (MSIC)method.Then，through an adaptive threshold gridding strategy，the study area is divided into independent hexagonal units，and the Otsu algorithm is applied to achieve fully automatic extraction of tidal flats in the Bohai Bay in 2023. [Findings] Experiments show that the MSIC-Otsu-IHG algorithm reduces the misclassification of confusing land cover types such as salt pans and suspended sediment.Compared with the MSIC-Otsu algorithm，the overall classification accuracy is improved by 8.4%，and the Kappa coefficient increases from 0.63 to 0.86.When comparing the tidal flat extraction results from two types of images，the overall accuracy and Kappa coefficient of Sentinel-2 data reach 93.60% and 0.86，respectively，which are superior to those of Landsat 8 (89.60% and 0.79). [Conclusions] The MSIC-Otsu-IHG algorithm not only effectively improves the accuracy of land cover separation but also exhibits good applicability in both medium and high spatiotemporal resolution images，providing methodological support for the automatic and refined extraction of tidal flats．

Lightweight monocular depth estimation model based on multi-kernel grouped convolution and adaptive feature fusion

Liu Yao, Huang He, Ma Yanjie, Ma Chaowei, Ren Boyang, Yang Junxing

2026, 0(8):  67-73,81.  doi:10.13474/j.cnki.11-2246.2026.0810

Asbtract ( )   PDF (5002KB) ( )  

References | Related Articles | Metrics

[Purposes] To address the problems of insufficient extraction of distant object edge details，weak global consistency of depth prediction，and the difficulty of meeting real-time 3D perception requirements in autonomous driving for self-supervised monocular depth estimation，this paper proposes a lightweight self-supervised monocular depth estimation model. [Methods] Firstly，a multi-kernel grouped depthwise separable convolution module is designed to effectively fuse local detail and global structural information，significantly enhancing feature representation capability.In addition，an attention-guided adaptive feature fusion module is introduced to dynamically adjust multi-scale feature weighting and enhance the overall consistency of depth prediction. [Findings] Experimental results show that，compared with classical methods Monodepth2，the proposed approach reduces parameters by 10.4% and computation by 34.7%，achieving 0.112 and 0.878 on key performance metrics Abs Rel and δ₁，respectively，and demonstrating superior generalization performance in cross-scene testing. [Conclusions] The proposed method effectively improves depth estimation accuracy under a lightweight design，providing a reliable solution for real-time 3D perception in autonomous driving．

Transmission line tower tilt detection under flooding conditions based on UAV LiDAR

Wei Pengcheng, Wang Xiulong, Wang Yuyang, Huang Haifeng, Guo Ruolan, Guo Fei, Liu Yi, Zhao Binbin, Wen Qingfeng, An Yang, Guo Wei

2026, 0(8):  74-81.  doi:10.13474/j.cnki.11-2246.2026.0811

Asbtract ( )   PDF (9693KB) ( )  

References | Related Articles | Metrics

[Purposes] To address the challenges of invisible tower bases and enhanced point cloud noise in flood-submerged environments，a novel UAV LiDAR-based method is proposed for detecting transmission tower inclination in this paper. [Methods] This method requires only a horizontal member cross-section at the tower bottom and the tower top position to complete inclination assessment，eliminating the need for tower base references and complete tower body data.First，the cross-sectional point cloud of horizontal members is extracted using a bounding box and denoised through statistical filtering.Then，a multi-iteration RANSAC combined with PCA strategy is employed to fit the cross-sectional plane，and a convex hull-rotating calipers-circumscribed square method is used to extract the accurate section center.Finally，the tilt rate is calculated by combining the tower top position with the tower height.An axial-rotation directional graded tilt simulation method is proposed to validate the algorithm performance. [Findings] Results show that within a 6.0‰ tilt rate range，the horizontal offset detection error is less than 6.0 mm，the tilt angle error is below 0.01°，and the tilt rate error is under 0.2‰，with all correlation coefficients exceeding 0.99，meeting the requirements of transmission line operation regulations. [Conclusions] This method provides an effective technical approach for tower inspection and post-disaster assessment in complex aquatic environments．

Selection and evaluation of pseudo-invariant calibration sites in the arid region of Northwest China based on spatio-temporal variability and spatial clustering

Yuan Ye, Chen Wei, Tang Hongzhao, Zhang Jiali, Zhao Jing

2026, 0(8):  82-88,111.  doi:10.13474/j.cnki.11-2246.2026.0812

Asbtract ( )   PDF (14538KB) ( )  

References | Related Articles | Metrics

[Purposes] This study aims to select and comprehensively evaluate potential pseudo-invariant calibration sites (PICS)in the arid region of Northwest China，addressing the current limitations of PICS studies in China，such as insufficient research coverage，poor regional representativeness，and the lack of a multi-scale evaluation framework. [Methods] Using Landsat 8 data from 2020 to 2024，the arid region of Northwest China was selected as the study area.The coefficient of variation (CV)index and the Getis-Ord G^*_i spatial clustering method were integrated to jointly evaluate the spatiotemporal stability and spatial consistency of surface reflectance for identifying and assessing potential PICS. [Findings] The results show that areas with high spatiotemporal stability and spatial consistency are primarily located in the Tarim Basin，Qaidam Basin，and other typical desert，gobi，and salt-flat regions.Twenty highly consistent candidate sites were identified，and eight optimal sites were selected for detailed analysis of surface reflectance characteristics.These optimal sites exhibited remarkable spectral stability and interannual consistency across multiple temporal scales. [Conclusions] The findings confirm the significant potential and advantages of the arid region of Northwest China as an ideal area for satellite radiometric calibration.

A stable multi-target tracking method for ships in complex sea conditions based on interactive multiple models

Tao Zengjie, Zhang Hui, Yao Chaoyun, Zhu Juanxi

2026, 0(8):  89-95.  doi:10.13474/j.cnki.11-2246.2026.0813

Asbtract ( )   PDF (2159KB) ( )  

References | Related Articles | Metrics

[Purposes] Aiming at the core problems of ship multi-target tracking in complex sea conditions,such as high false alarm rate caused by strong sea clutter interference,track fracture and tracking loss caused by diverse maneuvering behaviors,and time-space synchronization of multi-source data in dense scenes,this paper proposes a ship multi-target stable tracking method based on interactive multi model (IMM). [Methods] This method uses a hierarchical processing architecture,first completes the multi-source data preprocessing and space-time alignment,and then designs the IMM adaptive tracking filtering algorithm.Through the parallel interaction of the three models of constant speed,acceleration and rotation rate and the joint probability data association,it realizes the adaptive tracking of maneuvering targets,and develops a variable threshold adaptive detection strategy.According to the sea state level,it dynamically adjusts the constant false alarm rate threshold to suppress clutter. [Findings] The results showed that the root mean square error of the method was 4.2±0.3 m,the trajectory integrity rate was 96.5% ±1.2%, and the detection probability was 0.88±0.03 under the 6-level sea state;In the high-frequency maneuver scenario,the lag time of model conversion is less than 0.3 s,and the tracking accuracy is significantly better than that of single model and traditional tracking scheme. [Conclusions] This method can provide effective technical support for ship traffic safety under adverse sea conditions.

Identification of urban functional zones based on a multi-modal fusion framework

Zhang Yongchuan, Gao Jie, Zhang Zhiqing, Zhou Zhengqiang, Wang Luxiao, Guan Dongjie

2026, 0(8):  96-102.  doi:10.13474/j.cnki.11-2246.2026.0814

Asbtract ( )   PDF (3035KB) ( )  

References | Related Articles | Metrics

[Purposes] To address the limitations of traditional urban functional zone (UFZ)identification methods，such as insufficient multi-source data fusion and limited feature representation under complex terrain conditions. [Methods] This study proposes a deep learning-based multimodal fusion framework.The model extracts heterogeneous features from Sentinel-2 imagery，POI data，and DEM information.An auxiliary weighting layer (AWL)is designed to adaptively evaluate the contribution of each auxiliary data source.A dual-branch attention module (DBAM)dynamically balances the influence of remote sensing and auxiliary data through a cross-modal attention mechanism.Additionally，a multi-modal feature fusion module (MFFM)is constructed by integrating depthwise separable convolutions with Transformer encoders to enhance cross-modal semantic representation and improve classification of urban areas. [Findings] Experiments conducted in Nan'an district，Chongqing，using 2019 labeled samples across nine UFZ categories demonstrate that the proposed method achieves an overall accuracy of 88.02% and a Kappa coefficient of 0.847. [Conclusions] These results confirm the effectiveness of the framework in UFZ classification for complex urban environments．

Evaluation of landscape spatial quality in traditional villages driven by visual perception

Yao Xuhui, Li Ruijun, Hou Xiaoyu

2026, 0(8):  103-111.  doi:10.13474/j.cnki.11-2246.2026.0815

Asbtract ( )   PDF (25421KB) ( )  

References | Related Articles | Metrics

[Purposes] Landscape visual quality evaluation is of great significance for the protection and development of traditional villages' landscape spatial quality and features.How to establish the association between public landscape preference and landscape characteristics has become research focuses. [Methods] This study proposes to acquire physiological data through eye-tracking,combined with questionnaire surveys to obtain subjective preference scores.Image segmentation and the CatBoost-SHAP model were employed to analyze the contribution of each landscape element to visual preference. [Findings] Historical and cultural landscapes received the highest preference scores.Fixation hotspots were concentrated on artificial elements,among which architectural elements were the most favored.Saccade frequency and average pupil diameter showed a highly significant effect on preference (P<0.001),while fixation frequency showed a significant effect (P<0.01).Architectural features and the textural characteristics of raw-soil features were identified as core elements,and the artificial-to-natural ratio was also an important factor.Architecture promoted landscape preference,whereas non-intrinsic features exhibited a nonlinear effect with a certain threshold range,beyond which an inhibitory effect occurred. [Conclusions] The combination of eye-tracking analysis,subjective preference scores,and the CatBoost-SHAP model can effectively reveal the association mechanism between public preference and landscape characteristics,providing decision-making references for the adaptive protection and development of traditional villages.

Calibration method for hyperspectral line-scan camera mounting parameters based on integrated registration of high-precision point clouds and image-based point clouds under the same viewing geometry

Yin Yuting, Wang Ruiqing, Wei Zhanying, Zhang Weihong

2026, 0(8):  112-116.  doi:10.13474/j.cnki.11-2246.2026.0816

Asbtract ( )   PDF (7871KB) ( )  

References | Related Articles | Metrics

[Purposes] Hyperspectral cameras are important instruments for low-altitude data acquisition,and accurate calibration of camera mounting parameters is essential for generating high-precision DOMs.A high-precision calibration method is proposed. [Methods] By dynamically adjusting the mounting parameters,the exterior orientation elements of line-scan imagery are recovered in near real time,and image-based point clouds are generated.Under the same viewing geometry,these point clouds are registered with known high-precision point cloud data through integrated feature matching,enabling high-accuracy calibration of the camera mounting parameters. [Findings] The calibrated parameters were applied to image rectification and DOM generation.The resulting DOMs exhibited no obvious stitching errors between adjacent flight strips,eliminating the need for secondary processing and significantly improving operational efficiency. [Conclusions] Practical applications demonstrate that the proposed method is feasible and effective.The method can also be extended to the calibration of mounting parameters for other line-scan cameras．

Hyperspectral image classification algorithm based on multi-feature dynamic ensemble

Xu Hongxin, Yu Yao

2026, 0(8):  117-124,136.  doi:10.13474/j.cnki.11-2246.2026.0817

Asbtract ( )   PDF (3823KB) ( )  

References | Related Articles | Metrics

[Purposes] In order to use spectral information and spatial information in hyperspectral remote sensing images，a multi-feature dynamic ensemble method (MDE)based on hyperspectral remote sensing is proposed. [Methods] Firstly，the algorithm extracts spectral features，Gabor features，LBP features and EMAPs features of hyperspectral remote sensing images.Then,according to the characteristics of each test sample，the best feature prediction results are dynamically selected to participate in the integrated decision-making.Finally，two datasets of AVIRIS (airbone visible infrared Imaging spectrometer)sensor are used to evaluate the performance of the proposed algorithm. [Findings] The results show that the overall accuracy of MDE algorithm is up to 96.89% in Salinas dataset and 94.99% in Indian Pines dataset. [Conclusions] Compared with other methods，MDE can provide excellent and stable classification results．

Classification optimization algorithm for airborne LiDAR point cloud in complex terrain

Wang Tian, Li Qianli, Wang Xiping, Zhang Ningli, Wu Yanping

2026, 0(8):  125-129.  doi:10.13474/j.cnki.11-2246.2026.0818

Asbtract ( )   PDF (2167KB) ( )  

References | Related Articles | Metrics

[Purposes] This paper aims to address the technical bottlenecks of automatic classification in high-precision DEM production under complex terrain,including suboptimal data quality,unreasonable filtering effects,and excessive manual editing workload. [Methods] A dual-feature coupled point cloud classification optimization (DFC-PCACO)algorithm is proposed,which takes terrain geometric and point cloud spatial distribution features as dual core constraints.Adopting the process of “terrain classification-differentiated processing-closed-loop optimization”,it integrates multi-dimensional sub-features under the dual-feature framework to realize the secondary optimization of classification results. [Findings] Validated by experimental data from flat and densely forested mountainous areas,the DFC-PCACO algorithm significantly improves point cloud classification accuracy compared with automatic classification： total error is 2.37% in flat areas,11.92% in densely forested mountainous areas,and manual editing efficiency is increased by 35%～40%. [Conclusions] The DFC-PCACO algorithm provides an efficient technical solution for large-scale high-precision DEM production in complex terrain areas．

A water body segmentation algorithm integrating global-local contextual information

Long Fei, Wang Wenhui, Shi Yifan, Lin Siqun, Liu Wenzhuang, Shao Lin

2026, 0(8):  130-136.  doi:10.13474/j.cnki.11-2246.2026.0819

Asbtract ( )   PDF (3420KB) ( )  

References | Related Articles | Metrics

[Purposes] This study proposes a deep learning-based water body segmentation algorithm integrating global-local contextual information,aiming to improve the accuracy of water body extraction in complex environments. [Methods] A high-resolution remote sensing image dataset containing multiple types of water bodies was constructed,and a PCT-Net model was developed.By introducing an interactive attention-based feature fusion mechanism and a global-local contextual information fusion strategy,the proposed model enhances discrimination capability and segmentation stability in complex water body scenarios,including dark backgrounds,dense pond distributions,and water-background confusion.In addition,a semantic-aware dynamic upsampling module was designed to effectively alleviate boundary blurring and insufficient detail recovery of small-scale water bodies in high-resolution remote sensing imagery. [Findings] Experimental results demonstrate that the proposed PCT-Net model outperforms representative segmentation models such as U-Net,SeaFormer,DeepLabV3+,and PSPNet,achieving an overall accuracy of 99.08%, a Kappa coefficient of 0.981 3,a false positive rate of 0.91%, and a false negative rate of 0.93%. [Conclusions] The proposed method can be widely applied in water resource monitoring,flood disaster assessment,and related fields,providing technical support for regional sustainable development．

Intelligent cultivated land extraction from ZY1E imagery via dual-backbone deep learning network

Zhou Hong, Zhang Guohe, You Siqi, Wang Ran, Ding Pengfei

2026, 0(8):  137-144.  doi:10.13474/j.cnki.11-2246.2026.0820

Asbtract ( )   PDF (7013KB) ( )  

References | Related Articles | Metrics

[Purposes] Cultivated land is a fundamental resource crucial for ensuring national food security and ecological stability.With the rapid development of high-resolution satellite data,utilizing deep learning technology for high-precision automated identification of cultivated land has become a significant direction in agricultural remote sensing research. [Methods] This study takes the Suqian area of Jiangsu province as the research area.ZY1E images underwent preprocessing operations including radiometric calibration,atmospheric correction,fusion,and geometric registration.A high-quality training sample set was constructed through a multi-type data augmentation strategy.On this basis,a DB-DeepLabV3+semantic segmentation model integrating the dual-backbone networks of EfficientNet-B3 and ResNet-50 is proposed.In this model,the EfficientNet-B3 backbone network is used for efficient multi-scale feature extraction,while the ResNet-50 backbone enhances deep semantic representation capability； the two achieve structural complementarity and boundary optimization within the DeepLabV3+framework. [Findings] Results indicate that the DB-DeepLabV3+model demonstrates high accuracy and stability in cultivated land identification,achieving an overall accuracy (OA)of 90.91%, a Kappa coefficient of 0.818,and an intersection over union (IoU)of 0.832,outperforming the single-backbone DeepLabV3+,EfficientNet-B3+DeepLabV3+,and ResNet-50+DeepLabV3+models.The model maintained good performance consistency under different data split ratios,verifying its strong generalization ability. [Conclusions] The research findings demonstrate that the dual-backbone structure holds significant advantages for cultivated land extraction from high-resolution remote sensing imagery and provides effective technical support for the application of the domestic ZY1E satellite in agricultural intelligent recognition and land monitoring．

LDN-DETR:lightweight denoising DETR for efficient foreign object detection on power transmission lines

Gao Shuhan, Zhou Chao, Shen Hao, Jia Ran, Liu Hui, Liu Chuanbin, Liu Rong

2026, 0(8):  145-153.  doi:10.13474/j.cnki.11-2246.2026.0821

Asbtract ( )   PDF (3731KB) ( )  

References | Related Articles | Metrics

[Purposes] Object detection methods serve as the core technology in the process of detecting foreign objects on transmission lines, bringing significant convenience to line inspection. However, due to the complex and variable environment of transmission lines, existing methods still face issues such as slow inference speed, poor detection performance for occluded objects, and low precision for small object detection. To address these issues, research on the RT-DETR object detection method was conducted, leading to the proposal of LDN-DETR (lightweight denoising-DETR) for foreign object detection on transmission lines. [Methods] Firstly, a lightweight feature extraction network based on variable kernel convolution was constructed to enhance detection performance for objects of various shapes and sizes, while improving inference speed. Subsequently, a cross-scale feature fusion module based on region-sensitive attention was designed to improve detection accuracy for occluded objects. Finally, a Gaussian modulation-based IoU-aware query denoising training method was adopted to address the issue of poor detection precision for small objects. [Findings] To prove the effectiveness of the method, experiments were conducted in this paper on RailFOD23 and VisDrone 2019. The results show that LDN-DETR has a 4% improvement in mAP over RT-DETR on RailFOD23, a 3% improvement on VisDrone 2019, and an inference speed of 35 frames per second with post-processing. [Conclusions] This method outperforms various mainstream object detection methods, thereby meeting the needs for inspection of foreign objects.

Intelligent detection of subsurface defects in ground penetrating radar images based on an improved YOLO11 network

Huang Bin, Shi Guojie, Du Li, Wang Linke, Han Muyang, Peng Hongwei, Li Kefan, Li Jiaxi, Wang Jinguo, Hou Zhaoyang

2026, 0(8):  154-160,173.  doi:10.13474/j.cnki.11-2246.2026.0822

Asbtract ( )   PDF (6367KB) ( )  

References | Related Articles | Metrics

[Purposes] This paper aims to address the strong noise interference，blurred target edges，and high model complexity in the intelligent detection of ground penetrating radar (GPR)B-scan images. [Methods] This work proposed a lightweight underground defect detection method based on an improved YOLO11 framework.This approach integrates a dynamic convolution and adaptive down-sampling modules into the network architecture of YOLO11 to enhance feature extraction capability while preserving critical low-frequency information. [Findings] The experimental results demonstrate that the improved model of YOLO11_DA achieves a precision of 91.3%，recall of 86.1%，mAP _0.5 of 87.7%，and mAP _0.5：0.95 of 58.8%，while reducing parameters to 2.61×10⁶ and achieving an inference speed of 429.56 frames per second. [Conclusions] Compared to benchmark models including YOLOv5，YOLOv8，YOLOv10，and the original YOLO11，the proposed method maintains lightweight characteristics while exhibiting superior robustness and generalization capability in complex noise environments，weak echo conditions，and multi-scale target scenarios，which has provided an effective technical solution for intelligent detection of underground road defects．

A lightweight approach research to bridge BIM models based on improved QEM algorithm

Wang Qingguo, Wan Xiaowan

2026, 0(8):  161-167.  doi:10.13474/j.cnki.11-2246.2026.0823

Asbtract ( )   PDF (3118KB) ( )  

References | Related Articles | Metrics

[Purposes] Complete bridge BIM models typically contain a large amount of detailed information，leading to difficulties in real-time visualization，transmission，and processing.Lightweight processing of bridge BIM models is a practical necessity to achieve efficient model visualization，transmission，and processing operations.Based on this,This paper proposes a lightweight method for bridge BIM models based on an improved QEM algorithm. [Methods] The method addresses issues such as loss of feature information，missing boundary data，and the generation of narrow triangular surfaces encountered when applying the classic QEM algorithm for lightweight processing of bridge BIM models.Improvements are made by refining boundary collapse conditions and optimizing the processing method for triangular patch contraction. [Findings] Experimental results on lightweight continuous girder bridge BIM models show that the improved QEM algorithm significantly reduces model data volume while effectively suppressing the occurrence of surface ruptures and better preserving key geometric features. [Conclusions] This enhances data transmission efficiency and improves the visual quality of the model．

Blind zone correction method for nearshore measurement of single beam unmanned vessels with slope constraints

Chu Fuyu, Zang Binggui, Zhang Xinyi, Kang Jianrong, Hu Haifeng, Xiang Tao, Wang Chuanguang

2026, 0(8):  168-173.  doi:10.13474/j.cnki.11-2246.2026.0824

Asbtract ( )   PDF (1272KB) ( )  

References | Related Articles | Metrics

[Purposes] Single beam unmanned vessels serve as critical tools for water depth measurement and underwater topographic mapping.To address the challenges of incomplete data acquisition and the risk of hull grounding during nearshore operations，this study proposes a measurement path optimization method incorporating slope constraints. [Methods] The method establishes a mathematical model that considers the gradient of the bank slope，the geometric parameters of the vessel，and the detection range of the transducer to calculate the optimal safe distance between the hull and the shoreline.By integrating the river centerline and water level line，the measurement path is optimized to balance operational safety and data completeness. [Findings] The river cross-section measurement experiments results indicate that with fixed vessel dimensions and transducer detection limits，the slope of the bank is the primary factor influencing the safe distance.Compared to traditional measurement paths，the optimized path enables dynamic adjustment of the safety distance based on slope variations.Compared with the original measurement results，the data volume collected along the optimized path has increased by 0.039% to 0.797%，enabling more comprehensive data acquisition and effectively addressing the issue of data omission associated with traditional single beam sounding in complex terrains. [Conclusions] This method effectively balances hull safety and data completeness，making it suitable for nearshore environments with non-vertical slopes.It offers advantages such as low hardware cost and high operational efficiency，providing a scalable theoretical framework for nearshore water depth inversion using single beam unmanned vessels.Future research may integrate unmanned vessels with 3D modeling modules for real-time slope terrain analysis and employ artificial intelligence for nearshore underwater data prediction．

Improved large-scale real-scene 3D reconstruction method for coastal ancient villages based on CityGaussianV2

Liang Jinbao, Chen Jiai

2026, 0(8):  174-179.  doi:10.13474/j.cnki.11-2246.2026.0825

Asbtract ( )   PDF (5445KB) ( )  

References | Related Articles | Metrics

[Purposes] Aiming at the problems of complex street layout,intricate architectural structures and variable illumination in waterfront areas of coastal historic villages,which easily cause holes,rough edges and noise artifacts in large-scale 3D reconstruction,this paper takes Houguan village in Fuzhou as the research area and proposes an improved large-scale realistic 3D reconstruction framework based on CityGaussianV2. [Methods] Multi-view images are acquired via five-direction oblique photography using UAV.The prior weight of building edges is adopted to guide the splitting candidate optimization,so as to improve the reconstruction accuracy of building contours and component boundaries.Adaptive kernel density estimation（AKDE） is introduced for point cloud denoising to suppress outliers and noise artifacts. [Findings] Experimental results indicate that the overall RMSE values of Octree-GS,Mip-Splatting,LightGaussian and CityGaussianV2 are 4.2,4.0,3.7 and 4.6 cm respectively.The overall RMSE of the proposed method is 2.8 cm,with an error reduction ranging from 24.3% to 39.1% compared with all comparison methods. [Conclusions] The proposed method can provide a technical solution for large-scale and high-precision 3D modeling of historic villages,and satisfy the demands of digital archiving and smart tourism．

A rainfall-induced landslide early warning system integrating 3D monitoring and multi-model analysis

Deng Yuxin, Chen Guojun, Chen Jiai, Kong Qiuping, Nie Wen

2026, 0(8):  180-186.  doi:10.13474/j.cnki.11-2246.2026.0826

Asbtract ( )   PDF (7026KB) ( )  

References | Related Articles | Metrics

[Purposes] To address the insufficient integration of multi-source data in rainfall-induced landslide early warning,this study developed an early warning system integrating 3D monitoring and multi-model analysis with Longyan city,Fujian province as the study area. [Methods] Based on 635 landslide events and corresponding rainfall data from 1984 to 2012,an EI-D rainfall threshold curve was established with a validation accuracy of 93%.A regional landslide susceptibility model was constructed using the FR-SVM method,yielding an AUC of 0.85.UAV photogrammetry and a six-camera array were employed for 3D slope reconstruction and deformation monitoring.FLAC3D fluid-mechanical coupling simulation revealed that the factor of safety decreased from 1.676 to 1.230 within five days of typhoon-induced rainfall,indicating that rainfall duration is a critical variable for slope instability. [Findings] The integrated early warning system achieves multi-level synergy from regional susceptibility assessment to individual slope deformation monitoring. [Conclusions] This paper provides technical reference for landslide prevention and control in similar rainfall-prone areas．