Table of Content

25 April 2025, Volume 0 Issue 4

Previous Issue   

Visual-inertial odometry localization method based on NCC dynamic covariance adjustment

SUI Xin, BAI Jianzhou, WANG Changqiang, SHI Zhengxu, GAO Song, ZHAO Hongchao

2025, 0(4):  1-8.  doi:10.13474/j.cnki.11-2246.2025.0401

Asbtract ( )   PDF (2384KB) ( )  

References | Related Articles | Metrics

Visual-inertial odometry (VIO) is widely applied in UAV and robotics navigation. However, existing VIO systems often struggle with robustness when feature point matching quality is inconsistent. Most VIO algorithms assume a constant covariance matrix for the noise term in the observation model, overlooking variations in feature point matching quality. To address this issue, this paper proposes a VIO localization method based on the multi-state constraint Kalman filter (MSCKF), incorporating normalized cross-correlation (NCC) for dynamic covariance adjustment. This method constructs a new observation model by introducing tracking errors for feature points at the pixel level, with NCC employed as a metric for quantifying feature point matching quality. By calculating the NCC of the feature matches, the method adjusts the covariance matrix of the observation noise dynamically to reflect variations in the matching quality. This approach leads to more accurate and robust results, especially in complex environments with significant differences in feature point matching quality. Comparative experiments conducted on the EuRoC open dataset and a real-world underground parking lot dataset demonstrate that, compared to the traditional MSCKF algorithm, the proposed method reduces planar root mean square error (RMSE) by 41.8% and 33.8%, respectively. Furthermore, compared to the VINS-MONO algorithm, the proposed method reduces planar RMSE by 26.3% and 24.7%, respectively. These results demonstrate a significant improvement in both the robustness and localization accuracy of VIO systems under challenging feature matching conditions.

A visual SLAM method for merging semantic information in indoor dynamic scenes

WANG Yizhe, ZHANG Ruiju, WANG Jian, XIE Xinrui, HUANG Qicheng

2025, 0(4):  9-13.  doi:10.13474/j.cnki.11-2246.2025.0402

Asbtract ( )   PDF (2993KB) ( )  

References | Related Articles | Metrics

Visual SLAM is a core technology for autonomous perception and navigation in intelligent devices, playing a crucial role in AI and robotics. However, traditional visual SLAM algorithms suffer significantly in stability and localization accuracy when scenes contain moving objects. To address this, this paper proposes a SLAM scheme that integrates semantic information for indoor dynamic scenarios. Based on ORB-SLAM2, it introduces the GCNv2 network for deep feature extraction and YOLOv5 for semantic segmentation to identify dynamic objects. Combined with motion consistency analysis, it effectively eliminates dynamic interference, enhancing robustness. Tests on the TUM standard dataset show the improved algorithm significantly outperforms the original ORB-SLAM2 in dynamic indoor environments, with an average positioning accuracy improvement of 55.75%. This result demonstrates the proposed method's effectiveness, significantly boosting SLAM system performance in complex dynamic environments.

Robot visual SLAM algorithm based on multi-feature information localization

FAN Qiliang, DING Dukun

2025, 0(4):  14-19,26.  doi:10.13474/j.cnki.11-2246.2025.0403

Asbtract ( )   PDF (3357KB) ( )  

References | Related Articles | Metrics

The visual simultaneous localization and mapping (SLAM) algorithm has been widely applied in indoor service robots. However, current point cloud, plane-based, and semantic visual SLAM algorithms face issues such as single map structures and inaccurate localization. This paper proposes a multi-layered map construction SLAM (MFIL-SLAM)algorithm based on the classical ORB-SLAM2, incorporating plane and semantic information. The algorithm extracts feature points, planes, and semantic objects from visual and depth images, associates them with map landmarks, updates camera poses, and optimizes multi-layered maps through factor graph optimization. Experimental results demonstrate that the proposed algorithm outperforms existing ones in mapping quality, localization accuracy, and robustness.

Visual inertial navigation algorithm based on key frame selection and loopclosure constraint in complex scenes

HAO Chunting, LIU Fei, WANG Jian, HAN Houzeng, LI Yandong

2025, 0(4):  20-26.  doi:10.13474/j.cnki.11-2246.2025.0404

Asbtract ( )   PDF (2273KB) ( )  

References | Related Articles | Metrics

In order to solve the problem that the error of the previous frame will be propagated to the next frame when the unmanned vehicle moves for a long time in a complex scene, resulting in the error accumulation of the visual inertia odometry, a multi-state constrained Kalman filter visual inertia odometry algorithm based on key frame loopclosure constraints is proposed. Firstly, the pose of the key frame with fixed time interval is preserved to make full use of the image information and limit the state growth effectively.Then, the loop closure detection is carried out using the bag of words model to determine the key frame where the loop closure occurs, and the observations of loop closure constraintsare add to the feature track for measurement update.Finally, validation analysis is performed in both public datasets and real environments. Experimental results show that compared with MSCKF algorithm, the proposed algorithm can effectively reduce the positioning error and get closer to the real motion trajectory,with higher positioning accuracy and better robustness.

Real-time semantic SLAM algorithm based on dynamic scenes

FU Qiang, ZHONG Zhen, JI Yuanfa, REN Fenghua

2025, 0(4):  27-33.  doi:10.13474/j.cnki.11-2246.2025.0405

Asbtract ( )   PDF (3578KB) ( )  

References | Related Articles | Metrics

Aiming at the problems of traditional visual SLAM (simultaneous localization and mapping) in indoor dynamic environment with low localization accuracy, poor robustness and poor real-time performance after combining with deep learning as well as the inability to construct dense maps, this paper proposes an improved algorithm based on ORB-SLAM3. First, a lightweight SegFormer semantic segmentation network is used to identify dynamic objects present in the image, and then a mask image adaptive expansion method is added to automatically adjust the mask expansion range according to the total number of feature points to more effectively remove potential dynamic object feature points; second, the bag-of-words model is improved to enhance the loading and matching speed of the algorithm; and lastly, a dense map building thread is added to construct a map for removing dynamic features according to the mask information and keyframes to construct a dense point cloud map after removing dynamic features. The experimental results show that the algorithm in this paper can effectively remove dynamic object feature points in highly dynamic scenes, and improve the localization accuracy and robustness of the system,and the average processing speed is 20.3FPS, which basically meets the requirements of real-time operation.

Pipeline defect detection method for floating capsule robot image

ZHU Song, CHEN Antai, HUA Yuansheng, JIANG Wenyu, YUAN Pengpeng, ZHU Jiasong

2025, 0(4):  34-38.  doi:10.13474/j.cnki.11-2246.2025.0406

Asbtract ( )   PDF (2732KB) ( )  

References | Related Articles | Metrics

The safe operation and maintenance of urban underground water supply and drainage systems are crucial for the sustainable development of society and the economy. The drifting capsule robot, as a new type of automated inspection tool, can effectively address the high costs and low efficiency of traditional methods. However, factors such as water flow disturbances inside the pipeline and the self-heating of the equipment can cause water mist interference in the images collected by the capsule robot, severely affecting the accuracy of defect identification. Therefore, this paper designs a lightweight defect detection network based on dual-branch feature fusion dehazing to improve the image quality and defect identification accuracy of low-quality capsule robot images. The dual-branch dehazing module adaptively extracts and fuses the spatial structure and spectral features of the image, enhancing the image's dehazing performance. The lightweight detection module, with YOLOv5 as the backbone, identifies and locates pipeline defect types and damaged areas in the dehazed images. Experimental results on the SFCRI dataset show that the dehazing module improves the SSIM and PSNR of the images by 0.271 and 24.04, respectively, and increases the average precision of defect identification by nearly 12%, achieving a recognition speed of 120.3 frame/s. The proposed lightweight defect detection network based on the dual-branch dehazing module can effectively achieve high-efficiency and low-cost defect detection for urban underground water supply and drainage pipelines.

Road speed limit and height limit information extraction from traffic sign in street view images

JI Chen, LIU Lei, CAI Dong, CHENG Liang

2025, 0(4):  39-44.  doi:10.13474/j.cnki.11-2246.2025.0407

Asbtract ( )   PDF (4460KB) ( )  

References | Related Articles | Metrics

Aiming at the problem that the end-to-end detection model has poor effect on few-shot sample traffic signs,this paper constructs a method to firstly detect traffic signs in street view images and then extract the semantic information,as to update the speed limit and height limit attributes of the road where the sign is located. Firstly,the public sample dataset is integrated to train the detection model based on four typical networks. After comparison,the best Faster R-CNN is selected as the sign detection model.Then,the panoramic street view image is segmented to detect the speed limit and height limit signs in the sub-image.Finally,the PP-OCRv3 text detection model is combined to extract the speed limit and height limit values,and the results are cleaned by semantic constraints. The method in this paper is verified by using the public data of the transportation department,and the application experiment is carried out in Tainan city,Taiwan,China. The results show that this method is suitable for few-shotsample traffic signs,which can accurately extract speed limit and height limit information from street view images,and has application value in traffic sign management and speed limit and height limit road attribute verification and update.

Integrated monitoring of permafrost deformation in the Qinghai-Xizang Plateau using machine learning and SBAS-InSAR

YANG Yang, JIA Hongguo, BAI Zhenghang, LIU Yuchen

2025, 0(4):  45-50.  doi:10.13474/j.cnki.11-2246.2025.0408

Asbtract ( )   PDF (2902KB) ( )  

References | Related Articles | Metrics

Affected by global climate change, the degradation of permafrost and surface instability in the Qinghai-Xizang Plateau have been continuously intensifying, posing obstacles to the construction and maintenance of infrastructure as well as regional socio-economic development. In recent years, SBAS-InSAR technology has been widely applied in monitoring surface deformation of permafrost. However, due to severe decorrelation phenomena in some areas of the Qinghai-Xizang Plateau, the deformation monitoring results exhibit spatial discontinuity, making it difficult to obtain comprehensive and detailed monitoring outcomes. To address these issues, this paper proposes a method for monitoring permafrost deformation that integrates machine learning with SBAS-InSAR. Taking the Menshi Township in Ali, Xizang, as the study area, a total of 43 descending Sentinel-1A images from January 7, 2020, to June 6, 2021, are used to extract surface deformation information. After generating a training dataset by integrating multiple environmental factor data, a machine learning model is introduced to fit the intrinsic relationship between SBAS-InSAR monitoring results and environmental factors, thereby obtaining a continuous deformation rate map of the study area. The results indicate that the method combining the random forest model with SBAS-InSAR performs optimally. By interpolating the missing regions of permafrost deformation using this method, the monitoring coverage of the original SBAS-InSAR method can be significantly improved, with an average error and root mean square error of 0.459 and 0.739 mm/a, respectively, for the interpolation results.

Time series analysis and prediction of surface subsidence based on SBAS-InSAR technology

SHAO Yaxuan, MA Jing, HAN Lili, YAO Guanyu

2025, 0(4):  51-57,62.  doi:10.13474/j.cnki.11-2246.2025.0409

Asbtract ( )   PDF (4045KB) ( )  

References | Related Articles | Metrics

Wanbailin district is listed as one of the high-risk areas of geological disasters in Taiyuan city.In order to explore the surface subsidence in Wanbailin district of Taiyuan city,it is difficult to obtain long-term and large-scale surface deformation information by traditional measurement methods. In this paper,Wanbailin district of Taiyuan city is taken as the research area. Based on 36 scenes of Sentinel-1A image data,SBAS-InSAR technology is used to obtain the cumulative settlement and change rate of the study area from July 2017 to June 2020.LSTM neural network model and GM(1,1) model are used to simulate and predict the monitoring results of surface feature points. The results show that: ①The uneven settlement of Wanbailin area in Taiyuan city is more and more serious from east to west. The most obvious area is located near the western factory village,and the maximum settlement rate can reach -60 mm/a; ②The GM(1,1) model cannot effectively predict the surface feature points with large fluctuations during the settlement monitoring period. The LSTM neural network model can better realize the urban settlement prediction,and the prediction accuracy is higher.

Identification and analysis of landslide hazards in Mianning county using fusion of ascending and descending orbit SBAS-InSAR technology

SHANG Yiwei, XIONG Junnan, JIA Qian, LUO Siyuan, WANG Qisheng, CAO Yifan

2025, 0(4):  58-62.  doi:10.13474/j.cnki.11-2246.2025.0410

Asbtract ( )   PDF (6695KB) ( )  

References | Related Articles | Metrics

To address the issues of missed and misclassified landslide hazards using single-track InSAR technology, this paper proposes a method that integrates ascending and descending track data in the SBAS-InSAR technique. Taking Mianyang county, Sichuan province, as an example, a 2D deformation field was derived from 90 ascending and 80 descending track Sentinel-1A images from 2019 to 2021, combined with optical remote sensing imagery, identifying a total of 30 landslide hazards. In the vertical deformation field, the number of identified landslide hazards increased to 34, with 25 known disaster points and 9 new hazards. The results indicate that vertical deformation monitoring has a stronger observational capability for landslide hazard identification, improving accuracy and compensating for the limitations of single-track data. Three typical hazards are selected for spatiotemporal distribution analysis, optical imagery, and rainfall data analysis. The results show that the slope body is unstable and significantly influenced by rainfall, especially during the concentrated rainfall period from June to August, where deformation exhibited an accelerating trend.

Feature optimization and rapeseed lodging recognition based on particle swarm optimization and mutual information

WANG Kexiao, LI Bo

2025, 0(4):  63-67,74.  doi:10.13474/j.cnki.11-2246.2025.0411

Asbtract ( )   PDF (4509KB) ( )  

References | Related Articles | Metrics

To achieve the high-precision extraction of rapeseed lodging by visible remote sensing, this paper put forward a feature optimization and rapeseed lodging recognition algorithm based on particle swarm optimization-mutual information (PSO-MI) and stacking ensemble learning by the visible vegetation indices, principal component band texture features, and DSM of the experimental area. This paper proposed the particle swarm random loop search method, introduced the mutual information threshold adaptive strategy, and built the applicability function to screen the lodging characteristics of rapeseed. A stacking ensemble learning model based on optimized features was constructed using logistic regression (LR) meta learner combined with four base learners, namely, K-nearest neighbor (KNN), random forest (RF) and support vector machine (SVM), which realized the identification of lodging rapeseed and mapping, and effectively improved the accuracy of lodging rapeseed recognition. This study provides a target recognition method combining stochastic intelligent feature optimization and high-performance machine ensemble learning, which can provide a technical reference for remote sensing extraction of target features.

Extraction of gobi desert gravel layer based on U-ConvHDNet

MA Yubo, ZHANG Aiguo, WANG Haoyu, LIU Shuaiqi, JIN Jingyu, SHEN Zhanfeng, LI Junli

2025, 0(4):  68-74.  doi:10.13474/j.cnki.11-2246.2025.0412

Asbtract ( )   PDF (5112KB) ( )  

References | Related Articles | Metrics

The gravel layer is an essential component of the gobi desert ecosystem. Conducting large-scale remote sensing monitoring of the gravel layer is of great significance for protecting the gobi desert ecology. In response to the loose structure and strong heterogeneity of the gravel layer, this paper proposes an automatic information mapping method for the gravel layer based on the U-ConvHDNet semantic segmentation model. This method utilizes Sentinel-2 imagery from the entire Hami region captured in August 2023 to extract information on gobi gravel layer. The results indicate that the F1 score of the U-ConvHDNet model is 0.918, which is superior to that of the other seven semantic segmentation models. Ablation experiments demonstrate that the combined use of the improved backbone network, upsampling and downsampling modules effectively enhances the accuracy. The dual receptive field sliding window strategy optimizes the instability near stitching lines, enabling the extraction of the total area of the gobi gravel layer in Hami at 1.026×10⁵ km², with an information extraction precision of F1 score 0.921. This study provides technical support for monitoring of gobi gravel layer and the management of gobi ecosystems.

Chemical oxygen demand inversion in typical coastal breeding areas in Fujian based on Sentinel-2 data

CHEN Baofeng, CHEN Yunzhi, CHEN Hongmei

2025, 0(4):  75-81.  doi:10.13474/j.cnki.11-2246.2025.0413

Asbtract ( )   PDF (4633KB) ( )  

References | Related Articles | Metrics

Chemical oxygen demand(COD) is an important indicator for evaluating the ecological and eutrophication levels of the water environment. Timely monitoring of COD concentrations in coastal waters is significant for marine environmental protection. This study utilizes field-measured data and Sentinel-2 MSI satellite remote sensing image data to identify the best band combination for inverting COD. It applies an empirical model based on statistical analysis and a machine learning model for inversion. The optimal inversion model for COD, applicable to Zhao'an Bay and Dongshan Bay in Fujian province, is determined. The spatiotemporal characteristics of water quality in these areas are also analyzed. The results show that the COD index regression model, constructed using the band reciprocal difference (BRD) combination, is the best inversion model for Zhao'an Bay and Dongshan Bay. This model achieves a determination coefficient R² of 0.82, a mean square error (MSE) of 1.85%, and a mean absolute percentage error (MAPE) of 11.55%. The inversion results indicate that the COD concentration in the study area remains relatively stable from 2017 to 2022, with high-value areas concentrated in nearshore regions and river estuaries. The COD concentration decreases in 2023, which is attributed to changes in aquaculture density resulting from local ecological protection measures implemented between 2022 and 2023.

Building recognition in transmission line corridors based on high-resolution images and improved YOLOv7 model

YANG Guozhu, SUN Shirui, TIAN Maojie, SUN Huamin, HU Wei, LI Junlei

2025, 0(4):  82-89.  doi:10.13474/j.cnki.11-2246.2025.0414

Asbtract ( )   PDF (7348KB) ( )  

References | Related Articles | Metrics

Early disaster warning and safety assessment in transmission line corridors are among the priorities of smart grid construction. Therefore, it is very important to grasp the location and distribution of settlements in the transmission line corridor area to do a good job of disaster prevention and mitigation in mountainous areas. In recent years, with the continuous development of target detection technology, the fields it is applied to are becoming more and more extensive, and remote sensing target detection, as one of the application scenarios, is widely used in building information because of its wide coverage and the characteristics of covering many targets. The existing deep learning models have limitations regarding both recognition accuracy and detection speed in building identification and segmentation. Aiming at such problems, this study takes the GF-2 image as the database, labels the buildings in mountainous areas, establishes the sample dataset and divides it into the training set and the test set according to the ratio of 9∶1. Secondly, the standard version of YOLOv7 network is improved by adding a dual-attention module with GAM-CBAM synthesis in the neck part to reduce the feature loss of buildings, which improves the detection ability of the network. The results show that the improved YOLOv7 network achieves an average precision of 88.74% for segmentation and recognition of buildings in mountainous areas, which is also higher than other deep learning models in terms of precision and recall. Therefore, this method can provide data support for rapid and efficient acquisition of mountainous area settlement information, geographic information analysis in the process of disaster prevention and mitigation in mountainous areas, and the development of emergency plans.

Improved feature pyramid pooling for obstacle rxtraction in remote sensing images

SUN Kai, XU Qing, ZHANG Ruixin, SU Youneng

2025, 0(4):  90-95.  doi:10.13474/j.cnki.11-2246.2025.0415

Asbtract ( )   PDF (3176KB) ( )  

References | Related Articles | Metrics

The extraction of obstacles from high-resolution remote sensing images is one of the crucial bases for off-road path planning, as accurate obstacle locations can significantly reduce transit costs. Traditional surveying methods for obstacle extraction are inefficient and susceptible to human factors and terrain influences, making them unsuitable for complex battlefield environments. Current deep learning methods face issues such as feature loss and inadequate resolution when extracting obstacles like residential areas and water systems, particularly struggling with precision in identifying small-scale features, resulting in outputs that fail to meet requirements. To address these challenges, a method utilizing a feature pyramid attention network (ResT-PNet) for extracting features from remote sensing images has been proposed in this paper.Employing a feature pyramid pooling module to obtain global semantic information.Firstly, a feature fusion module has been constructed to integrate feature information across different scales, enhancing the feature extraction efficacy. Then, spatial and channel attention mechanisms have been introduced to minimize the loss of detail information and to integrate local and global features. Finally, comparative experiments and model applicability validation have been conducted. The results indicating that the proposed model achieves higher accuracy and better distinguishes small-scale obstacles, thereby providing support for off-road path planning.

The water level change of Namco Lake based on ICESat-2 height difference data

LIU Mingqi, CHEN Guodong, ZHANG Zhijie

2025, 0(4):  96-101.  doi:10.13474/j.cnki.11-2246.2025.0416

Asbtract ( )   PDF (2158KB) ( )  

References | Related Articles | Metrics

Lake water level monitoring is one of the important application scenarios of satellite altimetry technology. However, most studies on lake water level monitoring using satellite altimetry technology do not take into account the impact of geoid residuals on the results. To address this issue, this study selects ATL13 data from ICESat-2 from October 2018 to August 2023 to analyze the water level changes of Nam Co Lake using the height difference method. The results show that the water level changes of Nam Co Lake obtainedby using the intersection method and repeated orbit data can effectively reduce the impact of geoid residuals and improve the accuracy of water level monitoring, with a correlation coefficient of up to 0.92 with the measured water level. Through the analysis of the water level changes of Nam Co Lake, it is found that its water level rises in summer every year and is relatively stable in winter, with an overall upward trend. This study proves the accuracy of detecting lake water levels by combining satellite altimetry intersections and repeated orbit data, providing reliable data support for hydrological research on inland lakes.

River level monitoring correction model based on ICESat-2

WANG Ruikun, HE Rong, WU Leigang

2025, 0(4):  102-108.  doi:10.13474/j.cnki.11-2246.2025.0417

Asbtract ( )   PDF (2456KB) ( )  

References | Related Articles | Metrics

In light of the challenge in establishing a time series of water level for hydrological research caused by the slope of the river surface and the considerable distance in satellite monitoring data, this paper puts forward a correction approach to construct a river level correction model based on ICESat-2 altimeter data. The global altimeter data of ICESat-2 from 2018 to 2023 is employed. Taking the lower reaches of the Yellow River as the research area, a river level correction model is developed to compensate for the ground track variations of the ICESat-2 altimeter satellite. In combination with the monorail data of Sentinel-3A/B from 2016 to 2024, the water level time series at four hydrological stations in the study area is constructed, and the accuracy verification and change analysis are conducted. The results demonstrate that the accuracy of the multi-orbit data of ICESat-2 corrected by the corrected model is comparable to that of the Sentinel-3A/B monorail data, with the maximum correlation coefficient being 0.96 and the minimum root-mean-square error being 0.23 and 0.21 m respectively. The average time resolution of the constructed water level time series can reach 13 days. From 2016 to 2024, the water level in the Henan section of the lower reaches of the Yellow River exhibits a seasonal variation law of rising first and then decreasing, and rainfall and evaporation are the main factors causing the water level change.

Spatio-temporal analysis of surface subsidence along the Jinan subway line based on SBAS-InSAR technology

ZHOU Tianxiang, DONG Jinglong, SHEN Qiang

2025, 0(4):  109-113.  doi:10.13474/j.cnki.11-2246.2025.0418

Asbtract ( )   PDF (4815KB) ( )  

References | Related Articles | Metrics

In the process of construction and operation of Jinan subway, there are different degrees of subsidence, and it is of great significance to carry out subsidence monitoring in Jinan subway and its surroundings to ensure urban safety. In this paper, the deformation information of Jinan subway Lines 1, 2, and 3 is obtained by using SBAS-InSAR technology from 30 Sentinel-1A image data from March 2022 to April 2024. Combined with the Peck formula, the obtained results are modeled, and the variation characteristics of the settlement trough are explored. The results show that there are different degrees of deformation along the Jinan rail transit, and spatially, the subway in the eastern and central regions is relatively stable, and Line 1 is the most stable. The settlement near Yaoqiang airport station is the largest, with a maximum average annual deformation rate of -66 mm/a. In terms of time, the settlement during the construction of the subway is greater than that during the operation, and the settlement of the settlement trough near Yufuhe station, Lishan road station, Chuanliu station and Yaoqiang airport station in the center shows a gradual increasing trend. The radii of the settlement trough are 174, 151, 171 and 350 m, respectively.

Improved region merging algorithm combining elevation information with random forest model

WANG Rongkang, XIONG Junnan, TANG Haoran, TU Caisen, SONG Nanxiao

2025, 0(4):  114-119.  doi:10.13474/j.cnki.11-2246.2025.0419

Asbtract ( )   PDF (6826KB) ( )  

References | Related Articles | Metrics

With the wide application of object-oriented image analysis, image segmentation plays an important role in remote sensing image processing. At present, many image segmentation algorithms are based on region merging method, but these methods generally face the problem of limited feature scale, relying on a single optical image feature and subjective parameter setting, which limits the segmentation effect. To solve this problem, this paper proposes a machine learning region merging method that takes into account the elevation feature merging strategy. In this paper, the machine learning model based on random forest (RF) is used to assist the elevation feature merging strategy, and the region merging classifier is constructed by calculating the feature matrix of the adjacent region as the input feature variable. Transform the region merging problem into a classification problem of 0 and 1. Experimental results show that the region merging algorithm with 0.5 m spatial resolution elevation features achieves excellent segmentation results, with F1, accuracy, recall and crossover ratio reaching 90.5%, 89.98%, 91.02% and 82.64%, respectively. Compared with no elevation features, the proposed algorithm effectively improves segmentation accuracy. It is increased by about 3.4%, 6.8%, 1.1% and 6.2% respectively. Meanwhile, the importance of elevation features reached 32.5%, which is about 7% higher than that of optical features.

Analysis of the relationship among road navigation attributes based on multi-model features of road networks and crowd-sourced trajectories

ZHANG Caili, XIANG Longgang, LI Yali, ZHOU Yushi, LIU Zhen, LU Chunyang

2025, 0(4):  120-126.  doi:10.13474/j.cnki.11-2246.2025.0420

Asbtract ( )   PDF (2293KB) ( )  

References | Related Articles | Metrics

Geometrical and topological information about the road network is certainly important,but navigation attribute information such as road classes,number of lanes,and speed limits is also essential for the implementation of core road network applications,and route planning,vehicle navigation,and location services are typical cases. This research explores the hierarchical relationship among these three attributes and proposes potential multi-modal progressive classification methods considering upstream and downstream information for predicting road classes,number of lanes,and speed limits of road sections. First,we preprocessed trajectories and road networks and realized the connection between track points and road sections; then,we took a road section as the analysis unit and mine the multivariate and multi-order features of road networks and crowd-sourced trajectories; finally,potential methods are summarized and analyzed,and these methods are based on random forest algorithm that integrate these complementary features of current and adjacent road sections and considers hierarchical information to identify road classes,number of lanes,and speed limits based on the voting method. Experimental results in Wuhan and Xian show that our exploration has a certain reference value.

Dual attention for power corridor point cloud semantic segmentation

LI Jian, WANG Jian, WANG Lei, LI Min, YANG Like, ZHAO Yilong

2025, 0(4):  127-133.  doi:10.13474/j.cnki.11-2246.2025.0421

Asbtract ( )   PDF (4365KB) ( )  

References | Related Articles | Metrics

The point cloud scene of power corridors is unique and faces a number of challenges when using deep learning methods for semantic segmentation.There is a serious data skew problem in the scene, with background objects such as ground and vegetation occupying the majority, which may lead to degradation of network performance. In addition, when dealing with point cloud data of overhead power lines and power towers, the model has difficulty in extracting enough local features due to the insufficient number of points within the local radius, which reduces the segmentation accuracy of similar objects. For this reason, this paper designs a two-order semantic transmission line semantic segmentation method based on the dual attention mechanism. Firstly, in the data streamlining stage, the elevation difference between the power transmission equipment and the background is utilized to have remove a large number of background points, thus reducing the training burden of the subsequent neural network, accelerating the training process, and significantly alleviating the data skew problem. Then, the dual-attention model, which takes into account both global and local features, is proposed to enhance the differentiation of similar objects and improves the accuracy of point cloud segmentation. After testing, the data streamlining method in this paper can remove more than 63% of background points and partially solve the data skewing problem; the proposed dual-attention network outperforms other methods for segmentation of ground wires, conductors and insulators.

Motion amplification extraction method for vibration modes of power grid facilities based on principal component analysis

ZHANG Ke, TONG Yang, HUANG Wenli, HU Shang, TAO Tingye, DAI Ju

2025, 0(4):  134-138,151.  doi:10.13474/j.cnki.11-2246.2025.0422

Asbtract ( )   PDF (2366KB) ( )  

References | Related Articles | Metrics

In the safe operation and inspection of power systems, modal health monitoring of power infrastructure is very important. Due to the advantages of high precision, high efficiency, flexibility and reliability, the machine vision method has become an important means of structural health monitoring. In this paper, a vibration modal extraction method based on the principal component analysis and blind source complexity tracking method, aiming to solve the problem of frequency passband empirical setting in the machine vision motion magnification method, is proposed, which can extract the micro motion signals automatically. Meanwhile, the mean subtracted contrast normalized (MSCN) coefficient is proposed as an evaluation index of image quality, which can identify and remove the blurry image in real-time and exactly. It is verified by the experiments on the overhead cable support system that the performance of the proposed method show consistency with that of the vibration accelerometer data, and the absolute error of the extracted vibration frequency is within 0.2 Hz. This indicates that the proposed method in this paper can extract the high precision micro motion signals of the power facilities in real-time, thus can provide high-quality structural modal monitoring data for the automatic operation and inspection of power systems.

A Brovey fusion method based on image histogram matching

LIU Chaoqun, YU Shunchao, GU Zhujun, HE Yingqing, ZHAO Min, HE Qiuyin

2025, 0(4):  139-144.  doi:10.13474/j.cnki.11-2246.2025.0423

Asbtract ( )   PDF (3901KB) ( )  

References | Related Articles | Metrics

The paper discusses a Brovey fusion method based on image histogram matching(IHM Brovey fusion). This method solves the defects of Brovey fusion method limiting the number of multi-spectral bands and large variation of spectral characteristics of ground objects in fusion image results. The experimental results show that compared with Brovey fusion image, the average correlation coefficient between each band of IHM Brovey fusion image is reduced by 2.60%, the average information entropy is increased by 0.25%, and the average gradient is increased by 6.02%, which better preserves the rich spectral information of the original multi-spectral image and the fine spatial texture information of panchromatic image. It is found that the spectral offset of surface objects such as water, forest land and buildings in the fusion image results of this method is greatly reduced, especially for the whole remote sensing image of the test area, the spectral offset is as low as 0.46%, and the spectral variation of surface objects in the fusion image of IHM Brovey method is smaller and more stable.

Analysis of spatial and temporal variations and driving factors of ecological environment quality in Huangshi city

CHEN Zimou, WANG Xinchi, GUO Jing

2025, 0(4):  145-151.  doi:10.13474/j.cnki.11-2246.2025.0424

Asbtract ( )   PDF (12030KB) ( )  

References | Related Articles | Metrics

An accurate evaluation of the ecological quality of Huangshi is imperative for its urban transformation. Multi-period Landsat 5 TM, Landsat 8 OLI imagery, and other remotely sensed data products with 5 year intervals are selected through the implementation of a remotely sensed ecological index coupling moisture, greenness, dryness, and heat. This approach is selected to assess the ecological environment quality in Huangshi city. The Moran's index is employed to examine the spatial autocorrelation of ecosystem quality, and geoprobes are utilized to investigate the drivers affecting ecosystem quality.The results show that, the mean value of RSEI shows a trend of decreasing and then increasing, with the highest in 2004, reaching 0.563, and the lowest in 2009, with a value of 0.518.the ecological environment quality in the central and southern part of Huangshi city is more excellent, and the ecological environment quality in Tieshan district and Daye city is worse; and the spatial agglomeration degree of the ecological environment quality in Huangshi city firstly rises and then decreases, with the highest in 2009 and the lowest in 2019; Slope and land use dominate among the factors affecting ecological environment quality in Huangshi city, and the explanatory power of multi-factor interaction is higher than that of single factor. Ecological environmental protection policies in Huangshi have been found to be generally effective, with the quality of the ecological environment undergoing a gradual improvement. However, further efforts are necessary to address land remediation in the entire region and to optimize the spatial structure and layout of the national territory as a whole.

Spatio-temporal distribution of grassland degradation in the Qilian Mountain area from 2013 to 2020 based on Landsat 8 OLI data

TIAN Dan, BAI Xiao, QIN Kun

2025, 0(4):  152-157.  doi:10.13474/j.cnki.11-2246.2025.0425

Asbtract ( )   PDF (3854KB) ( )  

References | Related Articles | Metrics

To clarify the spatio-temporal distribution characteristics of grassland degradation in the Qilian Mountain area and to assess its degradation trends, this study uses Landsat 8 OLI image data in conjunction with geographical national condition monitoring results. By calculating the normalized difference vegetation index (NDVI) and grassland degradation index (GDI) from 2013 to 2020, a quantitative analysis of the spatial and temporal changes in grassland coverage and degradation in the area is conducted. The results show that: ①The grassland coverage area in the Qilian Mountain area is 121 039.75 km², accounting for 79.15% of the total area, with high coverage grasslands accounting for 28.25%.②The grassland degradation index in the Qilian Mountain area decreases from 1.59 in 2014 to 1.27 in 2020, indicating an overall improvement in the temporal trend.③The spatial distribution of degraded grasslands shows significant heterogeneity, with more concentrated degradation in the western region, while the central and eastern regions show no significance. The change trends in the grassland degradation index vary significantly among different watersheds, with the Huangshui/Datong River and Hexi Corridor areas showing a fluctuating downward trend, while the Qaidam Basin area has risen after 2018. The findings provide a scientific basis for assessing grassland degradation and vegetation restoration in the Qilian Mountain area, which is of great significance for regional ecological protection and sustainable development.

Application of UAV aerial survey and remote sensing technology in the improvement and renovation of existing railways: taking the Baotian section of the Longhai Railway as an example

CHEN Fuqiang, WEI Yujun, MA Bangchuang, LI Dan, LIU Yuxin, ZHANG Zhanzhong

2025, 0(4):  158-163.  doi:10.13474/j.cnki.11-2246.2025.0426

Asbtract ( )   PDF (7732KB) ( )  

References | Related Articles | Metrics

On the basis of analyzing the current application status of UAV aerial surveying and remote sensing technology,this article classifies and summarizes the technology. Taking the Baotian section of the Longhai Railway as a case study,the application and technical advantages of UAV aerial surveying and remote sensing technology are detailed from four aspects: UAV aerial surveying and mapping,UAV LiDAR existing railway resurveying,UAV 3D modeling and LiDAR point cloud assisted geological hazard analysis,and UAV aerial photography assisted investigation. Summarize the current shortcomings of this technology in terms of operational efficiency,intelligence level,absolute accuracy,etc. At the same time,it is believed that hardware development,data processing,and expanded applications are the main development directions for its future.The UAV remote sensing technology system adopted in this article can provide technical reference and guidance for the survey and design of similar mountainous railways.

Intelligent extraction method for linear road surface features of complex mountainous highways based on semantic segmentation network and feature matching

ZHANG Kaizhou, MA Ruifeng, JIA Xin

2025, 0(4):  164-169.  doi:10.13474/j.cnki.11-2246.2025.0427

Asbtract ( )   PDF (2383KB) ( )  

References | Related Articles | Metrics

Road boundaries and markings, as crucial linear elements of road surfaces, provide vital support for highway expansion, reconstruction, and intelligent asset management by offering accurate positioning and semantic information. Mobile LiDAR scanning (MLS) point clouds have emerged as a significant data source for identifying and updating road scene elements. Addressing shortcomings of existing methods in complex mountainous highway scenarios, this study proposes a progressive method for intelligent extraction of linear road surface features from vehicle-mounted LiDAR point clouds: Utilizing a deep learning network based on 3D point clouds, semantic segmentation of road surfaces and markings is performed to obtain coarse results. Subsequently, refined extraction and optimization processes are applied to road boundaries and markings. This includes extraction of road contour points, clustering of line structures, completion and vectorization of road boundary points, and semantic enhancement through feature attribute analysis and edge-based feature matching for precise extraction and semantic interpretation of markings. Experimental results demonstrate that the proposed method effectively addresses occlusion and noise challenges in complex mountainous highway scenes, achieving accurate extraction of linear road surface features.

Extraction of skyline based on LiDAR point cloud and statistical analysis of its relationship with building: taking Wuhan as an example

LIU Yanxia, LIU Tao, ZHENG Fengjiao, LIU Ying, YANG Xia, ZHOU Yue, LI Xianju

2025, 0(4):  170-175.  doi:10.13474/j.cnki.11-2246.2025.0428

Asbtract ( )   PDF (3417KB) ( )  

References | Related Articles | Metrics

The skyline is a symbol and concentrated reflection of urban landscape style.With the acceleration of urbanization,its research has important theoretical significance and practical value for urban planning,construction,etc.This article is based on point cloud data from Wuhan city,extracting skylines from 9 key development areas.Three indicators,namely contour shape,degree of building height change,and average turning point of contour lines are selected for quantitative evaluation and analysis.The results are compared with the aesthetic pleasure threshold values of urban skylines in related studies.Overall,the evaluation of urban skylines shows that some skyline shapes are very pleasant,while others evoke less intense aesthetic feelings.Finally,by combining building data within the skyline area,statistical analysis is conducted on the relationship between quantitative indicators and the distribution of buildings with different purposes.The results indicate that there is a strong correlation between the turning point and the proportion of public facilities and residential buildings,with absolute correlation coefficients ranging from 0.39 to 0.83.The plot ratio has a strong correlation with the proportion of commercial and service buildings,with correlation coefficients ranging from 0.38 to 0.43.The research results of this article provide highly current skyline data for Wuhan city,provide quantitative basis for urban planning and design,and can also provide reference for skyline extraction and analysis of other cities.

Preliminary study on machine map design for unmanned platform

MA Chao, WANG Qiang, JIANG Danni

2025, 0(4):  176-178,183.  doi:10.13474/j.cnki.11-2246.2025.0429

Asbtract ( )   PDF (1058KB) ( )  

References | Related Articles | Metrics

Autonomy is the future development trend of unmanned platforms, and machine map is an important tool for unmanned platforms to realize autonomous spatial cognition. Firstly, the characteristics of machine maps are analyzed and summarized in this paper.Then, from the perspective of maps and their role in people's geospatial cognition, the mode of action of machine maps in the spatial cognition of unmanned platforms is analyzed.Finally, according to the application mode of machine maps, the design principles of machine maps are sorted out, which can provide reference for the in-depth study of machine maps.

A method for tilt measurement and precision evaluation of the separation wall in single-hole shield tunnels with two-tracks

TANG Jimin

2025, 0(4):  179-183.  doi:10.13474/j.cnki.11-2246.2025.0430

Asbtract ( )   PDF (1658KB) ( )  

References | Related Articles | Metrics

The metro tunnel of Line 16 of Shanghai is a 10.4 mdiameter singletube shieldbored tunnel with separation wall,the interior of tunnel is separated by reinforced concrete wall panels.To ensure the safety of the separation walls, a regular, rapid, and full coverage monitoring method is required urgently.This article studies a moving-scanning monitoring method that can reflect the state of the separation wall through special relative tilt angle of the wall.It introduces the principles, methods, and theoretical model of the new monitoring technology for separation walls based on practical engineering cases. The accuracy of this method is estimated and verified through repeated scanning and total station testing. The results show that this method has good performance in quickly screening and locating with problem area of the separation wall. Also it has characteristics of high detection efficiency and high degree of automation, which can be referenced for tilt measurement of the separation wall in areas of traffic and municipal engineering.