Table of Content

25 June 2023, Volume 0 Issue 6

Previous Issue   

Method and application of dynamic generation of local relative position based on high precision map

FEI Wenkai

2023, 0(6):  1-5.  doi:10.13474/j.cnki.11-2246.2023.0159

Asbtract ( )   HTML ( )   PDF (2027KB) ( )  

References | Related Articles | Metrics

For the application of information sharing and transmission between roadside intelligent devices and self-driving vehicles equipped with high precision maps, this paper proposes a method of dynamic generation and expression of high precision relative positions. The method relies on the common key information of high precision map to encode and decode, and realizes relative location information transmission through geometric matching technology, which improves the security of high precision location information transmission and the compatibility between different maps. This paper gives a complete information generation method, encoding and decoding flow scheme, and obtains good experimental results.

Review from vehicle navigation to autonomous driving: the evolution of electronic map

ZHAO Mingshu

2023, 0(6):  6-10,92.  doi:10.13474/j.cnki.11-2246.2023.0160

Asbtract ( )   HTML ( )   PDF (4284KB) ( )  

References | Related Articles | Metrics

Taking the implement of electronic map in vehicle navigation and autonomous driving as the research object, this paper deeply analyzes the closd relationship between the evolution of electronic map and the development of driving technology. The significant characteristics of map data content and model structure at each stage are discussed along with the development of driving technology.In recent years, the iteration of autonomous driving technology has continuously puts forward new requirements for map data, which has further enriched the map content to extent even beyond the scope of traditional geographic information. It requires add more information to spatial dataframe, including environmental perception data, sensor data, driving experience data and so on. Which enables comprehensive management, query and sharing on massive information. This paper proves the rationality and inevitability of the above requirements by reviewing the geographic information transmission model extension. The international data standard NDS is taken as an example to introduce one of the practices. Finally, summarize the relevant progress in brief and presents an outlook of this field.

High precision map data engine model for autonomous driving

WANG Liyan, ZHOU Xun, HU Wei, LIU Lixin, LI Feixue

2023, 0(6):  11-14,19.  doi:10.13474/j.cnki.11-2246.2023.0161

Asbtract ( )  

References | Related Articles | Metrics

High precision map provides many lane level map elements beyond the sensing range for autonomous driving positioning, perception, planning and control modules, It is an indispensable part of L3 and above autonomous driving. High precision map has high precision and large data volume, but the computing power of the autonomous driving vehicle is limited, so it cannot be separated from the rapid analysis and dynamic extraction of the elements by the high precision map data engine. A high precision map data engine model for autonomous driving is proposed. Through the transformation of high precision map data, element acquisition, map matching, road network construction, data broadcasting, on-board Ethernet transmission, road network reconstruction and application adaptation, the static and dynamic element information required by autonomous driving can be continuously and efficiently provided to meet the real-time demand of high precision map during driving.

Automatic mapping of virtual lanes in intersections

LIU Dawei

2023, 0(6):  15-19.  doi:10.13474/j.cnki.11-2246.2023.0162

Asbtract ( )  

References | Related Articles | Metrics

High precision maps play a critical role in scenarios such as location finding, auto-navigation, and even autonomous driving. This paper classifies and summarizes the existing high precision mapping methods, and points out the problems of the existing mapping methods: the low efficiency of mapping, and no obstacle avoidance, which leads to the automatic marking of virtual lane lines in intersections, but it is not available in practice. Aiming at the problems of existing mapping methods, this paper designs a virtual lane automatic mapping method which can avoid obstacles (such as guard booth, traffic circle, etc.) in intersections. The virtual lane lines in several intersections in Beijing and Shanghai are extracted automatically, and the accuracy of the first automatic drawing extraction is recorded. According to the experimental results, it is proved that the method can be applied to various intersections such as cross, X-shaped, Y-shaped, T-shaped, and so on, and the virtual lane lines automatically marked can avoid obstacles in intersections, which can meet the requirements of automatic lane lines extraction in most intersections. Therefore, this method can meet the expected goal of automatic mapping of virtual lane lines in intersections. Through the automatic mapping method, the lane lines in intersections can be marked automatically and quickly, and the efficiency of the cartographers is greatly improved.

A satellite remote sensing method for detecting marine plastic debirs

LI Peng, ZHOU Hongli, LIN Shicong, WANG Houjie, LI Zhenhong

2023, 0(6):  20-26.  doi:10.13474/j.cnki.11-2246.2023.0163

Asbtract ( )   HTML ( )   PDF (13046KB) ( )  

References | Related Articles | Metrics

Marine floating plastic debris is widely distributed in surface waters such as coastal waters and ocean gyres, which seriously endangers marine life and the sustainable development of human society. Limited by the small-scale, small number of marine plastic target samples and the spatial resolution of satellite remote sensing sensors, accurately detect the spatial and temporal distribution characteristics of marine plastic debris has an important practical significance. Based on the known spectral characteristics of plastic and other floating objects in the sea, this study proposes a reflectance feature classification method based on Sentinel-2 satellite imagery. By combining the reflectance threshold and peak characteristics of different bands, it can effectively detect and identify floating plastic in multiple regions of the world, with an overall accuracy of 98% and an F-score of 0.85, which is better than the traditional machine learning classification method, and is beneficial for the change detection and impact mechanism research of marine plastic debris.

Dynamic monitoring of wetlands in the Yellow River delta based on multi-source remote sensing

FAN Yanguo, WANG Jie, FAN Bowen, XU Ziyao

2023, 0(6):  27-35.  doi:10.13474/j.cnki.11-2246.2023.0164

Asbtract ( )   HTML ( )   PDF (5532KB) ( )  

References | Related Articles | Metrics

The contradiction between the gradual urban and rural construction and natural ecological environmental protection is becoming more and more intense, and the state and government attach great importance to the development of the Yellow River delta wetland area. Therefore, it is urgent to monitor, extract and analyze the dynamic change information of the Yellow River delta in recent years. In this paper, macroscopic (land use) and microscopic (surface deformation) information of the Yellow River delta wetland region is extracted and analyzed by using 8-view Landsat optical images from 2000 to 2021 and 34-view Sentinel-1 downscaled VV polarized SAR images from 2015 to 2021, combined with socio-economic and natural environment data for driving factor analysis. The results show that the land use pattern of the Yellow River delta wetland region did not change significantly between 2000 and 2021, but the land use degree gradually increased; the spatial distribution of the surface deformation degree of the Yellow River delta wetland region is extremely uneven between 2015 and 2021, with the maximum sedimentation rate reaching -43.643 mm/a and the maximum uplift rate of 0.136 mm/a; human activities and socio human activities and social economy are the dominant drivers of surface information changes in the region.

Extraction of invasive plant Spartina alterniflora by combining vegetation phenological characteristics and machine learning supported by GEE

LIU Mingyue, ZHENG Hao, CHEN Xingtong, YANG Xiaowu, SONG Jingru, ZHANG Yongbin, MAN Weidong

2023, 0(6):  36-43.  doi:10.13474/j.cnki.11-2246.2023.0165

Asbtract ( )  

References | Related Articles | Metrics

The invasion of alien species threatens biodiversity and destroys ecosystem structure and function. As the only coastal salt marsh plant in the list of the first 16 invasive alien species in China, the rapid and accurate identification of Spartina alterniflora (S. alterniflora) is of great significance for the sustainable development and management of coastal wetlands. Based on time-series Sentinel-2 images on the GEE, HANTS algorithm is used to fit the NDVI time-series curves, then J-M distance is used to preferentially select the key phenological periods of the S. alterniflora. Sentinel-1, Sentinel-2, and DEM datasets during the key phenological periods are integrated to construct spectral, radar, topographic, and texture features. On this basis, four machine learning methods: SVM, CART, RF and GTB are applied to extract S. alterniflora in Yancheng coastal wetlands. The results show that: ①the key phenology extracted by HANTS algorithm and J-M distance optimization for S. alterniflora identification are maturity and early senescence (October—November), and the discrimination of S. alterniflora from native plants in the key phenology is significantly improved; ② 4 classifiers are trained based on multi-source features of the key phenology, the F1-score of S. alterniflora classification resulted from SVM, CART, RF and GTB classifiers are 0.95, 0.93, 0.97 and 0.95, respectively, and RF had the best classification result; ③The existing area of S. alterniflora in the core area of Yancheng Wetland Rare Bird National Nature Reserve (YNNR) is 3 741.86 hm², accounting for 16.56%. The patches of S. alterniflora show point-source and community boundary-source diffusion, which occupy the ecological niche of native plant communities and threaten the ecological balance of YNNR. Based on the GEE cloud platform, combined with vegetation phenology characteristics and machine learning algorithms, this study utilized multi-source remote sensing data to extract S. alterniflora information in coastal wetlands accurately, rapidly, and efficiently.

Analyzing the spatio-temporal distribution of Eichhornia crassipes based on Sentinel-2 remote sensing

WANG Dongmei, WU Yongfeng, SHI Yifan, LIANG Wenguang, WANG Yihong, PAN Siyuan

2023, 0(6):  44-49.  doi:10.13474/j.cnki.11-2246.2023.0166

Asbtract ( )  

References | Related Articles | Metrics

The large-scale outbreak of Eichhornia crassipes dramatically influence river and lake flood control, water supply security, and water ecology. In this study, the Lixia River area of Jiangsu province is chosen as the research area. We applied the Sentinel-2 image data to filter the optimal classification method of Eichhornia crassipes by compare and analyze the classification accuracy of three machine learning algorithms. Then, analysis the inter-annual spatio-temporal (2017—2021) characteristics and their spreading trends of Eichhornia crassipes by inversion of remote sensing data. The results showed that, the classification performance based on support vector machine (SVM) (overall accuracy is 84.81%～94.30%, Kappa coefficient is 0.70～0.89) is better than neural network (NN) and random forest (RF), The annual outbreak area of Eichhornia crassipes showed a trend of “increase and then decrease” from 2017 to 2021, the outbreak area of Eichhornia crassipes reached its peak in 2019. The outbreak hotspots are mainly located in the junctions of administrative region and dense water networks,such as southern Funing county, central Baoying county, and southern Xinghua city. The average Eichhornia crassipes area during 2017—2021 in the five districts and counties of Xinghua, Baoying, Gaoyou, Jiangdu, and Funing is above 3 km². Especially, the annual average Eichhornia crassipes outbreak in Xinghua is the most frequent, with an area of 6.85 km².

Remote sensing monitoring and coupling analysis on glacier and surface water in Sichuan-Tibet traffic corridor

WANG Lixuan, YE Chengming, SUI Tianbo, WEI Ruilong, LI Hongfu

2023, 0(6):  50-55.  doi:10.13474/j.cnki.11-2246.2023.0167

Asbtract ( )   HTML ( )   PDF (1909KB) ( )  

References | Related Articles | Metrics

Glaciers retreat caused by climate warming has rapidly increased surface water area in the Qinghai-Tibet Plateau based on 3952 Landsat 5 TM, Landsat 8 OIL remote sensing images.The paper uses DSWE method and FMask algorithm to obtain respectively glacier and surface water information in Sichuan-Tibet traffic corridor,and analyses spatio-temporal changes and coupling characteristic between glacier and surface water combining with DEM and watershed data.The glaciers in the Yarlung Zangbo River and Nujiang River basins have shrunk by more than 5000 km² in the past 35 years, with the annual shrinking rate increasing gradually. The other four basinsalso retreated as a whole. However, the retreat rate is lower and sporadic areas are increasing in the past decade. The surface water in Sichuan-Tibet traffic corridor is strongly affected by glacier melting, especially in Yarlung Zangbo River and Nujiang River basins whose surface water areas are expanded by 327 and 155 km² respectively. But the glaciers are relative stable which located in Lantsang River, Jinsha River. The rich precipitation in Minjiang River basin is the main reason for the surface water area expansion. The coupling of the change between glacier and surface water is good in different elevation intervals. Especially in the elevation range of 4500～5000 m and 5000～5500 m, the retreat of glacier is accelerated, accompanied by the accelerated expansion of surface water in the same period.

Remote sensing monitoring and climate response analysis of snow cover in Yunnan

YANG Jiaxin, LI Jiatian, LU Mei, HU Minghong, LI Wen, JIN Wei, HU Hao

2023, 0(6):  56-60.  doi:10.13474/j.cnki.11-2246.2023.0168

Asbtract ( )  

References | Related Articles | Metrics

Snow cover is highly sensitive to climate and can respond to climate change, and its change study is of great significance. Based on Landsat 7 ETM SLC and Landsat 8 OLI images, the object-oriented extraction and human-computer interaction interpretation methods of eCognition are jointly used for the extraction and analysis of snow distribution data in Yunnan in 2000 and 2020, combined with 23 meteorological stations and 50 interpolation points in 2000—2020 temperature, precipitation, snowfall, snow depth records and elevation data and other auxiliary data, using univariate linear regression analysis and Mann-Kendall trend test. By mutation detection method, the distribution status of snow cover in Yunnan, the change of snow cover in 2000—2020 and the climate response analysis are carried out. The results show that: ①In 2000, the total snow area in Yunnan was 1 533.1 km², in 2000, the total snow area in Yunnan was 987.8 km², and in 2000—2020, the snow area decreased by 545.3 km². Among them, the snow cover decreased by 945.6 km², the snow area increased by 400.300 2 km², and 587.463 3 km² remained unchanged. ②The distribution and change of snow cover in Yunnan showed significant spatial differences. The changes in snow cover area were concentrated in the latitude range of 28°N—28.5°N and the elevation range of 3.6～4.0 km, and the area decreased by 16.2% and 14.5%, respectively. ③The temperature and precipitation in Yunnan showed a significant upward trend, with linear tendency reaching 0.38℃ and 314.7 mm every ten years, respectively, which were negatively correlated with the change of snow cover, and the snowfall amount and snow depth showed a weak downward trend, which was positively correlated with the overall change of snow cover. This study provides supporting data for the spatial distribution and change of snow cover in Yunnan and the trend of climate change.

Building change detection in remote sensing images with fully convolutional neural network enhanced edge information

CHEN Jie, LIU Jiping, XU Shenghua

2023, 0(6):  61-67.  doi:10.13474/j.cnki.11-2246.2023.0169

Asbtract ( )  

References | Related Articles | Metrics

For many existing deep learning building change detection methods, it is difficult to obtain image structure features, which leads to the problem of low segmentation accuracy of building edge pixels. In this paper, a building change detection model based on enhanced edge information in remote sensing images is proposed. Firstly,the Canny algorithm and the probabilistic Hough transform algorithm are used to extract the linear edge feature map of the building in the bitemporal image as the image structure feature. Then the bi-temporal images and their corresponding edge feature maps are input into a fully convolutional neural network (FCN) that enhances edge information. Finally, the weighted combination function of Dice Loss and CrossEntropy Loss is used to measure the network model. Experiments show that the FCN network with enhanced edge information has certain advantages in accuracy evaluation and visual analysis.

Near-ultraviolet channel surface reflectance simulation based on XGBoost algorithm

AO Yong, LI Hongli, ZHANG Wenjuan, QIN Meng

2023, 0(6):  68-74,103.  doi:10.13474/j.cnki.11-2246.2023.0170

Asbtract ( )  

References | Related Articles | Metrics

The ultraviolet spectrum has significant applications in the fields of global auroral detection,marine oil spill,atmospheric glow,etc. Surface reflectance is important background data in the research. However,the existing satellite data resources are relatively insufficient to meet the application needs. In this study,a machine learning-based on XGBoost algorithm is proposed for simulating surface reflectance data in the near-ultraviolet (N-UV) (350～400 nm) spectral channel. Firstly,Sentinel-2 MSI 2,3 and 4 channels are selected as the data source and the spectral of vegetation,water,soil and other typical features are extract based on the USGS spectral database,then equivalently calculated to the corresponding channels. Secondly,the correlation analysis between the data source and the channel to be simulated is carried out. The correlation coefficients between Sentinel-2 MSI 2,3 and 4 channels and the channels to be simulated are all greater than 0.88,which indicates that the N-UV surface reflectance simulation can be carried out based on this data source. Thirdly,based on the typical spectral data set after the equivalent calculation construct XGBoost regression model to simulate the N-UV channel surface reflectance. Results indicate that the coefficient of determination (R²) of all the channel models is above 0.91,the root mean square error (RMSE) is less than 0.076,the mean absolute error percentage (MAPE) is within 20%,and the standard deviation of the above three accuracy indicators for different categories of samples is within 0.021 2,which shows that the model has high accuracy and robustness. Finally,based on the Sentinel-2 MSI 2,3 and 4 channels image data,the simulated images of surface reflectance at 355,365,375,385 and 395 nm are generated,and the images better reflect the spectral characteristics of the surface.

Establishment method of woodland refinement of DEM based on dense point cloud of UAV images

HE Rong, BAI Weisen, DAI Zhen, ZHAI Huipeng

2023, 0(6):  75-81.  doi:10.13474/j.cnki.11-2246.2023.0171

Asbtract ( )  

References | Related Articles | Metrics

Aiming at the problem that UAV photogrammetry is susceptible to vegetation influence and terrain loss when establishing DEM, this paper proposes a method to establish refinement of DEM in woodland based on ground point clouds fusing images of different routes. Firstly, the images are classified according to the UAV route, and the ground point clouds are extracted separately.Then the ICP algorithm with inverse distance weight constraint is used to construct the fused ground point cloud. Finally,the woodland refinement of DEM is established based on the fused ground point cloud. The results show that the number of fused ground point clouds is 2 182 740, and the density is 9612/m². The error in DEM established based on fused ground point cloud is 7.3 cm, and the correlation coefficient with the actual terrain reaches 0.925, and in different vegetation areas, the error in the DEM established by the fused ground point cloud is within 10 cm, and the correlation coefficient with the actual terrain is above 0.89. Experiments verify the feasibility and applicability of the proposed method, and provide a reference for the establishment of fine DEM of woodland by UAV photogrammetry.

Application of improved Harris Hawks optimization algorithm in patching cloud holes at matching ground points

ZHANG Yan, LIU Lilong, HE Guanghuan, XU Yong, MENG Jinlong

2023, 0(6):  82-87.  doi:10.13474/j.cnki.11-2246.2023.0172

Asbtract ( )  

References | Related Articles | Metrics

Aiming at the problem that there will be more holes in the UAV matching point cloud after ground point filtering,this paper proposes to use the improved Harris Hawks optimization algorithm to optimize the least squares support vector machine to repair the ground point cloud hole. Firstly,the eight-pronged tree structure method is used to extract ground feature points from the filtered point cloud data. Secondly,the Harris Eagle algorithm is improved by using nonlinear convergence factor and adaptive escape probability strategy,and it is used for parameter optimization of hole repair model of least squares support vector machine. Experimental results show that compared with the conventional least squares support vector machine,the hole repair accuracy of the combined model is improved by 34.3%,and its stability is also enhanced,which has a certain timeliness and reality.

An improved fuzzy Wishart-PSO polarimetric SAR image intelligent clustering algorithm

ZHU Teng, GAO Zhaozhong, SHEN Chen, HUANG Tielan, ZHOU Huiyuan

2023, 0(6):  88-92.  doi:10.13474/j.cnki.11-2246.2023.0173

Asbtract ( )  

References | Related Articles | Metrics

Aiming at the problems of low accuracy of polarized SAR image clustering, large data volume of polarization parameters and complicated calculation, this paper proposes an intelligent clustering method for particle swarm of PolSAR images based on improved fuzzy Wishart distance. The method improves the traditional Wishart clustering evaluation criterion by combining fuzzy division for PolSAR data distribution to reduce the influence of isolated point noise, then completes the initial division of clusters according to the polarization scattering mechanism, and finally introduces the particle swarm optimization framework in the iterative optimization search step to improve the effectiveness of clustering centers and classification accuracy. In the experimental part, the effectiveness of the fuzzy Wishart-PSO clustering algorithm is verified by using L-band AIRSAR data and X-band high-resolution polarized SAR data respectively, and the classification results are significantly more reasonable than the traditional H/α-Wishart method, and the clustering accuracy can reach 90%.

The improved method of U-type deep learning neural networkfor remote sensing in land type change detection

SHEN Xinsu, JI Ling

2023, 0(6):  93-97,103.  doi:10.13474/j.cnki.11-2246.2023.0174

Asbtract ( )  

References | Related Articles | Metrics

The change detection of multi-temporal remote sensing image is widely used in natural resource management such as survey and monitoring. According to the high construction cost and deep learning algorithms difficult of the sample library,this paper proposes multi-temporal change detection method to improve image change deep learning detection. This method take multi-temporal images as different band for information fusion,and transform the change detection task into image segmentation task, use land use vector data as label for model training and build deep learning sample library. Improve the structure of the original U-type deep learning neural network, and accelerated model training. Experimental results show that:①Multi-temporal change detection method is conducive to learn more features during model training and improving the feature extraction capability of the model,and finally getting the better prediction effect;②The recall rate and precision rate of the model is improved in a certain degree,and the whole prediction effect is obviously improved.

Experimental analysis of traffic-oriented GNSS high-precision positioning integrity

BAO Yeqing, LIU Hui, WANG Yixin, QIAN Chuang, TANG Jian

2023, 0(6):  98-103.  doi:10.13474/j.cnki.11-2246.2023.0175

Asbtract ( )  

References | Related Articles | Metrics

With the application of differential GNSS and other high-precision positioning technologies in land and marine fields gradually becoming popular, higher requirements for reliability, accuracy and continuity of high-precision positioning have been put forward, and how to develop an integrity system suitable for high-precision positioning applications based on aviation integrity theory and technology has become one of the current research hotspots. In this paper, we analyze the minimum performance requirements for the integrity of high-precision positioning service system and user terminal based on aviation integrity theory and the characteristics of domestic road traffic, design road experiments, and verify the GNSS integrity index of road traffic in a wide range of scenarios using low-cost high-precision modules. The test results show that the index availability of high-speed open/obscured scenes is 73.05%/52.42% and 79.27%/71.30% in the heading direction and left-right direction, respectively; the index availability of low-speed open/obscured scenes is 80.54%/65.65% and 81.09%/61.79%, respectively.

Indoor visible light fingerprint location method based on GF-KF and Improved-Bayes

GU Yaxiong, ZHONG Wen

2023, 0(6):  104-109,128.  doi:10.13474/j.cnki.11-2246.2023.0176

Asbtract ( )  

References | Related Articles | Metrics

In view of the problem that indoor ambient light, noise and other factors will interfere with the intensity of the visible light signal strength received by the mobile terminal and cause the positioning accuracy to be low, this paper proposes a visible light fingerprint positioning method that integrates Gaussian fitting and Kalman filtering (GF-KF) with Improved-Bayes. Firstly, the RSS date collected by GF-KF algorithm is corrected as fingerprint database data, and then the weight coefficient of the k-neighbor method is transformed and fused with Bayesian algorithm, which matches the RSS data of the point to be measured and the fingerprint point Finally, the position is calculated. Experimental results show that under the algorithm model, the average positioning error is 1.42 cm, and 92.83% of the test point positioning error is not more than 2 cm, which is more accurate and robust than the convolutional neural network algorithm, the weighted K nearest neighbor algorithm and the support vector machine method.

Track measurement system based on multi-sensor fusion

HAN Yulong, SUN Haili, DING Zhigang, ZHONG Ruofei, DU Zejun

2023, 0(6):  110-116,133.  doi:10.13474/j.cnki.11-2246.2023.0177

Asbtract ( )  

References | Related Articles | Metrics

Dynamic observation with multi-sensor set as the core and collaborative solution of multi-source data is an important development direction of track engineering measurement, and important progress has been made. However, due to technical protection and other reasons, there are few literatures on the research and discussion of the integrated multi-sensor track measurement system. This paper presents and introduces an orbit measurement system based on multi-sensor integration and its data processing scheme. The system takes total station and inertial navigation as the main sensors, and adopts mobile measurement-static point fixed-point correction in the acquisition mode. In the calculation of track geometric parameters, the research system gives two static point calculation schemes, adjustment solution and single point control, according to different accuracy requirements and taking into account efficiency. The dynamic trajectory is calculated smoothly by fusing fixed interval filtering, and the track position is solved with the help of inclination and attitude, so as to realize track dynamic measurement. The experiment shows that the system realizes the absolute measurement of track line, and its measurement accuracy of lateral deviation is 4 mm and vertical deviation is 3 mm, which meets the current index requirements, and provides a good reference scheme for the development of track dynamic measurement technology in the field of road and track operation detection and measurement.

Spatio-temporal difference analysis of carbon storage in Beihai secosystem based on FLUS-InVEST models

LI Xiaojun, CHE Liangge, HU Baoqing

2023, 0(6):  117-123,183.  doi:10.13474/j.cnki.11-2246.2023.0178

Asbtract ( )  

References | Related Articles | Metrics

By coupling FLUS-InVEST models and the current data of land use from 2010 to 2020,this study calculated and predicted the spatio-temporal difference of land and carbon storage in Beihai city, and predicted the impact of natural evolution scenarios and green intensive ecological development scenarios on land use and carbon storage in 2035. The spatial autocorrelation model was used to reveal the future spatial distribution trend, of which can provide scientific reference for land use management and land spatial planning under the “dual carbon” goal. The results show that: ① From 2010 to 2020, the overall transformation of land types was dominated by the conversion of land types with low carbon density to land types with high carbon density. The disordered flow of cultivated land to forest land is prominent; ②Over the course of the study, the overall carbon storage of Beihai city decreased first and then increased, with an overall increase of 4.01×10⁵ t in the past 10 years; ③By 2035, the predicted carbon reserves in Beihai city would further decrease in the natural evolution scenario. But in the green intensive ecological protection scenario, carbon reserves can still slowly recover under the premise of fully ensuring high-quality socio-economic development. In the next 15 years, carbon reserves can lose 1.36×10⁵ t less than in the natural change scenario.

Modeling methodology for fine DEM in urban based on multi-source data

ZHANG Longqi, DU Shuai, XIONG Xuping, ZHANG Yuqing

2023, 0(6):  124-128.  doi:10.13474/j.cnki.11-2246.2023.0179

Asbtract ( )  

References | Related Articles | Metrics

It is still a current technology problem to build a fine DEM in urban using single technology because of dense buildings,vegetation,heavy traffic,etc. This paper discusses a method to build a fine DEM based on large-scale digital topographic map,DOM and DEM generated by UAV photogrammetry or laser radar point cloud data (DSM). Firstly,contour and elevation points are extracted from large-scale digital topographic map,which can be directly used to construct DEM. Then some extraction is separately done to get terrain-related line elements,which are always 2D without elevation information,such as contour of buildings,road sidelines,square boundaries,scarps,lake sidelines and so on. Then some necessary process is done to build 3D-lines after obtaining from the correct elevation of general DEM and DSM. Based on these 3D lines and elevation points,a heterogeneous TIN model is constructed using constrained TIN methodology. Because of data source with different accuracy in different area,the modeling accuracy is also analyzed using theory and practice. Experiments show that the constructed DEM is so high consistency with the topographic map that can better meet the needs of flood analysis,urban fine management,and real three-dimensional construction.

Detection and extraction of tunnel steel arch using point cloud spatial projection features

SUN Senzhen, JING Liujie, WANG Li

2023, 0(6):  129-133.  doi:10.13474/j.cnki.11-2246.2023.0180

Asbtract ( )  

References | Related Articles | Metrics

The rapid identification and extraction of tunnel steel arches supporting structures based on 3D point cloud scanning is the key link in the process of intelligent perception for tunnel special equipment. Due to the limitations of tunnel construction environment and cost, the point cloud scanned by special equipment trolleys lacks the strength information, which increases the difficulty of extracting and segmenting the point cloud of steel arches. Aiming at this problem, a method for detecting and extracting tunnel steel arches based on point cloud space projection features is proposed. Firstly, the tunnel point cloud is projected spatially according to the tunnel axis and transverse plane, forming a projected image based on the tunnel cross-section. Secondly, based on the characteristics of the point cloud projection distribution of steel arches, combined with the digital image method, the projection area of steel arches is detected. Finally, the point cloud of the steel arch supporting structure corresponding to the projection area is extracted to analyze the position and state of steel arch. This method ensures the accuracy and flexibility of arch extraction based on single-site cloud scanning data by setting projection parameters, and can be applied to the identification and positioning of steel arches when covered by shotcrete during wet spraying.

Rural building information extraction using object-oriented and deep belief network

CHEN Qiaoyi, YAN Yufei, HUANG Yongfang

2023, 0(6):  134-137.  doi:10.13474/j.cnki.11-2246.2023.0181

Asbtract ( )  

References | Related Articles | Metrics

Mapping the information related to the number, area and location of rural buildings is the basis for rational planning of rural land and building beautiful and livable rural areas in the new era. However, due to the phenomenon that rural buildings are scattered and severely fragmented, it is still challenging to accurately extract the scattered rural buildings using remote sensing technology. In this study, we propose a rural building extraction method that combines object-oriented and deep confidence networks. Firstly, we use object-oriented scale segmentation based on the spectral, shape and texture features of rural buildings, and further use deep confidence networks to learn high-level semantic features such as texture and environment of different objects to extract rural building information. Compared with random forest commonly used image classification methods,the method in this paper performs better in rural building extraction, with clearer and more complete edge contours of the extracted patches, and better recognition of the distinction between the gap parts between different buildings, and less noise in the extraction results. The method can effectively and efficiently extract rural building information.

GF-7 and ZY-3 satellites jointly serve airport clearance monitoring

LIU Dongzhi, XU Qingling

2023, 0(6):  138-141.  doi:10.13474/j.cnki.11-2246.2023.0182

Asbtract ( )  

References | Related Articles | Metrics

According to the DSM production requirement of airport clearance monitoring, the DSM data is obtained by stereo matching of the GF-7 satellite image based on the high-precision control data in this paper. The matching effect of high-rise and super high-rise buildings is not good, and the manual stereo editing workload is large. Joint block adjustment is performed on the satellite images of GF-7 and ZY-3, and technologies such as multi-horizon epipolar image matching, globally optimized occlusion detection and parallax hole repair are adopted to obtain high-precision DSM data by stereo matching. The high-precision DSM can effectively extract high-rise and super-high-rise buildings, and provide basic spatial data for airport clearance monitoring.

Accuracy evaluation for mobile backpack-mounted scanning system based on standard baseline field

CUI Lei, LIU Yingjie, LU Hongbo, WANG Dongxu

2023, 0(6):  142-145,179.  doi:10.13474/j.cnki.11-2246.2023.0183

Asbtract ( )  

References | Related Articles | Metrics

After the long-term use, there will be the impact on the measurement accuracy and reliability of the mobile backpack-mounted scanning system, without no standard for system calibration and accuracy evaluation. The key technologies and performances indexes of backpack-mounted scanning system are studied. The comprehensive accuracy evaluation method is put forward. The homemade stereo reflection targets as feature points are distributed in an orderly way around the calibration pillars of Beijing standard length baseline field, and the comprehensive accuracy of the system is evaluated and analyzed by comparing the scanning measurement values of target characteristic points with the standard values. The experimental results show that the plane accuracy of the system is better than 3 cm and the elevation accuracy is better than 2 cm, which provides a reliable basis for the accuracy evaluation and reliability application of the backpack-mounted scanning system.

Extraction of shallow water bodies in Shenzhen using 0.5 m resolution satellite data

LIU Min, HUANG Jinhui, PENG Zhenhua, WU Zhenyu

2023, 0(6):  146-149,175.  doi:10.13474/j.cnki.11-2246.2023.0184

Asbtract ( )  

References | Related Articles | Metrics

Taking the domestic SuperView satellite images as the data source, the coastal shallow water area, reservoir shallow water area in Shenzhen are selected as the research objects,due to the influence of the substrate in the shallow water area,the reflectance of coastal water and coastal soil is close, which causes the difficulty and poor accuracy of water boundary extraction. Using the improved coefficient of variation method to screen the water index, combined with DEM data to modify the water boundary line and denoise, the shallow water boundary is obtained more accurately.The results show that the overall accuracy of discrimination of coastal, reservoir shallow water in the 0.5 m resolution satellite images is 98.75% and 91.10% respectively, the accuracy and stability of shallow water extraction are significantly improved.

Remote sensing analysis of the evolution of wandering trends in the lower Yellow River root-shaped channel

FAN Xin, DING Laizhong, LI Ying, WANG Wenjie, CHENG Ming, SONG Huichuan, GAO Shuang, GENG Liyan, LI Chunyi

2023, 0(6):  150-154.  doi:10.13474/j.cnki.11-2246.2023.0185

Asbtract ( )  

References | Related Articles | Metrics

River wading is a major cause of natural disasters in the lower reaches of the Yellow River, and the analysis of its wading pattern is of great significance for the protection of settlements and farmlands in the downstream beach areas. Aiming at the river loitering problem in the lower Yellow River, this paper proposes a PCA-SVM extraction method for the Yellow River channel by using the multi-source long time series of domestic satellites, and interpretes the river channel information of the Puyang section of the Yellow River from 2013 to 2022, and takes the Lotus node channel as an example to analyze its loitering trend. The results show that the Yellow River channel extracted by PCA-SVM method is complete and the sand bar is clear, which significantly improves the classification confusion of the river channel, wetland and tidal flat, and the interpretation accuracy is 84.17%. The Kappa coefficient is 0.613, and the accuracy is significantly higher than that of maximum likelihood classification, minimum distance classification and SVM. Through the analysis of the loitering trend of the Yellow River from 2013 to 2022, it can be seen that the loitering trend of the lotus node channel in the study area is obvious, and the main channel has migrated from the left bank to the right side, and there is still a loitering and migration trend to the right by August 2022. The erosion of residential areas and farmland in the right bank beach area is intensified, which is easy to cause dangerous workers and lead to flood disasters.

Application of improved random forest model in population spatialization

JIANG Xueli, XIONG Yongliang, GUO Hongmei, ZHAO Zhen, ZHANG Ying, MENG Yatian

2023, 0(6):  155-160.  doi:10.13474/j.cnki.11-2246.2023.0186

Asbtract ( )  

References | Related Articles | Metrics

The random forest model-based population spatialization method does not take into account the non-equilibrium of population spatial distribution, and the use of Bootstrap sampling exacerbates the unevenness of the sample, making it unrepresentative and resulting in low model prediction accuracy. For this problem,this study takes Chengdu city as an example, the characteristic factors of affecting the population distribution are extracted through correlation analysis, the data set is clustered based on the K-means++clustering algorithm, and then an equal amount of data from each cluster is fused as a training subset using the Bootstrap sampling method to construct an improved random forest model and compare it with the traditional random forest model. Finally, the population data of Chengdu city in 2020 is spatialized using an improved random forest model, and the results are compared with the WorldPop dataset for accuracy. The results show that the overall accuracy of the population spatialisation model based on the improved random forest reaches 80.5%, which is about 3.4% higher than before the improvement, indicating that the improved random forest model can effectively improve the model prediction accuracy. Compared to the WorldPop dataset, the population spatialisation results based on the improved random forest model are better in terms of fit and accuracy.

The construction of building monomer model framework based on real scene 3D model

SHI Yuzheng, CHEN Menghua, HUANG Yu, ZHANG Tongyun, ZHOU Xian

2023, 0(6):  161-166.  doi:10.13474/j.cnki.11-2246.2023.0187

Asbtract ( )  

References | Related Articles | Metrics

Due to a series of problems in the real scene 3D model, such as unreasonable data structure and difficulty to recognize and utilize, methods including automatic selection of building boundaries and refined model correction in the monomer model framework are proposed in this paper, which utilize the key technologies of chi-squared distribution, enhanced feature pyramid network, multi-scale progressive growth optimization algorithm, parent-child coding and so on. Experimental results show that the established monomer 3D model satisfies the requirement of a 0.05 m functional level model, and the number of patches decreased by 99.54% in comparison to the MESH model, which achieves the rapid monomerization and structurization of the real scene 3D model of buildings and provides technical support for the construction of real scene 3D China.

Urban safety production risk assessment supported by spatial big data

HE Keqin, CHENG Nanwei, DENG Min

2023, 0(6):  167-171.  doi:10.13474/j.cnki.11-2246.2023.0188

Asbtract ( )  

References | Related Articles | Metrics

Most of the existing studies on urban safety production risk assessment are based on statistical survey data to assess a single enterprise or park, and without sufficiently considering the spatial effect of safety production accidents, which is difficult to adapt to the urban scale safety production risk assessment. Based on the H-V risk assessment framework, this paper constructs the urban safety production risk assessment system by comprehensively analyzing regional hazard factors, regional target exposure and regional rescue adaptation. With the support of multi-source geospatial data, the urban safety production risk index is quantified based on spatial analysis to reveal the distribution pattern of urban safety production risk, which could provide decision-making knowledge service for the prevention and control of urban production safety. Taking Kunshan as the study area, the urban safety production risk assessment supported by spatial big data is realized. The experimental results show that the high-risk areas of safety production in Kunshan are mainly concentrated in the east and south of Yushan Town, and the risk of disaster factors is higher in the south of Yushan Town, while the risk of disaster factors and regional target exposure are higher in the east of Yushan Town.

Key methods of 3D visualization cartographic generalization in AIS based on 2D/3D platform

CHEN Guihua, PENG Wen

2023, 0(6):  172-175.  doi:10.13474/j.cnki.11-2246.2023.0189

Asbtract ( )  

References | Related Articles | Metrics

In view of the problems of 2D display and loading of massive data, data redundancy and single type of data faced by automatic identification system(AIS) accessed by various platforms in providing maritime ship information services, this paper proposes an optimized key method for 3D visualization cartographic generalization of AIS ships.The method is elaborated from the aspects of AIS ship 3D visualisation processing, dynamic data analysis, scaled display loading technology and ship attribute visualisation, which greatly improves the loading and updating speed of AIS data as well as maritime ship identification, while using the method of cartographic generalization to scientifically handle the amount of screen loading of AIS data and other chart elements at different display scales, effectively improving the identification of ship types and the visualisation effect of reasonable aesthetics.

Reform and exploration of the practical teaching mode of remote sensing course in the era of big data and artificial intelligence

CHEN Jie, DENG Min, HOU Dongyang

2023, 0(6):  176-179.  doi:10.13474/j.cnki.11-2246.2023.0190

Asbtract ( )  

References | Related Articles | Metrics

The development of big data and artificial intelligence has added new vitality to related disciplines. In the practice of training undergraduates majoring in remote sensing or geo-information, many universities still follow the traditional education model, ignore the innovation of remote sensing applications brought about by big data and artificial intelligence technology. To cope with the opportunities and challenges brought by big data and artificial intelligence technology, talent training needs to be reformed accordingly. To this end, we explore the practical teaching of remote sensing courses in the context of big data and artificial intelligence, aiming to combine professional characteristics and train undergraduate talents with strong practical ability and innovative consciousness.

Buildings extraction based on high-resolution remote sensing imagery

WANG Limei, WANG Yanzheng

2023, 0(6):  180-183.  doi:10.13474/j.cnki.11-2246.2023.0191

Asbtract ( )  

References | Related Articles | Metrics

High-resolution remote sensing images not only have rich spectrum, spatial distribution, shape and texture features, but also contain clear scene semantic information. Taking Zongyang town, Zongyang county, Anhui province as the research area, and using high-resolution images as the basic data source, the deep learning and object-oriented method in eCognition software is used to automatically extract buildings in this paper. The results show that the method of combining deep learning with object-oriented has a better effect of building extraction, and the overall classification accuracy reaches 96.8%, which can be used for building extraction production based on high-resolution images.