This paper introduces the basic situation and performance of the upgraded CORS system based on the construction project of the continuous operating reference station system in Xi'an,the system functions,real-time positioning accuracy,post-processing po-sitioning accuracy,time and space availability and compatibility of the data center are tested.The results show that the system per-formance can meet the design requirements,in the coverage of the CORS site and the extension of a certain extent it can provide users with centimeter-level real-time positioning services.
In order to accurately analyze the monitoring data of highway subgrade settlement and predict the deformation trend in a cer-tain period of time in the future according to the current monitoring data,this paper combines the advantages of least square support vector machines(LSSVM)model and improved particle swarm optimization(IPSO)algorithm in data prediction and parameter opti-mization,build a new IPSO-LSSVM combined prediction model.The combined prediction model continuously optimizes the penalty factor C and kernel function parameters in LSSVM model through IPSO algorithm to avoid the problem of low prediction accuracy caused by the randomness of parameter selection.The effectiveness and superiority of the proposed model are tested by using the meas-ured settlement data of a highway subgrade.The results show that the LSSVM model optimized by IPSO algorithm has higher prediction accuracy and better stability,which can provide a certain reference value for deformation prediction.
Oblique photogrammetry is one of the new data acquisition methods for surveying and mapping.It can obtain the surface tex-ture information and geometric information that can not be obtained by conventional aerial photography,and can quickly build a 3D model.It has the advantages of real modeling,rich texture,high precision,quick production process,less personnel demand and low cost.This article adopts UAV oblique photography method and uses software to achieve 3D real scene modeling of the research area.The entire technology application process is studied,and the data quality is analyzed.The results show that UAV oblique photography technology has obvious advantages in the process of 3D model construction,and the model accuracy can meet the actual production re-quirements,which can provide strong technical support for surveying and mapping projects with high accuracy requirements.
This paper takes SOKKIA NET05AXⅡ measurement robot as the basic platform,combined with the actual needs of slope monitoring,and based on the principle of"the intersection of the corners behind the work base point + polar coordinate measurement of the monitoring point"to design and develop a slope safety online monitoring system.The overall framework,topological structure and business flow of the system are designed in detail,so as to solve the problems such as poor timeliness of data in slope monitoring and inability to monitor in bad weather;And analyze and study the principle and error source of polar coordinate measurement,and conclude that the main sources of error include measurement robot angle and distance measurement errors,as well as the slant distance and vertical angle between the measurement robot and the monitoring point.Then,through a slope monitoring project,comprehensive-ly verify the stability and accuracy of the online monitoring system,and use manual monitoring methods to recheck the automated mo-nitoring data.The results show that the monitoring data of the system is basically consistent with the results of manual review,which meets the requirements of regulation and slope safety monitoring.
Visual presentation and expression of deformation monitoring results can help professionals provide reliable assistance and decision-making,and play an active role in smart city planning,urban disaster monitoring and urban disaster early warning.This pa-per combines GPS and InSAR technology to monitor surface deformation in some areas of Datong city,Shanxi province,and integrates the 3D geospatial platform Cesium under the B/S architecture to realize the loading and display of large-scale monitoring data.At the same time,statistical analysis and visualization are carried out through charts and other methods.The online warning function is com-pleted through deformation threshold setting,and a multi-source,multi-scale,and multi-functional deformation monitoring platform is constructed.
Aiming at the problems such as difficult to distinguish the boundary between changing pixels and unchanged pixels in the change detection of high-resolution remote sensing images,and easy to miss detection in the detection of changing small objects,a Si-amese network of combining dense connection and attention(CDASNet)is proposed in this paper.The network adopts the input meth-od of the Siamese network.In the encoding stage,the residual connection unit and dense connection are used to enhance the extraction of the underlying features of the original image.In the decoding stage,an attention module is introduced to assign new weights to the feature maps at different levels to pay attention to changes.Then,skip connections are used to fuse the shallow features in the enco-ding stage and the deep features in the decoding stage.Finally,change detection results is obtained through upsampling to restore the original image resolution.Experiments are performed on the datasets LEVIR-CD and CDD.The results indicate that the proposed method is superior to FC-Siam-conv,FC-Siam-diff,DSAMNet,SUN-CD in both precision and recall.
Aiming at the urgent demand of efficient and accurate extraction of illegal change spots in land law-enforcement,this paper uses AI technology to automatically detect the change position and extract the change spots from two remote sensing images covering the same region taken by satellites on different days,and then analyzes the legality of the change spots using GIS technology so as to deter-mine the legal changes and suspected illegal spots,providing accurate information for the land law-enforcement and supervision de-partment.Compared with the traditional artificial change extraction method,the AI+RS+GIS technology method proposed in this paper has the advantages of high efficiency,low cost and high target accuracy in the land law-enforcement and supervision work.
With the continuous deepening of the construction of the natural resource big data system,the realization of data fusion of the results of various elements in the system has become the basic work of the entire system construction.This paper introduces the im-plementation of the basic geographic information data fusion project,and combines the application of data classification,fusion rules,fusion processes and other technologies to provide a reference for the deep fusion of natural resource element data.
The JiLin-1 KF01C is a domestic super-wide high-resolution earth observation satellite launched in May 2022.The stand-ard level 1 image has a positioning accuracy of better than 20 m(CE90)without control.In order to further demonstrate the adaptabil-ity of its data products to high-precision applications,this article uses 24 images taken by the JiLin-1 KF01C satellite and processed by the Pixel Factory(PF)software.After aerial triangulation processing,the orthophoto is corrected and generated,and check points are uniformly collected on it.By collecting the check points to contrast the mean square errors from different land forms and sensors and use these data to check the plane accuracy of satellite images.The result shows that the plane accuracy of these 24 images from this project is better than 0.936 m and meets the standard for 1∶5 000 mapping projects.
The high precision and high resolution gravity data required for the construction of the current gravity background field can be obtained in a variety of ways.The combination of neural network and gravity field is in the ascendant and considering the character-istics of neural network usage,it is suitable for the interpolation and fitting process of gravity data processing.This paper proposes a new method for interpolating gravity data based on orthogonal polynomial neural networks to address the issue of reduced accuracy in the grid interpolation process of gravity discrete data.Legendre polynomial neural network(LPNN)model has complex nonlinear map-ping ability and can predict and model gravity data.Moreover,due to its single-layer structure,LPNN has low computational com-plexity.Compared with the existing methods,the accuracy and reliability of the results obtained by the proposed method are proved.Finally,high precision gravity data interpolation is performed in the South China Sea experiment area to further verify the feasibility of the method.
Taking a building as an example,this paper carries out 3D laser scanning,studies the system composition,working princi-ple and characteristics of 3D laser scanning,collects building point cloud data,processes building point cloud data and other data pro-cessing methods,analyzes the method of building 3D modeling and reconstructs the building model.
The digitalization of the underground parking lot is a direction for the construction of smart cities in the future.Combining precise ground navigation maps with high-precision digital parking lot maps can achieve intelligent navigation and positioning of park-ing lots.The traditional underground space measurement method has low efficiency and is greatly affected by light and passing vehi-cles.In this paper,a backpack-type 3D laser scanning equipment based on SLAM technology is used to collect data on an under-ground parking lot,and the data processing process is introduced in detail,and the final generation of large-scale and high-precision topographic maps of underground parking lots can provide technical reference for the future intelligent construction of complex scenes in underground spaces.Finally the underground parking lot is generated.The large-scale and high-precision topographic map of the field can be used as a technical reference for the intelligent construction of complex scenes in the underground space in the future.
In view of the problems existing in the quality inspection of provincial basic surveying and mapping data,such as low effi-ciency of human-computer interaction inspection,cumbersome data format conversion and projection and difficult to find errors.Based on Python and ArcPy,taking 1∶10 000 basic surveying and mapping data of Anhui province as an example,research new methods of automatic data quality inspection,develop automatic inspection tools and batch read GDB,greatly improve data production efficiency and reduce costs.At the same time,it provides a certain reference value for the future research on the quality inspection method of provincial basic surveying and mapping data.
In order to improve the working efficiency of short distance precise point position measurement,this paper tries to apply TS30 total station ATR technology to short distance precise point position measurement,and compares the measurement precision of manual measurement method and automatic observation method using ATR technology through measuring example.The test results show that the measurement time can be greatly shortened by using ATR technology,the coordinate results obtained by the two methods are close to each other,the one-circle-measuring mean square error of angle observation and mean square error of triangulation net-work angle observation by automatic observation method are slightly less than that by manual observation method.The research results can provide reference for rapid measurement of precision engineering.
In order to address the problem that traditional convolutional neural network is ineffective in building extraction,this paper adds an attention gate mechanism to the hopping structure of U-Net network and uses a strategy of hybrid cross-entropy loss function and Lovasz loss function to supervise training based on U-Net network.The above method can effectively reduce the problem of se-mantic loss of features after stitching due to semantic gaps when different layers of features are connected in the jump,and the strategy of hybrid loss function can also effectively integrate the advantages of several different hybrid loss functions,thus enhancing the robust-ness of the model.Both qualitative and quantitative experimental results show that the building extraction results of this paper have fe-wer errors and omissions,the building extraction is more complete,and the accuracy has certain advantages over other methods.
In order to improve the accuracy of a single BP(Back Propagation)neural network model in the prediction of building foun-dation pit settlement data,this paper introduces singular spectrum analysis(SSA)and Kalman filter(KF)into the prediction model,and constructs a KF-BP neural network prediction model based on SSA.The combined prediction model first extracts the trend and periodic terms from the original time series by using SSA;Secondly,KF-BP neural network model is used to predict the trend term and periodic term respectively;Finally,the final prediction result is obtained by reconstructing the trend term prediction result and the periodic term prediction result.The combined prediction model based on SSA proposed in this paper is applied to the prediction of building foundation pit settlement monitoring data.The results show that the prediction model proposed in this paper has higher overall prediction accuracy and more stable prediction results than BP neural network model and KF-BP neural network model.
At present,there are some problems in urban landscape greening management,such as scattered data resources,large a-mount of data and low sharing,poor business coordination and low level of informatization.Aiming at the above problems,this paper constructed a landscape greening management database by collecting and integrating relevant landscape greening resources,and devel-oped a landscape greening management system based on SOA service-oriented thinking by using advanced technologies such as Web-GIS,GPS,cloud computing and Internet of Things.It realizes the landscape data resources display,query,statistics and index analy-sis,landscape greening inspection and maintenance,greening engineering,dynamic monitoring,digital park and other business man-agement functions,so that the landscape management become more informatizational,refined and dynamic.
Aiming at the detection of multi-category targets in UAV remote sensing images,a single-stage deep learning new target detection model is proposed.In the feature extraction structure,a hole convolution kernel is first used to construct the basic extraction structure,so that the model can obtain a feature map with a larger receptive field and more adequate original information retention dur-ing the extraction process;for the problem of poor detection accuracy of small objects,adopting the multi-channel attention mecha-nism combining channel attention and spatial attention can improve the model's attention to the real target;on the basis of continuous up-sampling of feature maps,the feature maps from the same layer and high-level down-sampling are fused to obtain robustness.Feature maps with stronger stickiness and more semantic information implement the final detection.The dataset is composed of Vis-Drone,DLR-MVDA dataset and road-collected images,and is enhanced by dark channel prior methods to form a training dataset to train the model.The experimental results show that the model proposed in this paper can achieve good detection of various types of tar-gets in UAV remote sensing images,and its average accuracy is 8.56%,4.58%and 15.81%higher than the other three benchmark models,respectively,the detection speed can also reach the level of 25 frames per second,indicating that the proposed model can perform fast and accurate detection of multi-category targets in remote sensing images,and has good generalization ability.
In order to grasp the hot topics and evolutionary paths in the research field of cultural tourism integration in rural areas of China,this paper runs Citespace software to visualize and analyze the collected sample literature.The research results show that there is less cooperation among scholars and research institutions in this field,a close cooperation network has not been formed,and there is a lack of broad academic consensus;from the results of keyword clustering,the research on the integration of rural culture and tourism in China can be summarized into five major research themes;from the evolutionary path of the whole research field,it can be divided into three stages:initial development,rapid development and in-depth development.The results of this study can provide a reference for subsequent scholars' studies.
Building elevation view plays an important role in urban spatial information management,urban planning and construction.The traditional facade measurement is mainly based on total station and theodolite.Although the measurement points accuracy is high,but the workload of both indoor and field is large.UAV close-range photography has the advantages of simple data acquisition,low modeling cost and fast modeling speed,which is an ideal way to acquire 3D digital information of buildings.In this paper,taking an engineering practice as an example,the close-range photogrammetry technology is applied to the building facade measurement,and a set of practical facade measurement technology flow is developed.The research results have strong reference value for the effective col-lection of building facade data,the later management and the establishment of a perfect building archives.
