Modeled camera layout method and system

Abstract

The invention relates to the technical field of camera layout, in particular to a modeling camera layout method and system. The method comprises the following steps: performing laser radar scanning on an intersection area to obtain intersection three-dimensional point cloud data; performing static element identification on the intersection three-dimensional point cloud data to obtain an intersection base layer data graph; traffic flow and people flow collection is carried out on the intersection area to obtain an intersection dynamic layer data graph; performing hydrodynamic modeling on the intersection area based on the intersection basic layer data graph and the intersection dynamic layer data graph to obtain intersection traffic flow field data; performing flow field feature recognition on the intersection traffic flow field data to obtain an intersection flow field feature point distribution map; and performing speed density field simulation on the intersection area based on the intersection flow field feature point distribution diagram to obtain an intersection flow field dynamic distribution model. According to the invention, the accuracy and effectiveness of the camera layout scheme are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of camera layout, and particularly to a modeling method and system for camera layout. Background Art

[0002] In modern urban traffic monitoring systems, the optimization of camera layout is a key technical challenge. Taking urban intersections as an example, this is a typical complex monitoring scenario, characterized by large traffic flows, diverse vehicle types, frequent pedestrian crossings, and complex traffic signal changes. Existing camera layout methods often struggle to comprehensively capture these dynamic changes, resulting in the emergence of monitoring blind spots. Especially during peak hours, the rapid changes in vehicle and pedestrian flows make fixed camera layouts difficult to adapt, thus unable to effectively monitor traffic violations, affecting the efficiency and safety of traffic management. Further, considering the multi-directionality of urban intersections, existing camera layout methods have limitations in perspective selection. Due to the limited viewing angle and coverage range of cameras, it is difficult to simultaneously monitor vehicles and pedestrians in all directions, especially at complex multi-lane intersections. This limitation leads to the inability of the monitoring system to provide a comprehensive perspective during critical traffic events, such as traffic accidents or traffic congestion, thus affecting the rapid response and handling of events. Summary of the Invention

[0003] Based on this, it is necessary for the present invention to provide a modeling method and system for camera layout to solve at least one of the above technical problems.

[0004] To achieve the above object, a modeling method for camera layout includes the following steps:

[0005] Step S1: Perform lidar scanning on the intersection area to obtain three-dimensional point cloud data of the intersection; identify static elements from the three-dimensional point cloud data of the intersection to obtain a basic layer data map of the intersection; collect vehicle and pedestrian flows in the intersection area to obtain a dynamic layer data map of the intersection;

[0006] Step S2: Based on the basic layer data map of the intersection and the dynamic layer data map of the intersection, perform fluid dynamics modeling on the intersection area to obtain traffic flow field data of the intersection; identify flow field characteristics from the traffic flow field data of the intersection to obtain a distribution map of flow field characteristic points of the intersection; based on the distribution map of flow field characteristic points of the intersection, perform speed-density field simulation on the intersection area to obtain a dynamic distribution model of the flow field of the intersection;

[0007] Step S3: Evaluate the constraints of camera installation points in the intersection area to obtain a set of installation points in the intersection; perform spherical projection calculation on the cameras according to the set of installation points in the intersection to obtain a set of camera field-of-view angles; perform spherical coverage visualization on the cameras according to the set of camera field-of-view angles to obtain an initial coverage plan map;

[0008] Step S4: Obtain the historical traffic data of the intersection; perform heat map conversion on the historical traffic data of the intersection to obtain the distribution data of risk points at the intersection; perform spatial overlay analysis on the initial coverage plan map according to the distribution data of risk points at the intersection to obtain the layout verification result map; perform coverage difference adjustment on the initial coverage plan map according to the layout verification result map to obtain the layout optimization data set;

[0009] Step S5: Reconstruct and adjust the intersection installation point set according to the layout optimization data set to obtain the final camera layout plan.

[0010] The present invention obtains the 3D point cloud data of the intersection through lidar scanning, and combines the dynamic layer data map collected by vehicle and pedestrian flows, and can comprehensively and accurately construct the static and dynamic scenes of the intersection. Through hydrodynamic modeling and flow field feature recognition, the distribution and change rules of the traffic flow at the intersection are further clarified, so that the evaluation and selection of camera installation points can fully consider the actual dynamics of the traffic flow, thereby realizing all-round monitoring of all directions and regions of the intersection, effectively solving the problem of monitoring blind spots caused by limited viewing angles and coverage ranges in the existing methods, and ensuring that all vehicles and pedestrians in all directions can be included in the monitoring range. Through the speed-density field simulation, the spatio-temporal changes of key information such as the speed and density of the traffic flow at the intersection can be reflected. Through the installation point constraint evaluation link, multi-dimensional factors such as building boundaries, installation height, power supply, network transmission, lightning protection and grounding, and obstacles are comprehensively considered, and the candidate points are strictly screened. This ensures that the finally determined installation point set can not only achieve good monitoring effects theoretically, but also has high feasibility and stability in the actual installation and operation process, avoiding the abnormal operation of monitoring equipment caused by non-meeting installation conditions. By generating the distribution data of risk points, performing spatial overlay analysis and coverage difference adjustment on the initial coverage plan map, the camera layout can be accurately optimized for risk points such as accident-prone areas, violation-prone areas, and traffic congestion-prone areas. Through layout optimization, as well as the analysis of the dynamic distribution model of the intersection flow field and the mining of historical traffic data, the changing trend of traffic patterns can be timely discovered, and the camera layout can be adjusted and optimized accordingly.

[0011] Preferably, the present invention also provides a model-based camera layout system for executing the model-based camera layout method as described above. The model-based camera layout system includes:

[0012] A data acquisition module, configured to perform lidar scanning on the intersection area to obtain the 3D point cloud data of the intersection; perform static element recognition on the 3D point cloud data of the intersection to obtain the basic layer data map of the intersection; perform vehicle and pedestrian flow collection on the intersection area to obtain the dynamic layer data map of the intersection;

[0013] A model construction module, configured to perform hydrodynamic modeling on the intersection area based on the intersection basic layer data map and the intersection dynamic layer data map to obtain intersection traffic flow field data; perform flow field feature recognition on the intersection traffic flow field data to obtain an intersection flow field feature point distribution map; perform speed-density field simulation on the intersection area based on the intersection flow field feature point distribution map to obtain an intersection flow field dynamic distribution model.

[0014] A layout planning module, configured to evaluate the camera installation point constraints in the intersection area to obtain an intersection installation point set; perform spherical projection calculation on the cameras according to the intersection installation point set to obtain a camera field of view angle set; perform spherical coverage visualization on the cameras according to the camera field of view angle set to obtain an initial coverage plan map.

[0015] A verification and optimization module, configured to obtain the intersection historical traffic data; perform heat map conversion on the intersection historical traffic data to obtain intersection risk point distribution data; perform spatial overlay analysis on the initial coverage plan map according to the intersection risk point distribution data to obtain a layout verification result map; perform coverage difference adjustment on the initial coverage plan map according to the layout verification result map to obtain a layout optimization data set.

[0016] A scheme generation module, configured to reconstruct and adjust the intersection installation point set according to the layout optimization data set to obtain a final camera layout scheme.

[0017] Through the data acquisition module of the present invention, the three-dimensional point cloud data, the basic layer data map and the dynamic layer data map of the intersection can be quickly obtained. The model construction module and the layout planning module can automatically perform complex operations such as hydrodynamic modeling, flow field feature recognition, speed-density field simulation, and constraint evaluation and spherical projection calculation of camera installation points based on the collected data, and quickly generate an intersection flow field dynamic distribution model and an initial coverage plan map, significantly improving the efficiency and quality of layout planning. The model construction module can accurately construct an intersection traffic flow field model and accurately identify the distribution of flow field feature points, providing a scientific and reliable basis for camera layout and ensuring the accuracy and effectiveness of the layout scheme. The verification and optimization module uses the intersection historical traffic data for heat map conversion and spatial overlay analysis, and can perform coverage difference adjustment on the initial coverage plan map. The scheme generation module then reconstructs and adjusts the installation point set according to the optimization data set, and the finally obtained camera layout scheme is more reasonable and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Other features, objects, and advantages of the present invention will become more apparent by reading the detailed description with reference to the following drawings:

[0019] Figure 1 The step flow diagram of the modeled camera layout method according to an embodiment is shown.

[0020] Figure 2 Shows a detailed step - by - step flowchart of step S45 of an embodiment.

[0021] Figure 3 Shows a detailed step - by - step flowchart of step S5 of an embodiment. Specific embodiments

[0022] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those skilled in the art within the scope of the present invention without creative work belong to the scope of protection of the present invention.

[0023] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, so repeated descriptions of them will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0024] It should be understood that although terms such as "first", "second", etc. may be used here to describe each unit, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit can be called the second unit, and similarly the second unit can be called the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0025] To achieve the above - mentioned purpose, please refer to Figures 1 to 3 , the present invention provides a modeled camera layout method, including the following steps:

[0026] Step S1: Perform lidar scanning on the intersection area to obtain intersection three - dimensional point cloud data; perform static element recognition on the intersection three - dimensional point cloud data to obtain an intersection basic layer data map; collect vehicle and pedestrian flows in the intersection area to obtain an intersection dynamic layer data map;

[0027] Step S2: Based on the intersection basic layer data map and the intersection dynamic layer data map, perform hydrodynamic modeling on the intersection area to obtain intersection traffic flow field data; identify the flow field characteristics of the intersection traffic flow field data to obtain the distribution map of intersection flow field characteristic points; based on the distribution map of intersection flow field characteristic points, perform speed-density field simulation on the intersection area to obtain the dynamic distribution model of the intersection flow field.

[0028] Step S3: Evaluate the installation point constraints of the cameras in the intersection area to obtain the intersection installation point set; perform spherical projection calculation on the cameras according to the intersection installation point set to obtain the camera field of view angle set; perform spherical coverage visualization on the cameras according to the camera field of view angle set to obtain the initial coverage plan map.

[0029] Step S4: Obtain the historical traffic data of the intersection; perform heat map conversion on the historical traffic data of the intersection to obtain the distribution data of intersection risk points; perform spatial overlay analysis on the initial coverage plan map according to the distribution data of intersection risk points to obtain the layout verification result map; adjust the coverage difference of the initial coverage plan map according to the layout verification result map to obtain the layout optimization data set.

[0030] Step S5: Reconstruct and adjust the intersection installation point set according to the layout optimization data set to obtain the final camera layout plan.

[0031] In this embodiment, a Velodyne VLP-16 lidar is used to scan the intersection to obtain three-dimensional point cloud data. The CloudCompare software is used to identify static elements in the point cloud data and extract the basic layer data map of the intersection, including static information such as buildings and roads. At the same time, by deploying multiple high-definition cameras and radar sensors at the intersection, traffic flow and pedestrian flow data are collected to generate a dynamic layer data map of the intersection, recording the movement trajectories of vehicles and pedestrians. Based on the basic layer and dynamic layer data maps, the ANSYS Fluent software is used for fluid dynamics modeling to simulate the traffic flow field data of the intersection. By setting appropriate turbulence model parameters, such as the k-ε model, the distribution map of characteristic points in the flow field is identified, including the vortex core and critical points. Based on the distribution map of characteristic points, a velocity density field simulation is carried out to construct a dynamic distribution model of the intersection flow field. According to the physical environment and traffic flow characteristics of the intersection, the GIS software (such as QGIS) is used to evaluate the constraints of the camera installation positions, considering factors such as building occlusion, installation height, and power supply, and a suitable set of installation positions is selected. The Genetec Security Center software is used to calculate the spherical projection of the cameras to obtain the field of view angle set of each camera, and spherical coverage visualization is carried out to generate an initial coverage plan map. The historical traffic data of the intersection, including accident records and violation information, is obtained, and a heat map conversion is carried out using a Python script to generate the distribution data of risk points at the intersection. The distribution data of risk points is subjected to spatial overlay analysis with the initial coverage plan map, and the "layer overlay" function of QGIS is used to evaluate the coverage effect and identify areas with insufficient coverage and over-coverage. According to these analysis results, the initial coverage plan map is adjusted for coverage differences to generate a layout optimization data set. According to the layout optimization data set, the set of intersection installation positions is reconstructed and adjusted, and the Blender software is used for 3D modeling and perspective compensation calculation to ensure that the field of view of each camera can effectively cover the target area and reduce blind spots. Through the above steps, the final camera layout plan is obtained.

[0032] Preferably, step S1 includes the following steps:

[0033] Step S11: Use lidar to scan the intersection area to obtain three-dimensional point cloud data of the intersection;

[0034] Specifically, a lidar device of the Velodyne VLP-16 model can be used for lidar scanning. During the implementation process, the lidar device is installed on a mobile measurement vehicle, and the vehicle slowly drives through the intersection area at a speed of 10 kilometers per hour. During the scanning process, the lidar continuously emits laser pulses and receives the reflected signals. By calculating the flight time of the laser, the distance from each point to the lidar is accurately measured. At the same time, combined with the data of the vehicle's positioning system (such as GPS), the accurate position of each point in the geographical coordinate system is determined. After the scanning operation is completed, the three-dimensional point cloud data of the intersection area is obtained.

[0035] Step S12: Perform non-ground point clustering on the three-dimensional point cloud data of the intersection to obtain a set of candidate points for intersection feature targets;

[0036] Specifically, the density-based clustering algorithm DBSCAN (Density-Based Spatial Clustering of Applications with Noise) can be used for non-ground point clustering. The DBSCAN algorithm has two key parameters: radius ε (Epsilon) and minimum number of points MinPts. In this embodiment, according to the density distribution of the three-dimensional point cloud data of the intersection and the actual scene characteristics of the intersection, after multiple experimental adjustments, ε = 0.5 meters and MinPts = 10 are selected as the parameter values. In the specific operation, the point cloud data is imported into the point cloud processing software CloudCompare, and its built-in DBSCAN clustering function is used to analyze and process the point cloud data. The software will automatically identify the areas with higher density in the point cloud according to the set ε and MinPts parameters and divide them into different clusters. During the clustering process, the ground points are usually not assigned to any cluster because of their relatively uniform distribution and low density, while non-ground points such as vehicles, pedestrians, and traffic signs will form different clusters according to their spatial distribution. After the clustering process, multiple clusters containing non-ground points are obtained, and the points within each cluster have high spatial correlation. These clusters are the set of candidate points for intersection feature targets. For example, one cluster contains multiple points of a certain traffic sign, and another cluster contains the point cloud data of several vehicles waiting for the traffic light.

[0037] Step S13: Extract the building structure from the three-dimensional point cloud data of the intersection according to the set of candidate points for intersection feature targets to obtain the building structure data of the intersection environment;

[0038] Specifically, the 3D point cloud data of the intersection can be imported into the software Rapidform XOR, and the candidate point set of intersection feature targets can be loaded. In the software, a method based on plane fitting is adopted to identify the wall and roof structures of buildings. During specific operations, a building area in the point cloud is selected, and the plane fitting tool of Rapidform XOR is used to set the fitting accuracy to 0.01 meters. The software will automatically search for the best fitting plane within the selected area, and continuously optimize the plane parameters through the iterative closest point algorithm until the fitting accuracy requirements are met. For each wall and roof plane of the building, the above operations are repeated to gradually construct the main structure of the building. Finally, by integrating all the fitted planes and contour lines, the building structure data of the intersection environment is obtained.

[0039] Step S14: Identify road facilities from the candidate point set of intersection feature targets to obtain the location data of intersection infrastructure;

[0040] Specifically, a part of road facility samples, such as traffic signs, street lights, and signal lights, can be manually marked from the candidate point set of intersection feature targets as training data. These sample data include the geometric features of the point cloud (such as the coordinates and normal vectors of points) and statistical features (such as point cloud density and curvature). The support vector machine (SVM) is selected as the classifier, and the parameters of SVM are optimized through the grid search method. Finally, the penalty parameter C = 1.0 and the kernel function parameter gamma = 0.1 are determined. After the model training is completed, the entire candidate point set of feature targets is input into the trained SVM model for classification and recognition. The model will automatically judge whether each point belongs to a road facility according to the features of the point cloud and output the classification result. Finally, the location data of intersection infrastructure is obtained, and these data record the spatial locations and types of various road facilities.

[0041] Step S15: Perform feature fusion on the building structure data of the intersection environment and the location data of intersection infrastructure to obtain the basic layer data map of the intersection;

[0042] Specifically, the building structure data of the intersection environment and the location data of intersection infrastructure can be imported into ArcGIS respectively, and both of these data sets exist in the 3D vector format. In ArcGIS, using the "overlay analysis" function in the spatial analysis toolbox, the building structure data layer is overlaid with the infrastructure location data layer. During the overlay process, an appropriate buffer distance is set. For example, a 1-meter buffer is set for traffic signs. By setting the merging rules of attribute fields, the attribute information of the two layers is integrated. For example, the height information of the building and the functional type information of the infrastructure are merged into the same attribute table. Finally, the generated basic layer data map of the intersection not only contains the 3D structure of the building but also integrates the location and type information of the infrastructure.

[0043] Step S16: Conduct collaborative sensing on the intersection area to obtain the trajectory data of dynamic objects at the intersection, and extract spatio-temporal features from the trajectory data of dynamic objects at the intersection to obtain the dynamic layer data map of the intersection.

[0044] Specifically, one high-definition camera and one millimeter-wave radar can be installed in each of the four directions of the intersection. The resolution of the high-definition camera is 1920×1080 pixels, and the frame rate is 30fps, which can clearly capture the appearance features of vehicles and pedestrians; the detection distance of the millimeter-wave radar is 200 meters, and the angular resolution is 1°, which can obtain the speed and distance information of the target in real time. These sensors are synchronized and data-fused through a central control unit. During the data acquisition process, the video stream captured by the high-definition camera is sent into a computer vision algorithm for object detection and tracking. The YOLOv5 algorithm based on deep learning is used, which can quickly and accurately detect vehicles and pedestrians in the real-time video stream and assign a unique ID to each target. At the same time, the data of the millimeter-wave radar is converted into the speed and position information of the target through a radar signal processing algorithm. In the central control unit, the visual data of the camera and the sensing data of the radar are spatio-temporally aligned. Finally, the accurate trajectory data of the dynamic objects at the intersection is obtained, and this data includes information such as the ID, position, speed, and acceleration of the target. By extracting spatio-temporal features from these trajectory data, such as calculating traffic flow, average vehicle speed, and congestion index, the dynamic layer data map of the intersection is generated.

[0045] By obtaining the 3D point cloud data of the intersection, the present invention can accurately reflect the terrain, landform of the intersection, as well as the 3D shapes and positional relationships of various objects. Through the non-ground point clustering operation, it is possible to effectively screen out the candidate point sets of traffic-related feature targets from a large amount of point cloud data, such as vehicles, pedestrians, and traffic signs. By respectively identifying and extracting the building structures and road facilities, the physical environment framework of the intersection can be clearly outlined. Through feature fusion, the physical layout of the intersection is shown, and it can also provide a unified and comprehensive reference benchmark for traffic flow analysis, traffic planning, and camera layout. Through spatio-temporal feature extraction, the movement trajectories, speed changes, and flow distribution information of vehicles and pedestrians can be clearly depicted.

[0046] Preferably, step S2 includes the following steps:

[0047] Step S21: Rasterize the basic layer data map of the intersection to obtain the intersection environment calculation grid;

[0048] Specifically, the "rasterization" function of ArcGIS software can be used for rasterization to convert the intersection basic layer data map into a raster data format. During the operation, determine the resolution of the raster, and select a raster size of 0.5 m × 0.5 m according to the actual size of the intersection and the complexity of the traffic flow. During the rasterization process, the software will convert each element (such as buildings, roads, etc.) in the basic layer data map into raster cells and assign corresponding attribute values. For example, the raster cells in the road area are marked as 1, and the raster cells in the building area are marked as 0. Through the above steps, a computational grid of the intersection environment is obtained, and this grid covers the entire intersection area.

[0049] Step S22: Configure the boundary conditions for the intersection environment computational grid according to the intersection dynamic layer data map to obtain the initial condition dataset of the intersection flow field;

[0050] Specifically, the traffic engineering software VISSIM can be used to simulate the traffic flow at the intersection. In VISSIM, according to the actual traffic flow data and the geometric layout of the intersection in the intersection dynamic layer data map, set parameters such as the flow, vehicle speed, and vehicle type of each approach and exit lane. For example, for a main approach, set the flow during the peak period to 1800 vehicles / hour, the average vehicle speed to 30 km / h, and the vehicle type to mainly cars according to the traffic survey data. Map these traffic flow parameters to the corresponding boundaries of the intersection environment computational grid. During the specific operation, select the raster cells corresponding to the actual road boundaries in the computational grid and assign corresponding attribute values such as flow, vehicle speed, and vehicle type to these cells according to the VISSIM simulation results. For example, set the flow attribute of the raster cells at the approach boundary to 1800 vehicles / hour and the vehicle speed attribute to 30 km / h. According to the traffic signal control plan, set parameters such as the phase and green ratio of the traffic lights. Through the above steps, an initial condition dataset of the intersection flow field is obtained, and this dataset describes the initial state of the traffic flow at the intersection boundary.

[0051] Step S23: Set the turbulence model parameters for the initial condition dataset of the intersection flow field to obtain the turbulence characteristic data of the intersection flow field;

[0052] Specifically, the initial condition dataset of the intersection flow field can be imported into the software ANSYS Fluent. The k-ε turbulence model is selected to simulate the complexity and nonlinear characteristics of traffic flow. When setting the parameters of the turbulence model, according to the characteristics of traffic flow, appropriate turbulence intensity and length scale are selected. Specifically, the turbulence intensity is set to 5%, and the length scale is set to 1 meter. These parameter values are determined based on the observed data of actual traffic flow and empirical formulas. Through parameter setting, the turbulence characteristic data of the intersection flow field are obtained, including the kinetic energy and dissipation rate of turbulence. The turbulent kinetic energy TKE reflects the energy magnitude of the turbulent motion in traffic flow, while the dissipation rate ε describes the dissipation rate of turbulent energy. In ANSYS Fluent, the values of TKE and ε in each computational grid cell can be calculated by solving the turbulence model equations.

[0053] Step S24: Perform discretization solution of the flow field based on the turbulence characteristic data of the intersection flow field to obtain the numerical solution of the intersection traffic flow field;

[0054] Specifically, the computational grid of the intersection flow field can be discretized in ANSYS Fluent to convert the continuous flow field equation into a discrete algebraic equation system. In the discretization process, the finite volume method is adopted. Specifically in the operation, the second-order upwind scheme is selected to discretize the convection term. The SIMPLE algorithm (Semi-Implicit Method for Pressure-Linked Equations) is selected to couple the solution of velocity and pressure. In the solution process, the convergence criterion is set. When the residual drops to 10 -6 , the solution is considered to converge. Through iterative solution, the numerical solutions of the intersection traffic flow field are obtained, and these numerical solutions include the velocity field, pressure field, turbulence field, etc.

[0055] Step S25: Evaluate the convergence of the numerical solution of the intersection traffic flow field to obtain the stability evaluation result of the traffic flow field;

[0056] Specifically, the built-in convergence monitoring tool of the ANSYS Fluent software can be used for convergence evaluation. Multiple monitoring points are set in the solution process, and these monitoring points are distributed at key positions of the intersection, such as the entrance lane, exit lane, and intersection area. Specifically, 5 monitoring points are selected, which are located at the centerline of each entrance lane and the center position of the intersection area. In the solution settings of ANSYS Fluent, the residual monitoring function is enabled, and the residual threshold is set to 10 -6During the solution process, the software will calculate and display the residual values of each equation (such as the continuity equation, momentum equation, and turbulence equation) in real time. When the residual values of all equations drop below the set threshold and remain stable for 100 consecutive iterations, it is considered that the numerical solution has converged. In addition, the convergence can be further verified by observing the changing trends of the velocity and pressure at the monitoring points. If the changes in the velocity and pressure at the monitoring points are less than 0.1% in 100 consecutive iterations, it is considered that the numerical solution has also converged at these key positions. Through the above steps, the stability evaluation results of the traffic flow field are obtained.

[0057] Step S26: Extract the flow field characteristics of the numerical solution of the intersection traffic flow field according to the stability evaluation results of the traffic flow field to obtain the intersection traffic flow field data;

[0058] Specifically, the stability evaluation results of the traffic flow field can be referred to. Once it is confirmed that the numerical solution is stable, that is, the residual value drops below the set threshold and remains stable in multiple consecutive iterations, the flow field characteristics can be extracted. In the post-processing module of ANSYS Fluent, the "slice" function is used to extract the velocity field data. By creating multiple slices at different heights, velocity vector diagrams of different height levels at the intersection are generated. Then, the pressure field data is extracted to generate a pressure contour diagram. In addition, the turbulent kinetic energy and turbulent dissipation rate data are also extracted. Through the "isosurface" function, three-dimensional isosurface diagrams of the turbulent kinetic energy and turbulent dissipation rate are generated. Flow monitoring surfaces are set at the entrance and exit lanes. ANSYS Fluent automatically calculates and outputs the flow values of each monitoring surface. Finally, the extracted flow field characteristic data are exported as the intersection traffic flow field data.

[0059] Step S27: Identify the flow field characteristic points of the intersection traffic flow field data to obtain the distribution diagram of the intersection flow field characteristic points, and simulate the velocity density field of the intersection area based on the distribution diagram of the intersection flow field characteristic points to obtain the dynamic distribution model of the intersection flow field.

[0060] Specifically, for the detailed implementation process of this embodiment, please refer to the sub-steps of Step S27.

[0061] Through rasterization processing, the calculation of traffic flow can be carried out on more detailed units, enabling more accurate capture of the distribution and changes of traffic flow in various regions of the intersection. Through boundary condition configuration, the simulated traffic flow can better conform to the actual traffic operation rules, avoiding simulation errors caused by inaccurate boundary conditions and enhancing the reliability and credibility of the simulation results. Through the setting of turbulence model parameters, the complexity and nonlinear characteristics of traffic flow are considered. Through convergence evaluation, numerical instability problems occurring during the simulation process can be promptly detected, and corresponding measures can be taken for adjustment and optimization. Through flow field feature extraction, key feature information such as the speed, density, and flow rate of traffic flow can be screened out from complex traffic flow field data. Through the identification of flow field feature points, key nodes and regions in traffic flow can be clearly identified, such as traffic congestion points, accident-prone areas, and vehicle flow convergence areas. Through the simulation of the speed-density field, the speed and density distribution of traffic flow at the intersection and the changing trends of these distributions over time can be intuitively displayed.

[0062] Preferably, step S27 includes the following steps:

[0063] Step S271: Identify the vortex structure of the intersection traffic flow field data to obtain the intersection vortex core distribution map;

[0064] Specifically, the intersection traffic flow field data can be imported into the software Tecplot 360. In Tecplot 360, using its built-in vortex identification function, the Q-criterion is adopted to identify the vortex structure. The Q-criterion is a commonly used vortex identification method that identifies the vortex core by calculating the rotation intensity in the flow field. During specific operations, the threshold of the Q value is set to 100s -2 , and this threshold is determined based on the characteristics and experience of the actual flow field and is used to distinguish the vortex core and non-vortex regions. The software will automatically calculate the Q value of each point and identify the region with a Q value greater than the threshold as the vortex core. Through the above steps, the intersection vortex core distribution map is obtained, which intuitively shows the position and size of the vortex core within the intersection. For example, at the center of the intersection area, a relatively large vortex core is identified, indicating that there is an obvious rotation phenomenon in the traffic flow in this area.

[0065] Step S272: Identify the critical points of the intersection vortex core distribution map to obtain the position data of the flow field singularities at the intersection;

[0066] Specifically, based on the intersection vortex core distribution map, the critical point analysis function of Tecplot 360 software can be used, and the topological analysis method can be adopted to identify the singularities in the flow field. During the specific operation, calculate the velocity gradient tensor of the flow field, and determine the positions of the critical points by solving the eigenvalues and eigenvectors of the velocity gradient tensor. The critical points refer to the points where the velocity in the flow field is zero, and these points are the centers of the vortex cores or other important flow field characteristic points. In Tecplot 360, set the search range for the critical points, for example, search for critical points within each vortex core region. The software will automatically search and mark the positions of the critical points within the specified region, and output the coordinate information of the critical points. Through the above steps, the position data of the singularities in the intersection flow field is obtained, and these data record the spatial positions of each singularity. For example, within the vortex core at the center of the intersection area, a critical point is identified with coordinates (X: 50.0, Y: 30.0, Z: 0.0), which indicates that the traffic flow velocity is zero at this position and it is the center position of the vortex core.

[0067] Step S273: Classify the characteristics of the position data of the singularities in the intersection flow field to obtain the type data of the characteristic points in the intersection flow field;

[0068] Specifically, the position data of the singularities in the intersection flow field can be imported into MATLAB. In MATLAB, write a script to calculate the local flow field characteristics of each singularity, such as vorticity, divergence, and rotation direction. During the specific operation, use the gradient function in MATLAB to calculate the gradient of the velocity field, and then determine the characteristic types of each singularity by calculating the vorticity (curl of the velocity field) and divergence. According to the magnitude and direction of the vorticity, classify the singularities into types such as vortex centers, sinks, and sources. For example, the vorticity of the vortex center is relatively large and the direction is consistent, the divergence of the sink is negative, and the divergence of the source is positive. Set specific classification thresholds, such as singularities with vorticity greater than 50s -1 are classified as vortex centers, singularities with divergence less than -10s -1 are classified as sinks, and singularities with divergence greater than 10s -1 are classified as sources. Through the above operations, the type data of the characteristic points in the intersection flow field is obtained, and these data record the types and positions of each singularity. For example, within the vortex core at the center of the intersection area, a vortex center is identified with coordinates (X: 50.0, Y: 30.0, Z: 0.0), and the vorticity is 60s -1 and the direction is counterclockwise.

[0069] Step S274: Perform spatial clustering on the position data of the singularities in the intersection flow field according to the type data of the characteristic points in the intersection flow field to obtain the distribution map of the characteristic points in the intersection flow field;

[0070] Specifically, the data of the types of characteristic points of the intersection flow field and the position data of the singular points of the intersection flow field can be imported into MATLAB. In MATLAB, the k-means clustering algorithm is used to perform spatial clustering on the characteristic points. During specific operations, according to the actual size of the intersection and the distribution of the characteristic points, an appropriate number of clustering centers k is selected. For example, for a complex intersection, k = 6 is selected. The kmeans function in MATLAB is used for clustering, the input data is the coordinates of the characteristic points, and the output is the clustering label to which each characteristic point belongs. Through the above steps, a distribution map of the characteristic points of the intersection flow field is obtained, which intuitively shows the spatial distribution of different types of characteristic points within the intersection. For example, the vortex centers are mainly concentrated in the center of the intersection area, the sinks are mainly distributed at the ends of the approach roads, and the sources are mainly distributed at the starting positions of the departure roads.

[0071] Step S275: Perform velocity field interpolation on the distribution map of the characteristic points of the intersection flow field to obtain the distribution data of the intersection velocity field;

[0072] Specifically, the velocity data (including the velocity vectors of each characteristic point) in the distribution map of the characteristic points of the intersection flow field can be imported into MATLAB. In MATLAB, the griddata function is used for velocity field interpolation. During specific operations, the bilinear interpolation method is selected. The resolution of the interpolation grid is set, for example, a grid size of 0.5 m × 0.5 m is selected. Through the griddata function, the velocity data of the characteristic points is interpolated onto the grid of the entire intersection area to generate the distribution data of the intersection velocity field. These data are stored in the form of a two-dimensional array, and each array element corresponds to the velocity vector of a grid point.

[0073] Step S276: Derive the density field based on the distribution data of the intersection velocity field to obtain the distribution data of the intersection density field;

[0074] Specifically, the distribution data of the intersection velocity field can be imported into MATLAB. In MATLAB, the continuity equation is used to derive the density field distribution. The continuity equation describes the mass conservation of the fluid and has the form: where ρ is the density and u is the velocity vector. During specific operations, it is assumed that the traffic flow is steady-state, that is The finite difference method is used to discretize this equation, and the density field distribution is obtained by iterative solution. During the iteration process, the initial density distribution is set. For example, it is assumed that the density at the intersection entrance is 100 vehicles per square kilometer, and then the density value is gradually adjusted according to the velocity field distribution until the continuity equation is satisfied. The convergence criterion is set. When the change in the density field is less than 0.1%, the solution is considered to have converged. Through the above steps, the density field distribution data of the intersection are obtained, and these data are stored in the form of a two-dimensional array, where each array element corresponds to the density value of a grid point. For example, at a grid point in the center of the intersection area, the derived density value is 150 vehicles per square kilometer, indicating a high density of traffic flow at this location.

[0075] Step S277: Perform temporal evolution on the intersection velocity field distribution data and the intersection density field distribution data to obtain the time-varying characteristic data of the intersection flow field;

[0076] Specifically, the intersection velocity field distribution data and the intersection density field distribution data can be imported into the Python environment. These data are stored in the form of a time series, and each time step corresponds to a two-dimensional array of the velocity field and the density field. The time step is set to 1 second, and the simulation duration is 60 seconds. At each time step, using the array operation function in the NumPy library, the velocity field and density field data are superimposed and analyzed to calculate the change rates of the velocity and density at each grid point. For example, at the 10th second, the velocity of a grid point in the center of the intersection area changes from (u: 2.5 m / s, v: -1.8 m / s) to (u: 2.2 m / s, v: -2.0 m / s), and the density changes from 150 vehicles per square kilometer to 160 vehicles per square kilometer. Through the above steps, the time-varying characteristic data of the intersection flow field are obtained, and these data record the velocity and density changes of each grid point in the time series.

[0077] Step S278: Perform mode decomposition on the time-varying characteristic data of the intersection flow field to obtain the dominant mode data of the intersection flow field, and perform reconstruction integration on the dominant mode data of the intersection flow field to obtain the dynamic distribution model of the intersection flow field.

[0078] Specifically, the SciPy library in Python can be used, especially the singular value decomposition (SVD) function in its signal processing module, for mode decomposition. The time-varying feature data of the intersection flow field (including the time series of speed and density) is organized into a matrix, where each row represents a time step and each column represents the eigenvalue of a grid point. Perform SVD decomposition on this matrix to extract the main singular values and the corresponding singular vectors. By setting a threshold, for example, retaining the singular values with a cumulative contribution rate reaching 90%, the dominant modes of the intersection flow field are screened out. These dominant modes represent the main movement patterns and trends of traffic flow. For example, the first dominant mode represents the congestion pattern of traffic flow during peak hours, and the second dominant mode represents the rapid change pattern of traffic flow when the traffic signal changes. Use the data of these dominant modes for reconstruction integration. By recombining the screened singular values and singular vectors, a dynamic distribution model of the intersection flow field is generated. This model shows the main dynamic characteristics of traffic flow in the form of a time series, and each time step corresponds to a reconstructed speed field and density field distribution. Finally, use the Matplotlib library to visualize the reconstructed dynamic distribution model of the flow field and generate a dynamic distribution map.

[0079] Through vortex structure identification and critical point identification, the present invention can deeply reveal complex flow phenomena in intersection traffic flow, such as vortices and stagnation points. Through feature classification and spatial clustering, the distribution of different types of feature points in the intersection flow field can be clearly shown. Through velocity field interpolation and density field derivation, the changes in the speed and density of traffic flow can be intuitively shown. Through time series evolution analysis, the change trend of traffic flow over time can be captured. Through mode decomposition and reconstruction integration, the main movement patterns and trends of traffic flow can be highlighted.

[0080] Preferably, step S3 includes the following steps:

[0081] Step S31: Evaluate the camera installation point constraints in the intersection area to obtain the intersection installation point set;

[0082] Specifically, for the detailed implementation process of this embodiment, please refer to the sub-steps of step S31.

[0083] Step S32: Evaluate the field of view accessibility of the intersection installation point set to obtain the intersection space visibility data;

[0084] Specifically, in the Genetec Security Center software, according to the set of installation points at intersections, input the specific parameters of each camera, such as focal length and aperture size, and combine with the 3D model of the intersection to simulate the field of view of each camera. The software will automatically calculate the visible area of each camera. Set the evaluation criteria for field of view accessibility. For example, the visible area of each camera should cover at least one key area of the intersection, such as a crosswalk, traffic lights, or the main traffic flow direction. At the same time, also consider the situation of field of view overlap to ensure that there is at least two cameras' field of view overlap in the key area. Through the visualization function of the software, the visible area of each camera can be seen, and the installation points that do not meet the requirements can be adjusted. Finally, obtain the intersection space visible area data, which record the visible area range of each camera.

[0085] Step S33: Configure the camera parameters for the intersection area according to the set of installation points at intersections and the intersection space visible area data to obtain the camera optical parameter set, and perform spherical projection modeling on the intersection area based on the camera optical parameter set to obtain the single-point projection data set of the monitoring area;

[0086] Specifically, the set of installation points at intersections and the visible area data can be imported into the camera configuration Milestones XProtect software. In the software, according to the specific location and visible area range of each installation point, configure appropriate optical parameters for each camera. For example, for a camera installed at a higher position and needing to cover a wider area, select a wide-angle lens with a horizontal field of view angle set to 120 degrees and a vertical field of view angle set to 90 degrees; for a camera installed at a lower position and needing to monitor a specific area, select a standard lens with a horizontal field of view angle set to 70 degrees and a vertical field of view angle set to 50 degrees. At the same time, also adjust the aperture size and focal length according to the lighting conditions and the size of the monitoring target. After configuration, obtain the camera optical parameter set, and these parameters include information such as the lens type, field of view angle, aperture size, and focal length of each camera. Use 3D modeling software (such as 3ds Max) to perform spherical projection modeling on the intersection area. Import the camera optical parameter set into the 3D modeling software, and create a spherical projection model according to the installation position and optical parameters of each camera. In the model, the field of view of each camera is projected onto a sphere to form the single-point projection data set of the monitoring area. These data sets record the projection position and coverage range of each camera on the sphere.

[0087] Step S34: Calculate the field of view angle according to the single-point projection data set of the monitoring area to obtain the monitoring area field of view coverage data set, and perform spatial registration transformation on the monitoring area field of view coverage data set to obtain the camera field of view angle set;

[0088] Specifically, the single-point projection data set of the monitoring area can be imported into GIS software. These data include the projection positions and coverage ranges of each camera on the spherical surface. According to the optical parameters and projection positions of the cameras, the actual field of view angles of each camera are calculated. Specifically, using spherical geometry formulas and combining the horizontal and vertical field of view angles of the camera, the actual coverage angles on the spherical surface are calculated. For example, for a camera with a horizontal field of view angle of 120 degrees and a vertical field of view angle of 90 degrees, the actual coverage angles on the spherical surface are calculated to be 100 degrees (horizontal) and 80 degrees (vertical). Through the above steps, the field of view coverage data set of the monitoring area is obtained, and these data record the actual field of view angles and coverage ranges of each camera. Next, spatial registration transformation is performed on the field of view coverage data set of the monitoring area. In GIS software, spatial registration tools are used to match the field of view angle data of each camera with the geographic coordinate system of the intersection. According to the installation positions and projection positions of the cameras, the field of view angle data is converted into polygon layers in the geographic coordinate system. For example, some areas are overlapped and covered by the fields of view of multiple cameras, while some areas have insufficient coverage. Through the above steps, the set of camera field of view angles is obtained, and these data include not only the field of view angle information of each camera, but also the distribution of the field of view in the geographic space.

[0089] Step S35: Perform spherical dissection of the observation space of the intersection area according to the set of camera field of view angles to obtain the spherical grid data of the monitoring area, and calculate the coverage of the spherical grid data of the monitoring area to obtain the coverage rate distribution map of the monitoring area;

[0090] Specifically, the set of camera field of view angles can be imported into 3D modeling software (such as Blender). These data include the field of view angles and coverage ranges of each camera. In Blender, a spherical model with the intersection center as the center of the sphere is created, and the sphere is dissected according to the set of camera field of view angles. During specific operations, the "knife cut" tool in Blender is used to cut the sphere into multiple small regions according to the field of view boundaries of each camera. Each small region represents the coverage range of a camera, forming the spherical grid data of the monitoring area. Next, a custom Python script is used to calculate the coverage of each spherical grid. The script will traverse each spherical grid and count how many cameras cover the grid. For example, if a spherical grid is covered by 3 cameras, the coverage of the grid is 3. Through the above steps, the coverage information of the spherical grid data of the monitoring area is obtained. Finally, the coverage information is imported into GIS software (such as QGIS) to generate the coverage rate distribution map of the monitoring area.

[0091] Step S36: Perform three-dimensional visualization on the intersection area based on the coverage rate distribution map of the monitoring area to obtain the initial coverage plan map.

[0092] Specifically, the monitoring area coverage rate distribution map can be imported into ArcGIS Pro. In ArcGIS Pro, using the "3D Scene" function, the coverage rate distribution map is converted into a three-dimensional scene model. This model is based on the geographic coordinate system of the intersection, and the coverage information of each spherical grid is represented in the form of height. Next, this three-dimensional coverage topographic map is imported into Unity 3D for further visualization processing. In Unity 3D, using the "Material" and "Lighting" tools, realistic material effects and lighting effects are added to the three-dimensional coverage topographic map. For example, bright-colored materials are added to areas with higher coverage, dark-colored materials are added to areas with lower coverage, and by adjusting the lighting angle and intensity, the three-dimensional model becomes more realistic. At the same time, a camera model is added, and according to the actual installation position and field of view angle, the direction and perspective of the camera model are adjusted. Through the above steps, the initial coverage plan map is obtained.

[0093] Through the installation point constraint evaluation of the present invention, various actual factors such as building occlusion, installation height, power supply, and network transmission can be comprehensively considered to screen out a set of eligible installation points. Through the field of view accessibility evaluation, the visible range of each installation point can be determined. Through the camera parameter configuration, the most suitable optical parameters for each camera, such as focal length and aperture, can be set. Through the spherical projection modeling and field of view angle calculation, the monitoring range of each camera can be accurately determined, and through the coverage rate calculation of the spherical grid data in the monitoring area, it helps to intuitively display the coverage effect of the monitoring system and discover potential monitoring blind spots. Through the three-dimensional visualization, the effect of the camera layout and the coverage of the monitoring area can be intuitively displayed.

[0094] Preferably, step S31 includes the following steps:

[0095] Step S311: Obtain the three-dimensional point cloud data of the intersection, perform spatial partition slicing on the three-dimensional point cloud data of the intersection to obtain the intersection area slice data, and perform building boundary extraction on the intersection area slice data to obtain the intersection building contour data;

[0096] Specifically, the 3D point cloud data of the intersection collected in step S11 can be retrieved from a database. The 3D point cloud data of the intersection is imported into CloudCompare, and the "cutting" function of the software is used to horizontally slice the point cloud data at a preset height interval (for example, every 1 meter), thereby obtaining point cloud slice data of multiple different height layers. Then, building boundary extraction is performed on each slice of data. In CloudCompare, the "plane fitting" and "edge detection" tools are used. By setting the fitting accuracy to 0.05 meters, the wall and roof planes of the building are identified, and the contour lines of the building are extracted. These contour lines are saved in vector format to form the contour data of the intersection building.

[0097] Step S312: Identify the elevation features of the intersection area based on the contour data of the intersection building to obtain the building surface feature data of the intersection;

[0098] Specifically, the contour data of the intersection building can be imported into the CloudCompare software, and a specific building elevation can be selected for analysis. The "feature extraction" function of the software is used, which can identify geometric features in the point cloud data, such as windows, doors, and balconies. During the operation, the parameters for feature recognition are set. For example, the minimum size of a window is 0.5 meters × 0.5 meters, and the minimum size of a door is 0.8 meters × 2.0 meters. With these parameters, the software can automatically identify the positions and sizes of the windows and doors on the building elevation. In addition, the "curvature analysis" tool can be used to identify the protruding parts on the elevation, such as balconies. By setting the curvature threshold to 0.1, the software can highlight the areas with large curvature changes on the elevation, and these areas usually correspond to balconies or other protruding structures. The identified features are saved in the form of a point cloud subset, and their geometric center coordinates and size information are extracted to form the building surface feature data of the intersection. For example, it is identified that there are multiple windows on a building elevation, and the center coordinates and sizes of each window are recorded. These data include not only the positions of the windows but also the aspect ratios and areas of the windows.

[0099] Step S313: Constrain the installation height of the camera based on the building surface feature data of the intersection to obtain the installation height constraint data, and select candidate points in the intersection area based on the installation height constraint data to obtain the candidate set of intersection installation points;

[0100] Specifically, the building surface feature data can be imported into the GIS software. This data includes the positions and dimensions of features such as windows, doors, and balconies. According to the field of view angle and monitoring range requirements of the camera, the minimum installation height of the camera is set to 3 meters to ensure that the traffic conditions on the ground can be covered and at the same time avoid being blocked by the heads of pedestrians. For cameras installed on buildings, the height of the windows is also considered to ensure that the installation position of the camera is at least 0.5 meters above the upper edge of the window to avoid privacy issues. During specific operations, a Python script is written to traverse each building surface feature and calculate the eligible installation height range. For example, for a building surface with a window height of 2 meters, the installation height of the camera is restricted between 2.5 meters and 5 meters. Through the above steps, the installation height constraint data is obtained, and a candidate set of eligible installation points is generated in the GIS software. These candidate points are presented in the form of a point layer, and each point is attached with its installation height information.

[0101] Step S314: Conduct an assessment of the power supply for the candidate set of intersection installation points to obtain a dataset of the power supply conditions for the installation points;

[0102] Specifically, the power system planning software and the data from on-site surveys can be used. The candidate set of intersection installation points is imported into the power system planning software, which can analyze the distribution and power supply capacity of the power network. According to the positions of the power facilities and the cable routes obtained from the on-site survey, the evaluation parameters for power supply are set. For example, the maximum power of each camera is 50 watts, and the allowable voltage drop is 5%. The software will automatically calculate the distance from each candidate point to the nearest power facility and the voltage drop of the cable, and determine whether the point can meet the power supply requirements. During specific operations, the power supply conditions of each candidate point are evaluated. If the voltage drop of a point exceeds the allowable value, or the distance from the power facility exceeds 100 meters (considering the cable laying cost and safety), then this point is marked as having poor power supply conditions. Through the above steps, a dataset of the power supply conditions for the installation points is obtained, and these data record the power supply feasibility of each candidate point. For example, several candidate points located at a corner of the intersection have poor power supply conditions due to their long distances from the power facilities, and additional power supply solutions such as solar power supply or laying new cables need to be considered.

[0103] Step S315: Conduct a network transmission assessment on the candidate set of intersection installation points based on the dataset of the power supply conditions for the installation points to obtain a dataset of the network coverage for the installation points, and conduct a lightning protection and grounding assessment on the candidate set of intersection installation points based on the installation height constraint data to obtain a dataset of the lightning protection conditions for the installation points;

[0104] Specifically, the power supply condition data set of the installation points can be imported into the network planning software CISCO NetworkPlanner, which can analyze the network coverage and signal strength. According to the location of the on-site network infrastructure, such as base stations and fiber optic nodes, evaluation parameters for network transmission are set. For example, the network bandwidth requirement for each camera is 10 Mbps, and the allowed signal attenuation is 30 dB. The software will automatically calculate the distance and signal attenuation from each candidate point to the nearest network facility, and determine whether this point can meet the network transmission requirements. During specific operations, a network coverage assessment is performed on each candidate point. If the signal attenuation of a point exceeds the allowed value, or the distance from the network facility exceeds 200 meters (considering signal strength and transmission cost), then this point is marked as having poor network coverage. At the same time, a lightning protection and grounding assessment is performed on the candidate set of intersection installation points according to the installation point height constraint data. Using the lightning protection design software Lightning Protection Designer, according to the local lightning activity data and the lightning protection level of the building, evaluation parameters for lightning protection and grounding are set. For example, the grounding resistance should be less than 10 ohms. The software will automatically calculate the grounding resistance of each candidate point and determine whether this point meets the lightning protection and grounding requirements. Through the above steps, the network coverage data set of the installation points and the lightning protection condition data set of the installation points are obtained. For example, it is found that several candidate points located at high positions at intersections have poor network coverage due to their long distance from network facilities. At the same time, these high-position points have poor lightning protection conditions due to their high grounding resistance.

[0105] Step S316: Based on the intersection 3D point cloud data and according to the installation point height constraint data, perform occlusion object recognition on the candidate set of intersection installation points to obtain the installation point occlusion interference data set;

[0106] Specifically, the intersection 3D point cloud data and the installation point height constraint data can be imported into the CloudCompare software. In the software, use the "line of sight analysis" tool, which can calculate whether the line of sight from each candidate point to the preset monitoring area is occluded. During specific operations, set the boundaries of the monitoring area and the key monitoring points, and then perform line of sight analysis on each candidate point. For example, set the monitoring area as the vehicle passing area and the crosswalk at the intersection, and the key monitoring points as traffic lights and zebra crossings. The software will automatically calculate the line of sight paths from each candidate point to these key monitoring points and mark the occluded lines of sight. If the line of sight of a candidate point is occluded by more than 50% of the key monitoring points, then this point is marked as having severe occlusion interference. Through the above steps, the installation point occlusion interference data set is obtained. For example, it is found that several candidate points located at the corners of buildings have severe occlusion interference due to being occluded by the building itself or other obstacles.

[0107] Step S317: Comprehensively screen the candidate set of intersection installation points based on the power supply condition dataset of installation points, the network coverage dataset of installation points, the lightning protection condition dataset of installation points, and the occlusion interference dataset of installation points to obtain the intersection installation point set.

[0108] Specifically, the power supply condition dataset of installation points, the network coverage dataset of installation points, the lightning protection condition dataset of installation points, and the occlusion interference dataset of installation points can be imported into the GIS software. These datasets respectively contain the power supply feasibility, network coverage, lightning protection conditions, and occlusion interference conditions of each candidate point. In the GIS software, a comprehensive evaluation model is created, and this model scores each dataset according to the preset weights. Specifically, the power supply condition accounts for 30% of the weight, the network coverage accounts for 30% of the weight, the lightning protection condition accounts for 20% of the weight, and the occlusion interference accounts for 20% of the weight. The score range of each dataset is from 0 to 100 points, where 100 points represents the best condition. For example, if a candidate point has good power supply conditions (score 90 points), good network coverage (score 85 points), average lightning protection conditions (score 60 points), and less occlusion interference (score 80 points), then its comprehensive score is calculated as follows: Comprehensive score = 0.3×90 + 0.3×85 + 0.2×60 + 0.2×80 = 79.5; Set a threshold for the comprehensive score, such as 70 points. Only the candidate points with a comprehensive score higher than 70 points are considered qualified installation points. Through the above steps, the qualified intersection installation point set is screened out. For example, several candidate points located above the center of the intersection have high comprehensive scores due to good power supply and network conditions and no occlusions, and are selected as the final installation points. While several candidate points located on the back of the building are excluded due to severe occlusion interference and low comprehensive scores. Finally, the screened intersection installation point set is exported as a GIS layer file, and each point is attached with evaluation information, including power supply conditions, network coverage, lightning protection conditions, and occlusion interference conditions.

[0109] Through spatial partitioning slicing and building boundary extraction, the present invention can accurately locate the building contour of the intersection. Through elevation feature recognition, the surface features of the building, such as windows and balconies, can be understood. By restricting the installation height of the camera, it can ensure that the camera is installed in a suitable position, which can not only obtain the best monitoring perspective but also ensure the safety and stability of the installation. Through the comprehensive evaluation of power supply, network transmission, lightning protection grounding, and occlusions, various conditions in actual installation and operation can be comprehensively considered, ensuring that the selected installation points are not only technically feasible but also sustainable in terms of economy and maintenance, reducing monitoring system failures and maintenance costs caused by insufficient infrastructure. Through comprehensive screening, the most suitable installation points can be selected from numerous candidate points.

[0110] Preferably, step S4 includes the following steps:

[0111] Step S41: Obtain the historical traffic data of the intersection, and perform effective accident extraction on the historical traffic data of the intersection to obtain the effective accident data of the intersection;

[0112] Specifically, all traffic accident reports of this intersection in the past year can be obtained from the traffic management department. These reports are provided in the form of spreadsheets and contain information such as accident time, location, accident type, and number of injured people. Import these data into GIS software (such as ArcGIS), and use its spatial analysis function to match the accident data with the geographical coordinates of the intersection. Next, screen these accident data to extract the effective accident data. The definition of an effective accident is those accidents that result in casualties or property losses exceeding a certain amount (for example, property losses exceeding 5000 yuan). In ArcGIS, use the "Select" tool to screen out the accident records that meet the conditions according to the accident type and property loss fields. These screened accident records are saved as the effective accident data of the intersection.

[0113] Step S42: Extract violation information from the preset traffic violation database to obtain the effective violation data of the intersection;

[0114] Specifically, all traffic violation records of this intersection in the past year can be obtained from the traffic enforcement department. These records are stored in the database and contain information such as violation time, location, violation type, and penalty result. Use the SQL query tool to connect to this database and write SQL query statements to extract specific types of violation records. For example, the violation types of concern include speeding, running a red light, and not driving in the prescribed lane. The query results contain the information of each violation record. Import these data into GIS software (such as QGIS) for spatial analysis to ensure that each violation record can be accurately located at the specific location of the intersection. Through the above steps, the effective violation data of the intersection is obtained.

[0115] Step S43: Extract the intersection traffic flow data from the preset real-time traffic monitoring system to obtain the intersection traffic flow statistics data;

[0116] Specifically, one can log in to a traffic flow monitoring system (such as VantagePoint of Peek Traffic), which collects traffic flow data in real time through induction coils and video cameras installed at intersections. In the system, select a specific intersection and time range (for example, the morning rush hour in the past month, that is, from 7 am to 9 am every day). In the VantagePoint system, use the built-in data export function to export the traffic flow data within the selected time period. This data includes information such as the number of vehicle passages on each lane, vehicle types (such as cars, trucks, buses), and average vehicle speed. For example, export the traffic flow data of a certain intersection during the morning rush hour. The data shows that during this period, the average traffic flow on each lane is 1,200 vehicles per hour, of which cars account for 70%, trucks account for 20%, and buses account for 10%. Import the file into data processing software (such as Microsoft Excel or the Pandas library in Python) for preliminary data cleaning and sorting. Calculate statistical data such as total traffic flow, traffic flow of each vehicle type, and average vehicle speed to form intersection traffic flow statistical data.

[0117] Step S44: Identify traffic flow changes in the intersection traffic flow statistical data to obtain intersection traffic flow change data;

[0118] Specifically, the intersection traffic flow statistical data can be imported into the Python environment. Use the Pandas library to read the CSV file and convert the data into a time series format. When operating specifically, write a Python script to calculate the traffic flow change rate for each time period (for example, every 15 minutes). The calculation formula for the traffic flow change rate is: For example, the traffic flow at a certain intersection during the morning rush hour increased from 1,000 vehicles per hour to 1,200 vehicles between 7:30 and 7:45, and the change rate is 20%. Through the above operations, intersection traffic flow change data is obtained. Next, use the Matplotlib library to visualize these traffic flow change data and generate a traffic flow change curve graph. In the curve graph, the horizontal axis represents time, and the vertical axis represents the traffic flow change rate.

[0119] Step S45: Perform spatial distribution fusion on the intersection effective accident data, intersection effective violation data, and intersection traffic flow change data to obtain intersection risk point distribution data, and perform spatial overlay analysis on the initial coverage plan map based on the intersection risk point distribution data to obtain a layout verification result map; perform coverage difference adjustment on the initial coverage plan map according to the layout verification result map to obtain a layout optimization data set.

[0120] Specifically, for the detailed implementation process of this embodiment, please refer to the sub-steps of Step S45.

[0121] By extracting effective accident and violation information, the present invention can identify the high-incidence areas of accidents and violations at intersections. By identifying the changing trends of traffic flow, it can make the monitoring layout better adapt to the dynamic changes of traffic flow. Through spatial distribution fusion, it can accurately locate the risk areas at intersections. Through spatial overlay analysis, it can visually display the matching degree between the monitoring layout and the risk points. Through coverage difference adjustment, the monitoring layout can be dynamically adjusted according to the changes in the actual traffic conditions. Through the monitoring system, potential traffic problems and risk points can be identified in advance to achieve preventive monitoring. Through risk point location and layout optimization, over-configuration of monitoring devices in low-risk areas is avoided.

[0122] Preferably, step S45 includes the following steps:

[0123] Step S451: Perform spatial density mapping on the effective accident data of the intersection to obtain the accident heat data of the intersection;

[0124] Specifically, the effective accident data of the intersection can be imported into QGIS. In QGIS, use the "Heatmap" plugin, which can generate a heatmap based on point data. During specific operations, set the parameters of the heatmap, such as the search radius (Radius) and the decay coefficient (Decay). The search radius determines the range affected by each accident point. According to the size of the intersection and the density of accident distribution, 100 meters is selected as the search radius. The decay coefficient determines the decay rate of the influence of the accident point on the surrounding area. The Gaussian decay function is selected, and its decay coefficient is set to 0.5. Through parameter setting, the heatmap plugin of QGIS will automatically calculate the accident density at each location and generate a continuous heatmap. The generated heatmap represents the accident density by the depth of color, and the darker the color, the denser the accidents. Save this heatmap as a GeoTIFF format file to form the accident heat data of the intersection.

[0125] Step S452: Perform spatial clustering segmentation on the effective violation data of the intersection to obtain the violation heat data of the intersection;

[0126] Specifically, the effective intersection violation data can be imported into the Python environment, and the DBSCAN algorithm in the Scikit-learn library can be used for spatial clustering. Set the parameters of the DBSCAN algorithm, such as the radius ε (Epsilon) and the minimum number of points MinPts. According to the distribution density of the violation data, select 50 meters as the radius ε and 5 as the minimum number of points MinPts. Through parameter setting, the DBSCAN algorithm will automatically divide the violation data into multiple clusters, and each cluster represents a violation hot spot area. After clustering is completed, import the clustering results into QGIS, and use the "heat map" plugin to generate a violation heat map. Similarly, set the parameters of the heat map, such as the search radius and the attenuation coefficient. Select 100 meters as the search radius, and set the attenuation coefficient of the Gaussian attenuation function to 0.5. The generated heat map represents the density of violations by the depth of color, and the darker the color, the denser the violations. Save this heat map as a GeoTIFF format file to form the intersection violation heat data.

[0127] Step S453: Conduct a congestion assessment on the intersection traffic flow change data to obtain the intersection congestion heat data, and perform a spatial overlay on the intersection accident heat data, the intersection violation heat data, and the intersection congestion heat data to obtain the intersection risk point distribution data;

[0128] Specifically, the intersection traffic flow change data can be imported into the Synchro software. In Synchro, use the built-in congestion assessment model, which calculates the congestion index based on the traffic flow change rate and the average vehicle speed. The calculation formula for the congestion index is: For example, the congestion index of a certain intersection during the morning rush hour is relatively high, indicating that the traffic congestion is relatively serious during this period. Through the above steps, the intersection congestion heat data is obtained, and these data record the congestion situation in each time period. Import the intersection accident heat data, the intersection violation heat data, and the intersection congestion heat data into QGIS. In QGIS, use the "layer overlay" function to perform a spatial overlay on these three heat maps. Specifically, when operating, set the weight of each heat map. For example, the weight of the accident heat map is 0.4, the weight of the violation heat map is 0.3, and the weight of the congestion heat map is 0.3. Through weighted overlay, a comprehensive risk point distribution map is generated. This map represents the level of risk by the depth of color, and the darker the color, the higher the risk. Save this comprehensive risk point distribution map as a GeoTIFF format file to form the intersection risk point distribution data.

[0129] Step S454: Evaluate the matching degree of the initial coverage plan map based on the intersection risk point distribution data to obtain the intersection coverage matching data;

[0130] Specifically, the initial coverage plan map and the distribution data of intersection risk points can be imported into QGIS. In QGIS, use the "Spatial Analysis" tool to perform a spatial overlay analysis on the initial coverage plan map and the risk point distribution map. During the specific operation, calculate the risk point density within the coverage area of each camera. For example, define a risk point density threshold, such as 50 risk points per square kilometer. If the risk point density within the coverage area of a camera exceeds this threshold, it is considered that the coverage matching degree of this camera is relatively high; if it is lower than this threshold, the matching degree is considered low. Next, calculate the matching degree score according to the risk point density within the coverage area of each camera. The calculation formula for the matching degree score is: For example, the risk point density within the coverage area of a certain camera is 60 per square kilometer, exceeding the threshold of 50, and its matching degree score is 120%. Through the above steps, the matching degree scores of each camera are obtained to form the intersection coverage matching data.

[0131] Step S455: Perform a difference evaluation on the intersection coverage matching data to obtain the intersection coverage difference data;

[0132] Specifically, the intersection coverage matching data can be imported into statistical analysis software (such as R). In R, use the descriptive statistical analysis method to calculate the mean, standard deviation, minimum value, and maximum value of the matching degree scores. For example, the mean of the matching degree scores of all cameras is 80%, the standard deviation is 15%, the minimum value is 40%, and the maximum value is 120%. Define the threshold for difference evaluation. For example, mark the cameras with a matching degree score lower than 60% as under-covered, and mark the cameras with a score higher than 100% as over-covered. Import the evaluation results into QGIS and visualize them in combination with the initial coverage plan map. In QGIS, use the "Classification Symbols" function to classify and display the cameras according to the coverage difference. For example, mark the under-covered cameras in red, the over-covered cameras in blue, and the cameras with a moderate matching degree in green. Through the above steps, the intersection coverage difference data is obtained.

[0133] Step S456: Adjust the initial coverage plan map according to the intersection coverage difference data to obtain the layout optimization data set.

[0134] Specifically, the intersection coverage difference data can be imported into QGIS. In QGIS, use the "Edit" function to add new camera points in areas with insufficient coverage and adjust the orientation of cameras or reduce the number of cameras in areas with excessive coverage. For example, in areas with insufficient coverage, new cameras are added at locations with a high density of risk points, and the orientation of adjacent cameras is adjusted. In areas with excessive coverage, some cameras are removed. The adjusted camera layout plan is imported into Genetec Security Center for further simulation and verification. In Genetec Security Center, according to the adjusted layout plan, the field of view of each camera is simulated. Through the above operations, a layout optimization dataset is obtained, which records the adjusted camera positions, orientations, and coverage ranges.

[0135] By generating accident heat data and violation heat data, the present invention can intuitively display the high-incidence areas of accidents and violations. By spatially overlaying the accident, violation, and congestion heat data, the traffic risk status of intersections can be comprehensively reflected. Through the matching degree evaluation, the degree of fit between the existing monitoring layout and the actual risk points can be accurately evaluated. This helps to identify the deficiencies in the monitoring layout, such as monitoring blind spots or areas with insufficient coverage. Through the difference evaluation, the optimization plan for the monitoring layout can be further refined. By identifying areas with large coverage differences, the installation positions, angles, or numbers of cameras can be adjusted accordingly to ensure that monitoring resources can be more reasonably allocated to high-risk areas. Through the coverage difference adjustment, dynamic adjustment can be made according to changes in the actual traffic conditions.

[0136] Preferably, step S5 includes the following steps:

[0137] Step S51: Evaluate the spatial distribution of the intersection installation point set according to the layout optimization dataset to obtain installation point distribution feature data;

[0138] Specifically, the layout optimization dataset can be imported into QGIS. In QGIS, use the "Kernel Density Estimation" (KDE) tool, which can calculate the camera density at each location. During specific operations, set the parameters of the kernel density estimation, such as the search radius and resolution. According to the size of the intersection and the distribution density of the cameras, select 50 meters as the search radius and 1 meter as the resolution. Through parameter settings, the kernel density estimation tool in QGIS will automatically calculate the camera density at each location and generate a continuous density map. The generated density map represents the camera density by the depth of color, and the darker the color, the denser the cameras. Save this density map as a GeoTIFF format file to form the installation point distribution feature data.

[0139] Step S52: Evaluate the redundancy of the intersection area based on the installation point distribution feature data to obtain the monitoring coverage redundancy data;

[0140] Specifically, the installation point distribution feature data can be imported into QGIS. In QGIS, use the "buffer analysis" tool to generate a buffer for the coverage range of each camera. When operating specifically, set the radius of the buffer according to the field of view angle and installation height of the camera. For example, for a camera with a horizontal field of view angle of 90 degrees and an installation height of 10 meters, set the buffer radius to 50 meters. Through the above operations, generate the coverage range layer of each camera. Overlay the coverage range layers of all cameras and calculate the coverage times at each location. For example, if a location is overlapped by the coverage ranges of 3 cameras, the coverage times at this location is 3. Export these coverage times data as a CSV file and perform further analysis using a Python script. In Python, calculate the statistical quantities of the coverage times, such as the mean, standard deviation, minimum value, and maximum value. For example, the mean of the coverage times at all locations is 2, the standard deviation is 1, the minimum value is 1, and the maximum value is 5. According to these statistical quantities, define the threshold for redundancy evaluation. For example, mark the area where the coverage times exceed 3 times as the redundant coverage area. Through the above steps, obtain the monitoring coverage redundancy data.

[0141] Step S53: Refine the installation point set at the intersection according to the monitoring coverage redundancy data to obtain the optimized intersection installation point set;

[0142] Specifically, the monitoring coverage redundancy data can be imported into QGIS. In QGIS, use the "selection" tool to select the cameras whose coverage times exceed the threshold (for example, 3 times). These cameras are considered redundant and can be considered for removal or adjustment. Write a point refinement algorithm using a Python script. This algorithm automatically selects which cameras can be removed according to the redundancy data and the coverage range of the cameras. For example, if multiple cameras cover the same area, the algorithm will preferentially retain the cameras with better positions and wider coverage ranges and remove other redundant cameras. When operating specifically, set the parameters of the refinement algorithm, such as the minimum coverage overlap area ratio (for example, 50%). If the coverage areas of two cameras overlap by more than 50%, then one of them can be considered for removal. Through the above steps, obtain the optimized intersection installation point set.

[0143] Step S54: Calculate the perspective compensation for the optimized intersection installation point set to obtain the monitoring perspective compensation parameters;

[0144] Specifically, the optimized intersection installation point set can be imported into Genetec Security Center. This data includes the location, orientation, and coverage range of each camera. In Genetec Security Center, use the "Field of View Analysis" tool to simulate the field of view range of each camera and calculate the overlapping area between the fields of view. When operating specifically, input the optical parameters of each camera, such as the horizontal field of view angle, vertical field of view angle, and focal length. The software will automatically calculate the field of view boundaries of each camera and generate a field of view coverage map. By analyzing the field of view coverage map, identify the areas where the fields of view overlap and calculate the overlap ratio of each area. Use Blender for 3D modeling. According to the 3D model of the intersection and the installation positions of the cameras, adjust the orientation of the cameras. When operating specifically, create a spherical model in Blender to represent the field of view range of the camera. By adjusting the position and orientation of the spherical model, simulate the coverage effects from different perspectives. For example, if the fields of view of two cameras overlap by more than 30%, by fine-tuning the orientation of one of the cameras, reduce the overlapping area to less than 20%. Through the above steps, obtain the monitoring perspective compensation parameters.

[0145] Step S55: Supplement the blind spots in the intersection area according to the monitoring perspective compensation parameters to obtain the optimized data for monitoring blind spots;

[0146] Specifically, the monitoring perspective compensation parameters can be imported into Blender. In Blender, according to the 3D model of the intersection and the installation positions of the cameras, adjust the orientation and field of view range of the cameras. Use the "Animation" function of Blender to simulate the changes in the field of view of the cameras before and after adjustment. For example, if the orientation of a certain camera needs to be rotated clockwise by 15 degrees, create an animation in Blender to show the field of view coverage before and after rotation. Through the above operations, identify the blind spots that still exist after adjustment. In the software, according to the blind spots identified in Blender, add new camera positions or adjust the parameters of existing cameras to supplement the blind spot coverage. For example, if there is a blind spot in a corner of the intersection, add a new camera near that corner and adjust its orientation and field of view angle. Through the above steps, obtain the optimized data for monitoring blind spots.

[0147] Step S56: Dynamically adjust the optimized intersection installation point set according to the optimized data for monitoring blind spots to obtain the final camera layout plan.

[0148] Specifically, the optimized data for monitoring blind spots can be imported into QGIS. In QGIS, use the "Edit" function to adjust the installation positions of cameras according to the blind spot optimization data. Based on the positions of the blind spots, move or add camera positions and adjust their orientations and field of view angles. For example, if a new camera is needed to cover a certain blind spot, add a new position in QGIS and set its orientation and field of view angle. At the same time, adjust the parameters of the existing cameras according to the blind spot optimization data. Import the adjusted set of installation positions into Genetec Security Center for further simulation and verification. In Genetec Security Center, simulate the field of view range of each camera according to the adjusted layout plan. Through the above steps, obtain the final camera layout plan, and these data record the positions, orientations, and coverage ranges of each camera.

[0149] Through spatial distribution evaluation, the present invention can understand the distribution characteristics of each installation position. Through redundancy evaluation, it can identify the overlapping areas and unnecessary installation positions in the monitoring coverage. Through point position reduction, it can remove the redundant camera installation positions and reduce unnecessary equipment investment. Through perspective compensation calculation, it can ensure that the perspective of each camera can effectively cover the target area and reduce monitoring blind spots. Through blind spot supplementation, the monitoring layout can be further optimized to ensure that the monitoring system can comprehensively cover all key areas of the intersection and improve the integrity and reliability of monitoring. Through dynamic adjustment, the monitoring layout can be flexibly adjusted according to the changes in the actual traffic conditions.

[0150] The data acquisition module is used to perform lidar scanning on the intersection area to obtain the three-dimensional point cloud data of the intersection; perform static element recognition on the three-dimensional point cloud data of the intersection to obtain the basic layer data map of the intersection; perform vehicle and pedestrian flow acquisition on the intersection area to obtain the dynamic layer data map of the intersection.

[0151] The model construction module is used to perform hydrodynamic modeling on the intersection area based on the basic layer data map and the dynamic layer data map of the intersection to obtain the traffic flow field data of the intersection; perform flow field feature recognition on the traffic flow field data of the intersection to obtain the distribution map of flow field feature points of the intersection; perform speed density field simulation on the intersection area based on the distribution map of flow field feature points of the intersection to obtain the dynamic distribution model of the flow field of the intersection.

[0152] The layout planning module is used to evaluate the constraints of the camera installation positions in the intersection area to obtain the set of installation positions at the intersection; perform spherical projection calculation on the cameras according to the set of installation positions at the intersection to obtain the set of camera field of view angles; perform spherical coverage visualization on the cameras according to the set of camera field of view angles to obtain the initial coverage plan diagram.

[0153] A verification and optimization module, configured to obtain historical traffic data of intersections; perform heat map conversion on the historical traffic data of intersections to obtain intersection risk point distribution data; perform spatial overlay analysis on the initial coverage plan map according to the intersection risk point distribution data to obtain a layout verification result map; perform coverage difference adjustment on the initial coverage plan map according to the layout verification result map to obtain a layout optimization data set;

[0154] A scheme generation module, configured to reconstruct and adjust the intersection installation point set according to the layout optimization data set to obtain a final camera layout scheme.

[0155] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the application document are intended to be included in the present invention.

[0156] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A modeled camera layout method, characterized in that: The following steps are involved: Step S1: Scan the intersection area with a laser radar to obtain three-dimensional point cloud data of the intersection; Static elements are identified on the three-dimensional point cloud data of the intersection to obtain the basic layer data map of the intersection; traffic and pedestrian flow are collected in the intersection area to obtain the dynamic layer data map of the intersection; Step S2: Based on the intersection base layer data map and the intersection dynamic layer data map, a fluid dynamics model is performed on the intersection area to obtain the intersection traffic flow field data; flow field characteristics are identified on the intersection traffic flow field data to obtain a flow field characteristic point distribution map; based on the intersection flow field characteristic point distribution map, a speed density field simulation is performed on the intersection area to obtain a flow field dynamic distribution model of the intersection; Step S3: Perform camera installation point constraint evaluation on the intersection area to obtain an intersection installation point set; perform spherical projection calculation on the camera based on the intersection installation point set to obtain a camera field of view angle set; perform spherical coverage visualization on the camera based on the camera field of view angle set to obtain an initial coverage plan diagram; Step S4: Obtain historical traffic data of the intersection; Perform heat map conversion on the historical traffic data of the intersection to obtain the distribution data of the risk points at the intersection; perform spatial overlay analysis on the initial coverage plan map based on the distribution data of the risk points at the intersection to obtain the layout verification result map; perform coverage difference adjustment on the initial coverage plan map based on the layout verification result map to obtain the layout optimization data set; Step S5: reconstruct and adjust the intersection installation point set according to the layout optimization data set to obtain the final camera layout solution.

2. The modeled camera layout method according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Scan the intersection area with a laser radar to obtain three-dimensional point cloud data of the intersection; Step S12: clustering non-ground points on the intersection three-dimensional point cloud data to obtain a candidate point set of intersection feature targets; Step S13: extracting building structures from the intersection three-dimensional point cloud data according to the intersection feature target candidate point set to obtain intersection environment building structure data; Step S14: identifying road facilities on the intersection feature target candidate point set to obtain intersection infrastructure location data; Step S15: feature fusion of the intersection environment building structure data and the intersection infrastructure location data to obtain the intersection base layer data map; Step S16: Perform sensor collaborative perception on the intersection area to obtain intersection dynamic target trajectory data, and perform spatiotemporal feature extraction on the intersection dynamic target trajectory data to obtain an intersection dynamic layer data map.

3. The modeled camera layout method according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: rasterizing the intersection base layer data map to obtain an intersection environment calculation grid; Step S22: configuring the boundary conditions of the intersection environment calculation grid according to the intersection dynamic layer data map to obtain the intersection flow field initial condition data set; Step S23: setting turbulence model parameters for the initial condition data set of the intersection flow field to obtain turbulence characteristic data of the intersection flow field; Step S24: Discretize the flow field and solve it according to the turbulence characteristic data of the intersection flow field to obtain a numerical solution of the intersection traffic flow field; Step S25: performing convergence evaluation on the numerical solution of the traffic flow field at the intersection to obtain a traffic flow field stability evaluation result; Step S26: extracting flow field features from the numerical solution of the intersection traffic flow field according to the traffic flow field stability evaluation result to obtain intersection traffic flow field data; Step S27: identifying flow field feature points of the intersection traffic flow field data to obtain a distribution map of the intersection flow field feature points, and simulating the speed density field of the intersection area based on the distribution map of the intersection flow field feature points to obtain a dynamic distribution model of the intersection flow field.

4. The modeled camera layout method according to claim 3, characterized in that: Step S27 includes the following steps: Step S271: performing vortex structure recognition on the intersection traffic flow field data to obtain an intersection vortex core distribution map; Step S272: identifying critical points of the intersection vortex core distribution map to obtain intersection flow field singular point position data; Step S273: performing feature classification on the intersection flow field singular point position data to obtain intersection flow field feature point type data; Step S274: spatially clustering the intersection flow field singular point position data according to the intersection flow field characteristic point type data to obtain an intersection flow field characteristic point distribution map; Step S275: performing velocity field interpolation on the intersection flow field characteristic point distribution map to obtain intersection velocity field distribution data; Step S276: Derivation of density field according to intersection speed field distribution data to obtain intersection density field distribution data; Step S277: performing time-series evolution on the intersection velocity field distribution data and the intersection density field distribution data to obtain time-varying characteristic data of the intersection flow field; Step S278: performing mode decomposition on the time-varying characteristic data of the intersection flow field to obtain the dominant modal data of the intersection flow field, and reconstructing and integrating the dominant modal data of the intersection flow field to obtain a dynamic distribution model of the intersection flow field.

5. The modeled camera layout method according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: performing a camera installation point constraint evaluation on the intersection area to obtain an intersection installation point set; Step S32: Performing a visual field accessibility assessment on the intersection installation point set to obtain spatial visual area data of the intersection; Step S33: configuring camera parameters for the intersection area according to the intersection installation point set and the intersection spatial visual area data to obtain a camera optical parameter set, and performing spherical projection modeling on the intersection area based on the camera optical parameter set to obtain a single-point projection data set of the monitoring area; Step S34: Calculate the field of view angle according to the single-point projection data set of the monitoring area to obtain the field of view coverage data set of the monitoring area, and perform spatial registration conversion on the field of view coverage data set of the monitoring area to obtain the camera field of view angle set; Step S35: performing spherical segmentation of the observation space of the intersection area according to the camera field of view angle set to obtain spherical grid data of the monitoring area, and performing coverage calculation on the spherical grid data of the monitoring area to obtain a distribution map of the monitoring area coverage; Step S36: Perform three-dimensional visualization of the intersection area based on the monitoring area coverage distribution map to obtain an initial coverage plan map.

6. The modeled camera layout method according to claim 5, characterized in that: Step S31 includes the following steps: Step S311: Acquire the three-dimensional point cloud data of the intersection, and spatially partition and slice the three-dimensional point cloud data of the intersection to obtain intersection area slice data, and extract the building boundary of the intersection area slice data to obtain intersection building outline data; Step S312: performing elevation feature recognition on the intersection area according to the intersection building outline data to obtain intersection building surface feature data; Step S313: constraining the camera installation height according to the surface feature data of the building at the intersection to obtain installation height constraint data, and selecting candidate points in the intersection area according to the installation height constraint data to obtain a candidate set of intersection installation points; Step S314: Performing power supply evaluation on the candidate set of intersection installation points to obtain a data set of power supply conditions for the installation points; Step S315: Performing a network transmission evaluation on the candidate set of intersection installation points according to the installation point power supply condition data set to obtain an installation point network coverage data set, and performing a lightning protection grounding evaluation on the candidate set of intersection installation points according to the installation point height constraint data to obtain an installation point lightning protection condition data set; Step S316: Based on the intersection three-dimensional point cloud data and the installation point height constraint data, the intersection installation point candidate set is identified to obtain the installation point occlusion interference data set; Step S317: Comprehensively screen the intersection installation point candidate set according to the installation point power supply condition data set, the installation point network coverage data set, the installation point lightning protection condition data set and the installation point shielding interference data set to obtain the intersection installation point set.

7. The modeled camera layout method according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: Obtain historical traffic data of the intersection, and extract valid accidents from the historical traffic data of the intersection to obtain valid accident data of the intersection; Step S42: extracting violation information from a preset traffic violation database to obtain valid violation data at the intersection; Step S43: extracting intersection traffic data from a preset real-time traffic monitoring system to obtain intersection traffic flow statistics; Step S44: identifying traffic flow changes on the intersection traffic flow statistics data to obtain intersection traffic flow change data; Step S45: spatially integrate the valid accident data, valid violation data and traffic flow change data of the intersection to obtain the risk point distribution data of the intersection, and perform spatial overlay analysis on the initial coverage plan map based on the risk point distribution data of the intersection to obtain a layout verification result map; adjust the coverage difference of the initial coverage plan map based on the layout verification result map to obtain a layout optimization data set.

8. The modeled camera layout method according to claim 7, characterized in that: Step S45 includes the following steps: Step S451: Perform spatial density mapping on the effective accident data at the intersection to obtain thermal data of the accident at the intersection; Step S452: performing spatial clustering segmentation on the valid violation data at the intersection to obtain thermal data of the violation at the intersection; Step S453: Conduct congestion assessment on the intersection traffic flow change data to obtain intersection congestion thermal data, and spatially superimpose the intersection accident thermal data, intersection violation thermal data and intersection congestion thermal data to obtain intersection risk point distribution data; Step S454: evaluating the matching degree of the initial coverage plan map according to the intersection risk point distribution data to obtain intersection coverage matching data; Step S455: evaluating the difference of the intersection coverage matching data to obtain intersection coverage difference data; Step S456: adjusting the initial coverage plan map based on the intersection coverage difference data to obtain a layout optimization data set.

9. The modeled camera layout method according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: performing spatial distribution evaluation on the intersection installation point set according to the layout optimization data set to obtain installation point distribution feature data; Step S52: performing redundancy evaluation on the intersection area according to the installation point distribution characteristic data to obtain monitoring coverage redundancy data; Step S53: simplifying the intersection installation point set according to the monitoring coverage redundant data to obtain an optimized intersection installation point set; Step S54: performing viewing angle compensation calculation on the optimized intersection installation point set to obtain monitoring viewing angle compensation parameters; Step S55: supplementing the blind spot of the intersection area according to the monitoring view angle compensation parameter to obtain monitoring blind spot optimization data; Step S56: dynamically adjust the optimized intersection installation point set according to the monitoring blind spot optimization data to obtain the final camera layout plan.

10. A modeled camera layout system, characterized in that: For executing the modeled camera layout method as claimed in claim 1, the modeled camera layout system comprises: The data acquisition module is used to perform laser radar scanning on the intersection area to obtain the three-dimensional point cloud data of the intersection; perform static element recognition on the three-dimensional point cloud data of the intersection to obtain the basic layer data map of the intersection; and collect the traffic and pedestrian flow in the intersection area to obtain the dynamic layer data map of the intersection; The model building module is used to perform fluid dynamics modeling on the intersection area based on the intersection basic layer data map and the intersection dynamic layer data map to obtain the intersection traffic flow field data; perform flow field feature recognition on the intersection traffic flow field data to obtain the intersection flow field feature point distribution map; perform speed density field simulation on the intersection area based on the intersection flow field feature point distribution map to obtain the intersection flow field dynamic distribution model; The layout planning module is used to evaluate the camera installation point constraints in the intersection area to obtain the intersection installation point set; perform spherical projection calculation on the camera based on the intersection installation point set to obtain the camera field of view angle set; perform spherical coverage visualization on the camera based on the camera field of view angle set to obtain the initial coverage plan diagram; The verification and optimization module is used to obtain historical traffic data of the intersection; perform heat map conversion on the historical traffic data of the intersection to obtain the distribution data of risk points at the intersection; perform spatial overlay analysis on the initial coverage plan map based on the distribution data of risk points at the intersection to obtain the layout verification result map; perform coverage difference adjustment on the initial coverage plan map based on the layout verification result map to obtain the layout optimization data set; The solution generation module is used to reconstruct and adjust the intersection installation point set according to the layout optimization data set to obtain the final camera layout solution.

Images