Road network perception device layout method, device, equipment, medium and program product

Abstract

The application provides a road network perception device layout method, device, equipment, medium and program product. The method comprises the following steps: constructing a directed weighted connected graph with intersections as nodes for a road network comprising a plurality of intersections; for a plurality of paths between two nodes, calculating a characteristic index of each path based on the characteristic value of the target node; based on the calculated characteristic index, determining whether the plurality of paths can be uniquely distinguished; if not, extracting the indistinguishable paths, selecting nodes that meet the preset conditions in terms of the number of occurrences and the weight difference between different paths from the nodes passed by the extracted paths as target nodes, and re-executing the determination until the plurality of paths can be uniquely distinguished. In this way, the key intersections in the road network that can be used to distinguish different paths are used as the layout points of the perception devices, so that the data collected by the perception devices at these points can accurately restore the passing trajectories between intersections.

Description

Technical Field

[0001] This application relates to the field of road network monitoring technology, and in particular to a method, device, equipment, medium and program product for the layout of road network sensing equipment. Background Technology

[0002] Deploying cameras and other sensing devices in urban road networks is of great significance for maintaining public safety. Given the complexity of road network structures and the role of intersections in diverting traffic, the deployment points of sensing devices in the road network should, as far as possible, cover intersection nodes that play a crucial role in trajectory reconstruction.

[0003] Currently, the deployment of sensing devices in road networks mainly adopts either full regional coverage or deployment in key areas based on subjective experience. However, road networks have numerous intersections with complex connections. Full regional coverage is costly and results in low equipment utilization, while deployment based on experience can easily lead to problems such as redundant equipment at some intersections and omissions of intersections where equipment should be deployed. Therefore, a reasonable deployment scheme capable of reconstructing the trajectories between intersections is needed. Summary of the Invention

[0004] This application provides a method, apparatus, device, medium, and program product for laying out road network sensing devices, which aims to accurately reconstruct the trajectory between intersections by optimizing the placement of device locations while using as few sensing devices as possible.

[0005] In a first aspect, embodiments of this application provide a method for deploying road network sensing devices, including:

[0006] For a road network containing multiple intersections, a directed weighted connected graph is constructed with intersections as nodes. The weights of the directed weighted connected graph are the road length or the road travel time.

[0007] For multiple paths between two nodes, the feature index of each path is calculated based on the feature value of the target node. The target node is the node with a determined layout of sensing devices. The feature value is determined based on the number of times the node appears in multiple paths and the weight of the node in the corresponding path.

[0008] Based on the calculated feature indicators, it is determined whether multiple pathways can be uniquely distinguished.

[0009] If a path cannot be uniquely distinguished, the path that cannot be distinguished is extracted. From the nodes traversed by the extracted path, the nodes whose frequency of occurrence and the weight difference between different paths meet the preset conditions are selected as target nodes, and the judgment is re-executed until multiple paths can be uniquely distinguished.

[0010] The target node is located at the intersection where the sensing devices are deployed.

[0011] In some possible implementations, one node is the starting node and the other is the ending node. For multiple paths between the two nodes, feature indicators for each path are calculated based on the feature values ​​of the target node, including:

[0012] For multiple paths between two nodes, construct a path weight matrix. Each row of the path weight matrix corresponds to all the nodes along a path except for the starting node and the ending node. The elements in the path weight matrix are the cumulative weights of the nodes in the corresponding path starting from the starting node.

[0013] Based on the number of times the target node appears in the path weight matrix, and the degree of difference in the corresponding element values ​​of the same target node in different rows of the path weight matrix, the feature value of the target node for each path is determined.

[0014] For any path, the feature values ​​of all target nodes in the path are summed to obtain the feature index.

[0015] In some possible implementations, based on calculated feature indicators, it is determined whether multiple pathways can be uniquely distinguished, including:

[0016] Clustering of characteristic indicators from multiple pathways;

[0017] If each cluster in the clustering results contains only one pathway, then it is determined that multiple pathways can be uniquely distinguished.

[0018] Otherwise, multiple paths are determined to be indistinguishable, and paths within the same cluster are considered indistinguishable.

[0019] In some possible implementations, from the nodes traversed by the extracted path, nodes whose frequency of occurrence and the weight difference between different paths satisfy preset conditions are selected as target nodes, including:

[0020] For the extracted pathways, obtain the number of times each node along the pathway appears in these pathways, and the degree of difference in the weight of the same node in different pathways;

[0021] Prioritize selecting the node with the greatest degree of difference. If the degree of difference is the same, select the node that appears most frequently as the target node.

[0022] In some possible implementations, before selecting the chosen node as the target node, the following steps are also included:

[0023] Determine whether the selected node, the existing target node, and the two nodes form a direct closed loop in the directed weighted connected graph.

[0024] If a direct closed loop is formed, the selected node will not be used as the target node, and other nodes that meet the preset conditions will be selected again.

[0025] If a direct closed loop is not formed, the selected node will be used as the target node.

[0026] In some possible implementations, before calculating the feature indexes of each path based on the feature values ​​of the target node, the following is also included:

[0027] Calculate the topological importance of each node in a directed weighted connected graph. The topological importance is determined based on at least one of the following: the degree of the node, the rate of change of the average path length of the road network after deleting the node, and the betweenness centrality of the node. The degree of the node represents the size of the road connected by the intersection corresponding to the node, and the betweenness centrality is determined based on the number of times the current node is located on the shortest path between other node pairs.

[0028] Nodes whose topological importance meets the preset conditions are identified as target nodes.

[0029] In some possible implementations, before calculating the feature indexes of each path based on the feature values ​​of the target node, the following is also included:

[0030] Calculate the business importance of each node in the directed weighted connected graph. The business importance is determined based on at least one of the following: traffic flow at the intersection corresponding to the node, historical traffic case data, and key perception paths.

[0031] Nodes whose business importance meets the preset conditions are identified as target nodes.

[0032] In some possible implementations, the intersection layout sensing devices corresponding to the target node include:

[0033] Based on the node type attribute of the target node, determine the type of sensing device to be deployed at the corresponding intersection;

[0034] The node type attribute includes at least one of the following: topological key node, node whose traffic flow meets preset conditions, node located within a preset range of key locations, node located on the security path, and highway entrance / exit node.

[0035] The types of sensing devices include at least one of the following: devices for capturing facial images, devices for capturing vehicle license plates, devices for capturing panoramic images, and devices for collecting terminal identification information.

[0036] Secondly, embodiments of this application provide a road network sensing device deployment apparatus, comprising:

[0037] The mapping module is used to construct a directed weighted connected graph with intersections as nodes for a road network containing multiple intersections. The weights of the directed weighted connected graph are the road length or the road travel time.

[0038] The calculation module is used to calculate the feature index of each path based on the feature value of the target node for multiple paths between two nodes. The target node is the node of the determined layout sensing device. The feature value is determined based on the number of times the node appears in multiple paths and the weight of the node in the corresponding path.

[0039] The judgment module is used to determine whether multiple paths can be uniquely distinguished based on the calculated feature indicators.

[0040] The iteration module is used to extract indistinguishable paths when they cannot be uniquely distinguished. From the nodes traversed by the extracted paths, the nodes whose frequency of occurrence and the weight difference between different paths meet the preset conditions are selected as target nodes, and the judgment is re-executed until multiple paths can be uniquely distinguished.

[0041] The layout module is used to deploy sensing devices according to the intersection corresponding to the target node.

[0042] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;

[0043] The memory stores instructions that the computer executes;

[0044] The processor executes computer execution instructions stored in memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.

[0045] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.

[0046] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.

[0047] The road network sensing device layout method, apparatus, device, medium, and program product provided in this application construct a directed weighted connected graph with intersections as nodes for road networks containing multiple intersections. For multiple paths between two nodes, feature indicators of each path are calculated based on the feature values ​​of the target node. Based on the calculated feature indicators, it is determined whether multiple paths can be uniquely distinguished. If they cannot be uniquely distinguished, the indistinguishable paths are extracted, and nodes whose frequency of occurrence and weight differences between different paths meet preset conditions are selected from the nodes traversed by the extracted paths as target nodes. The judgment is repeated until multiple paths can be uniquely distinguished. Sensing devices are then deployed according to the intersections corresponding to the target nodes. By abstracting the urban road network into a directed weighted connected graph and establishing path feature indicators around the determined or gradually supplemented target nodes, and then using an iterative screening mechanism for indistinguishable paths to supplement nodes with distinguishing capabilities, a sensing device layout scheme corresponding to the unique identification requirements of paths can be formed. This allows key intersections in the road network that can be used to distinguish different paths to be used as deployment points for sensing devices, thereby enabling accurate reconstruction of traffic trajectories between intersections through data collected by the sensing devices at these points. Attached Figure Description

[0048] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0049] Figure 1 A flowchart illustrating a road network sensing device deployment method provided as an example in this application;

[0050] Figure 2 This application provides an exemplary flowchart for determining a target node.

[0051] Figure 3 This is a partial schematic diagram of a road network processing result provided in an embodiment of this application;

[0052] Figure 4 This is a schematic diagram of a local gridding process for a road network, provided as an embodiment of this application.

[0053] Figure 5 This is a schematic diagram of the structure of a road network sensing device layout apparatus provided in an embodiment of this application;

[0054] Figure 6 This is a schematic diagram of the structure of an electronic device provided in this application.

[0055] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0056] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0057] In the fields of urban security, traffic management and sensing equipment monitoring, sensing devices such as cameras are usually deployed at multiple intersections of the urban road network to collect traffic data of vehicles or pedestrians passing through the intersections. Establishing continuous sensing links between multiple intersections can serve applications such as trajectory tracking and traffic monitoring.

[0058] Currently, most road network sensing device deployment schemes adopt either full regional coverage or priority deployment in key areas. The former requires too many sensing devices, which is difficult to deploy at every intersection due to the increasing complexity of road networks and the growing number of intersections. On the other hand, the effectiveness of priority deployment in key areas is highly uncertain. Its basic idea is to select deployment points based on empirical rules or local business needs, and then use the data collected from adjacent intersections to infer the movement trajectories of vehicles or people in the road network.

[0059] The aforementioned methods do not adequately consider the path connectivity of intersections within the overall road network, making it difficult to accurately identify the key nodes that truly affect path discrimination. This often leads to duplicate deployments at some ordinary intersections and omissions at some critical intersections. Especially when multiple reachable paths exist between two intersections, relying solely on the data collected from existing nodes often fails to effectively distinguish between these paths. Consequently, the calculated and reconstructed trajectories frequently present multiple possible paths, failing to accurately reconstruct the true trajectory. This deficiency not only reduces the accuracy of trajectory analysis but also creates a mismatch between equipment investment and actual sensing effects. On the one hand, it increases deployment costs and maintenance burdens; on the other hand, it weakens the ability to discriminate continuous traffic trajectories in complex road network scenarios, making it difficult to balance deployment economy with monitoring effectiveness.

[0060] To address the aforementioned issues, this application proposes a technical concept: For road networks containing multiple intersections, a directed weighted connected graph is constructed with intersections as nodes. When multiple paths exist between two nodes, characteristic indicators of each path are calculated based on the target nodes of the determined sensing devices, and the uniqueness of these paths is determined accordingly. If uniqueness is not possible, nodes meeting preset conditions are selected from the nodes traversed by the relevant paths as new target nodes, until multiple paths can be uniquely distinguished. Finally, sensing devices are deployed according to the intersections corresponding to the target nodes. This method, supported by road network topology data, enables device deployment that balances trajectory discrimination accuracy and deployment cost.

[0061] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0062] Figure 1 This is a flowchart illustrating a road network sensing device deployment method provided in an embodiment of this application, as shown below. Figure 1 As shown, the method includes:

[0063] Step S101: For a road network containing multiple intersections, construct a directed weighted connected graph with intersections as nodes.

[0064] The road network can be a dataset of urban roads and their intersection connections. A directed weighted connected graph can be used to represent the directional connectivity between intersections and the travel costs of different roads. Each node in the graph corresponds to an intersection in the road network, and the weights of the edges between nodes can be road length or road travel time. Road travel time can refer to the travel time under free-flowing road conditions.

[0065] When constructing a directed weighted connected graph, for bidirectional roads, they can be split into two directed edges with opposite directions; for one-way roads, only the directed edges in the allowed direction can be retained. Each intersection corresponds to a node in the graph, and the roads between intersections correspond to weighted edges. The start and end points of the edges can be determined by the actual direction of traffic.

[0066] In some possible implementations, road length can be extracted from the road segment mileage field of existing road network basic data. If the road network basic data does not directly provide road length, the geometric length of the road can be calculated based on the coordinates of adjacent intersections or map projection coordinates, and this length can be written into the weight field of the corresponding edge. In some other possible implementations, when road travel time is used as an edge weight, the initial duration can be obtained by dividing the road length by the baseline travel speed of the corresponding road, or the duration parameter can be generated based on historical floating traffic data or time-segmented traffic operation data, and then written into the connected graph model according to time periods. For intersections with rules such as traffic light control, a turning penalty time can be added to the edge weight to make the directed weighted connected graph more closely resemble the actual travel path.

[0067] The core of this step lies in transforming the original road network topology into a computable graph structure, enabling all subsequent paths to be expressed using unified nodes, edges, directions, and weights. This graph structure allows for the analysis of multiple possible paths between different intersections within a unified framework, providing foundational data for subsequent target node feature value calculations and path uniqueness determination.

[0068] Step S102: For multiple paths between two nodes, calculate the feature index of each path based on the feature value of the target node.

[0069] It should be noted that the execution entity in this application embodiment can maintain a target node set, which includes all nodes of the currently determined layout sensing device. Feature values ​​can be determined based on the number of times a node appears in multiple paths and the weight of the node in the corresponding path. Feature indicators can characterize the identifiability of the entire path under the current target node set.

[0070] There are several ways to determine target nodes. For example, in practical applications, some intersections within a road network have already been equipped with sensing devices, so the nodes corresponding to these intersections can be directly identified as target nodes. In other examples, target nodes can be determined manually or based on factors such as the importance of the intersection's traffic flow.

[0071] In this embodiment of the application, a path can represent a reachable path between two nodes in a connected graph, corresponding to a route from one starting intersection to another ending intersection in an actual road network. In this case, the two nodes in step S102 can be referred to as the "starting node" and the "ending node," respectively, and these two nodes can be collectively referred to as the "starting and ending nodes." The path between the starting node and the ending node can pass through other nodes in the graph, including the target node.

[0072] For example, a road network contains 26 intersections: A, B, C, D, ..., Y and Z. In a directed weighted connected graph, this corresponds to 26 nodes: A through Z. For nodes A and Z, m (where m is a positive integer) paths can be extracted between them according to the road network's traffic rules. These paths can pass through nodes other than A and Z. For instance, the first path could be [A, C, D, Z] (meaning starting from A, passing through C and D sequentially, and arriving at Z), the second path could be [A, D, Z], and the third path could be [A, B, E, Z]. The weight of node D in the first path is the sum of the weights of edges AC and CD (i.e., the edge between node A and node C, plus the edge between node C and node D). The weight of node D in the second path is the weight of edge AD. If there are only these 3 paths between A and Z, and node D appears in the 1st and 2nd paths but not in the 3rd path, then node D appears 2 times in the multiple paths between A and Z.

[0073] In some possible implementations, the characteristic indicators of the computational path can be calculated based on certain rules, such as the weight of the target node in the path and the number of times the target node appears, so that each path is converted into a quantitative indicator associated with the current set of target nodes.

[0074] This step transforms the path enumeration between intersections into feature indicators for device deployment decisions. Instead of relying solely on the total length or duration of the path, it comprehensively considers the occurrence of the target node of the proposed sensing device in different paths and the relative weight differences, so that subsequent distinctions can reflect the degree of support of the current sensing device layout for path identification.

[0075] Step S103: Based on the calculated feature indicators, determine whether multiple pathways can be uniquely distinguished.

[0076] For example, for m paths between node A and node Z, the distinction can be made by comparing the characteristic indicators of these paths. If two or more paths have the same characteristic indicator, or their difference is less than a preset distinction threshold, these paths can be marked as indistinguishable. If the difference in characteristic indicators between any two paths is not less than the preset distinction threshold, then these m paths are determined to be uniquely distinguishable. The distinction threshold can be set according to the weighting units and the data fluctuation range. After comparing the characteristic indicators of the m paths, all paths marked as indistinguishable can be placed into a set of indistinguishable paths.

[0077] In some possible implementations, the feature metrics can be normalized before comparison, mapping them to a numerical range of 0 to 1.

[0078] In some possible implementations, the judgment process can be performed using a pairwise comparison method. That is, for any two paths a and b between the same set of start and end nodes, |Ta-Tb| is calculated, where Ta and Tb represent the characteristic indices of path a and path b, respectively. When the calculation result is less than a threshold, path a and path b are recorded as belonging to the same set of indistinguishable relationships. After all comparisons are completed, if no indistinguishable relationships exist, then the multiple paths between this set of start and end nodes are determined to be uniquely distinguishable. This method can not only make binary judgments on whether paths between start and end nodes are distinguishable or indistinguishable, but can also include sets of indistinguishable paths and sets of differences between paths.

[0079] This step, by performing a unified discrimination on the quantitative indicators formed in step S102, transforms the path ambiguity problem, which originally relied on manual experience analysis, into a computable uniqueness determination problem. When all paths between the start and end nodes form distinguishable independent indicator expressions under the current target node conditions, it indicates that the corresponding sensing device intersection layout scheme has the basis for path identification; when there are indistinguishable indicators, it indicates that the current sensing device layout is still insufficient to support the unique reconstruction of the trajectory between the start and end intersections, and further sensing devices need to be deployed at other intersections to participate in the analysis.

[0080] Step S104: If a path cannot be uniquely distinguished, extract the path that cannot be distinguished. From the nodes traversed by the extracted path, select the nodes whose frequency of occurrence and the weight difference between different paths meet the preset conditions as target nodes, and re-execute the judgment until multiple paths can be uniquely distinguished.

[0081] Among them, preset conditions can be used to filter nodes that contribute to path differentiation.

[0082] In step S103, pathways deemed indistinguishable after comparing feature indicators can be marked. Based on this, in this step, all pathways marked as indistinguishable between the same pair of start and end nodes can be extracted and placed into a set of indistinguishable pathways. The nodes traversed by each pathway in these sets are then summarized to form a candidate node set. Next, the candidate nodes can be statistically analyzed to obtain the frequency of each node in the current set of indistinguishable pathways, as well as the weight difference of each node between different pathways. If the frequency of occurrence and the weight difference between different pathways of a candidate node meet preset conditions, then that candidate node can be added as a new target node.

[0083] After adding a new target node, it can be merged into the current target node set. Then, step S102 is executed again to calculate the feature indicators of multiple paths, and step S103 is executed for judgment. If there are still paths that cannot be distinguished, the next round of node screening and supplementation is carried out according to this step; if all paths can be uniquely distinguished, the iteration is terminated.

[0084] This step constitutes the closed-loop optimization process of the entire method. Its function is to quantify the contribution of nodes whose layout sensing devices have not yet been determined in the current set of paths where path ambiguity still exists. By filtering nodes that meet preset conditions from indistinguishable paths, adding target nodes can gradually widen the differences in path characteristic indicators, thereby achieving unique differentiation of multiple paths with a smaller number of nodes.

[0085] Step S105: Layout sensing devices according to the intersection corresponding to the target node.

[0086] In this embodiment of the application, the intersection corresponding to the target node is the location where the sensing device is deployed. The sensing device is used to collect traffic data when vehicles, people or other traffic objects pass through the corresponding intersection.

[0087] Steps S102 to S104 above can be used to uniquely distinguish multiple paths between any two intersections in the road network. By traversing all the node pairs of starting and ending intersections in the road network, the intersections in the entire road network where sensing devices should be deployed can be obtained.

[0088] For example, after step S104 is completed to make all paths distinguishable, a list of sensing device deployments can be generated based on the current target node set. The list may include intersection signs, coordinates, road direction, associated path groups, and the recommended number of devices to be installed. If the target node is directly derived from a node with a pre-determined sensing device deployment, it is retained as an existing deployment point; if the target node is a newly selected node during the iteration process, a new sensing device deployment task can be added at the corresponding intersection.

[0089] In the above embodiments, a directed weighted connected graph with intersections as nodes is constructed for a road network containing multiple intersections. For multiple paths between two nodes, feature indicators of each path are calculated based on the feature values ​​of the target node. Based on the calculated feature indicators, it is determined whether multiple paths can be uniquely distinguished. If they cannot be uniquely distinguished, the indistinguishable paths are extracted, and nodes whose frequency of occurrence and weight differences between different paths meet preset conditions are selected from the nodes traversed by the extracted paths as target nodes. The judgment is repeated until multiple paths can be uniquely distinguished. Sensing devices are deployed according to the intersections corresponding to the target nodes. By abstracting the urban road network into a directed weighted connected graph and establishing path feature indicators around the determined or gradually supplemented target nodes, and then using an iterative screening mechanism for indistinguishable paths to supplement nodes with distinguishing capabilities, a sensing device deployment scheme corresponding to the unique identification requirements of paths can be formed. This allows key intersections in the road network that can be used to distinguish different paths to be used as deployment points for sensing devices, thereby enabling accurate reconstruction of traffic trajectories between intersections through data collected by sensing devices at these points.

[0090] In one embodiment, for multiple paths between two nodes, calculating the feature index of each path based on the feature value of the target node may include:

[0091] For multiple paths between two nodes, construct a path weight matrix; based on the number of times the target node appears in the path weight matrix and the degree of difference of the corresponding element values ​​of the same target node in different rows of the path weight matrix, determine the feature value of the target node for each path; for any path, sum the feature values ​​of all target nodes in the path to obtain the feature index.

[0092] In this matrix, each row of the path weight matrix corresponds to all the nodes along a path except for the starting node and the ending node, and the elements in the path weight matrix are the cumulative weights of the nodes in the corresponding path starting from the starting node.

[0093] The embodiments of this application can be used to calculate characteristic indicators of pathways between any two intersections in a road network. Specifically, the two intersections are first determined as the starting node and the ending node, respectively. For all reachable pathways between these two nodes, the pathway nodes other than the two endpoints are extracted from each pathway and written into the corresponding row of a pathway node matrix according to the pathway order.

[0094] For example, the path node matrix MatrixAZ between node A and node Z can be represented by the following equation (1):

[0095] (1)

[0096] The matrix above represents 7 paths between node A and node Z. Each row corresponds to one of these paths, and the letter elements in each row represent the nodes the path passes through. The element 0 is empty (i.e., no node exists there). For example, the first row indicates that the path starts from node A, passes through node C, and directly reaches node Z. The fifth row indicates that the path starts from node A, passes through nodes B, D, and F in sequence, and reaches node Z. The seventh row indicates that the path starts from node A, passes through nodes D, E, F, H, and I in sequence, and reaches node Z.

[0097] Based on the above path node matrix, the weight of the node in the current path can be introduced to generate the path weight matrix MatrixQAZ, which can be expressed as the following equation (2).

[0098] (2)

[0099] In the above formula, B1, C1, D1, D2, E1, E2, E3, etc., represent the weights of each node in the matrix MatrixAZ in the current path. For example, C1 in the first column of the matrix MatrixQAZ indicates that C is the first node passed after starting from the starting node A in the first four paths; G4 in the fourth column of the matrix MatrixQAZ indicates that G is the last node passed before reaching node Z from the starting node A in the sixth path.

[0100] For m paths between the start and end nodes, m×n nodes can be extracted to form a path node matrix and a path weight matrix, where n can be the number of nodes along the longest path among the m paths. It should be noted that the matrix can contain empty nodes. Taking the start node A and end node Z as an example, 7×5 nodes can be extracted using the above method to form the aforementioned matrices MatrixAZ and MatrixQAZ.

[0101] After obtaining the path node matrix and path weight matrix, we can use the matrices to determine the frequency of a node's occurrence in each of the m paths and the node's weight in different paths as eigenvalues. For example, in the matrix MatrixAZ, there are 4 elements named C, indicating that node C appears 4 times in these 7 paths.

[0102] In some possible implementations, before calculating the eigenvalues ​​and eigenindices of the pathway, the elements in the pathway weight matrix can be normalized, mapping the node weights in the pathway to the numerical range (0,1). The normalization formula can be: Nor = real / max. Where Nor is the normalized value of the node weight, real is the true weight of the node in the current pathway, and max is the maximum value in the pathway weight matrix.

[0103] For example, the feature values ​​of a node in the path can characterize the fusion result of the node's frequency and weights, and the frequency can be represented by the number of times the node appears. Optionally, the nodes in the current path node matrix are sorted in descending order of their frequency of appearance, with each node numbered starting from 1, and nodes with the same frequency of appearance having the same number.

[0104] If we denote the feature value of a node in the path as y, then y = f(a, w), where f represents the fusion function, the value of a is equal to the node number, and w can represent the weight of the node in the path.

[0105] After obtaining the feature values ​​of the nodes in the path, for any path, the feature values ​​of all target nodes in the path can be summed, and the sum can be used as the feature index of the path.

[0106] In the above embodiments, by expressing the cumulative weight distribution of nodes in multiple paths in a structured matrix form, and incorporating the frequency of node occurrence and weight differences into the feature value calculation, the comprehensive characterization result of each path can simultaneously reflect the coverage of the target node and the differences between paths. The resulting feature indicators can be directly used to determine whether different paths can be distinguished, and provide a basis for subsequent target node updates and the deployment of sensing devices.

[0107] In one embodiment, determining whether multiple pathways can be uniquely distinguished based on calculated feature indicators includes:

[0108] Clustering is performed on the feature indicators of multiple pathways; if each cluster contains only one pathway in the clustering results, it is determined that multiple pathways can be uniquely distinguished; otherwise, it is determined that multiple pathways cannot be uniquely distinguished, and pathways in the same cluster are considered as indistinguishable pathways.

[0109] Specifically, the feature indicators of each pathway can be used as the clustering input vector. Distance-based clustering methods are employed for grouping, such as determining the proximity between pathways based on Euclidean distance or cosine similarity between feature indicators, and clustering is completed according to a preset threshold. The clustering model can employ K-means clustering or density clustering, combining the number of pathways and the dispersion of feature indicators to determine clustering parameters, ensuring that the feature differences between pathways within the same cluster are less than a preset discrimination threshold. After clustering, if any cluster contains only one pathway, it indicates that each pathway is independent in the current feature space and can form a unique identification result. If a cluster contains multiple pathways, the pathways within that cluster are directly marked as indistinguishable pathways and output to the subsequent node filtering process to continue supplementing target nodes and recalculating feature indicators.

[0110] For example, if the characteristic index of a path is labeled Y, then DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering analysis can be performed on the Y values ​​corresponding to each row of the path weight matrix for the target node to determine whether these Y values ​​belong to the same cluster. If the number of Y values ​​in each cluster is 1, it means that each path in the current matrix can be distinguished by the frequency and weight of the nodes. After deploying sensing devices at the intersections in the target node set, a relatively accurate trajectory can be reconstructed between the starting and ending intersections using the data collected by the sensing devices, the node weights, and the frequency of the nodes.

[0111] In the above embodiments, by clustering and merging feature indicators, the path distinguishability judgment is transformed into a verification process of the number of paths within a cluster. This allows the judgment result to directly reflect the degree of ambiguity of multiple paths under the current sensing device deployment conditions, and provides a clear basis for subsequent selection of new target nodes. Using this method, sets of paths that cannot be uniquely distinguished can be identified without adding additional manual rule judgments, thus making the path discrimination process more stable and helping to focus subsequent layout adjustments on ambiguous paths, improving the accuracy of road network sensing device deployment.

[0112] In one embodiment, selecting nodes from the extracted path nodes that meet preset conditions in terms of frequency of occurrence and weight differences between different paths as target nodes includes:

[0113] For the extracted pathways, obtain the frequency of each node along the pathway in these pathways, as well as the degree of difference in the weight of the same node in different pathways; prioritize the node with the largest degree of difference, and if the degree of difference is the same, select the node with the most frequency of occurrence as the target node.

[0114] In this embodiment, for m paths between start and end nodes, r indistinguishable paths can be identified first, and the nodes along these r paths can be extracted to form a candidate node set. Then, the number of times each node in the set is included in the r paths is counted, and the degree of weight difference for the same node in different paths is calculated. The degree of difference is then used as the first ranking criterion, and among nodes with the same degree of difference, the frequency of occurrence is used as the second ranking criterion for further selection, ultimately determining the target node.

[0115] In some possible implementations, the weight span can be used to represent the degree of difference in weights corresponding to nodes in different paths. Specifically, for the path weight matrix, DBSCAN clustering can be performed on the normalized weights of nodes with the same name in the matrix, and the number of clusters is the span of the weights of nodes with the same name. Then, the nodes in the matrix can be sorted in descending order of frequency of occurrence and weight span, with the sorting priority being weight span > frequency of occurrence, and the node with the largest combined weight span and frequency of occurrence can be selected as the target node.

[0116] In some possible implementations, before selecting a node as the target node, the following steps are taken: determining whether the selected node, the existing target node, and the two nodes form a direct closed loop in the directed weighted connected graph; if a direct closed loop is formed, the node is not selected as the target node, and other nodes that meet the preset conditions are selected; if a direct closed loop is not formed, the node is selected as the target node.

[0117] When selecting a target node from candidate nodes, if a candidate node is to be determined as the target node, it can first be checked for connectivity with the existing target node, starting node, and ending node, and the direction and weight relationships of the edges can be combined to determine whether a direct closed path is formed. If there is a directly reachable closed loop connection between the candidate node and the existing target node or between the two nodes, the candidate node is marked as not meeting the target node condition, and the remaining nodes that meet the conditions of occurrence frequency and weight difference between different paths are further filtered; if no closed loop connection exists, the candidate node is added to the target node set and used for subsequent calculation of path characteristic indicators.

[0118] The direct connection closed-loop check is used to complete rule constraints before adding target nodes, ensuring that the selected nodes do not form direct closed relationships with existing target nodes and related nodes. This maintains the structural characteristics of the target node set suitable for distinguishing multiple paths. By adding direct connection closed-loop verification before selecting target nodes, the final added target nodes better meet the requirements of road network path discrimination, reduce the interference of unsuitable nodes on path feature calculation, and make the subsequent discrimination results of multiple paths more stable.

[0119] In the above embodiments, selecting new target nodes based on the weight differences of nodes in different paths allows target nodes to preferentially fall in positions that have a stronger distinguishing effect on multiple paths, and the occurrence frequency constraint ensures that the node can cover more paths. Once the target node is determined, its corresponding intersection can be used for subsequent deployment of sensing devices and participate in the recalculation of path feature indicators. Since the selected node has both a large weight difference and a high occurrence frequency, it can improve the feature separation degree on which path discrimination depends and reduce the repeated selection of nodes with weak distinguishing ability.

[0120] In one embodiment, before calculating the feature index of each path based on the feature value of the target node, the following may also be included:

[0121] Calculate the topological importance of each node in a directed weighted connected graph; determine the nodes whose topological importance meets the preset conditions as target nodes.

[0122] Topological importance is used to screen road network nodes for structural considerations. It can be determined based on at least one of the following: node degree, rate of change of average path length in the road network after node deletion, and node betweenness centrality. The degree of a node represents the size of the roads connected to the intersection corresponding to the node, and the betweenness centrality of a node is determined based on the number of times the current node is located on the shortest path between other node pairs.

[0123] This application's embodiments can analyze the importance of nodes from a macroscopic, global perspective, and determine the target nodes for deploying sensing devices based on the topological importance of the nodes. Specifically, the travel network structure and traveler behavior have a significant impact on node importance, which can be measured by dimensions such as node degree, average path length change rate, and node betweenness centrality.

[0124] For example, the degree of a node can be the average of the sum of the number of lanes of each approach and exit lane of the road segment connected to the intersection, which can be used to assess the importance of the intersection node according to the road grade. If the degree of the node is denoted as k, then k can be calculated by the following formula (3):

[0125] (3)

[0126] In the above formula, n is the number of approach lanes at the intersection (i.e., the road junction); k j0 k is the number of lanes at the intersection exit. ji This represents the number of lanes at the intersection's approach. The degree of a node describes its connectivity; a higher degree indicates a more important node in the network.

[0127] If the rate of change of the average path length of the road network after deleting a node is denoted as ΔL, it can be calculated by the following formula (4):

[0128] (4)

[0129] In the above formula, L0 is the average path length under free-driving conditions; L i Let L be the average path length of the road network after any node i is deleted. The average path length L is the average distance between all pairs of nodes in the traffic network, which can be calculated using the following formula (5):

[0130] (5)

[0131] In the above formula, N is the number of nodes in the road network; d ij Let V be the shortest distance from node i to node j; V is the set of nodes in the road network. The greater the rate of change of the average path length, the more important the node is in the traffic network. This path length can also be replaced by other weights such as travel time.

[0132] Node betweenness centrality based on travel time can be defined as the number of times a node at an intersection in a road traffic network lies on the shortest path between other intersection pairs. Here, the shortest path can be either the shortest distance or the shortest travel time. Node betweenness centrality can be calculated using the following formula (6):

[0133] (6)

[0134] In the above formula, od(y,z,x) represents the starting node, z represents the ending node, and x represents the node currently used to calculate the betweenness centrality. If x is on the shortest path from y to z, the value of od(y,z,x) is 1; if x equals y but not z, the value of od(y,z,x) is 1 / 2; if x equals z but not y, the value of od(y,z,x) is 1 / 2; if x equals and equals y, the value of od(y,z,x) is 1 / 3; otherwise, the value of od(y,z,x) is 0.

[0135] For example, the degree, average path length change rate, and betweenness centrality of the nodes can be normalized individually. The normalized results are then weighted and subjected to intersection and union analysis to output a set of topological key nodes Rt as the target nodes.

[0136] In the above embodiments, by calculating the degree and other indicators that characterize the topological importance of nodes, the selection of target nodes can be more closely related to the road network topology, thereby improving the effectiveness of the path characteristic indicators, making the differentiation results of multiple subsequent paths more stable, and concentrating the deployment of sensing devices at key structural intersections, thus improving the targeting of road network sensing device deployment.

[0137] In one embodiment, before calculating the feature index of each path based on the feature value of the target node, the method further includes:

[0138] Calculate the business importance of each node in the directed weighted connected graph; determine the nodes whose business importance meets the preset conditions as target nodes.

[0139] The importance of the business is determined based on at least one of the following: traffic flow at the intersection corresponding to the node, historical traffic case data, and key perception paths.

[0140] This application embodiment can select target nodes based on the business importance of nodes. The business importance of nodes mainly considers the traffic flow characteristics of nodes, whether they are located in areas with a high incidence of incidents, whether they are located near key locations, and whether they are located on important security routes.

[0141] For example, node traffic flow characteristics can be calculated based on vehicle trajectory data, such as floating car data / checkpoint trajectory data. Taking floating car trajectories as an example, vehicle trajectories for one week are extracted, each trip is treated as a trajectory, and road network matching is performed on each trajectory. The number of trajectories passing through each intersection is taken as the intersection's traffic flow. Statistical analysis is performed on the above calculation results to generate a statistical graph, with the horizontal axis representing the number of intersections and the vertical axis representing traffic flow. Nodes corresponding to intersections with traffic flow greater than f are extracted to generate the first node set Rf.

[0142] For example, it is also possible to acquire and analyze historical case data of the region, analyze areas with high incidence of historical cases, and extract surface data of these high-incidence areas. Key unit POIs can be acquired, buffer analysis can be performed on these key units, and the covered streets can be extracted. Important security routes can be acquired, and the road segments they pass through can be extracted. Overlay analysis can be performed on data such as intersections, high-incidence areas of cases, streets near key units, and important security routes to extract the nodes corresponding to intersections falling within these areas and generate a second node set Rp.

[0143] In some possible implementations, the first node set Rf and the second node set Rp can be combined to determine the third node set β with higher business importance. For example, the first node set Rf and the second node set Rp can be combined into a union.

[0144] After obtaining the third node set β, we can perform a weighted intersection and union operation on this set and the currently determined target node set RC to obtain the final node set R for deploying intelligent sensing devices.

[0145] In this way, target nodes can prioritize coverage of intersections with high traffic volume, concentrated historical cases, or located on key sensing paths, making the calculation of path characteristic indicators closer to the actual business focus areas, thereby making the subsequent differentiation results of multiple paths more in line with the deployment needs of traffic management and public safety sensing equipment.

[0146] Figure 2 This is a schematic diagram illustrating a process for determining a target node, provided as an embodiment of this application. Figure 2 As shown, in some specific examples, the intersections where sensing devices need to be deployed can be determined by combining multiple dimensions such as the topological importance of the intersection nodes, business importance, and path distinguishability.

[0147] First, after obtaining the road network data, we can remove roads below the township / village level or roads within the park that do not need attention. On this basis, we can also remove intersections that only connect two roads, so that the remaining intersection nodes connect at least three roads.

[0148] Figure 3 This is a partial schematic diagram of a road network processing result provided in an embodiment of this application. Figure 3 As shown, after processing the road segments and intersections of the road network, each intersection such as 8065, 8051, 8066 and 3869 in the road network area connects at least three roads.

[0149] Based on the road network topology, a directed weighted connected graph with intersections as nodes can be constructed. The structure of the directed weighted connected graph is graph G: [nodes, links (start point, end point), weights, etc.].

[0150] Then, path calculations can be performed on each pair of nodes in the directed weighted connected graph to obtain the path matrix MatrixG. The calculation process can be referenced in the following table:

[0151]

[0152] In the table above, TL n The Nth path between two points consists of an ordered set of node numbers (e.g., 1, 2, 3, 4, 5, 6, 7). For example, the Nth path between points 1 and 3 is {1, 4, 2, 7, 6, 3}. The N paths between two points can be the first N paths with the shortest travel time.

[0153] For intersections in the road network that have already been equipped with sensing devices, these intersections can be matched to nodes in a directed weighted connected graph. These intersection nodes equipped with sensing devices can be transformed into target nodes and combined with the set of topological key nodes Rt to generate the current target node set.

[0154] In some possible implementations, if the road network area is large and the number of intersections within the road network is large, the road network topology can be processed into a grid.

[0155] Figure 4 This is a schematic diagram illustrating the result of localized gridding of a road network, as provided in an embodiment of this application. Figure 4As shown, the original road network can be divided into multiple analysis regions (Cells) through meshing. For each analysis region, the method described in the above embodiment can be executed to determine the target nodes. The nodes of the analysis region can be defined as nodes that fall directly inside the analysis region and neighboring nodes outside the region contained in road segments that intersect with the analysis region. Even if the road that crosses the boundary of the Cell extends outward to the nearest intersection node and is included in the Cell, although the nodes at the edge may be calculated twice, the calculation error at the edge can be reduced.

[0156] In one embodiment, the intersection layout sensing device corresponding to the target node includes:

[0157] Based on the node type attribute of the target node, determine the type of sensing device to be deployed at the corresponding intersection.

[0158] The node type attribute includes at least one of the following: a topological key node, a node whose traffic flow meets preset conditions, a node located within a preset range of key locations, a node located on a security path, or a highway entrance / exit node. In some possible implementations, the node type attribute may also include nodes with monitoring deployment requirements, such as railway stations, bus stations, and public service halls.

[0159] The types of sensing devices include at least one of the following: devices for capturing facial images, devices for capturing vehicle license plates, devices for capturing panoramic images, and devices for collecting terminal identification information.

[0160] In this embodiment, after determining the target nodes where sensing devices need to be deployed, the type of sensing devices to be deployed can be further determined based on the node's own type attributes. Specifically, the steps of topology importance calculation and business importance calculation in the above embodiments can not only filter out key intersection nodes, but also reflect the type labels of the nodes, such as topological key nodes, nodes with high traffic volume, nodes located near key locations or on security routes, highway entrance and exit nodes, etc. Based on this, these type labels can be matched and associated with the functional characteristics of the sensing devices: for intersections where vehicle recognition and license plate capture are the main requirements, devices for capturing vehicle license plates are prioritized; for intersections where face capture and pedestrian feature recognition are the main requirements, devices for capturing face images are prioritized; for intersections requiring large-scale situational awareness, devices for capturing panoramic images are deployed; and for areas requiring the collection of mobile terminal or communication identification information, devices for collecting terminal identification information are deployed.

[0161] For example, the characteristics of different types of sensing devices can be represented by Table 1 below.

[0162] Table 1:

[0163]

[0164] For example, based on the types of sensing devices described above, matching can be performed according to node type. For instance, sensing devices of different node types can be assigned according to Tables 2 and 3 below.

[0165] Table 2:

[0166]

[0167] Table 3:

[0168]

[0169] When actually deploying sensing devices, nodes that have already been deployed with sensing devices in the target node set can be removed, and a set of intersection nodes that need to have new sensing devices can be output. The set can record information such as intersection name, intersection geometry, recommended type of sensing device to be deployed, and typical deployment location.

[0170] In some possible implementations, we can analyze the number of intersections that have already deployed smart sensing devices, the number of intersections that need to be newly equipped with smart sensing devices, and the percentage of intersections that do not need to be equipped with smart sensing devices among all intersections.

[0171] In the above embodiments, based on the already determined deployment points, the specific equipment types to be deployed at each point are further clarified, so that the entire layout scheme is not only streamlined and efficient in terms of the number and location of points, but also achieves a precise match between functions and requirements in terms of equipment selection, avoiding resource waste or perception blind spots caused by uniformly deploying similar equipment.

[0172] In one example, a road network within a specific area can be selected, and the effectiveness of this solution can be verified using vehicle-mounted GPS trajectories as a benchmark. Vehicle-mounted GPS has a high sampling frequency and collects coordinates with high accuracy. Multiple floating car GPS trajectories are collected, verified, and noise is removed before road network matching is performed on these trajectories. The matched trajectory paths serve as the verification benchmark. For each trajectory, the intersection points and passage times are extracted, and the intersection points present in the target node set are retained to generate simulated trajectory data. For each simulated trajectory data, the vehicle trajectory is reconstructed using information such as travel time and the probability of passing intersections. The overlap ratio between the reconstructed trajectory and its respective benchmark trajectory is calculated; trajectories with an overlap ratio greater than 90% are considered to have high reconstruction accuracy.

[0173] The formula for calculating the effectiveness of netting is Ef=N match / N sample , where N sample N can represent the number of sample trajectories collected. matchEf represents the number of reconstructed trajectories with a greater than 90% overlap with the sample trajectories, and Ef represents the proportion of trajectories with a greater than 90% overlap among all sample trajectories. If Ef is greater than 80%, it indicates that the road network-based intersection sensing device deployment point recommendation algorithm meets expectations and can accurately reconstruct vehicle / personnel trajectories.

[0174] In one example, the layout method of this application embodiment is applied to a local road network in a certain area to verify the execution effect.

[0175] In actual implementation, the weighted connectivity graph of the corresponding road network area can be constructed first according to the method in the above embodiments, the paths between nodes can be analyzed, and the intersection nodes where sensing devices need to be deployed can be determined. Then, based on the node type and sensing device type of the intersection, the suggested sensing devices for each intersection node can be further refined, and the layout points corresponding to different sensing devices can be output.

[0176] The results of the suggested deployment of sensing devices in the aforementioned region are as follows:

[0177] (1) Recommended number of intersections to be equipped with intelligent sensing devices: 587.

[0178] (2) Recommended number of intersections where no intelligent sensing devices are installed: 426.

[0179] (3) It is recommended that the proportion of intersections without intelligent sensing devices in the total number of intersections in the road network of this area be 426 / 1013=0.42.

[0180] (4) Number of intersections where intelligent sensing devices have been deployed: 267.

[0181] (5) The number of intersections where intelligent sensing devices are recommended to be deployed is 91.

[0182] (6) The number of intersections where smart sensing devices are recommended to be installed is 496.

[0183] The results were verified by extracting vehicle trajectories, and the results showed that there were 78 baseline trajectories and 70 similar trajectories, with a similarity rate of 84%.

[0184] The layout method implemented in this application embodiment can have the following effects:

[0185] First, the trajectory data collected by sensing devices can be used to reconstruct a vehicle / person's trajectory between two intersections as accurately as possible. This reduces the time that staff spend judging and selecting the more likely trajectory from multiple uncertain trajectories, and enhances the accuracy of the vehicle's spatiotemporal collision range.

[0186] Second, by analyzing the topological importance of nodes, a group of key nodes with high importance were extracted. Through key node calculations between any two intersections, a set of important nodes in the path was extracted. These key nodes were then further reduced and extracted based on key business priorities, such as areas with high crime rates or areas near important locations, thus reducing the number of nodes deployed and the associated costs.

[0187] Third, the deployment plan for intelligent sensing devices is refined based on their types, uses, and common installation locations, providing suggested types and locations of sensing devices for intersections. Fully utilizing device characteristics in the deployment plan reduces deployment costs while still meeting the goals of service and public safety.

[0188] Figure 5 This application provides a structural schematic diagram of a road network sensing device layout apparatus, as shown below. Figure 5 As shown, the road network sensing device layout device 500 provided in this embodiment includes:

[0189] The mapping module 501 is used to construct a directed weighted connected graph with intersections as nodes for a road network containing multiple intersections. The weights of the directed weighted connected graph are the road length or the road travel time.

[0190] The calculation module 502 is used to calculate the feature index of each path based on the feature value of the target node for multiple paths between two nodes. The target node is the node of the determined layout sensing device. The feature value is determined based on the number of times the node appears in multiple paths and the weight of the node in the corresponding path.

[0191] The judgment module 503 is used to determine whether multiple paths can be uniquely distinguished based on the calculated feature indicators.

[0192] The iteration module 504 is used to extract the indistinguishable paths when they cannot be uniquely distinguished. From the nodes traversed by the extracted paths, the nodes whose occurrence frequency and weight differences between different paths meet the preset conditions are selected as target nodes, and the judgment is re-executed until multiple paths can be uniquely distinguished.

[0193] The layout module 505 is used to lay out sensing devices according to the intersection corresponding to the target node.

[0194] In some possible implementations, the calculation module 502 can also be used to: construct a path weight matrix for multiple paths between two nodes, where each row of the path weight matrix corresponds to all the nodes along a path except for the starting node and the ending node, and the elements in the path weight matrix are the cumulative weights of the nodes in the corresponding path starting from the starting node; determine the feature values ​​of the target node for each path based on the number of times the target node appears in the path weight matrix and the degree of difference in the corresponding element values ​​of the same target node in different rows of the path weight matrix; and sum the feature values ​​of all target nodes in any path to obtain the feature index.

[0195] In some possible implementations, the judgment module 503 can also be used to: perform clustering processing on the feature indicators of multiple pathways; if each cluster in the clustering result contains only one pathway, then the multiple pathways are determined to be uniquely distinguishable; otherwise, the multiple pathways are determined to be uniquely distinguishable, and the pathways in the same cluster are regarded as indistinguishable pathways.

[0196] In some possible implementations, the iteration module 504 can also be used to: for the extracted path, obtain the number of times each node along the path appears in these paths, and the degree of difference in the weight of the same node in different paths; prioritize the node with the largest degree of difference, and if the degree of difference is the same, select the node with the most occurrences as the target node.

[0197] In some possible implementations, the iteration module 504 can also be used to: determine whether the selected node and the existing target node, as well as the two nodes, form a direct closed loop in the directed weighted connected graph; if a direct closed loop is formed, the selected node is not used as the target node, and other nodes that meet the preset conditions are reselected; if a direct closed loop is not formed, the selected node is used as the target node.

[0198] In some possible implementations, the iteration module 504 can also be used to: calculate the topological importance of each node in the directed weighted connected graph, the topological importance being determined based on at least one of the following: the degree of the node, the rate of change of the average path length of the road network after deleting the node, and the betweenness centrality of the node; wherein, the degree of the node represents the size of the road connected by the intersection corresponding to the node, and the betweenness centrality is determined based on the number of times the current node is located on the shortest path between other node pairs; and determine the nodes whose topological importance satisfies the preset conditions as target nodes.

[0199] In some possible implementations, the iteration module 504 can also be used to: calculate the business importance of each node in the directed weighted connected graph, the business importance being determined based on at least one of the traffic flow at the intersection corresponding to the node, historical traffic case data, and key perception paths; and determine the nodes whose business importance meets the preset conditions as target nodes.

[0200] In some possible implementations, the layout module 505 can also be used to: determine the type of sensing device deployed at the corresponding intersection based on the node type attribute of the target node; wherein the node type attribute includes at least one of the following: topological key node, node whose traffic flow meets preset conditions, node located within a preset range of key locations, node located on the security path, and highway entrance / exit node; the type of sensing device includes at least one of the following: device for capturing facial images, device for capturing vehicle license plates, device for capturing panoramic images, and device for collecting terminal identification information.

[0201] The road network sensing device layout device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0202] Figure 6 This is a schematic diagram of the structure of an electronic device provided in this application. Figure 6 As shown, the electronic device 60 provided in this embodiment includes at least one processor 601 and a memory 602. Optionally, the device 60 further includes a communication component 603. The processor 601, memory 602, and communication component 603 are connected via a bus 604.

[0203] In a specific implementation, at least one processor 601 executes computer execution instructions stored in memory 602, causing at least one processor 601 to perform the above-described method.

[0204] The specific implementation process of processor 601 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0205] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0206] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0207] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0208] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0209] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0210] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0211] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0212] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0213] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0214] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0215] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0216] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0217] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for deploying road network sensing devices, characterized in that, include: For a road network containing multiple intersections, a directed weighted connected graph is constructed with intersections as nodes. The weights of the directed weighted connected graph are the road length or the road travel time. For multiple paths between two nodes, feature indicators of each path are calculated based on the feature values ​​of the target node, where the target node is a node with a determined layout of sensing devices, and the feature values ​​are determined based on the number of times the node appears in the multiple paths and the weight of the node in the corresponding path. Based on the calculated feature indicators, it is determined whether the multiple pathways can be uniquely distinguished. If a path cannot be uniquely distinguished, the path that cannot be distinguished is extracted. From the nodes traversed by the extracted path, the nodes whose frequency of occurrence and the weight difference between different paths meet the preset conditions are selected as target nodes. The judgment is then re-executed until the multiple paths can be uniquely distinguished. The target node is located at the intersection where the sensing devices are deployed.

2. The method according to claim 1, characterized in that, Of the two nodes, one is the starting node and the other is the ending node. For multiple paths between the two nodes, the characteristic indicators of each path are calculated based on the characteristic values ​​of the target node, including: For multiple paths between two nodes, a path weight matrix is ​​constructed. Each row of the path weight matrix corresponds to all the nodes along a path except for the starting node and the ending node. The elements in the path weight matrix are the cumulative weights of the nodes in the corresponding path starting from the starting node. Based on the number of times the target node appears in the path weight matrix, and the degree of difference in the corresponding element values ​​of the same target node in different rows of the path weight matrix, the feature value of the target node for each path is determined; For any path, the feature values ​​of all target nodes in the path are summed to obtain the feature index.

3. The method according to claim 1 or 2, characterized in that, The determination of whether the multiple pathways can be uniquely distinguished based on the calculated feature indicators includes: Clustering is performed on the feature indicators of the multiple pathways; If each cluster in the clustering results contains only one pathway, then the multiple pathways are determined to be uniquely distinguishable. Otherwise, it is determined that the multiple paths cannot be uniquely distinguished, and the paths within the same cluster are considered indistinguishable paths.

4. The method according to claim 1 or 2, characterized in that, The step of selecting target nodes from the nodes traversed by the extracted pathways, based on their frequency of occurrence and the weight differences between different pathways meeting preset conditions, includes: For the extracted pathways, obtain the number of times each node along the pathway appears in these pathways, and the degree of difference in the weight of the same node in different pathways; Prioritize selecting the node with the greatest degree of difference. If the degree of difference is the same, select the node that appears most frequently as the target node.

5. The method according to claim 1 or 2, characterized in that, Before selecting the node as the target node, the following steps are also included: Determine whether the selected node, the existing target node, and the two nodes form a direct closed loop in the directed weighted connected graph. If a direct closed loop is formed, the selected node will not be used as the target node, and other nodes that meet the preset conditions will be selected again. If a direct closed loop is not formed, the selected node will be used as the target node.

6. The method according to claim 1 or 2, characterized in that, Before calculating the feature indexes of each path based on the feature values ​​of the target node, the following steps are also included: Calculate the topological importance of each node in the directed weighted connected graph. The topological importance is determined based on at least one of the following: the degree of the node, the rate of change of the average path length of the road network after deleting the node, and the betweenness centrality of the node. The degree of the node represents the size of the road connected by the intersection corresponding to the node, and the betweenness centrality is determined based on the number of times the current node is located on the shortest path between other pairs of nodes. Nodes whose topological importance meets the preset conditions are identified as target nodes.

7. The method according to claim 1 or 2, characterized in that, Before calculating the feature indexes of each path based on the feature values ​​of the target node, the following steps are also included: Calculate the business importance of each node in the directed weighted connected graph, wherein the business importance is determined based on at least one of the following: traffic flow at the intersection corresponding to the node, historical traffic case data, and key perception paths; Nodes whose business importance meets the preset conditions are identified as target nodes.

8. The method according to claim 1 or 2, characterized in that, Based on the intersection layout sensing devices corresponding to the target node, including: Based on the node type attribute of the target node, determine the type of sensing device deployed at the corresponding intersection; The node type attribute includes at least one of the following: topological key node, node whose traffic flow meets preset conditions, node located within a preset range of key locations, node located on a security path, and highway entrance / exit node. The types of sensing devices include at least one of the following: devices for capturing facial images, devices for capturing vehicle license plates, devices for capturing panoramic images, and devices for collecting terminal identification information.

9. A road network sensing device layout apparatus, characterized in that, include: The mapping module is used to construct a directed weighted connected graph with intersections as nodes for a road network containing multiple intersections. The weights of the directed weighted connected graph are the road length or the road travel time. The calculation module is used to calculate the feature index of each path based on the feature value of the target node for multiple paths between two nodes. The target node is a node of the determined layout sensing device. The feature value is determined based on the number of times the node appears in the multiple paths and the weight of the node in the corresponding path. The judgment module is used to determine whether the multiple paths can be uniquely distinguished based on the calculated feature indicators; The iterative module is used to extract indistinguishable paths when they cannot be uniquely distinguished. From the nodes traversed by the extracted paths, the nodes whose frequency of occurrence and the weight difference between different paths meet the preset conditions are selected as target nodes, and the judgment is re-executed until the multiple paths can be uniquely distinguished. The layout module is used to deploy sensing devices according to the intersection corresponding to the target node.

10. An electronic device / computer-readable storage medium / computer program product, characterized in that, The electronic device includes: a processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 8; And / or, The computer-readable storage medium stores computer-executable instructions, which, when executed, are used to implement the method as described in any one of claims 1 to 8; And / or, The computer program product includes a computer program that, when executed, implements the method of any one of claims 1 to 8.

Images