Intelligent arrangement method of elevated monitoring based on graph evolution algorithm

Abstract

The application discloses an elevated monitoring intelligent arrangement method based on a graph evolution algorithm, and relates to the technical field of image communication. The application adopts double-layer graph parameterization, comprehensively represents the structural features of an elevated structure, and provides accurate parameter basis for intelligent optimization; and based on PPM derived from imaging geometry, the application realizes accurate adaptation of different monitoring tasks (face recognition, license plate recognition, etc.); and the graph evolution algorithm adopted by the application efficiently explores the optimal solution space through crossover mutation and mutation operation, and takes into account coverage effect, imaging accuracy and cost minimization; the setting of the application does not need artificial experience intervention, is suitable for multi-scene elevated monitoring arrangement, and improves engineering safety and management efficiency.

Description

Technical Field

[0001] This invention relates to the field of image communication technology, specifically to an intelligent deployment method for elevated monitoring based on graph evolution algorithms. Background Technology

[0002] In scenarios such as pedestrian overpasses, large construction sites, and port terminals, gantry cranes and their surrounding areas typically require the deployment of numerous monitoring devices to achieve real-time detection of personnel safety and full monitoring of cargo movement during construction and handling. This not only helps ensure the safety of workers but also allows for dynamic statistics on construction progress, assisting in project management and enabling more efficient scheduling and decision-making. Figure 1 As shown.

[0003] However, traditional surveillance deployment methods often rely on manual experience, leading to problems such as unreasonable placement, insufficient coverage, or wasted resources. Therefore, establishing an efficient intelligent deployment method for elevated surveillance systems, as an upstream task in the entire surveillance system, is particularly important.

[0004] Therefore, a new solution is needed to address the above problems. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent deployment method for elevated monitoring based on graph evolution algorithm, so as to solve the technical problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an intelligent deployment method for elevated monitoring based on graph evolution algorithm, comprising at least the following steps:

[0007] S1: Two-layer graph parameterization, which uses graph data to parameterize buildings, achieves comprehensive parameterization of all design scenarios, and further proposes a two-layer graph parameterization method using planning and decision-making layers;

[0008] S2: Target setting. Employing a camera field-of-view analysis and sharpness assessment method based on imaging geometry, the target precision and coverage area for monitoring people or objects are set according to different usage scenarios. Camera field-of-view parameters are calculated based on imaging geometry, deriving the pixel density (PPM) at the boundary of the visible ground area. An optimization model is then constructed, aiming to minimize the total cost of the monitoring equipment while satisfying the condition that the PPM at any point within the target monitoring area is not lower than a preset pixel density threshold. Surveillance coverage Not lower than the preset coverage threshold ;

[0009] S3: Graph optimization algorithm, which is based on genetic algorithm to design graph chromosome operations. The graph chromosome operations include crossover and mutation operations and mutation operations. The optimal monitoring layout scheme is obtained through multiple generations of iteration by fitness function sorting and elite retention strategy.

[0010] S4: Map the optimized design results obtained through the graph optimization algorithm to the actual elevated structure to obtain the layout results.

[0011] Furthermore, the parameterization of the structure using graph data involves discretizing and abstracting long and wide elevated structures (such as truss beams or main beams of bridge erecting machines) into graph structure data.

[0012] Key structural points on the diagram (such as gusset plates, weld junctions, or component connections) are considered as nodes of the diagram. The geometric connection relationship between components is regarded as edge feature. Thus, undirected ;

[0013] in, A set of nodes consisting of key construction points. It is the set of edges formed by the geometric connection relationships of the components.

[0014] Furthermore, the specific construction method of the planning layer is as follows:

[0015] Each node is equipped with a set of feature vectors The feature vector Includes camera-related parameters The relative position of the leftmost node ;

[0016] Edge features The location of the camera is coded. and related camera parameters , which are variables for optimization design, where i and j are nodes i and j, respectively.

[0017] Furthermore, the specific method for constructing the decision-making layer is as follows:

[0018] Each node and each edge Each has a characteristic parameter Gate i and Gate ij ;

[0019] The value of Gate is 0 or 1, which respectively represent a scheme that does not use the planning layer and a method that uses the planning layer.

[0020] Furthermore, the camera field of view analysis and sharpness evaluation method based on imaging geometry is as follows:

[0021] First, based on the installation height of the surveillance camera Pitch angle and horizontal / vertical field of view Position the camera's field of view relative to the ground plane (e.g., ... Perform geometric intersection to calculate the closest distance to the visible area on the ground. Longest distance and the corresponding horizontal width , Based on this, the two-dimensional corner coordinates of the ground field of view trapezoid are constructed. and its three-dimensional extended coordinates ;

[0022] Based on this, combined with the horizontal resolution of the image Furthermore, the PPM at different distances was derived.

[0023] Furthermore, the construction of the optimization model includes at least the following steps:

[0024] In a given target monitoring area and pixel density lower limit threshold Under the condition that any point in the region is required The imaging pixel density satisfies the following formula:

[0025]

[0026] Among them, let For the target monitoring area, The combined ground coverage area for all selected monitoring devices is defined as follows:

[0027]

[0028] The monitoring coverage rate is no less than the preset coverage requirement, achieving coverage of the area. Full coverage or coverage rate not less than As shown in the following formula:

[0029]

[0030] Under the premise that the above constraints on imaging clarity and spatial coverage are met, the total cost of the monitoring equipment is used as the optimization objective to minimize the total price of the equipment selection and deployment scheme, as shown in the following formula:

[0031]

[0032] in, For the first The unit price of each candidate surveillance device; This is the decision variable for whether or not the device is selected.

[0033] Furthermore, the crossover mutation operation includes at least the following steps:

[0034] By exchanging the variable values ​​on the corresponding edges of two chromosomes in a graph with a certain probability, it is also possible to exchange the variable values ​​of nodes in the graph.

[0035] Suppose two parent chromosome diagrams are respectively , above Each gene (which can be a variable of an edge or a variable of a node) is respectively ;

[0036] For each gene locus, a crossover probability is introduced. and uniform random numbers Then, the two offspring chromosomes undergo crossing over variation, as shown in the following formula: .

[0037] Furthermore, the mutation operation includes at least the following steps:

[0038] The variable values ​​on the edges of a single chromosome or the variable values ​​of a node are changed with a certain probability within the feasible region;

[0039] Let the single-line chromosome be the first The current value of each gene is Its feasible region is ;

[0040] Introducing mutation probability and random numbers The mutation operation can be defined as:

[0041]

[0042] in, In the feasible region A mutation function that performs numerical perturbation or resampling on edge or node variables.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] 1. This invention employs a dual-layer graph parameterization to comprehensively characterize the structural features of elevated structures, providing a precise parameter basis for intelligent optimization; and based on imaging geometry derivation of PPM, it achieves precise adaptation to different monitoring tasks (face recognition, license plate recognition, etc.); and the graph evolution algorithm used efficiently explores the optimal solution space through crossover mutation and mutation operations, taking into account coverage effect, imaging accuracy and cost minimization.

[0045] 2. The setup of this invention requires no human experience intervention, is applicable to elevated monitoring deployment in multiple scenarios, and improves engineering safety and management efficiency. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart of the method steps of the present invention;

[0048] Figure 2 This is a schematic diagram of the two-layer graph parameterization method of the present invention;

[0049] Figure 3 This is a schematic diagram of the camera's field of view according to the present invention;

[0050] Figure 4 This is a schematic diagram of the evolutionary algorithm for the elevated monitoring layout of the present invention;

[0051] Figure 5 This is a schematic diagram of the monitoring layout design result of the present invention. Detailed Implementation

[0052] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0053] Existing technology

[0054] Please see Figure 1 The intelligent deployment method for elevated monitoring based on graph evolution algorithm includes at least the following steps:

[0055] See Figure 2 S1: Dual-layer graph parameterization uses graph data to parameterize the building, achieving comprehensive parameterization of all design scenarios. Furthermore, a dual-layer graph parameterization method using planning and decision layers is proposed. The proposed dual-layer graph parameterization method can comprehensively and efficiently characterize the combined features of elevated buildings, realize convenient and efficient digitization, and provide a parameterization basis for intelligent optimization design methods.

[0056] S2: Target setting. Employing a camera field-of-view analysis and sharpness assessment method based on imaging geometry, the target precision and coverage area for monitoring people or objects are set according to different usage scenarios. Camera field-of-view parameters are calculated based on imaging geometry, deriving the pixel density (PPM) at the boundary of the visible ground area. An optimization model is then constructed, aiming to minimize the total cost of the monitoring equipment while satisfying the condition that the PPM at any point within the target monitoring area is not lower than a preset pixel density threshold. Surveillance coverage Not lower than the preset coverage threshold ;

[0057] See Figure 4 S3: Graph optimization algorithm. The graph optimization algorithm is based on the design of graph chromosome operations using genetic algorithms. Graph chromosome operations include crossover and mutation operations and mutation operations. Through fitness function sorting and elite retention strategy, the optimal monitoring layout scheme is obtained through multiple generations of iteration.

[0058] S4: Mapping the optimized design results obtained through the graph optimization algorithm to the actual elevated structure can yield results such as... Figure 5 The layout results shown

[0059] The method of parameterizing a structure using graph data involves discretizing and abstracting long and wide elevated structures (such as truss beams or the main beams of bridge erecting machines) into graph structure data.

[0060] Specifically:

[0061] Key structural points on the diagram (such as gusset plates, weld junctions, or component connections) are considered as nodes of the diagram. The geometric connection relationship between components is regarded as edge feature. Thus, undirected ;

[0062] in, A set of nodes consisting of key construction points. It is the set of edges formed by the geometric connection relationships of the components.

[0063] The specific construction method of the planning layer is as follows:

[0064] Each node is equipped with a set of feature vectors eigenvectors Includes camera-related parameters The relative position of the leftmost node ;

[0065] Edge features The location of the camera is coded. and related camera parameters , which are variables for optimization design, where i and j are nodes i and j, respectively.

[0066] The specific method for constructing the decision-making layer is as follows:

[0067] Each node and each edge Each has a characteristic parameter Gate i and Gate ij ;

[0068] The value of Gate is 0 or 1, which respectively represent a scheme that does not use the planning layer and a method that uses the planning layer.

[0069] The camera field of view analysis and sharpness assessment method based on imaging geometry is as follows:

[0070] First, based on the installation height of the surveillance camera Pitch angle and horizontal / vertical field of view Position the camera's field of view relative to the ground plane (e.g., ... Perform geometric intersection to calculate the closest distance to the visible area on the ground. Longest distance and the corresponding horizontal width , Based on this, the two-dimensional corner coordinates of the ground field of view trapezoid are constructed. and its three-dimensional extended coordinates ;

[0071] Based on this, combined with the horizontal resolution of the image Furthermore, the PPM at different distances can be derived. Figure 3 As shown, pixel density is used to quantitatively characterize the imaging accuracy and recognition capability level per unit length in an image, providing a basis for camera parameter configuration in different monitoring tasks, such as face recognition and license plate recognition applications. Among these, the type of monitoring... Installation height Pitch angle These are the optimization variables for this invention.

[0072] Building an optimization model includes at least the following steps:

[0073] In a given target monitoring area and pixel density lower limit threshold Under the condition that any point in the region is required The imaging pixel density satisfies the following formula:

[0074]

[0075] Among them, let For the target monitoring area, The combined ground coverage area for all selected monitoring devices is defined as follows:

[0076]

[0077] The monitoring coverage rate is no less than the preset coverage requirement, achieving coverage of the area. Full coverage or coverage rate not less than As shown in the following formula:

[0078]

[0079] Under the premise that the above constraints on imaging clarity and spatial coverage are met, the total cost of the monitoring equipment is used as the optimization objective to minimize the total price of the equipment selection and deployment scheme, as shown in the following formula:

[0080]

[0081] in, For the first The unit price of each candidate surveillance device; This serves as a decision variable for whether the device is selected, thereby minimizing the construction cost of the monitoring system while ensuring imaging quality and coverage requirements within the specified area.

[0082] Crossover mutation operations include at least the following steps:

[0083] By exchanging the variable values ​​on the corresponding edges of two chromosomes in a graph with a certain probability, it is also possible to exchange the variable values ​​of nodes in the graph.

[0084] Suppose two parent chromosome diagrams are respectively , above Each gene (which can be a variable of an edge or a variable of a node) is respectively ;

[0085] For each gene locus, a crossover probability is introduced. and uniform random numbers Then, the two offspring chromosomes undergo crossing over variation, as shown in the following formula: .

[0086] Mutation operations include at least the following steps:

[0087] The variable values ​​on the edges of a single chromosome or the variable values ​​of a node are changed with a certain probability within the feasible region;

[0088] Let the single-line chromosome be the first The current value of each gene is Its feasible region is ;

[0089] Introducing mutation probability and random numbers The mutation operation can be defined as:

[0090]

[0091] in, In the feasible region The mutation function, which performs numerical perturbation or resampling on edge or node variables, makes local random changes to edge / node variables on the graph chromosome with a certain probability, thereby enabling local exploration of the solution space.

[0092] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A method for intelligent deployment of elevated monitoring based on graph evolution algorithm, characterized in that: At least the following steps are included: S1: Two-layer graph parameterization, which uses graph data to parameterize buildings and provides comprehensive parameterization for all design scenarios, proposes a two-layer graph parameterization method using planning and decision-making layers; The parameterization of the structure using graph data involves discretizing and abstracting the long and wide elevated structure into graph structure data. Treat the key constructs in the graph structure as nodes of the graph. The geometric connection relationship between components is regarded as edge feature. Thus, undirected ;in, A set of nodes consisting of key construction points. It is the set of edges formed by the geometric connection relationships of the components; S2: Target setting. Employing a camera field-of-view analysis and sharpness assessment method based on imaging geometry, the target precision and coverage area for monitoring people or objects are set according to different usage scenarios. Camera field-of-view parameters are calculated based on imaging geometry, deriving the pixel density (PPM) at the boundary of the visible ground area. An optimization model is then constructed, aiming to minimize the total cost of the monitoring equipment while satisfying the condition that the PPM at any point within the target monitoring area is not lower than a preset pixel density threshold. Surveillance coverage Not lower than the preset coverage threshold ; S3: Graph optimization algorithm, which is based on genetic algorithm to design graph chromosome operations. The graph chromosome operations include crossover and mutation operations and mutation operations. The optimal monitoring layout scheme is obtained through multiple generations of iteration by fitness function sorting and elite retention strategy. The crossover mutation operation includes at least the following steps: By exchanging the variable values ​​on the corresponding edges of two chromosomes in a graph with a certain probability, it is also possible to exchange the variable values ​​of nodes in the graph. Suppose the two parent chromosomes are respectively , above The genes are respectively ; For each gene locus, a crossover probability is introduced. and uniform random numbers Then, the two offspring chromosomes undergo crossing over variation, as shown in the following formula: ; The mutation operation includes at least the following steps: The variable values ​​on the edges of a single chromosome or the variable values ​​of a node are changed with a certain probability within the feasible region; Let the single-line chromosome be the first The current value of each gene is Its feasible region is ; Introducing mutation probability and random numbers The mutation operation can be defined as: in, In the feasible region A mutation function that performs numerical perturbation or resampling on edge or node variables; S4: Map the optimized design results obtained through the graph optimization algorithm to the actual elevated structure to obtain the layout results.

2. The elevated monitoring intelligent deployment method based on graph evolution algorithm according to claim 1, characterized in that: The specific construction method of the planning layer is as follows: Each node is equipped with a set of feature vectors The feature vector Includes camera-related parameters The relative position of the leftmost node ; Edge features The location of the camera is coded. and related camera parameters , which are variables for optimization design, where i and j are nodes i and j, respectively.

3. The intelligent deployment method for elevated monitoring based on graph evolution algorithm according to claim 2, characterized in that: The specific method for constructing the decision-making layer is as follows: Each node and each edge Each has a characteristic parameter Gate i and Gate ij ; The value of Gate is 0 or 1, which represents a scheme that does not use the planning layer and a scheme that uses the planning layer, respectively.

4. The elevated monitoring intelligent deployment method based on graph evolution algorithm according to claim 1, characterized in that: The camera field-of-view analysis and sharpness assessment method based on imaging geometry is as follows: First, based on the installation height of the surveillance camera Pitch angle and horizontal / vertical field of view By geometrically intersecting the camera's view frustum with the ground plane, the closest distance to the visible area on the ground is calculated. Longest distance and the corresponding horizontal width , Based on this, the two-dimensional corner coordinates of the ground field of view trapezoid are constructed. and its three-dimensional extended coordinates ; Based on this, combined with the horizontal resolution of the image Furthermore, the PPM at different distances was derived.

5. The elevated monitoring intelligent deployment method based on graph evolution algorithm according to claim 4, characterized in that: The construction of the optimization model includes at least the following steps: In a given target monitoring area and pixel density lower limit threshold Under the condition that any point in the region is required The imaging pixel density satisfies the following formula: in, For the target monitoring area, The combined ground coverage area for all selected monitoring devices is defined as follows: The monitoring coverage rate is no less than the preset coverage requirement, achieving coverage of the area. Full coverage or coverage rate not less than As shown in the following formula: Under the premise that both imaging clarity and spatial coverage constraints are satisfied, the total cost of the monitoring equipment is used as the optimization objective to minimize the total price of the equipment selection and deployment scheme, as shown in the following formula: in, For the first The unit price of each candidate surveillance device; This is the decision variable for whether or not the device is selected.

Images