Urban comprehensive law enforcement management and control system based on deep learning

Abstract

The invention belongs to the technical field of city law enforcement management and control, and particularly discloses and provides a city comprehensive law enforcement management and control system based on deep learning, which comprises a city region division module, a road occupation behavior judgment module, a road occupation behavior analysis module and a danger matching feedback terminal, according to the method, the road network topological relation graph is constructed, the basic monitoring area is dynamically divided in combination with the historical road occupation behavior data, the initial monitoring area is subjected to secondary division based on the road occupation point distribution characteristics, and the road occupation high-incidence area is accurately identified, so that the monitoring blind area is remarkably reduced, the flexibility and adaptability of area division are improved, and the method is suitable for large-scale popularization and application. Meanwhile, the road occupying behavior is dynamically identified and judged by adopting a real-time image data acquisition and feature extraction technology, a passive response mechanism which depends on historical abnormal frequency accumulation currently is avoided, the danger level and the position coordinates are pushed to the law enforcement terminal in real time, and traffic jam and potential safety hazards caused by the road occupying behavior are effectively restrained.

Description

Technical Field

[0001] The present invention belongs to the technical field of urban law enforcement and control, and relates to an urban comprehensive law enforcement and control system based on deep learning. Background Art

[0002] Urban law enforcement and control is the process of supervising and handling illegal and irregular activities in urban public spaces in accordance with laws and regulations. Its core purpose is to maintain order and the public interest. Businesses occupying the road can hinder traffic, disrupt public order, increase safety hazards, and lead to imbalanced management and order. Therefore, precise control of business activities occupying the road is an inevitable choice to ensure the smooth operation of cities and improve the level of governance.

[0003] For example, the Chinese invention patent with publication number CN118968322A discloses a method and system for fusion analysis of urban governance data, including: obtaining street image data of shops collected in real time from fixed monitoring equipment and mobile video acquisition equipment. Based on historical image data, evaluation parameters for urban management and business operations are established, and the evaluation parameters include the area of ​​shop occupation, road occupation time period, road occupation location and road occupation object type. The collected street image data is identified and abnormal data that deviates from the evaluation parameters is recorded. When the number of abnormalities detected in a shop accumulates to a preset value, a rectification notice is sent to the shop manager. The corresponding monitoring level is generated based on the frequency of rectification notices sent to the shop during the evaluation period. The monitoring strategy is dynamically adjusted according to the monitoring level.

[0004] The above existing technologies have the following deficiencies: 1. Currently, rectification notices are triggered only by the cumulative number of abnormalities, and they rely on historical data rather than real-time image analysis. They are unable to extract and judge the real-time features of road-blocking business behaviors, resulting in delayed law enforcement responses and difficulty in quickly curbing the safety hazards caused by road-blocking behaviors.

[0005] 2. Currently, risk assessment is based solely on static parameters such as road occupation area and time period, without considering multi-dimensional dynamic factors such as traffic density, blind spot interference, and the impact of road type. This makes it impossible to accurately quantify the actual danger level of road occupation, which in turn affects law enforcement priority decisions. Summary of the Invention

[0006] In view of this, in order to solve the problems raised in the above background technology, a city comprehensive law enforcement and control system based on deep learning is proposed.

[0007] The purpose of the present invention can be achieved through the following technical solutions: The present invention provides a city comprehensive law enforcement and control system based on deep learning, including: an urban area division module, which divides the urban area into several basic monitoring areas based on historical road occupation behavior data and road network topology relationships.

[0008] The road-occupying behavior determination module collects image data from each basic monitoring area in real time and determines the road-occupying business behavior in each basic monitoring area through feature extraction.

[0009] The road-occupying behavior analysis module marks the basic monitoring areas where road-occupying business behaviors exist as illegal road-occupying areas. Based on the location coordinates of the illegal road-occupying areas and combined with image data, it analyzes traffic density, blind spot interference, and road type impact, and then determines the danger level of road-occupying business behaviors in each illegal road-occupying area.

[0010] The danger matching feedback terminal matches the corresponding danger level based on the danger level of the road-occupying business behavior in each illegal road-occupying area, and pushes the danger level and location coordinates to the law enforcement terminal.

[0011] Compared with the existing technology, the beneficial effects of the present invention are as follows: (1) The present invention constructs a road network topology diagram, combines historical road occupation behavior data to dynamically divide the basic monitoring area, and divides the initial monitoring area secondary based on the distribution characteristics of road occupation points, thereby solving the current problem of rigid regional division caused by reliance on a fixed monitoring range. By accurately identifying areas with high incidence of road occupation, the present invention significantly reduces monitoring blind spots and improves the flexibility and adaptability of regional division.

[0012] (2) The present invention uses real-time image data acquisition and feature extraction technology to dynamically identify and judge road-occupying behavior, avoiding the current passive response mechanism that relies on the accumulation of historical abnormalities, and pushes the danger level and location coordinates to the law enforcement terminal in real time, greatly shortening the law enforcement response time, and effectively curbing traffic congestion and safety hazards caused by road-occupying behavior.

[0013] (3) The present invention analyzes traffic density, blind spot interference, and road type impact based on location coordinates and image data, and then determines the danger level of road-occupying business behavior, accurately quantifies the actual threat that road-occupying behavior poses to traffic order, and solves the current one-sided evaluation problem caused by relying on static parameters.

[0014] (4) The present invention determines the spatial impact range of road-occupying objects by combining the road unit boundary and dynamically affects the weight, thereby making up for the shortcomings of the current risk assessment of visual blind spots and improving the comprehensiveness and accuracy of safety hazard identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1This is a schematic diagram of the connection of various modules of the system of the present invention.

[0017] Figure 2 This is a connection diagram of the steps of dividing urban areas in the present invention.

[0018] Figure 3 This is a connection diagram of the traffic density analysis steps of the present invention. DETAILED DESCRIPTION

[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0020] See also Figure 1 As shown, the present invention provides an urban comprehensive law enforcement and control system based on deep learning, which includes: an urban area division module, a road occupying behavior determination module, a road occupying behavior analysis module and a danger matching feedback terminal.

[0021] In the above, the road-occupying behavior determination module is connected to the urban area division module and the road-occupying behavior analysis module respectively, and the road-occupying behavior analysis module is also connected to the danger matching feedback terminal.

[0022] The urban area division module divides the urban area into several basic monitoring areas based on historical road occupation behavior data and road network topology.

[0023] See also Figure 2 As shown, exemplarily, the urban area division module includes: Q1, dividing the road into sections with the intersections in the road network topology map as the dividing points, and using the sections as the initial monitoring areas.

[0024] Q2. Based on the historical road-occupying behavior data, identify the number of historical road-occupying behaviors in each initial monitoring area.

[0025] Q3. If the number of historical road-occupying behaviors in the initial monitoring area is less than or equal to 1, the initial monitoring area will be used as the basic monitoring area. Otherwise, it will be determined whether the historical road-occupying behaviors in the initial monitoring area are in the same location.

[0026] It should be added that the determination of whether the historical road-occupying behaviors in the initial monitoring area are at the same location is as follows: the location coordinates of each historical road-occupying behavior in the initial monitoring area are extracted from the historical road-occupying behavior data, and each historical road-occupying behavior is combined in pairs to obtain the distance between each historical road-occupying group. If the distances are all less than the preset distance threshold, the historical road-occupying behaviors in the initial monitoring area are at the same location; otherwise, the historical road-occupying behaviors in the initial monitoring area are not at the same location.

[0027] Q4. If the historical road occupation behavior in the initial monitoring area is at the same location, the initial monitoring area will be used as the basic monitoring area. Otherwise, the initial monitoring area will be divided into two parts to obtain the secondary divided monitoring areas, which will be used as the basic monitoring areas.

[0028] Furthermore, the secondary division of the initial monitoring area includes: Q4-1, within the initial monitoring area, extracting the location coordinates of each historical road occupying behavior from the historical road occupying behavior data, and marking them as road occupying points in the road network topology map.

[0029] Q4-2. Divide the initial detection area into road units according to the preset distance, and then determine whether each road occupation point is located in the same road unit.

[0030] Q4-3. If a road occupation point and its adjacent road occupation point are located in the same road unit, the road unit is divided into two sub-areas.

[0031] Q4-4. If a road-occupying point and its adjacent road-occupying points are not located in the same road unit, but the minimum road connection distance between the road-occupying points is less than a preset connection threshold, the road units are merged and used as secondary sub-areas.

[0032] It should be added that the preset connectivity threshold represents the maximum distance allowed for the distribution of road occupation points.

[0033] Q4-5. If the road-occupying point and its adjacent road-occupying point are not located in the same road unit and the minimum road connection distance between the road-occupying points is greater than or equal to the preset connection threshold, the road units to which the road-occupying point and its adjacent road-occupying point belong are respectively used as secondary division areas.

[0034] The road-occupying behavior determination module collects image data of each basic monitoring area in real time, and determines the road-occupying business behavior in each basic monitoring area through feature extraction.

[0035] It should be added that the image data is collected by a camera carried by a drone.

[0036] Exemplarily, the determination of road-occupying business behavior in each basic monitoring area includes: extracting the residence time of road-occupying objects from the image data of each basic monitoring area, and comparing it with a preset residence time threshold of the corresponding road type.

[0037] It should be added that the road types include but are not limited to: main roads, secondary roads, and branch roads. The preset stay time thresholds corresponding to the road types are shown in Table 1.

[0038] Table 1: Schematic diagram of preset stop time thresholds corresponding to road types

[0039]

[0040] Trunk roads are primarily for transportation, as illegal parking can easily trigger cascading congestion. Therefore, lower thresholds are set, allowing only temporary parking for emergency vehicles. Secondary trunk roads, serving as transitional routes between transportation and daily life, allow short periods of temporary parking, such as for merchants unloading goods or customers parking. However, if the parking exceeds 30 minutes, pedestrians may be forced to detour via motor vehicle lanes, increasing the risk of collisions. Branch roads restrict mobile vendors from staying there for extended periods, which can cause traffic jams. By setting differentiated thresholds, precise adaptation is achieved for different road functions.

[0041] If the stay time exceeds the preset stay time threshold, the coverage area of ​​the road-occupying object and the road area are extracted from the image data of each basic monitoring area, and the ratio of the two is calculated as the road-occupied area ratio.

[0042] The road occupation area ratio is compared with the set reference threshold. If the road occupation area ratio exceeds the set reference threshold, it is determined as road occupation business behavior. Otherwise, it is not determined as road occupation business behavior.

[0043] The embodiment of the present invention constructs a road network topology relationship diagram, dynamically divides the basic monitoring area based on historical road occupation behavior data, and performs secondary division of the initial monitoring area based on the distribution characteristics of the road occupation points. It solves the current problem of rigid regional division caused by reliance on a fixed monitoring range, and significantly reduces monitoring blind spots by accurately identifying areas with high incidence of road occupation, thereby improving the flexibility and adaptability of regional division.

[0044] The road-occupying behavior analysis module marks the basic monitoring area where there is road-occupying business behavior as an illegal road-occupying area, and analyzes the traffic density, blind spot interference and road type influence based on the location coordinates of the illegal road-occupying area combined with image data, and then determines the danger level of the road-occupying business behavior in each illegal road-occupying area.

[0045] See also Figure 3As shown, exemplarily, the analysis of traffic density includes: W1, obtaining the total number of vehicles passing through the illegal road occupation area within a preset time period from the image data, taking the ratio of the total number of vehicles to the length of the preset time period as the traffic flow, and combining the road area of ​​the illegal road occupation area to calculate the traffic flow density.

[0046] It should be added that the preset time period is a fixed time interval set in advance for analyzing the traffic impact of illegal road occupation areas. It is used to unify the time granularity of data collection and ensure the consistency and comparability of the calculation of indicators such as vehicle flow and pedestrian flow.

[0047] It should be added that the calculation formula for traffic flow density is: traffic flow density = traffic flow / road area.

[0048] W2. Based on the entry and exit times of each vehicle in the illegal road-occupying area and the road length within a preset time period in the image data, calculate the speed of each vehicle and take the average as the vehicle speed in the illegal road-occupying area.

[0049] It should be added that the calculation formula for the passing speed is: , where is the passing speed, is the road length, and The exit time and entry time respectively.

[0050] W3. Compare and analyze the traffic flow density and vehicle speed to obtain the initial congestion degree.

[0051] It should be added that the analysis formula for initial congestion is: , where for, and are the traffic density and vehicle speed, respectively. and are the preset traffic density and vehicle speed, and are the weights of traffic density ratio and vehicle speed ratio, , .

[0052] It should be added that the preset traffic flow density is determined by using the 85th percentile method in statistics, which extracts the historical traffic flow density data of the illegal road occupation area and sorts the historical traffic flow density from large to small, and then extracts the 85th percentile traffic flow density as the , meaning that the historical traffic density during 85% of the historical time periods was below this value. The preset vehicle speed is set based on the legal speed limit set by the traffic management department for that road, representing the theoretical upper limit of the safe driving speed for vehicles on that road section.

[0053] It should be added that It reflects the relative relationship between the traffic flow density and the preset traffic flow density. The greater the traffic flow density, the greater the contribution of this part to the congestion degree. It reflects the degree of deviation between the vehicle speed and the preset vehicle speed. The lower the vehicle speed, the larger the value of this part and the greater the contribution to congestion.

[0054] It should be added that traffic density directly reflects the degree of road space occupation and is the source indicator of congestion. When the density exceeds the road capacity, the safe distance between vehicles is compressed and traffic efficiency is inevitably reduced. Speed ​​is easily affected by non-congestion factors such as pedestrian crossing, driver driving habits, and weather changes, and density is dominated by structural factors such as road occupation and lane reduction, and the data is more stable. Therefore, the setting , in order to facilitate analysis, The specific value can be 0.6. The specific value can be 0.4.

[0055] W4. Based on the vehicle information within a preset time period in the image data, the lane occupancy rate of the illegal road occupation area is determined, and the initial congestion degree is dynamically adjusted based on the occupancy rate to obtain a corrected congestion degree.

[0056] It should be added that the analysis process of the corrected congestion degree is: match the lane occupancy rate of the illegal road occupation area with the lane occupancy rate interval corresponding to each lane occupation correction factor to obtain the lane occupation correction factor of the illegal road occupation area, and then calculate the product of the lane occupation correction factor and the initial congestion degree as the corrected congestion degree.

[0057] It's important to note that the lane occupation correction factor quantifies the extent to which lane occupation contributes to traffic congestion. It's a nonlinear amplification factor set based on the lane occupancy rate. The higher the lane occupancy rate, the larger the lane occupation correction factor, and the greater the amplification effect on the initial congestion level.

[0058] Furthermore, the determining of the lane occupancy rate of the illegally occupied area includes: taking the difference between the exit time and the entry time of each passing vehicle in the vehicle information as the total stay time of each passing vehicle.

[0059] The average calculation result of the ratio of the total stay time of each passing vehicle to the length of the preset time period is used as the time dimension occupancy rate of the illegal road occupation area.

[0060] The preset time period is divided into several continuous sub-windows, and the vehicle coverage area and road area of ​​each sub-window are identified. The middle value is selected from the calculation results of the ratio of the vehicle coverage area to the road area of ​​each sub-window as the spatial dimension occupancy rate of the illegal road occupation area.

[0061] The average of the time dimension occupancy rate and the space dimension occupancy rate of the illegal road occupation area is taken as the lane occupancy rate of the illegal road occupation area.

[0062] W5. Obtain pedestrian data within a preset time period from the image data to calculate pedestrian flow and obtain the pedestrian flow impact.

[0063] Furthermore, the calculation of the pedestrian flow impact includes: W5-1, obtaining the total number of pedestrians in a preset time period from pedestrian data, calculating the total pedestrian flow based on the length of the preset time period, and taking the ratio of the total pedestrian flow to its preset pedestrian flow as the basic impact.

[0064] It should be added that the current time period is obtained from the image data, and then the historical pedestrian flow data of the illegal road occupation area corresponding to the current time period is extracted from the historical data, and the maximum value is extracted as the preset pedestrian flow.

[0065] It should be added that the total pedestrian flow = the total number of pedestrians in the preset time period / the length of the preset time period.

[0066] W5-2. Count the number of pedestrians in each sub-window, and then calculate the pedestrian flow in each sub-window. Filter out the sub-window corresponding to the maximum pedestrian flow, and use the ratio of the pedestrian flow in the sub-window to the total pedestrian flow as the peak ratio.

[0067] W5-3. The product of the basic impact and the peak ratio is used as the pedestrian flow impact.

[0068] W6. Multiply the pedestrian flow impact and the corrected congestion degree to generate traffic density.

[0069] It should be added that the calculation formula for traffic density is: , where For traffic density, is the pedestrian flow impact, To correct congestion.

[0070] Exemplarily, the analysis of blind spot interference includes: obtaining the width and vertical height of the road-occupying object from the image data, performing product calculation, and obtaining the blind spot area of ​​the illegal road-occupying area.

[0071] It should be added that the width is obtained by converting the image pixel coordinates in the image data into actual geographic coordinates, and then calculating the projection width of the road-occupying object on the ground and using it as the width.

[0072] It should be added that the vertical height is obtained by detecting the vertical pixel difference between the bottom and top of the road-occupying object in the image based on the flight altitude and pitch angle of the drone, and calculating the actual vertical viewing angle difference by multiplying the vertical pixel difference and the pixel angular resolution. Calculate the vertical height of objects occupying the road , where is the flight altitude of the UAV, is the pitch angle, is the actual vertical viewing angle difference.

[0073] The ratio of the blind spot area of ​​the illegally occupied road area to the preset field of view area is used as the blind spot interference ratio.

[0074] It should be noted that the preset visual field area is the area of ​​unobstructed vision required for pedestrians or vehicles to pass normally on the road, calculated using geometric optics principles based on the road width and the height of buildings on both sides of the road. This area is used to measure the degree to which road users' vision is obstructed by road occupation activities.

[0075] The vertical height of the road-occupying object is matched with the vertical height interval corresponding to each height influence weight to obtain the height influence weight of the road-occupying object.

[0076] It should be added that the vertical height interval corresponding to each height impact weight refers to different vertical height ranges pre-divided according to the potential impact of road-occupying objects on traffic order, pedestrian safety and urban appearance. Each range corresponds to a specific impact weight value, which is used to quantitatively assess the hazard level of road-occupying behavior.

[0077] The blind spot interference degree is calculated by multiplying the blind spot interference ratio of the illegal road occupation area by the height impact weight.

[0078] The embodiment of the present invention determines the spatial impact range of road-occupying objects in combination with road unit boundaries and dynamically affects the weight, thereby making up for the shortcomings of current risk assessment of visual blind spots and improving the comprehensiveness and accuracy of safety hazard identification.

[0079] Exemplarily, the analyzing the influence of road types includes: importing the location coordinates of the illegal road-occupying area into a road network topology map to obtain the road type of the illegal road-occupying area.

[0080] The road type of the illegal road occupation area is matched with the road type corresponding to the influence degree of each road type to obtain the road type influence degree of the illegal road occupation area.

[0081] It should be added that the road type corresponding to the impact degree of each road type refers to the classification of roads and the assignment of different impact degree values ​​based on the functional positioning and traffic flow of the roads, which are used to evaluate the comprehensive impact of illegal road occupation on the traffic system and urban operations.

[0082] Illustratively, determining the risk level of road-occupying business activities in each illegal road-occupying area includes normalizing the traffic density to obtain the normalized traffic density.

[0083] It should be added that the normalized traffic density is: , where is the normalized traffic density, For traffic density, and are the minimum and maximum traffic densities set as references respectively. If the traffic density is greater than the maximum traffic density, the normalized traffic density is taken as 1.

[0084] It should be added that and The setting method is as follows: based on the time point of the preset time period, the historical traffic density of the illegal road occupation area at the same time point is extracted from the historical data, and the maximum and minimum values ​​are selected as the maximum traffic density and minimum traffic density for setting reference respectively.

[0085] The normalized traffic density, blind spot interference, and road type impact are weighted and fused with their preset impact weights to obtain the risk level of road-occupying business behavior in each illegal road-occupying area.

[0086] It should be added that the analysis formula for the risk of occupying the road for business is: , where To determine the risk level of road occupation business activities, For traffic density, is the blind zone interference degree, 、 and are the influence weights of normalized traffic density, blind spot interference, and road type influence, respectively. , .

[0087] It should be added that traffic density directly determines the intensity of risk and directly threatens traffic operations. Traffic density is a direct quantitative indicator of the impact of road occupation on traffic order. Road occupation in high-density areas will directly lead to lane narrowing, a sharp drop in traffic efficiency, and even cause accidents such as rear-end collisions and scratches. The risk is immediate and explicit, so its weight is the largest. Blind spots have the problem of line of sight obstruction. Road occupation will further compress the driver's reaction time, leading to sudden accidents such as "ghost heads" and "blind spot collisions". The consequences of such accidents are often more serious and difficult to predict. Blind spot interference does not directly cause accidents, but it will form a superimposed risk with traffic density, so its weight is second. Road type essentially reflects the design function and basic risk level of the road, but these characteristics have been indirectly reflected through traffic density and blind spot distribution patterns, so its weight is the lowest. Therefore, it is set , in order to facilitate analysis, The specific value can be 0.5. The specific value can be 0.3, The specific value can be 0.2.

[0088] The embodiments of the present invention analyze traffic density, blind spot interference, and road type impact based on location coordinates combined with image data, and then determine the danger level of road-occupying business behavior, accurately quantify the actual threat of road-occupying behavior to traffic order, and solve the current problem of one-sided evaluation caused by reliance on static parameters.

[0089] The risk matching feedback terminal matches the corresponding risk level based on the risk degree of the road-occupying business behavior in each illegal road-occupying area, and pushes the risk level and location coordinates to the law enforcement terminal.

[0090] It should be added that the matching method of the danger level is: match the danger level of the road-occupying business behavior with the danger level interval of the road-occupying business behavior corresponding to each danger level to obtain the danger level of the illegal road-occupying area.

[0091] The embodiment of the present invention adopts real-time image data acquisition and feature extraction technology to dynamically identify and judge the road-occupying behavior, avoiding the current passive response mechanism that relies on the accumulation of historical abnormal times, and pushing the danger level and location coordinates to the law enforcement terminal in real time, greatly shortening the law enforcement response time, and effectively curbing the traffic congestion and safety hazards caused by the road-occupying behavior.

[0092] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0093] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0094] Those skilled in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0095] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0096] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0097] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A deep learning-based urban comprehensive law enforcement control system, characterized by: The system includes: The urban area division module divides the urban area into several basic monitoring areas based on historical road occupation behavior data and road network topology; The road occupation behavior determination module collects image data from each basic monitoring area in real time and determines the road occupation business behavior in each basic monitoring area through feature extraction; The road occupation behavior analysis module marks basic monitoring areas where there is road occupation as illegal road occupation areas. Based on the location coordinates of the illegal road occupation areas and combined with image data, it analyzes traffic density, blind spot interference, and road type impact, and then determines the risk level of road occupation in each illegal road occupation area. The danger matching feedback terminal matches the corresponding danger level based on the danger level of the road-occupying business behavior in each illegal road-occupying area, and pushes the danger level and location coordinates to the law enforcement terminal.

2. The urban comprehensive law enforcement control system based on deep learning according to claim 1 is characterized by: The urban area division module includes: Q1. Divide the road into sections using the intersections in the road network topology as segmentation points, and use the sections as initial monitoring areas; Q2. Based on historical road-occupying behavior data, identify the number of historical road-occupying behaviors in each initial monitoring area; Q3. If the number of historical road-occupying behaviors in the initial monitoring area is less than or equal to 1, the initial monitoring area is used as the basic monitoring area. Otherwise, it is determined whether the historical road-occupying behaviors in the initial monitoring area are at the same location. Q4. If the historical road occupation behavior in the initial monitoring area is at the same location, the initial monitoring area will be used as the basic monitoring area. Otherwise, the initial monitoring area will be divided into two parts to obtain the secondary divided monitoring areas, which will be used as the basic monitoring areas.

3. The urban comprehensive law enforcement control system based on deep learning according to claim 2 is characterized by: The secondary division of the initial monitoring area includes: In the initial monitoring area, the location coordinates of each historical road-occupying behavior are extracted from the historical road-occupying behavior data and marked as road-occupying points in the road network topology map; The initial detection area is divided into road units according to the preset distance, and then it is determined whether each road occupation point is located in the same road unit; If the road occupation point and its adjacent road occupation point are located in the same road unit, the road unit is used as a secondary sub-area; If a road occupation point and its adjacent road occupation points are not located in the same road unit, but the minimum road connection distance between the road occupation points is less than a preset connection threshold, the road units are merged and used as secondary sub-regions; If the road-occupying point and its adjacent road-occupying point are not located in the same road unit and the minimum road connection distance between the road-occupying points is greater than or equal to the preset connection threshold, the road units to which the road-occupying point and its adjacent road-occupying point belong are respectively used as secondary division areas.

4. The urban comprehensive law enforcement control system based on deep learning according to claim 1 is characterized by: The determination of road occupation business behavior in each basic monitoring area includes: Extract the dwell time of road-occupying objects from the image data of each basic monitoring area and compare it with the preset dwell time threshold of the corresponding road type; If the dwell time exceeds the preset dwell time threshold, the coverage area of ​​the road-occupying object and the road area are extracted from the image data of each basic monitoring area, and the ratio of the two is calculated as the road-occupying area ratio; The road occupation area ratio is compared with the set reference threshold. If the road occupation area ratio exceeds the set reference threshold, it is determined as road occupation business behavior. Otherwise, it is not determined as road occupation business behavior.

5. The urban comprehensive law enforcement control system based on deep learning according to claim 1 is characterized by: The analysis of traffic density includes: W1. Obtain the total number of vehicles passing through the illegally occupied area within a preset time period from the image data. The ratio of the total number of vehicles to the duration of the preset time period is used as the traffic volume. Combined with the road area of ​​the illegally occupied area, the traffic volume density is calculated. W2. Calculate the speed of each vehicle passing through the illegal road occupation area based on the entry and exit times of each vehicle passing through the illegal road occupation area and the road length during a preset time period in the image data, and take the average value as the vehicle speed of the illegal road occupation area; W3. Compare and analyze the traffic flow density and vehicle speed to obtain the initial congestion degree; W4. Determine the lane occupancy rate of the illegally occupied area based on the vehicle information within a preset time period in the image data, and dynamically adjust the initial congestion level based on the occupancy rate to obtain a revised congestion level; W5. Obtain pedestrian data within a preset time period from the image data to calculate pedestrian flow and obtain the pedestrian flow impact; W6. Multiply the pedestrian flow impact and the corrected congestion degree to obtain the traffic density.

6. The urban comprehensive law enforcement control system based on deep learning according to claim 5 is characterized by: Determining the lane occupancy rate of the illegally occupied area includes: The difference between the exit time and the entry time of each passing vehicle in the vehicle information is used as the total stay time of each passing vehicle; The average of the ratio of the total dwell time of each passing vehicle to the duration of the preset time period is calculated as the time dimension occupancy rate of the illegal road occupation area; Divide the preset time period into several consecutive sub-windows, identify the vehicle coverage area and road area of ​​each sub-window, and select the middle value from the calculated ratio of the vehicle coverage area to the road area of ​​each sub-window as the spatial dimension occupancy rate of the illegal road occupation area; The average of the time dimension occupancy rate and the space dimension occupancy rate of the illegal road occupation area is taken as the lane occupancy rate of the illegal road occupation area.

7. The urban comprehensive law enforcement control system based on deep learning according to claim 6 is characterized by: The calculation of the pedestrian flow impact includes: Obtain the total number of pedestrians in a preset time period from pedestrian data, calculate the total pedestrian flow based on the length of the preset time period, and use the ratio of the total pedestrian flow to the preset pedestrian flow as the basic impact; Count the number of pedestrians in each sub-window, and then calculate the pedestrian flow in each sub-window. Filter out the sub-window corresponding to the maximum pedestrian flow, and use the ratio of the pedestrian flow in this sub-window to the total pedestrian flow as the peak ratio. The product of the basic impact and the peak ratio is calculated as the pedestrian flow impact.

8. The urban comprehensive law enforcement control system based on deep learning according to claim 1 is characterized by: The analyzing the blind area interference degree includes: Obtain the width and vertical height of the road-blocking object from the image data and multiply them to obtain the blind spot area of ​​the illegal road-blocking area; The ratio of the blind spot area of ​​the illegal road occupation area to the preset field of view area is used as the blind spot interference ratio; Match the vertical height of the road-occupying object with the vertical height interval corresponding to each height influence weight to obtain the height influence weight of the road-occupying object; The product of the blind spot interference ratio of the illegally occupied area and the height impact weight is calculated as the blind spot interference degree.

9. The urban comprehensive law enforcement control system based on deep learning according to claim 1 is characterized by: The analysis of the impact of road types includes: Import the location coordinates of the illegal road-occupying area into the road network topology map to obtain the road type of the illegal road-occupying area; The road type of the illegal road occupation area is matched with the road type corresponding to the influence degree of each road type to obtain the road type influence degree of the illegal road occupation area.

10. The urban comprehensive law enforcement control system based on deep learning according to claim 9 is characterized by: Determining the risk level of road occupation business activities in each illegal road occupation area includes: Normalizing the traffic density to obtain the normalized traffic density; The normalized traffic density, blind spot interference, and road type impact are weighted and fused with their preset impact weights to obtain the risk level of road-occupying business behavior in each illegal road-occupying area.

Images