Smart park security data early warning method and system based on big data

Abstract

The invention discloses a smart park security data early warning method and system based on big data, and belongs to the technical field of security early warning. The system comprises a data acquisition module, a security analysis module, an image management module and a risk early warning module. The data acquisition module is used for acquiring a traffic route map of a park and historical logs of vehicles, and acquiring video images in the park in real time through monitoring equipment; the security analysis module defines a target object through the vehicle and the pedestrian in the video image, calculates a behavior coefficient according to the historical record of the vehicle object, predicts the moving tracks of the vehicle object and the pedestrian object, divides an attention area and calculates a risk coefficient; the image management module calculates a transaction index according to the pixel change condition, defines a risk area according to the transaction index, adjusts parameters of the monitoring equipment, and improves the picture ratio of the risk area; and the risk early warning module is used for displaying the video image of each monitoring device, highlighting the real-time video image of the risk area and performing automatic early warning.

Description

Technical Field

[0001] The present invention relates to the field of security early warning technology, and in particular to a smart park security data early warning method and system based on big data. Background Art

[0002] As the park expands in size and functions, people and vehicles are mixed in the park. There is a lack of traffic control facilities such as traffic lights, resulting in a lack of clear traffic rules and order between people and vehicles, which makes it very easy to cause collisions, scrapes and other accidents. Traditional security measures such as security patrols are difficult to effectively deal with the complex traffic conditions of people and vehicles, and it is difficult to ensure the safety of people and vehicles in the park.

[0003] At present, for the early warning of park security, dynamic video detection combined with manual judgment is usually used for abnormal identification. This method has certain disadvantages. On the one hand, the dynamic detection in the existing technology is just a simple pixel threshold comparison and judgment, which cannot accurately distinguish the difference between normal driving of people and vehicles and traffic accidents. Relying solely on dynamic detection is prone to a large number of false alarms, which can easily affect the work enthusiasm of managers. On the other hand, there is an inverse relationship between the shooting range of the camera and the local clarity. In order to reduce the blind spot of monitoring, the maximum shooting range is usually selected by default. This also leads to the inability to clearly observe the specific details of the local picture even if a traffic accident occurs. It is impossible to make a risk warning in advance, and it is impossible to adjust the shooting angle and optical focal length in time to enlarge the accident picture, which makes it difficult to provide evidence in the future and the accident cannot be clearly determined. Therefore, a more intelligent and efficient security data early warning technology solution is needed at this stage to solve the above problems. Summary of the invention

[0004] The purpose of the present invention is to provide a smart park security data early warning method and system based on big data to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides a smart park security data early warning method based on big data, comprising the following steps:

[0006] S100: Collect the traffic route map of the park and the historical logs of vehicles, and collect video images in real time through the monitoring equipment installed in the park.

[0007] S200, analyzing vehicles and pedestrians in the video image and defining target objects, calculating behavior coefficients according to the historical records of vehicle objects, predicting movement trajectories of vehicle objects and pedestrian objects, dividing focus areas and calculating risk coefficients.

[0008] S300, calculating the anomaly index through the risk coefficient of the concerned area and the pixel change, defining the risk area according to the anomaly index, and adjusting the shooting angle and focal length change of the monitoring equipment to increase the screen proportion of the risk area.

[0009] S400, the security monitoring room displays the video images of each monitoring device in real time through a visual interface, highlights the real-time video images of risk areas, and automatically warns management personnel for processing.

[0010] In S100, the traffic route map refers to a plan view for displaying the roads within the park. The historical log refers to the entry and exit records of vehicles. The entry and exit records refer to the information records from the vehicle entering the park to leaving the park. Each entry and exit record includes a license plate number and a driving video set. The driving video set consists of video clips of the same vehicle at different locations within the park collected by different monitoring devices; the monitoring devices are installed on both sides of the driving roads in the park to collect video images of the driving roads in real time.

[0011] Multiple vehicles may appear in the video screen of the monitoring device at the same time, and video clips are captured according to the time period when each vehicle exists in the video screen.

[0012] Each vehicle will be photographed by different monitoring devices when driving at different locations in the park, and the vehicle object in the video image will be identified and its identity information will be confirmed by the license plate number.

[0013] The video clips of the same vehicle captured by different monitoring devices are associated according to the license plate number, and the multiple video clips generated during the period from when each vehicle enters the park to when it leaves the park are combined into a driving video set.

[0014] In S200, the specific steps are as follows:

[0015] S201. Use target recognition algorithms to analyze the video images collected by each camera, treat pedestrian targets and vehicle targets as pedestrian objects and vehicle objects respectively, and use YOLO algorithm and SVM algorithm to detect the license plate of each vehicle object and recognize the license plate number. Mark the location of each camera on the traffic route map, obtain the video image of the camera closest to the entrance and exit of the park, and analyze and mark the vehicle objects entering the park.

[0016] S202, obtain the marked vehicle object DX sign License plate number LPN sign , retrieve the license plate number LPN in the historical log sign Corresponding to all the entry and exit records, split all the driving video sets under these entry and exit records into video clips. Analyze the marked vehicle objects DX in the video clips sign According to the driving conditions, the driving distance is divided by the driving time to obtain the driving speed; a speed line graph is drawn according to the time period of each video clip and the driving speed at different times, and each video clip corresponds to a speed line graph.

[0017] The time is used as the horizontal axis of the speed line graph, and the driving speed is used as the vertical axis of the speed line graph. The speed line graph can intuitively represent the change in the driving speed of the marked vehicle object in the corresponding video clip.

[0018] S203, set the speed change duration C, mark the time nodes when the driving speed changes in the speed line graph, and combine multiple time nodes where the driving speed increases or decreases continuously and the interval duration is less than C into a speed change zone. Divide the speed range of all time nodes in the speed change zone by the duration of the speed change zone to obtain the speed change, and calculate the average speed change of all speed change zones in the speed line graph And the total duration ct n , substitute it into the formula to calculate the marked vehicle object DX sign The behavior coefficient WX sign :

[0019]

[0020] Where m is the marked vehicle object DX sign Number of video clips, JL n Mark the vehicle object Dx in the nth video clip sign The total driving distance, zt n is the duration of the nth video clip. The behavior coefficient of each marked vehicle object is calculated separately.

[0021] A time node refers to the time point in the speed line graph where the driving speed increases or decreases. The direction of the driving speed change at all time nodes in the speed change zone is the same, either increasing or decreasing. The driving speed range refers to the change in the vehicle's driving speed within the corresponding time period in the speed change zone, and the speed change is the average acceleration of the speed change zone.

[0022] S204, analyze the driving direction and driving speed of the marked vehicle object in the current video image of each monitoring device, set the deceleration time Q, and multiply the driving speed by Q to obtain the moving distance mD. Take the current position of the marked vehicle object as the starting point, and set a route with a distance of MD along the driving road on the traffic route map according to the driving direction as the predicted moving trajectory. Analyze the moving direction of the pedestrian objects on both sides of the predicted moving trajectory in the video image, and establish a vector according to the position and moving direction of the pedestrian objects. Marker Vector Pedestrian objects that intersect with the predicted moving trajectory.

[0023] The deceleration time is set in advance by the management personnel, and the specific value is based on the maximum time required for the vehicle to stop completely from the start of braking. The length of the predicted moving trajectory is calculated based on the current driving speed and deceleration time of the marked vehicle object, and the specific shape is determined by the corresponding road shape on the traffic route map.

[0024] S205. Set the distance U, take the marked pedestrian object as the center, and establish a circular area with U as the diameter. Combine the intersecting circular areas as the focus area. Count the number of predicted moving trajectories j that intersect with the focus area, obtain the behavior coefficients of the marked vehicle objects corresponding to these predicted moving trajectories, and substitute them into the formula to calculate the risk coefficient of each focus area:

[0025]

[0026] Where FX is the risk factor, h is the number of marked pedestrian objects in the area of ​​interest, and WX i is the behavior coefficient of the marked vehicle object corresponding to the i-th predicted moving trajectory intersecting the area of ​​interest.

[0027] In S300, the specific steps are as follows:

[0028] S301, marking the location of the focus area in the video image of each monitoring device, using OpenCV technology to continuously extract time-continuous single-frame images in the video image, and obtaining the pixel point number p of the single-frame image sum And the number of pixels p in each area of ​​interest fx .

[0029] When using OpenCV technology to continuously extract single-frame images from video images, it is necessary to ensure that the images are extracted at an appropriate frequency. The frequency should not be too high to waste computing resources and cause invalid detection, nor should the frequency be too low to cause the loss of important video information and reduce detection accuracy.

[0030] S302: Analyze and mark the pixels in the single-frame image whose colors have changed compared with the previous single-frame image, and count the number of all marked pixels in the single-frame image. sch And the number of marked pixels p in each area of ​​interest fch , substitute into the formula to calculate the abnormal index of each focus area:

[0031]

[0032] Where YZ is the abnormal index and α is a constant.

[0033] S303, set the index threshold e and the sampling number k, and set the focus area with the abnormal index greater than e in k consecutive single-frame images as the risk area. Mark the camera with the risk area in the video screen, and the marked camera automatically adjusts the shooting angle and focal length so that the images of all risk areas maximize the entire video screen.

[0034] In S400, the security monitoring room displays the video images taken by each monitoring device in real time through a visual interface, with each area of ​​concern marked. When an area of ​​concern becomes a risk area, the real-time video image of the risk area is highlighted at the first time, and the image is automatically alerted to the manager's smart terminal device.

[0035] A smart park security data early warning system based on big data includes a data acquisition module, a security analysis module, an image management module and a risk early warning module.

[0036] The data acquisition module is used to collect the traffic route map of the park and the historical log of vehicles, and to collect video images in the park in real time through monitoring equipment.

[0037] The security analysis module defines the target objects through the vehicles and pedestrians in the video images, calculates the behavior coefficient according to the historical records of the vehicle objects, predicts the movement trajectories of the vehicle objects and pedestrian objects, divides the focus area and calculates the risk coefficient.

[0038] The image management module calculates the anomaly index based on pixel changes, defines the risk area based on the anomaly index, and adjusts the parameters of the monitoring equipment to increase the screen ratio of the risk area.

[0039] The risk warning module is used to display the video images of each monitoring device, highlight the real-time video images of the risk area and automatically issue warnings.

[0040] The data acquisition module includes a video image acquisition unit, a history log acquisition unit and a traffic data acquisition unit.

[0041] The video image acquisition unit is used to acquire video images, and acquires video images of the driving road in real time through monitoring equipment installed in the park.

[0042] The historical log collection unit is used to collect vehicle entry and exit records. Entry and exit records refer to the information records from when a vehicle enters the park to when it leaves the park. Each entry and exit record includes a license plate number and a driving video set. The driving video set consists of video clips of the same vehicle at different locations within the park collected by different monitoring devices.

[0043] The traffic data collection unit is used to collect the traffic route map of the park. The traffic route map refers to the plan view of the internal driving roads of the park.

[0044] The security analysis module includes a vehicle analysis unit and a danger prediction unit.

[0045] The vehicle analysis unit is used to calculate the behavior coefficients of the vehicle object.

[0046] First, identify pedestrians and vehicles in the video image and detect the license plate number of the vehicle object. Analyze the position of each camera in the traffic route map and mark the vehicle objects entering the park in the video image of the camera closest to the park entrance and exit. Obtain the marked vehicle object DX sign License plate number LPN sign , retrieve the license plate number LPN in the historical log sign Corresponding to all entry and exit records, the driving video sets under these entry and exit records are all split into video clips.

[0047] Secondly, analyze the marked vehicle objects DX in the video clips sign Draw a speed line graph based on the time period of each video clip and the driving speed at different times. Set the speed change duration C, mark the time nodes when the driving speed changes in the speed line graph, and combine multiple time nodes where the driving speed increases or decreases continuously and the interval duration is less than C into a speed change zone. Divide the speed range of all time nodes in the speed change zone by the duration of the speed change zone to get the speed change, and calculate the average speed change of all speed change zones in the speed line graph. And the total duration ct n .

[0048] Finally, according to the formula Calculate the marked vehicle object DX sign The behavior coefficient WX sign , where m is the marked vehicle object DX sign Number of video clips, JL n Mark the vehicle object DX in the nth video clip sign The total driving distance, zt n is the duration of the nth video clip. The behavior coefficient of each marked vehicle object is calculated separately.

[0049] The risk prediction unit is used to divide the area of ​​concern and calculate the risk factor.

[0050] First, the driving direction and speed of the marked vehicle object in the current video image of each monitoring device are analyzed, the deceleration time Q is set, and the moving distance MD is obtained by multiplying the driving speed by Q. Taking the current position of the marked vehicle object as the starting point, a route with a distance of MD is set along the driving road on the traffic route map according to the driving direction as the predicted moving trajectory.

[0051] Secondly, the moving directions of pedestrian objects on both sides of the predicted moving trajectory in the video image are analyzed, and vectors are established based on the positions and moving directions of the pedestrian objects. Marker Vector Pedestrian objects that intersect with the predicted moving trajectory. Set the distance U, take the marked pedestrian object as the center, and establish a circular area with U as the diameter. Combine the intersecting circular areas as the focus area.

[0052] Finally, the number of predicted moving trajectories j that intersect with the focus area is counted, and the behavior coefficients of the marked vehicle objects corresponding to these predicted moving trajectories are obtained. According to the formula Calculate the risk factor FX for each area of ​​interest. Where h is the number of marked pedestrian objects in the area of ​​interest, WX i is the behavior coefficient of the marked vehicle object corresponding to the i-th predicted moving trajectory intersecting the area of ​​interest.

[0053] The image management module includes an image analysis unit and a parameter adjustment unit.

[0054] The image analysis unit is used to calculate the abnormal movement index of each area of ​​interest.

[0055] Firstly, the location of the area of ​​interest is marked in the video image of each monitoring device, and the 0penCV technology is used to continuously extract time-continuous single-frame images in the video image.

[0056] Secondly, get the number of pixels p of a single frame image sum And the number of pixels p in each area of ​​interest fx ; Then analyze and mark the pixels in the single frame image whose color has changed compared with the previous single frame image.

[0057] Finally, count the number of all marked pixels p in a single frame image sch And the number of marked pixels p in each area of ​​interest fch , according to the formula Calculate the abnormal index of each focus area, where α is a constant.

[0058] The parameter adjustment unit is used to adjust the camera parameters. The index threshold e and the sampling times k are set, and the focus area whose abnormality index is greater than e in k consecutive single-frame images is set as the risk area.

[0059] Mark the camera with risk areas in the video screen. The marked camera automatically adjusts the shooting angle and focal length so that the images of all risk areas maximize the entire video screen.

[0060] The risk warning module displays the video images taken by each monitoring device in real time through a visual interface, with each focus area marked. When the focus area becomes a risk area, the image will be warned to the manager's smart terminal device in the first time.

[0061] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0062] Efficient anomaly analysis: In this application, the behavior coefficient of each marked vehicle object is calculated by analyzing historical records. The movement trajectory of vehicle objects and pedestrian objects is predicted to divide the area of ​​interest, and the risk coefficient of the area of ​​interest is calculated according to the number of pedestrian objects and the behavior coefficient of vehicle objects in the area of ​​interest. Compared with the existing technology of indiscriminate detection of all monitoring images, anomaly analysis is more targeted and efficient.

[0063] Intelligent danger prediction: In this application, the pixel changes in the video images of each monitoring device are analyzed and the risk coefficient of the focus area is combined to calculate the abnormal movement index. The risk area is defined according to the abnormal movement index and the shooting angle and focal length of the monitoring device are adjusted in advance to increase the screen ratio of the risk area. Compared with the existing technology of manual adjustment of the monitoring screen or post-retrieval, it is more efficient and the danger prediction is more predictable and accurate.

[0064] In summary, compared with traditional technologies, the present invention has the advantages of efficient anomaly analysis and intelligent danger prediction, and can improve the security monitoring and early warning capabilities and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0066] Figure 1 It is a flow chart of a smart park security data early warning method based on big data of the present invention;

[0067] Figure 2 It is a structural schematic diagram of a smart park security data early warning system based on big data of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0069] See also Figure 1 The present invention provides a smart park security data early warning method based on big data, comprising the following steps:

[0070] S100: Collect the traffic route map of the park and the historical logs of vehicles, and collect video images in real time through the monitoring equipment installed in the park.

[0071] S200, analyzing vehicles and pedestrians in the video image and defining target objects, calculating behavior coefficients according to the historical records of vehicle objects, predicting movement trajectories of vehicle objects and pedestrian objects, dividing focus areas and calculating risk coefficients.

[0072] S300, calculating the anomaly index through the risk coefficient of the concerned area and the pixel change, defining the risk area according to the anomaly index, and adjusting the shooting angle and focal length change of the monitoring equipment to increase the screen proportion of the risk area.

[0073] S400, the security monitoring room displays the video images of each monitoring device in real time through a visual interface, highlights the real-time video images of risk areas, and automatically warns management personnel for processing.

[0074] In S100, the traffic route map refers to a plan view for displaying the roads within the park. The historical log refers to the entry and exit records of vehicles. The entry and exit records refer to the information records from the vehicle entering the park to leaving the park. Each entry and exit record includes a license plate number and a driving video set. The driving video set consists of video clips of the same vehicle at different locations within the park collected by different monitoring devices; the monitoring devices are installed on both sides of the driving roads in the park to collect video images of the driving roads in real time.

[0075] Multiple vehicles may appear in the video screen of the monitoring device at the same time, and video clips are captured according to the time period when each vehicle exists in the video screen.

[0076] Each vehicle will be photographed by different monitoring devices when driving at different locations in the park, and the vehicle object in the video image will be identified and its identity information will be confirmed by the license plate number.

[0077] The video clips of the same vehicle captured by different monitoring devices are associated according to the license plate number, and the multiple video clips generated during the period from when each vehicle enters the park to when it leaves the park are combined into a driving video set.

[0078] In S200, the specific steps are as follows:

[0079] S201. Use target recognition algorithms to analyze the video images collected by each camera, treat pedestrian targets and vehicle targets as pedestrian objects and vehicle objects respectively, and use YOLO algorithm and SVM algorithm to detect the license plate of each vehicle object and recognize the license plate number. Mark the location of each camera on the traffic route map, obtain the video image of the camera closest to the entrance and exit of the park, and analyze and mark the vehicle objects entering the park.

[0080] S202, obtain the marked vehicle object DX sign License plate number LPN sign , retrieve the license plate number LPN in the historical log signCorresponding to all the entry and exit records, split all the driving video sets under these entry and exit records into video clips. Analyze the marked vehicle objects DX in the video clips sign According to the driving conditions, the driving distance is divided by the driving time to obtain the driving speed; a speed line graph is drawn according to the time period of each video clip and the driving speed at different times, and each video clip corresponds to a speed line graph.

[0081] The time is used as the horizontal axis of the speed line graph, and the driving speed is used as the vertical axis of the speed line graph. The speed line graph can intuitively represent the change in the driving speed of the marked vehicle object in the corresponding video clip.

[0082] S203, set the speed change duration C, mark the time nodes when the driving speed changes in the speed line graph, and combine multiple time nodes where the driving speed increases or decreases continuously and the interval duration is less than C into a speed change zone. Divide the speed range of all time nodes in the speed change zone by the duration of the speed change zone to obtain the speed change, and calculate the average speed change of all speed change zones in the speed line graph And the total duration ct n , substitute it into the formula to calculate the marked vehicle object DX sign The behavior coefficient WX sign :

[0083]

[0084] Where m is the marked vehicle object DX sign Number of video clips, JL n Mark the vehicle object DX in the nth video clip sign The total driving distance, zt n is the duration of the nth video clip. The behavior coefficient of each marked vehicle object is calculated separately.

[0085] A time node refers to the time point in the speed line graph where the driving speed increases or decreases. The direction of the driving speed change at all time nodes in the speed change zone is the same, either increasing or decreasing. The driving speed range refers to the change in the vehicle's driving speed within the corresponding time period in the speed change zone, and the speed change is the average acceleration of the speed change zone.

[0086] S204, analyze the driving direction and driving speed of the marked vehicle object in the current video image of each monitoring device, set the deceleration time Q, and multiply the driving speed by Q to obtain the moving distance MD. Take the current position of the marked vehicle object as the starting point, and set a route with a distance of MD along the driving road on the traffic route map according to the driving direction as the predicted moving trajectory. Analyze the moving direction of the pedestrian objects on both sides of the predicted moving trajectory in the video image, and establish a vector according to the position and moving direction of the pedestrian objects. Marker Vector Pedestrian objects that intersect with the predicted moving trajectory.

[0087] The deceleration time is set in advance by the management personnel, and the specific value is based on the maximum time required for the vehicle to stop completely from the start of braking. The length of the predicted moving trajectory is calculated based on the current driving speed and deceleration time of the marked vehicle object, and the specific shape is determined by the corresponding road shape on the traffic route map.

[0088] S205. Set the distance U, take the marked pedestrian object as the center, and establish a circular area with U as the diameter. Combine the intersecting circular areas as the focus area. Count the number of predicted moving trajectories j that intersect with the focus area, obtain the behavior coefficients of the marked vehicle objects corresponding to these predicted moving trajectories, and substitute them into the formula to calculate the risk coefficient of each focus area:

[0089]

[0090] Where FX is the risk factor, h is the number of marked pedestrian objects in the area of ​​interest, and WX i is the behavior coefficient of the marked vehicle object corresponding to the i-th predicted moving trajectory intersecting the area of ​​interest.

[0091] In S300, the specific steps are as follows:

[0092] S301, marking the location of the focus area in the video image of each monitoring device, using OpenCV technology to continuously extract time-continuous single-frame images in the video image, and obtaining the pixel point number p of the single-frame image sum And the number of pixels p in each area of ​​interest fx .

[0093] When using OpenCV technology to continuously extract single-frame images from video images, it is necessary to ensure that the images are extracted at an appropriate frequency. The frequency should not be too high, wasting computing resources and causing invalid detection, nor should the frequency be too low, resulting in the loss of important video information and a decrease in detection accuracy.

[0094] S302: Analyze and mark the pixels in the single-frame image whose colors have changed compared with the previous single-frame image, and count the number of all marked pixels in the single-frame image. sch And the number of marked pixels p in each area of ​​interest fch , substitute into the formula to calculate the abnormal index of each focus area:

[0095]

[0096] Where YZ is the abnormal index and α is a constant.

[0097] S303, set the index threshold e and the sampling number k, and set the focus area with the abnormal index greater than e in k consecutive single-frame images as the risk area. Mark the camera with the risk area in the video screen, and the marked camera automatically adjusts the shooting angle and focal length so that the images of all risk areas maximize the entire video screen.

[0098] In S400, the security monitoring room displays the video images taken by each monitoring device in real time through a visual interface, with each area of ​​concern marked. When an area of ​​concern becomes a risk area, the real-time video image of the risk area is highlighted at the first time, and the image is automatically alerted to the manager's smart terminal device.

[0099] See also Figure 2 The present invention provides a smart park security data early warning system based on big data, including a data acquisition module, a security analysis module, an image management module and a risk early warning module.

[0100] The data acquisition module is used to collect the traffic route map of the park and the historical log of vehicles, and to collect video images in the park in real time through monitoring equipment.

[0101] The security analysis module defines the target objects through the vehicles and pedestrians in the video images, calculates the behavior coefficient according to the historical records of the vehicle objects, predicts the movement trajectories of the vehicle objects and pedestrian objects, divides the focus area and calculates the risk coefficient.

[0102] The image management module calculates the anomaly index based on pixel changes, defines the risk area based on the anomaly index, and adjusts the parameters of the monitoring equipment to increase the screen ratio of the risk area.

[0103] The risk warning module is used to display the video images of each monitoring device, highlight the real-time video images of the risk area and automatically issue warnings.

[0104] The data acquisition module includes a video image acquisition unit, a history log acquisition unit and a traffic data acquisition unit.

[0105] The video image acquisition unit is used to acquire video images, and acquires video images of the driving road in real time through monitoring equipment installed in the park.

[0106] The historical log collection unit is used to collect vehicle entry and exit records. Entry and exit records refer to the information records from when a vehicle enters the park to when it leaves the park. Each entry and exit record includes a license plate number and a driving video set. The driving video set consists of video clips of the same vehicle at different locations within the park collected by different monitoring devices.

[0107] The traffic data collection unit is used to collect the traffic route map of the park. The traffic route map refers to the plan view of the internal driving roads of the park.

[0108] The security analysis module includes a vehicle analysis unit and a danger prediction unit.

[0109] The vehicle analysis unit is used to calculate the behavior coefficients of the vehicle object.

[0110] First, identify pedestrians and vehicles in the video image and detect the license plate number of the vehicle object. Analyze the position of each camera in the traffic route map and mark the vehicle objects entering the park in the video image of the camera closest to the park entrance and exit. Obtain the marked vehicle object DX sign License plate number LPN sign , retrieve the license plate number LPN in the historical log sign Corresponding to all entry and exit records, the driving video sets under these entry and exit records are all split into video clips.

[0111] Secondly, analyze the marked vehicle objects DX in the video clips sign Draw a speed line graph based on the time period of each video clip and the driving speed at different times. Set the speed change duration C, mark the time nodes when the driving speed changes in the speed line graph, and combine multiple time nodes where the driving speed increases or decreases continuously and the interval duration is less than C into a speed change zone. Divide the speed range of all time nodes in the speed change zone by the duration of the speed change zone to get the speed change, and calculate the average speed change of all speed change zones in the speed line graph. And the total duration ct n .

[0112] Finally, according to the formula Calculate the marked vehicle object DX sign The behavior coefficient WX sign , where m is the marked vehicle object DX sign Number of video clips, JL n Mark the vehicle object DX in the nth video clip sign The total driving distance, zt n is the duration of the nth video clip. The behavior coefficient of each marked vehicle object is calculated separately.

[0113] The risk prediction unit is used to divide the area of ​​concern and calculate the risk factor.

[0114] First, the driving direction and speed of the marked vehicle object in the current video image of each monitoring device are analyzed, the deceleration time Q is set, and the moving distance MD is obtained by multiplying the driving speed by Q. Taking the current position of the marked vehicle object as the starting point, a route with a distance of MD is set along the driving road on the traffic route map according to the driving direction as the predicted moving trajectory.

[0115] Secondly, the moving directions of pedestrian objects on both sides of the predicted moving trajectory in the video image are analyzed, and vectors are established based on the positions and moving directions of the pedestrian objects. Marker Vector Pedestrian objects that intersect with the predicted moving trajectory. Set the distance U, take the marked pedestrian object as the center, and establish a circular area with U as the diameter. Combine the intersecting circular areas as the focus area.

[0116] Finally, the number of predicted moving trajectories j that intersect with the focus area is counted, and the behavior coefficients of the marked vehicle objects corresponding to these predicted moving trajectories are obtained. According to the formula Calculate the risk factor FX for each area of ​​interest. Where h is the number of marked pedestrian objects in the area of ​​interest, WX i is the behavior coefficient of the marked vehicle object corresponding to the i-th predicted moving trajectory intersecting the area of ​​interest.

[0117] The image management module includes an image analysis unit and a parameter adjustment unit.

[0118] The image analysis unit is used to calculate the abnormal movement index of each area of ​​interest.

[0119] Firstly, the location of the area of ​​interest is marked in the video image of each monitoring device, and the 0penCV technology is used to continuously extract time-continuous single-frame images in the video image.

[0120] Secondly, get the number of pixels p of a single frame image sum And the number of pixels p in each area of ​​interest fx ; Then analyze and mark the pixels in the single frame image whose color has changed compared with the previous single frame image.

[0121] Finally, count the number of all marked pixels p in a single frame image sch And the number of marked pixels p in each area of ​​interest fch , according to the formula Calculate the abnormal index of each focus area, where α is a constant.

[0122] The parameter adjustment unit is used to adjust the camera parameters. The index threshold e and the sampling times k are set, and the focus area whose abnormality index is greater than e in k consecutive single-frame images is set as the risk area.

[0123] Mark the camera with risk areas in the video screen. The marked camera automatically adjusts the shooting angle and focal length so that the images of all risk areas maximize the entire video screen.

[0124] The risk warning module displays the video images taken by each monitoring device in real time through a visual interface, with each focus area marked. When the focus area becomes a risk area, the image will be warned to the manager's smart terminal device in the first time.

[0125] Embodiment 1:

[0126] Assume that there are two areas of interest, A1 and A2, in a single-frame image, the number of pixels in the single-frame image is 1920×1080, and the number of pixels in the two areas of interest, A1 and A2, are 100×250 and 200×150 respectively; when the number of all marked pixels in a single-frame image is 300×400, the number of marked pixels in the two areas of interest, A1 and A2, are 30×50 and 60×80 respectively, the risk coefficients are 1.2 and 1.8 respectively, and the constant α is 1.1, substitute the formula to calculate the anomaly index of each area of ​​interest:

[0127] A1 Focus Area:

[0128] A2 Focus Area:

[0129] When the index threshold is 3 and the sampling times is 5, the A2 focus area with abnormality index greater than 3 in 5 consecutive single-frame images is set as the risk area.

[0130] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0131] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A smart park security data early warning method based on big data, characterized by: The method comprises the following steps: S100, collecting the traffic route map of the park and the historical logs of vehicles, and collecting video images in real time through the monitoring equipment installed in the park; S200, analyzing vehicles and pedestrians in the video image and defining target objects, calculating behavior coefficients based on the historical records of vehicle objects, predicting movement trajectories of vehicle objects and pedestrian objects, dividing areas of interest and calculating risk coefficients; S300, calculating the abnormal movement index by the risk coefficient of the concerned area and the pixel change, defining the risk area according to the abnormal movement index, and adjusting the shooting angle and focal length change of the monitoring device to increase the screen proportion of the risk area; S400, the security monitoring room displays the video images of each monitoring device in real time through a visual interface, highlights the real-time video images of risk areas, and automatically warns management personnel for processing.

2. According to the big data-based smart park security data early warning method of claim 1, it is characterized by: In S100, the traffic route map refers to a plan view used to display the internal driving roads of the park; the historical log refers to the entry and exit records of vehicles, and the entry and exit records refer to the information records of the vehicle entering the park and leaving the park, and each entry and exit record includes a license plate number and a driving video set; the driving video set is composed of video clips of the same vehicle at different locations inside the park collected by different monitoring devices; the monitoring equipment is installed on both sides of the driving roads in the park, and is used to collect video images of the driving roads in real time.

3. According to the big data-based smart park security data early warning method of claim 2, it is characterized by: In S200, the specific steps are as follows: S201, using a target recognition algorithm to analyze the video images collected by each camera, treating pedestrian targets and vehicle targets as pedestrian objects and vehicle objects respectively, using the YOLO algorithm and the SVM algorithm to detect the license plate of each vehicle object and recognize the license plate number; marking the location of each camera in the traffic route map, obtaining the video image of the camera closest to the entrance and exit of the park, and analyzing and marking the vehicle objects entering the park; S202, obtain the marked vehicle object DX sign License plate number LPN sign , retrieve the license plate number LPN in the historical log sign Corresponding to all the entry and exit records, split all the driving video sets under these entry and exit records into video clips; analyze the marked vehicle objects DX in the video clips sign According to the driving conditions, the driving distance is divided by the driving time to obtain the driving speed; a speed line graph is drawn according to the time period of each video clip and the driving speed at different times, and each video clip corresponds to a speed line graph; S203, set the speed change duration C, mark the time nodes when the driving speed changes in the speed line graph, and combine multiple time nodes where the driving speed increases or decreases continuously and the interval duration is less than C into a speed change zone; divide the driving speed range of all time nodes in the speed change zone by the duration of the speed change zone to obtain the speed change, and calculate the average speed change of all speed change zones in the speed line graph And the total duration ct n , substitute it into the formula to calculate the marked vehicle object DX sign The behavior coefficient WX sign : Where m is the marked vehicle object DX sign Number of video clips, JL n Mark the vehicle object DX in the nth video clip sign The total driving distance, zt n is the duration of the nth video clip; Calculate the behavior coefficient of each marked vehicle object separately; S204, analyzing the driving direction and driving speed of the marked vehicle object in the current video image of each monitoring device, setting a deceleration time Q, and multiplying the driving speed by Q to obtain a moving distance MD; Taking the current position of the marked vehicle object as the starting point, a route with a distance of MD is set along the driving road on the traffic route map according to the driving direction as the predicted moving trajectory; the moving direction of the pedestrian objects on both sides of the predicted moving trajectory in the video image is analyzed, and a vector is established according to the position and moving direction of the pedestrian objects. Marker Vector Pedestrian objects that intersect with the predicted moving trajectory; S205, setting a distance U, taking the marked pedestrian object as the center and U as the diameter to establish a circular area, and combining the intersecting circular areas as the focus area; Count the number of predicted moving trajectories j that intersect with the area of ​​interest, obtain the behavior coefficients of the marked vehicle objects corresponding to these predicted moving trajectories, and substitute them into the formula to calculate the risk coefficient of each area of ​​interest: Where FX is the risk factor, h is the number of marked pedestrian objects in the area of ​​interest, and WX i is the behavior coefficient of the marked vehicle object corresponding to the i-th predicted moving trajectory intersecting the area of ​​interest.

4. According to the big data-based smart park security data early warning method of claim 3, it is characterized by: In S300, the specific steps are as follows: S301, marking the location of the focus area in the video image of each monitoring device, using OpenCV technology to continuously extract time-continuous single-frame images in the video image, and obtaining the pixel point number p of the single-frame image sum And the number of pixels p in each area of ​​interest fx ; S302: Analyze and mark the pixels in the single-frame image whose colors have changed compared with the previous single-frame image, and count the number of all marked pixels in the single-frame image. sch And the number of marked pixels p in each focus area fch , substitute into the formula to calculate the abnormal index of each focus area: Where YZ is the abnormal index, α is a constant; S303, setting the index threshold e and the number of sampling times k, and setting the focus area whose abnormality index is greater than e in k consecutive single-frame images as the risk area; marking the camera in the video screen where there is a risk area, and the marked camera automatically adjusts the shooting angle and focal length so that the images of all risk areas maximize the entire video screen.

5. According to the big data-based smart park security data early warning method of claim 4, it is characterized by: In S400, the security monitoring room uses a visual interface to display the video images captured by each monitoring device in real time, with emphasis on marking each area of ​​concern; when an area of ​​concern becomes a risk area, the real-time video image of the risk area is highlighted at the first time, and the image is automatically alerted to the manager's smart terminal device.

6. A smart park security data early warning system based on big data, characterized by: The system includes data acquisition module, security analysis module, image management module and risk warning module; The data acquisition module is used to collect the traffic route map of the park and the historical logs of vehicles, and to collect video images in the park in real time through monitoring equipment; The security analysis module defines the target objects through vehicles and pedestrians in the video images, calculates the behavior coefficient according to the historical records of the vehicle objects, predicts the movement trajectories of the vehicle objects and pedestrian objects, divides the focus area and calculates the risk coefficient; The image management module calculates the abnormal index based on pixel changes, defines the risk area based on the abnormal index, and adjusts the parameters of the monitoring equipment to increase the screen share of the risk area; The risk warning module is used to display the video images of each monitoring device, highlight the real-time video images of the risk area and automatically issue warnings.

7. The smart park security data early warning system based on big data according to claim 6 is characterized by: The data acquisition module includes a video image acquisition unit, a history log acquisition unit and a traffic data acquisition unit; The video image acquisition unit is used to acquire video images, and acquires video images of the driving road in real time through monitoring equipment installed in the park; The historical log collection unit is used to collect vehicle entry and exit records. Entry and exit records refer to the information records of vehicles entering and leaving the park. Each entry and exit record includes a license plate number and a driving video set. The driving video set consists of video clips of the same vehicle at different locations within the park collected by different monitoring devices. The traffic data collection unit is used to collect the traffic route map of the park. The traffic route map refers to the plan view of the internal driving roads of the park.

8. According to the big data-based smart park security data early warning system of claim 7, it is characterized by: The security analysis module includes a vehicle analysis unit and a danger prediction unit; The vehicle analysis unit is used to calculate the behavior coefficient of the vehicle object; First, identify pedestrian objects and vehicle objects in the video image, and detect the license plate number of the vehicle object; analyze the position of each camera in the traffic route map, mark the vehicle object entering the park in the video image of the camera closest to the park entrance and exit; obtain the marked vehicle object DX sign License plate number LPN sign , retrieve the license plate number LPN in the historical log sign Corresponding to all entry and exit records, split all driving video sets under these entry and exit records into video clips; Secondly, analyze the marked vehicle objects DX in the video clips sign The speed of the vehicle is determined by the time period of each video clip and the speed at different times. A speed line graph is drawn according to the time period of each video clip and the speed at different times. The speed change duration C is set, and the time nodes at which the speed changes in the speed line graph are marked. Multiple time nodes at which the speed increases or decreases continuously and the interval duration is less than C are combined into a speed change zone. The speed change is obtained by dividing the speed range of all time nodes in the speed change zone by the duration of the speed change zone, and the speed change average value of all speed change zones in the speed line graph is calculated. And the total duration ct n ; Finally, according to the formula Calculate the marked vehicle object DX sign The behavior coefficient WX sign , where m is the marked vehicle object DX sign Number of video clips, JL n Mark the vehicle object DX in the nth video clip sign The total driving distance, zt n is the duration of the nth video clip; calculate the behavior coefficient of each marked vehicle object respectively; The risk prediction unit is used to divide the area of ​​concern and calculate the risk factor; First, the driving direction and speed of the marked vehicle object in the current video image of each monitoring device are analyzed, the deceleration time Q is set, and the moving distance MD is obtained by multiplying the driving speed by Q; taking the current position of the marked vehicle object as the starting point, a route with a distance of MD is set along the driving road on the traffic route map according to the driving direction as the predicted moving trajectory; Secondly, the moving directions of pedestrian objects on both sides of the predicted moving trajectory in the video image are analyzed, and vectors are established based on the positions and moving directions of the pedestrian objects. Marker Vector Pedestrian objects that intersect with the predicted moving trajectory; set the distance U, take the marked pedestrian object as the center, and establish a circular area with U as the diameter, and combine the intersecting circular areas as the focus area; Finally, the number of predicted moving trajectories j that intersect with the focus area is counted, and the behavior coefficients of the marked vehicle objects corresponding to these predicted moving trajectories are obtained. According to the formula Calculate the risk factor FX for each area of ​​interest; where h is the number of marked pedestrian objects in the area of ​​interest, WX i is the behavior coefficient of the marked vehicle object corresponding to the i-th predicted moving trajectory intersecting the area of ​​interest.

9. The smart park security data early warning system based on big data according to claim 8 is characterized by: The image management module includes an image analysis unit and a parameter adjustment unit; The image analysis unit is used to calculate the abnormal movement index of each area of ​​interest; First, the location of the area of ​​interest is marked in the video image of each monitoring device, and OpenCV technology is used to continuously extract single-frame images that are continuous in time from the video image; Secondly, get the number of pixels p of a single frame image sum And the number of pixels p in each area of ​​interest fx ; Analyze and mark the pixels in the single frame image whose color has changed compared with the previous single frame image; Finally, count the number of all marked pixels p in a single frame image sch And the number of marked pixels p in each area of ​​interest fch , according to the formula Calculate the abnormal index of each focus area; where α is a constant; The parameter adjustment unit is used to adjust the camera parameters; set the index threshold e and the sampling number k, and set the focus area whose abnormality index is greater than e in k consecutive single-frame images as the risk area; Mark the camera with risk areas in the video screen. The marked camera automatically adjusts the shooting angle and focal length so that the images of all risk areas maximize the entire video screen.

10. The smart park security data early warning system based on big data according to claim 9 is characterized by: The risk warning module uses a visual interface to display the video images captured by each monitoring device in real time, with emphasis on marking each area of ​​concern; when an area of ​​concern becomes a risk area, the image warning is immediately sent to the manager's smart terminal device.

Images