A kind of accident detection early warning system based on target abnormal behaviour trajectory analysisTechnical field
The present embodiments relate to Traffic monitoring fields, and in particular to a kind of based on the different of target abnormal behaviour trajectory analysisNormal event detection early warning system.
Background technique
There are detection accuracy for existing road accident detection equipment or system not enough, partial event type can not detect reportAlert shortcomings, such as occur collapsing on road, landslide, have barrier or undetectable hazard event on road.MeshPreceding field of traffic there are many kinds of event detection technology, for detect in road, bridge, the vehicle in tunnel or region have it is without exceptionThe appearance of event or dangerous situation.Such as based on video and graphic analysis or the Video Events detection system of video structural, baseRadar event detection system in millimetre-wave radar target following technology, the laser radar event based on laser radar scanning technologyDetection system, the vehicle active alarm system (vehicle SOS) positioned based on vehicle GPS/Beidou etc..
Existing traffic incident detecting system and equipment mostly use the mode of video and graphic analysis to be detected, and pass throughThe image that the video camera of road installation provides analyzes the abnormal vehicle in the presence or absence of road or abnormal road with image processing techniquesThe appearance of condition simultaneously provides corresponding warning message.Since image analysis processing and video structural side are used in Video Events detectionFormula has accident event without exception to analyze in picture, since this technology is stronger to video image dependence and defeated to frontThe camera installation locations of image, height and ambient enviroment are more demanding out, and image comparison technology is for special circumstances, thanSuch as shedding object, falling article, the object detection effect of motion profile is not poor after especially falling, can not be accurate rightThis kind of event is detected and issues alarm.It is (smog, sand and dust) and meteorological especially in the environment of illumination is weaker or eveningIn the case where inclement condition (greasy weather, snowy day, rainy day, haze), Video Events detection detection accuracy is greatly reduced, or even can not be rightEvent is detected and is alarmed.Considerably increase the risk that subsequent traffic accident occurs.
Millimetre-wave radar event detection system, this system be based on radar by transmitting radio wave by echo comeThe multidate information for obtaining target, since existing radar equipment majority is all using Doppler technology, which can be to the height of movementFast target is interested, and the object of object at a slow speed and stopping lower than 5Km/h will can not be tracked and be detected, such as: stopVehicle only, sheds the events such as object, road surface collapsing, landslide at pedestrian.
Based on vehicle GPS/BEI-DOU position system accident detection, the system mainly positioning by being installed on vehicleSystem and call for service system are completed jointly, and when vehicle occurs abnormal or driver and conductor issues when calling for help, vehicle monitoring centerThe warning message of event accident vehicle could be obtained, but other event accidents of traffic information and non-vehicle can not be obtainedEarly warning and alarming information.
Summary of the invention
For this purpose, the embodiment of the present invention provides a kind of accident detection early warning system based on target abnormal behaviour trajectory analysisSystem, to solve on existing road, common accident detection system detection precision is low, partial event type can not detect, can notThe problem of giving warning in advance.
To achieve the goals above, the embodiment of the present invention provides the following technical solutions:
The embodiment of the invention discloses a kind of accident detection early warning system based on target abnormal behaviour trajectory analysis,The early warning system includes: goal behavior track acquisition device, PTZ camera and server, the goal behavior track acquisitionReal time data is sent to server after the action trail of device acquisition target, the server sends out goal behavior acquisition deviceThe data sent carry out real-time analysis processing, and server compares real time data by trained abnormal event alarming modelIt is right, detect that goal behavior or track then issue warning information in the presence of abnormal, the PTZ camera acquisition abnormity event occursThe image in region confirms that the road real-time pictures of PTZ camera acquisition are stored in server, described to warning informationWarning information after verifying is sent to monitoring center by server, and the warning information that monitoring center sends over server carries outConfirm and be shown and early warning is issued to the public or third party system.
Further, goal behavior track acquisition device includes radar sensor, and the radar sensor setting existsRoad side, radar sensor obtain the multidate information of each target point in overlay area in real time by way of high frequency sweep,It include: instantaneous velocity, driving direction, place lane, latitude and longitude information, target type, the coordinate bit relative to radar sensorIt sets, region, distributing position, the data information of acquisition is sent to server by radar sensor.
Further, the PTZ camera is disposed in an evenly spaced relation in road side, the shooting model of adjacent PTZ cameraIt encloses boundary mutually to cover, guarantees that, without shooting blind area, the PTZ camera is connect with server, server control holder is taken the photograph360 ° of rotary takings of camera, content of shooting includes video, photo, and content of shooting is sent to server.
Further, it is slow that analyses and comparison module, control module, data transmission module, data are provided in the serverStoring module and vehicle running state large database concept, the vehicle running state large database concept include: limit of the present road to vehicleFast threshold value distinguishes the speed limit of station wagon, motorbus and lorry；Acceleration when station wagon, motorbus and lorry accelerateAcceleration when degree and deceleration；The lane quantity of present road, distinguishes fast traffic lane, slow lane, Emergency Vehicle Lane, and each vehicle corresponds to rowThe lane sailed；The corresponding relationship of speed and vehicle fleet size on road, it is close to be less than vehicle for average speed in the case that traffic density is bigSpend it is small in the case where average speed；
The data transmission module receives the data of radar sensor monitoring, is cached in data cache module, instituteIt states data of the analyses and comparison module based on normal vehicle operation state in vehicle running state large database concept and radar sensor is supervisedThe data of survey are analysed and compared, and analyses and comparison result is sent to background server, the control mould by data transmission moduleBlock controls PTZ camera and shoots road picture.
Further, the analyses and comparison result and correspondence of the vehicle running state large database concept typing analyses and comparison moduleRadar sensor detection data, database is carried out it is perfect, analyses and comparison module carry out constantly training study, promoted analysis sentencesDisconnected accuracy.
Further, the server includes multiple storage units, server to analyses and comparison result and road image intoRow stores and is sent to monitoring center.
Further, the monitoring center includes display module and alarm module, and monitoring center receives what server was sentAnalyses and comparison result and road image, according to analysing and comparing as a result, alarm module sounds an alarm, display module pop-up shows roadImage reminds staff to carry out checking processing in time.
Further, the early warning detection system the following steps are included:
Lane segmentation in coverage area is multiple regions, the traveling of monitoring vehicle in each region by radar sensorState；
Radar sensor acquires the real time kinematics track of all vehicles in each section, the vehicle of the every amount vehicle of acquisition, movementSpeed, acceleration, longitude and latitude and direction angle information；
The data of monitoring are sent to server by radar sensor, detect that vehicle abnormality slows down in a certain region, brake,Lane change or closeness increase, and server is by the number of normal vehicle operation in the data of monitoring and vehicle running state large database conceptAccording to analysing and comparing, anomalous event type is judged；
The driving status for analyzing all vehicles in above-mentioned zone adjacent area judges the anomalous event occurred in above-mentioned zoneIt whether is incident；
The nearest PTZ camera shooting anomalous event generation area in server command range anomalous event generation areaThe road image of video and picture, shooting is sent to server；
The longitude and latitude of position is occurred server for event time, event type, event and road image stores；
The information of storage is sent to monitoring center by server, and monitoring center is alarmed, and pops up road image pop-up,Remind staff to check and handle, in front of the anomalous event generation area on advices plate prompt front region have anomalous event orAccident occurs, and pays attention to evacuation of slowing down, gives warning in advance, reduce the generation of accident.
The embodiment of the present invention has the advantages that
The embodiment of the invention discloses a kind of accident detection early warning system based on target abnormal behaviour trajectory analysis,By the way that radar sensor is arranged in road side, the operating condition of vehicle is monitored, monitoring accuracy is promoted, monitoring data is sent toServer is compared with normal vehicle running data, judges whether to be abnormal event, and judge event type, realizes abnormalThe preliminary judgement of event, the road picture shot by PTZ camera, confirms the generation of anomalous event, server sentences comparisonDisconnected result and road image are stored, and are sounded an alarm by monitoring center, and pop-up shows road image, staff according toActual conditions give warning in advance in incident region fore, reduce accident rate.
Detailed description of the invention
It, below will be to embodiment party in order to illustrate more clearly of embodiments of the present invention or technical solution in the prior artFormula or attached drawing needed to be used in the description of the prior art are briefly described.It should be evident that the accompanying drawings in the following description is onlyIt is merely exemplary, it for those of ordinary skill in the art, without creative efforts, can also basisThe attached drawing of offer, which is extended, obtains other implementation attached drawings.
Structure depicted in this specification, ratio, size etc., only to cooperate the revealed content of specification, forThose skilled in the art understands and reads, and is not intended to limit the invention enforceable qualifications, therefore does not have technicalEssential meaning, the modification of any structure, the change of proportionate relationship or the adjustment of size are not influencing the function of the invention that can be generatedUnder effect and the purpose that can reach, should all still it fall in the range of disclosed technology contents obtain and can cover.
Fig. 1 is a kind of accident detection early warning system flow chart provided in an embodiment of the present invention；
Fig. 2 is a kind of road anomalous event schematic diagram provided in an embodiment of the present invention；
Fig. 3 is another road anomalous event schematic diagram provided in an embodiment of the present invention；
Fig. 4 is another road anomalous event schematic diagram provided in an embodiment of the present invention；
In figure: 1- radar sensor, 2- PTZ camera, 3- server.
Specific embodiment
Embodiments of the present invention are illustrated by particular specific embodiment below, those skilled in the art can be by this explanationContent disclosed by book is understood other advantages and efficacy of the present invention easily, it is clear that described embodiment is the present invention oneSection Example, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not doingEvery other embodiment obtained under the premise of creative work out, shall fall within the protection scope of the present invention.
Embodiment 1
With reference to Fig. 1, present embodiment discloses a kind of accident detection early warning based on target abnormal behaviour trajectory analysisSystem, comprising: goal behavior track acquisition device, PTZ camera 2 and server 3, goal behavior track acquisition deviceReal time data is sent to server 3 after the action trail of acquisition target, the server 3 sends goal behavior acquisition deviceData carry out real-time analysis processing, server 3 is compared real time data by trained abnormal event alarming model,Detect that goal behavior or track then issue warning information, the 2 acquisition abnormity event generating region of PTZ camera in the presence of abnormalThe image in domain confirms that the road real-time pictures that PTZ camera 2 acquires are stored in server, the clothes to warning informationWarning information after verifying is sent to monitoring center by business device 3, and the warning information that monitoring center sends over server carries outConfirm and be shown and early warning is issued to the public or third party system.
Goal behavior track acquisition device includes radar sensor 1, and the radar sensor 1 is arranged in road side, thunderObtain the multidate information of each target point in overlay area in real time by way of high frequency sweep up to sensor 1, comprising: i.e. speed per hourDegree, driving direction, place lane, latitude and longitude information, target type, the coordinate position relative to radar sensor 1, locationThe data information of acquisition is sent to server 3 by domain, distributing position, radar sensor 1.
PTZ camera 2 is disposed in an evenly spaced relation in road side, and the coverage boundary of adjacent PTZ camera 2 is mutualCovering guarantees that, without shooting blind area, the PTZ camera 2 is connect with server 3, and server 3 controls 360 ° of PTZ cameraRotary taking, content of shooting include video, photo, and content of shooting is sent to server 3.Analyses and comparison are provided in server 3Module, control module, data transmission module, data cache module and vehicle running state large database concept, the vehicle driving shapeState large database concept includes: speed limit threshold value of the present road to vehicle, distinguishes the speed limit of station wagon, motorbus and lorry；
Acceleration when acceleration when station wagon, motorbus and lorry accelerate and deceleration；
The lane quantity of present road, distinguishes fast traffic lane, slow lane, Emergency Vehicle Lane, and the lane of the corresponding traveling of each vehicle is smallCar walks fast traffic lane under normal circumstances, and lorry walks slow lane under normal circumstances；
The corresponding relationship of speed and vehicle fleet size on road, average speed is less than traffic density in the case that traffic density is bigAverage speed in the case where small；
The data transmission module receives the data that radar sensor 1 monitors, and is cached in data cache module, instituteIt states data of the analyses and comparison module based on normal vehicle operation state in vehicle running state large database concept and radar sensor 1 is supervisedThe data of survey are analysed and compared, and the control module control PTZ camera 2 shoots road picture.
Server 3 includes multiple storage units, and 3 pairs of analyses and comparison results of server and road image are stored and sentTo monitoring center.Monitoring center includes display module and alarm module, and monitoring center receives the analyses and comparison knot that server is sentFruit and road image, according to analysing and comparing as a result, alarm module sounds an alarm, display module pop-up shows road image, remindsStaff carries out checking processing in time.
The analyses and comparison result and corresponding radar sensor 1 of vehicle running state large database concept typing analyses and comparison moduleDetection data, database is carried out it is perfect, analyses and comparison module carry out constantly training study, promote the accuracy of analytical judgment,Faster judge the anomalous event occurred on road.
Embodiment 2
Present embodiment discloses a kind of accident detection early warning system based on target abnormal behaviour trajectory analysis, referencesIn Fig. 2, region C occur parking or shed, fall event when.Front vehicle normal pass is influenced, back car can select to becomeRoad avoids current obstacle, and is possible to only lane change without slowing down.Single car lane change is not as Rule of judgment, if met pre-The continuous vehicle first set is provided simultaneously with this behavior, for example the two vehicles that are connected all carry out lane change in this region, pass through comparisonThe judgement output alert data model set in database compares judgement and analyzes normal vehicle traveling and lane change behavior railMark, this behavior judge comparison condition as one.
The judgement of server 3 has a special event herein, and trigger event is alarmed while to drive PTZ camera 2 to shoot specialThe video and picture of event generation area are checked.The road that 3 pairs of analyses and comparison results of server and PTZ camera 2 are shotPicture data carries out event information storage, and the information of storage is sent to monitoring center by server 3, and monitoring center is alarmed,And road image pop-up is popped up, it reminds staff to check and handle, is prompted on advices plate in front of the anomalous event generation areaFront region has anomalous event or accident, pays attention to evacuation of slowing down, gives warning in advance, reduce the generation of accident.
With reference to Fig. 2, when occur in the C of region parking or shed, fall event when.Front vehicle normal pass is influenced,Front vehicle is possible to that deceleration and lane change can be selected during evacuation.When radar sensor 1 finds single target in this areaSlow down in domain and after lane change.Radar sensor 1 is by vehicle lane change data transmission to server 3, and server 3 is by lane change dataIt is compared with normal vehicle operation status data in vehicle running state large database concept, judges that current region has anomalous event hairIt is raw.It can be alarmed at this time with trigger event, server 3 drives the video and figure of the shooting of PTZ camera 2 special event generation areaPiece is checked.It will count simultaneously, server 3 carries out event information storage, and the information of storage is sent in monitoring by server 3The heart, monitoring center are alarmed, and pop up road image pop-up, and staff is reminded to check and handle, and are occurred in anomalous eventIt prompts front region to have anomalous event or accident on region fore advices plate, pays attention to evacuation of slowing down, give warning in advance, reduce thingTherefore generation.
Embodiment 3
Present embodiment discloses a kind of accident detection early warning system based on target abnormal behaviour trajectory analysis, referencesFig. 3, slow lane driving vehicle is lorry and motorbus in vehicle running state large database concept.In front of slow lane in the C of regionWhen closely without vehicle, while when thering are two or two or more carts to drive into fast traffic lane, and lane change rear close second simultaneouslyCart has brake behavior and follows lane change.Then there may be the barrier to fall among the current region C, the more vehicles in rear is caused to becomeRoad evacuation.Radar sensor 1 is by vehicle lane change data transmission to server 3, and server 3 is by lane change data and vehicle running stateNormal vehicle operation status data is compared in large database concept, it is possible to determine that in the region slow lane C, there may be anomalous events.
The video and figure that 3 trigger event of server alarms while PTZ camera 2 being driven to shoot special event generation areaPiece is checked.The road picture data that 3 pairs of analyses and comparison results of server and PTZ camera 2 are shot carries out event information and depositsThe information of storage is sent to monitoring center by storage, server 3, and monitoring center is alarmed, and pops up road image pop-up, is remindedStaff checks and handles, and prompt front region has anomalous event or accident on advices plate in front of the anomalous event generation areaOccur, pays attention to evacuation of slowing down, give warning in advance, reduce the generation of accident.
With reference to Fig. 3, detections of radar into region C traffic density relative to two neighboring region B, D larger, average speedIt is relatively low.Radar sensor 1 is by vehicle lane change data transmission to server 3, and server 3 is by lane change data, speed data and vehicleDensity data is compared with normal vehicle operation status data in vehicle running state large database concept, to determine current regionFor improper phenomenon, it is understood that there may be anomalous event.
The video and figure that 3 trigger event of server alarms while PTZ camera 2 being driven to shoot special event generation areaPiece is checked.The road picture data that 3 pairs of analyses and comparison results of server and PTZ camera 2 are shot carries out event information and depositsThe information of storage is sent to monitoring center by storage, server 3, and monitoring center is alarmed, and pops up road image pop-up, is remindedStaff checks and handles, and prompt front region has anomalous event or accident on advices plate in front of the anomalous event generation areaOccur, pays attention to evacuation of slowing down, give warning in advance, reduce the generation of accident.
Embodiment 4
Present embodiment discloses a kind of accident detection early warning system based on target abnormal behaviour trajectory analysis, referencesFig. 4, when region C is there is a situation where collapsing, all vehicles in rear can all follow same track and be hidden to carry out lane change.It canTo analyze multiple types vehicle by server 3, the same or similar track of a variety of quantity vehicle lane changes judges on roadWhether abnormal conditions are had.
The video and figure that 3 trigger event of server alarms while PTZ camera 2 being driven to shoot special event generation areaPiece is checked.The road picture data that 3 pairs of analyses and comparison results of server and PTZ camera 2 are shot carries out event information and depositsThe information of storage is sent to monitoring center by storage, server 3, and monitoring center is alarmed, and pops up road image pop-up, is remindedStaff checks and handles, and prompt front region has anomalous event or accident on advices plate in front of the anomalous event generation areaOccur, pays attention to evacuation of slowing down, give warning in advance, reduce the generation of accident.
When front vehicles exist go slowly but not up to go slowly standard when.The more vehicles in rear are hidden, such as slow downOr lane change.PTZ camera 2 can be driven to check the slow-moving vehicle of current region with the presence or absence of unusual condition at this time.If hairRaw abnormal conditions can be tracked and early warning, data are sent to server 3 simultaneously and are stored, and monitoring center is reportedIt is alert, and real-time video picture is popped up, remind operator on duty to check and handle.
The Parking of region occurrence of large-area before the same area and this region, and rear region do not detectIn this region major accident or event may occur for vehicle at this time.Server 3 drives PTZ camera 2 to check this regionIt transmits data to server 3 simultaneously to be stored, monitoring center is alarmed, and pops up real-time video picture, is reminded on dutyPersonnel check and handle.
A variety of models are existed simultaneously on highway, the anxious acceleration of cart and trolley and the time suddenly slowed down and form are notIt is logical.It can whether there is trail change by judging the whether normal acceleration and deceleration of vehicle in combination with vehicle, can be used as judgement itemPart judges whether behavior generation area has special event.Driving PTZ camera 2 checks this region at this time, is trackedAnd the road picture that early warning, 1 detection data of radar sensor and PTZ camera 2 are shot is sent to the progress of server 3 simultaneouslyIt stores and alarms, alarm in monitoring center and carry out real-time video broadcasting, operator on duty is reminded to check and handle.
The above several scenes can be included in the storage of vehicle behavior judgement output alert data model in the database, after systemPhase carries out self-perfection to judgement output alert data model by constantly learning.And it can be used as new comparison references objectCome with judgement output alert data model using when occurring later, anomalous event accident pattern or vehicle similar or that appearance is new are non-Normal behaviour can be judged quickly.Vehicle normal behaviour data and the improper behavioral data of vehicle are constantly improve.KnotThe warning message that feedback arrives work station is closed, so that judgement output alert data model is constantly learnt and is improved, the event of can be improved is sentencedIt disconnected accuracy rate and shortens event and judges the time.
Realization detection type existing for event detection system is lacked, detection content lacks, undetectable target andBlind area is supplemented.Increase event recall rate is predicted that may be present in advance it is possible thereby to reduce the traffic accident rate on roadRisk promotes traffic safety, makes safety high speed.
Although above having used general explanation and specific embodiment, the present invention is described in detail, at thisOn the basis of invention, it can be made some modifications or improvements, this will be apparent to those skilled in the art.Therefore,These modifications or improvements without departing from theon the basis of the spirit of the present invention are fallen within the scope of the claimed invention.