Summary of the invention
The object of the present invention is to provide a kind of intelligent bayonet system, this system can realize vehicle detection, number plate of vehicle and automatically identify to collect with vehicle of deploying to ensure effective monitoring and control of illegal activities and look into, and the speed of a motor vehicle measures transmit and remote maintenance, data retrieval, failure tolerant and anti-theft alarm function with overspeed alarming, vehicle flowrate, data.
The technical solution realizing the object of the invention is: a kind of public security intelligent bayonet system, comprising:
Shooting and floor light unit: comprise high-definition camera and light compensating lamp, catch vehicle and license board information for round-the-clock, every bar track at least configures a feature video camera, and each travel direction configures a panoramic camera;
Vehicle detection unit: comprise and detect main frame and ground induction coil, for completing numeration to the detection of vehicle, the mensuration of the speed of a motor vehicle and vehicle flowrate and statistics; When vehicle is by ground induction coil, ground induction coil produces induced signal and sends it to detection main frame, detects main frame and measures the speed of a motor vehicle and statistical vehicle flowrate according to these induced signals;
Image acquisition and processing unit: comprise industrial computer, video frequency collection card and control module, video frequency collection card, control module is installed in industrial computer, video frequency collection card receives and passes the video flowing of coming by video camera, control module comprises image acquisition and processing submodule and license auto-recognition system submodule, the analog video image signals of video camera is converted to digital picture by video frequency collection card, when detection Host Detection is to when having vehicle to pass through, image acquisition submodule now intercepts a width feature in the video flowing of feature video camera, in the video flowing of panoramic camera, intercept a Zhang Quanjing picture simultaneously, feature completes license board information identification through Car license recognition submodule,
Transmission unit: comprise remote network communication module, transmission equipment optical transmitter and receiver and transmission line optical fiber; For vision signal that shooting and floor light unit are gathered and the car test Signal transmissions of vehicle detection unit inspection to industrial computer;
Central manage-ment unit: comprise database server, the network switch and inquiry workstation, inquiry workstation and database server direct interconnection network switch, central manage-ment unit adopts C/S and B/S structure, for completing Query Information, deploy to ensure effective monitoring and control of illegal activities, warning function; Database server receives the license board information that also storage figure identifies as Acquire and process unit, and the feature that image acquisition submodule intercepts and panoramic pictures are stored in database server;
Shooting and floor light unit output video signal, vehicle detection unit exports car test signal, the car test signal exported and vision signal are input to the signal input part of image acquisition and processing unit, the signal output part of image acquisition and processing unit exports license board information, the license board information exported is input to the signal input part of central manage-ment unit, and license board information is deposited in database by central manage-ment unit.
The present invention compared with prior art, its remarkable advantage: the present invention adopts the Recognition Algorithm of License Plate based on intelligent decision and on-line study, not only increase the discrimination of system and the correctness of recognition result and robustness, and the ability that the system that improves adapts to new applied environment fast.The present invention adopts Long-distance Transmission Technologies, industrial computer is adopted house dog and integrated mainboard, not only increases the reliability and maintainability of system.The present invention, by burying the mode of two ground induction coils underground in each track, not only increases the accuracy of detection, and can detect the speed of a motor vehicle and reverse driving.The present invention is provided with central manage-ment unit, the internal user in public security net can Query Information, issue and to deploy to ensure effective monitoring and control of illegal activities in real time information and warning.The present invention by several functions set, meets multiple demands.
Embodiment
A kind of public security intelligent bayonet system, comprising:
Shooting and floor light unit: comprise high-definition camera and light compensating lamp, vehicle and license board information is caught for round-the-clock, every bar track at least configures a feature video camera, simultaneously in order to see the vehicle of vehicle, color, profile, loading condition and road conditions clearly, each travel direction configures a panoramic camera, and adopt CDM floodlight light filling, reduce the sense of discomfort of driver;
Vehicle detection unit: comprise and detect main frame and ground induction coil, for completing numeration to the detection of vehicle, the mensuration of the speed of a motor vehicle and vehicle flowrate and statistics; Adopt the vehicle detection mode of coil checker, two ground induction coils are buried in each track underground, for detecting vehicle and the speed of a motor vehicle, and can send alerting signal to over-speed vehicles and reverse fleet vehicles; Detecting main frame adopts toggle switch to arrange detection sensitivity, and system power failure never loses parameter, and reliability is high.When vehicle is by ground induction coil, ground induction coil produces induced signal and sends it to detection main frame, detects main frame and measures the speed of a motor vehicle and statistical vehicle flowrate according to these induced signals; Two ground induction coils are buried in the described each track of vehicle detection unit underground, for detecting vehicle and the speed of a motor vehicle, and can send alerting signal to over-speed vehicles and reverse fleet vehicles; Detecting main frame adopts toggle switch to arrange detection sensitivity.:
Image acquisition and processing unit: adopt industrial computer, video frequency collection card and corresponding control module to complete the acquisition and processing of image, comprise License Plate, car plate cutting, Car license recognition, license plate recognition result decision-making.Wherein, License Plate adopts the algorithm of locating license plate of vehicle that the various features of feature based conspicuousness merges to complete License Plate, be applicable to the background environment of various complexity and different camera angle, the historical record (comprising recognition result, recognition credibility, track record, similarity record etc.) that license plate recognition result decision-making module utilizes a car plate to stay in the process in the visual field, carries out intelligentized decision-making to recognition result, video frequency collection card, control module is installed in industrial computer, video frequency collection card receives and passes the video flowing of coming by video camera, control module comprises image acquisition and processing submodule and license auto-recognition system submodule, the analog video image signals of video camera is converted to digital picture by video frequency collection card, when detection Host Detection is to when having vehicle to pass through, image acquisition submodule now intercepts a width feature in the video flowing of feature video camera, in the video flowing of panoramic camera, intercept a Zhang Quanjing picture simultaneously, feature completes license board information identification through Car license recognition submodule,
Transmission unit: comprise remote network communication module, transmission equipment optical transmitter and receiver and transmission line optical fiber; For vision signal that shooting and floor light unit are gathered and the car test Signal transmissions of vehicle detection unit inspection to industrial computer;
Central manage-ment unit: comprise database server, the network switch and inquiry workstation, inquiry workstation and database server direct interconnection network switch, central manage-ment unit adopts C/S and B/S structure, for completing Query Information, deploy to ensure effective monitoring and control of illegal activities, warning function; Database server receives the license board information that also storage figure identifies as Acquire and process unit, and the feature that image acquisition submodule intercepts and panoramic pictures are stored in database server; Each user terminal has corresponding authority respectively, is the internal user in public security net, carries out differentiated control.
Shooting and floor light unit output video signal, vehicle detection unit exports car test signal, the car test signal exported and vision signal are input to the signal input part of image acquisition and processing unit, the signal output part of image acquisition and processing unit exports license board information, the license board information exported is input to the signal input part of central manage-ment unit, license board information is deposited in database by central manage-ment unit, externally provides the functions such as inquiry, mainly completes inquiry, deploys to ensure effective monitoring and control of illegal activities, warning function.
Public security intelligent bayonet system of the present invention, the license board information recognition methods of the acquisition and processing unit employing of described image, concrete steps are as follows:
The first step, analyzes video flowing, follows the tracks of the vehicle in image, intercepts feature;
Second step, adopts the algorithm of locating license plate of vehicle that the various features of feature based conspicuousness merges to complete License Plate;
3rd step, adopt the character cutting algorithm based on connected domain to complete the cutting of number-plate number character, concrete steps are as follows:
3.1 number-plate number character heights are estimated;
1) car plate is converted into white gravoply, with black engraved characters;
2) car plate vertical edge detects, and first use Sobel operator completes the rim detection to car plate, and then use OTSU method to obtain the vertical mapping of car plate, edge pixel black replaces, and the white replacement of other pixels, finds the vertical edge of car plate;
3) horizontal projection, finds coboundary and the lower boundary of number-plate number character;
3.2 number-plate number character durations are estimated;
1) use OTSU algorithm to find out threshold value, complete license plate image binaryzation;
2) vertical projection, is greater than 6 pixels for width and is less than the block of the widest block, using the estimated value of mean value as character duration of these blocks;
3.3 by connected domain extract distinguish character block, comprise connected domain detect and connected domain check two parts;
1) connected domain detection algorithm;
First, a label counting is created and is initialized as 1, then, binary picture by from left to right, scanning from top to bottom;
1., for each pixel, check whether the northeast of 8 – connected domains, north, northwest and western pixel are foreground pixels;
2. if the neighbours do not met, then the value of region-counter is assigned to this region, this region-counter adds 1;
3. if only have the neighbours met, pixel is distributed to this region, if multiple neighbours meet and are the same areas, pixel value is distributed to this region;
4., if multiple neighbours are mated and in zones of different, pixel are assigned to any one in these regions, recording all these regions is equivalent region;
5. scan image, and give identical regional value by the region of all equivalences;
6. scanning stops, until the pixel do not indicated in image;
2) connected domain inspection, rule criterion is as follows:
Rule 1: if the coboundary of a connected domain is less than overall coboundary, if or its coboundary be greater than overall lower boundary, then this connected domain is not containing character;
Rule 2: if the depth-width ratio of a connected domain is greater than 0.8 or be less than 0.3, then this connected domain is assumed that not containing character;
Rule 3: if the area of this connected domain is greater than 1/6th of the size of general image, then need further segmentation;
Rule 4: if the connected domain quantity successfully detected is more than 7 or lower than 3, then need to reexamine;
The connected domain not meeting above-mentioned standard is regarded as not containing character, therefore deleted; After all connected domains of given car plate are examined, this process completes;
4th step, adopts ORC to complete the identification of the number-plate number;
5th step, above-mentioned Real time identification is carried out to every frame video image, decision-making is carried out to recognition result, namely the synthetic reliability evaluation of this car plate is obtained by calculating observation frame number, recognition result stability, trajectory stability, speed stability, average credibility and measuring similarity value, thus determine it is continue to follow the tracks of this car plate, still export recognition result, or refuse this result.
Public security intelligent bayonet system of the present invention, described industrial computer built in hardware house dog, and adopt integrated mainboard.(system in case of system halt can autoboot, and another industrial computer adopts high, the integrated mainboard of current arithmetic speed, makes system processing speed fast, and reduce many boards plug the harmful effect caused).
Below in conjunction with accompanying drawing, the present invention is described in further detail.
Shown in Fig. 1, Fig. 2, Fig. 3 and Fig. 4, public security intelligent bayonet system comprises: vehicle detection unit, front end shooting and floor light unit, image variants unit, transmission unit and central manage-ment unit.In native system front end system, video camera is in real-time photography state (25 frames per second) always, image remote is transferred in the video frequency collection card in industrial computer, the analog video image signals of video camera is converted to digital picture by video frequency collection card, when detection Host Detection is to when having vehicle to pass through, image acquisition submodule now intercepts picture (car plate feature) in the video flowing of feature video camera, in the video flowing of panoramic camera, intercept panoramic pictures simultaneously, feature completes number-plate number identification through Car license recognition module, control module registration of vehicle is by the time of control point, place, Vehicle Speed, travel direction, length of wagon and vehicle image etc., when the speed of a motor vehicle recorded exceedes the speed limit of default, onsite alarming immediately, notify that operator on duty is noted, the above number-plate number, vehicle image, test the speed the images such as value, data message is treated is uploaded to central manage-ment unit by transmission network, and stored in database server, central manage-ment unit background system can accept the warning message of front end, and forward workstation on duty, issue blacklist etc., background system processes the image uploaded, data message, can carry out inquiring about (inquiry of deploying to ensure effective monitoring and control of illegal activities, illegal inquiry, passing inquiry), statistics (vehicle flowrate, illegal statistics, car statistics) etc., when need deploy to ensure effective monitoring and control of illegal activities vehicle time, typing can to deploy to ensure effective monitoring and control of illegal activities information, reach system front end simultaneously.
Front end shooting and floor light unit: be made up of panoramic camera, feature video camera and light compensating lamp.As shown in Figure 5, in the installation of double-lane road upper latch notch system front end in the same way, need two feature video cameras (one, every track) and a panoramic camera, be arranged on same " L " bar, light compensating lamp is arranged on another " L " bar.
Vehicle detection unit: whether ground induction coil detects has vehicle to pass through, signal is sent to detection main frame, detect main unit vehicle by time of front and back coil, judge vehicle whether reverse driving, and the distance before and after utilizing through between the time and coil of front and back coil, calculates the speed of a motor vehicle; Detect the time of host computer vehicle by single coil, be multiplied by current vehicle speed and obtain vehicle commander's information, result is sent to image acquisition and processing unit.Native system adopts model: JL810ZJ wagon detector, and built-in 6 ground, road sense testing circuits, can connect 6 ground induction coils, ground induction coil sense cycle is less than 1ms, coil sensitivity: 8 grades.
Image acquisition and processing unit: image variants is completed by 1 industrial computer.Being configured to of industrial control computer: CPU P4 more than 2.8, internal memory >=512M, built in video capture card 1 piece.Native system at least stores characteristic image 1 and panoramic picture 1 calculating by every portion vehicle, 2 pictures account for disk space 80K, the image storing 1,500,000 vehicles accounts for disk space 144G, specifically depending on control point vehicle flowrate, generally at least can store above information of vehicles half a year.When hard drive space is not enough, automatically to writing time information of vehicles the earliest and picture carry out circulation covering.Control module comprises image acquisition and processing submodule, license auto-recognition system submodule and database.Wherein license auto-recognition system submodule comprises: License Plate, car plate cutting, Car license recognition, license plate recognition result decision-making.
License Plate module: this module is the basis of follow-up link, and the impact of its accuracy on overall system performance is huge.Native system adopts the algorithm of locating license plate of vehicle that the various features of feature based conspicuousness merges to complete License Plate, is applicable to the background environment of various complexity and different camera angle.
Car plate cutting module: adopt the character cutting algorithm based on connected domain, on the basis of basic UNICOM's domain lookup algorithm, adds the detection in UNICOM territory, and gives 4 check criteria.
Car license recognition module: native system adopts ORC to complete the identification of car plate.
License plate recognition result decision-making module: native system can carry out Real time identification to every frame video image, therefore in the process of a car by the visual field, native system will obtain some identical or different recognition results.This just needs the decision-making module of a recognition result, specifically, the historical record (comprising recognition result, recognition credibility, track record, similarity record etc.) that decision-making module utilizes a car plate to stay in the process in the visual field, intelligentized decision-making is carried out to recognition result, the synthetic reliability evaluation of this car plate is obtained by calculating observation frame number, recognition result stability, trajectory stability, speed stability, average credibility and similarity isometry value, thus determine it is continue to follow the tracks of this car plate, still export recognition result, or refuse this result.The final recognition result of a car plate is the recognition result by analyzing all frames, carries out intelligentized classification and ballot to them, and forms in conjunction with certain grammatical informix.This method has fully utilized the information of all frames, decreases in the past based on the accidental error that the recognition methods of single image brings, and substantially increases the discrimination of system and the correctness of recognition result and robustness.
Transmission unit: each control point design and installation a pair fiber optical transceiver (10/100M self-adaptation) or multi-channel video multiplexed optical terminal.
Central manage-ment unit: native system central manage-ment unit mainly comprises database server, the network switch and inquiry workstation.Each user terminal has corresponding authority respectively, is the internal user in public security net, carries out differentiated control, and whole system adopts C/S and B/S structure, mainly completes inquiry, deploys to ensure effective monitoring and control of illegal activities, warning function.