CN110164139A - A kind of side parking detection identifying system and method - Google Patents
A kind of side parking detection identifying system and method Download PDFInfo
- Publication number
- CN110164139A CN110164139A CN201910466245.6A CN201910466245A CN110164139A CN 110164139 A CN110164139 A CN 110164139A CN 201910466245 A CN201910466245 A CN 201910466245A CN 110164139 A CN110164139 A CN 110164139A
- Authority
- CN
- China
- Prior art keywords
- license plate
- picture
- vehicle
- magnetic field
- mcu
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/017—Detecting movement of traffic to be counted or controlled identifying vehicles
- G08G1/0175—Detecting movement of traffic to be counted or controlled identifying vehicles by photographing vehicles, e.g. when violating traffic rules
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/04—Detecting movement of traffic to be counted or controlled using optical or ultrasonic detectors
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/042—Detecting movement of traffic to be counted or controlled using inductive or magnetic detectors
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/14—Traffic control systems for road vehicles indicating individual free spaces in parking areas
- G08G1/149—Traffic control systems for road vehicles indicating individual free spaces in parking areas coupled to means for restricting the access to the parking space, e.g. authorization, access barriers, indicative lights
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Computer Security & Cryptography (AREA)
- Traffic Control Systems (AREA)
- Character Discrimination (AREA)
Abstract
It stops the invention discloses a kind of side and detects identifying system and method.The changes of magnetic field data that the present invention is driven into using three axis geomagnetic sensors detection vehicles, are handled data using the algorithm of finite state, obtain the signal that vehicle drives into, and are taken pictures with this signal driving camera to license plate part.One-dimension array is converted by the image data comprising license plate taken pictures, the array content is uploaded to server end in the form of data flow using the method that NB-IoT module sends TCP data packet.Data flow is finally reduced to picture format by specific processing in server end, identifies that final result is stored in server end in the form of picture and character string using content of the specific Recognition Algorithm of License Plate to license plate.Compared with the prior art, the present invention has the characteristics that side parking scene specific aim and accuracy of identification are high, realizes the automation and intelligence of side parking information management.
Description
Technical field
The invention belongs to vehicle information management fields, and in particular to a kind of side parking detection identifying system and method.
Background technique
With economic development, the problem of number of vehicles sustainable growth, parking difficulty, is urgently to be resolved.Urban area is existing very at present
More large parking lots are for vehicle parking, but its position is fixed, and are distributed sparse, it is difficult to cover whole parking demands.Vast
In the daily life of resident, the roadside side parking scene of more commonly relatively dispersion.
The vehicle information management system of existing some electronizations large-scale application in parking lot, such as Entrance
Automatic lifting railing and license board information identification etc. of the place based on camera.But it stops under scene in the side in roadside, due to vehicle
Mobility is high, and parking stall uniformity is lower and scale is smaller, and the relevant technology in such parking lot is difficult to be applicable in.Therefore research side is stopped
Car test detection identifying method and system are necessary.
Summary of the invention
It stops the purpose of the present invention is to provide a kind of side and detects identifying system and method, realize that the avris that satisfies the need puts parking
The automation of information, intelligent management.
For achieving the above object, the present invention adopts the following technical scheme: a kind of parking of side detects identifying system,
Be characterized in that, including geomagnetism detecting module, MCU (Microcontroller Unit, micro controller unit), camera module,
NB-IoT (Narrow Band Internet of Things, narrowband Internet of Things) module and server.
The geomagnetism detecting module is used to acquire the magnetic field of tri- axis direction of X, Y, Z under the rectangular coordinate system in space of side parking stall
Data, and acquisition data are sent to MCU；
The MCU receives the magnetic field data that geomagnetism detecting module is sent, and is obtained by finite state machine and threshold detection algorithm
To whether there is or not the signals that vehicle drives into, when having detected that vehicle drives into, driving camera module is taken pictures；And receive camera module
The headstock license plate picture of acquisition is transmitted to server by NB-IoT module with stream socket；
The camera module is for shooting the headstock license plate image for driving into vehicle；
The server restores the data flow of picture, and the license plate knot identified using Recognition Algorithm of License Plate
Fruit.
Further, geomagnetism detecting module includes three pieces of TMR2012 tunnel resistor magnetic sensor chips, and placing direction is
Tri- axis direction of X, Y, Z under rectangular coordinate system in space collectively constitutes three axis geomagnetic sensors.The placement position of the geomagnetism detecting module
It is set at the rectangle border vertices on side parking stall, on roadbed stone and tiltedly to the side parking stall for needing to detect.Wherein
For rectangular coordinate system in space using the position of earth magnetism sensing module as origin, X-axis refers to the vehicle commander direction on side parking stall；Y-axis refers to same
Direction that is vertical with X-axis and being directed toward vehicle width in one ground level；Z axis refers to upwardly direction perpendicular to the ground.The geomagnetism detecting module
Detect the magnetic field data of placement location, output analog signal to MCU.
Further, the output signal of three axis geomagnetic sensors passes through finite state machine and threshold test inside MCU
Algorithm obtains detecting the presence of the signal that vehicle drives into, and the algorithm steps are as follows:
If MRIt is the current magnetic field data result calculated at each moment, calculation formula MR=aMx+bMy+cMz, wherein
A, b, c respectively correspond the weight of X, Y, Z axis TMR2012 output analog signal, it is desirable that a > b > c.
If T1 is that have vehicle (vehicle drives into parking stall completely) and the magnetic field differential threshold without vehicle, T2 is (vehicle institute when vehicle drives away
There is tire to sail out of parking stall completely) magnetic field and background magnetic field (parking stall is without vehicle) differential threshold allow T2 there are the relationship of T1 > T2
=0, T1 and T2 select different empirical values according to usage scenario.
If MRBAnd MRCRespectively without vehicle and magnetic field data with vehicles.
The operational process of finite state machine is described as follows:
(1) it powers on, obtain background magnetic field data and saves in three axis geomagnetic sensor of car-free status, be denoted as MRB；
(2) if being currently at the state of no vehicle, three axis geomagnetic sensors can obtain M with certain frequencyRIf: MR> MRBAnd
MR-MRB> T1 then exports vehicle input signal car in, with season MRC=MR。
(3) if being currently at the state of vehicle, three axis geomagnetic sensors can obtain M with certain frequencyRIf: MR< MRCAnd
MR-MRB< T2 then exports vehicle and sails out of signal car departure, with season MRC=0.
Further, the signal driving camera module that drives into detected is taken pictures, and the camera module is
ATK-OV7725.The picture format comprising license plate taken pictures is the RGB565 pixel square of QVGA (resolution ratio 320*240)
Battle array.
Further, picture pixels matrix conversion is data flow by MCU, using NB-IoT module to send TCP
The mode of (Transmission Control Protocol, transmission control protocol) data packet is transmitted to server.
Further, server end is following steps to the data flow restoring method of picture:
(1) the useless letter of all additional IP (Internet Protocol, Internet protocol) addresses and port numbers is deleted
Breath；
(2) all TCP data packet backwards received are saved；
(3) the TCP data packet that all backwards save is connected as one-dimension array；
(4) one-dimension array is divided into the picture element matrix of 320*240 according to the format of RGB565；
(5) picture element matrix is saved with the format of jpg to server specified folder.
Further, the Recognition Algorithm of License Plate of server end includes car plate detection and character recognition two parts.
The step of Detection of License, is as follows:
(1) License Plate
The picture comprising license plate taken pictures is used into following steps:
A. weight is calculated using Gaussian function to the eight neighborhood of each pixel, then further according to the neighborhood of respective weights
Calculated for pixel values goes out Gaussian Blur result of the mean value as current pixel point；
B. gray value is converted by the RGB565 value of each pixel；
C. convolution is carried out using picture element matrix of the Soble operator to picture；
D. the average value for taking all gray values is threshold value, by the picture element matrix binaryzation of picture；
E. the picture element matrix of binaryzation is first expanded into post-etching, obtains connected domain, then take the profile of connected domain；
F. the connected domain for excluding to be unlikely to be license plate obtains candidate license plate connected domain；
G. candidate license plate connected domain is subjected to stretching conversion, obtains the rectangular domain that predetermined size is 136*36 size.
(2) license plate differentiates
Using the machine learning method of SVM (support vector machines), real license plate figure is selected from candidate license plate rectangular domain
Piece includes the following steps:
A. the side parking stall shooting picture by 1000 by algorithm of locating license plate of vehicle processing has sticked license plate and without license plate
Two kinds of labels；
B. labelled picture is obtained into 172 (136+36) a dimensions using Histogram statistics, uses RBF core (Radial
Basis Function, radial basis function) it is trained；
C. the particular model that training obtains is saved as into xml document；
D. use list as output, output result be current image whether there is or not license plate and including blue board, yellow card, green board vehicle
Board color result.
The label application process has used successive iteration automated tag method, i.e., selects the picture training of 1% quantity first
To a naive model, the picture for then further taking out 3% is predicted using naive model, carries out hand to the result of prediction error
It is dynamic to correct, it is then trained again using above-mentioned 1% and 3% whole pictures, obtains the model that precision is promoted, gradually change
In generation, obtains the pinpoint accuracy model that can classify automatically to the end.
The step of character recognition algorithm, is as follows:
(1) Character segmentation
The rectangle segment of single character is intercepted from real license plate picture, steps are as follows for specific algorithm:
A. color judgement is carried out using the template matching method in hsv color space；
B. the progress of Otsu threshold method is carried out using different parameters according to blue board, yellow card, green board different colours judging result
Binaryzation；
C. to the independent contouring of Chinese character in license plate, English and numerical character unify contouring, after being divided
Character segment.
(2) character differentiates
Using the machine learning method of ANN (artificial neural network), the character segment after segmentation is identified, list is obtained
A character result, and combine and obtain the final result of Car license recognition.Input layer, hidden layer and the output layer of the ANN model
Neuron number be respectively 120,40 and 65.
It is a kind of to carry out side parking detection knowledge method for distinguishing using above system, method includes the following steps:
(1) magnetic field of tri- axis direction of X, Y, Z under the rectangular coordinate system in space of side parking stall is acquired by geomagnetism detecting module
Data, and acquisition data are sent to MCU；
(2) MCU receives the magnetic field data that geomagnetism detecting module is sent, and is obtained by finite state machine and threshold detection algorithm
Whether there is or not the signals that vehicle drives into, when having detected that vehicle drives into, driving camera module work；
(3) after camera module activation, the headstock license plate picture of vehicle is driven into shooting, and is sent to MCU；
(4) MCU receives the headstock license plate picture of camera module acquisition, is transmitted by NB-IoT module with stream socket
To server；
(5) server restores the data flow of headstock license plate picture, and identified using Recognition Algorithm of License Plate
License plate result.
Compared with prior art, the invention has the following advantages:
(1) present system and method compared with the prior art, can detect identification for the Scene realization of side parking, into
One step reaches the information management purpose of curb parking.
(2) present system and method compared with the prior art, use the detection of specific geomagnetic sensor chip composition
Circuit and the finite state machine algorithm of detection are higher to the detection efficiency of vehicle.
(3) present system and method compared with the prior art, are uploaded using the picture comprising license plate that will be taken pictures
Method for distinguishing is known to server end, terminal hardware resource is not take up, greatly reduces the cost of terminal hardware, convenient for device in side
The large-scale use of square parking stall scene.
(4) present system and method compared with the prior art, are fully considered in the Recognition Algorithm of License Plate of server end
Side parking scene get off board inclination the characteristics of, realize higher recognition accuracy.
Detailed description of the invention
Fig. 1 is side of the present invention parking detection identifying system structure chart；
Fig. 2 is side of the present invention parking detection identification device placement position schematic diagram；
Fig. 3 is geomagnetism detecting module diagram of the present invention；
Fig. 4 is side of the present invention parking detection finite state machine algorithm schematic diagram；
Fig. 5 is data flow restoring method flow chart of the server end of the present invention to picture；
Fig. 6 is the Recognition Algorithm of License Plate flow chart of server end of the present invention.
Specific embodiment
With reference to the accompanying drawings and detailed description, invention is further described in detail.
A kind of side parking detection identifying system provided by the invention, including geomagnetism detecting module, MCU
(Microcontroller Unit, micro controller unit), camera module, NB-IoT (Narrow Band Internet of
Things, narrowband Internet of Things) module and server.
Geomagnetism detecting module is used to acquire the magnetic field number of tri- axis direction of X, Y, Z under the rectangular coordinate system in space of side parking stall
According to, and acquisition data are sent to MCU；
MCU receives the magnetic field data that geomagnetism detecting module is sent, and is had by finite state machine and threshold detection algorithm
The signal that no vehicle drives into, when having detected that vehicle drives into, driving camera module is taken pictures；And receive camera module acquisition
Headstock license plate picture, server is transmitted to stream socket by NB-IoT module；
Camera module is for shooting the headstock license plate image for driving into vehicle；
Server restores the data flow of picture, and the license plate result identified using Recognition Algorithm of License Plate.
The system construction drawing is as shown in Fig. 1.
Wherein geomagnetism detecting module is to be bi-directionally connected in MCU, and reason is that geomagnetism detecting module is needed the simulation of output
Signal passes to MCU, and the algorithm for detecting the finite state machine of vehicle is run in MCU, then obtains the letter that vehicle drives into
Number.
MCU is also to be bi-directionally connected with camera module, and reason is that the signal that MCU drives into vehicle passes to camera mould
Block, as the enable signal that camera module is taken pictures, the picture comprising license plate taken pictures is temporarily stored in MCU by camera module
The area SRAM.
MCU same as NB-IoT is to be bi-directionally connected, and reason is that MCU passes image data stream by certain subcontract forms
NB-IoT module is passed, the data packet divided is uploaded to server end by the mode that NB-IoT module reuses transmission TCP data,
After being sent completely, the instruction that NB-IoT module is returned to MCU to be sent, MCU can restore to detect and empty camera caching
Data.
The device of side parking detection identifying system is placed on the foundation stone of roadside, five-pointed star in detail location such as attached drawing 2
It is shown.
Further, geomagnetism detecting module includes three pieces of TMR2012 tunnel resistor magnetic sensor chips, and placing direction is
Tri- axis direction of X, Y, Z under rectangular coordinate system in space collectively constitutes three axis geomagnetic sensors.The placement position of the geomagnetism detecting module
It is set at the rectangle border vertices on side parking stall, on roadbed stone and tiltedly to the side parking stall for needing to detect.Wherein
For rectangular coordinate system in space using the position of earth magnetism sensing module as origin, X-axis refers to the vehicle commander direction on side parking stall；Y-axis refers to same
Direction that is vertical with X-axis and being directed toward vehicle width in one ground level；Z axis refers to upwardly direction perpendicular to the ground.The geomagnetism detecting module
Detect the magnetic field data of placement location, output analog signal to MCU.
The three axis geomagnetic sensor schematic diagram is as shown in Fig. 3.
Further, the output signal of three axis geomagnetic sensors passes through finite state machine and threshold test inside MCU
Algorithm obtains detecting the presence of the signal that vehicle drives into, and the algorithm steps are as follows:
If MRIt is the current magnetic field data result calculated at each moment, calculation formula MR=aMx+bMy+cMz, wherein
A, b, c respectively correspond the weight of X, Y, Z axis TMR2012 output analog signal, it is desirable that a > b > c, such as in this default
X, under tri- axis direction of Y, Z, weight design is respectively 0.625,0.125,0.25.
If T1 is that have vehicle (vehicle drives into parking stall completely) and the magnetic field differential threshold without vehicle, T2 is (vehicle institute when vehicle drives away
There is tire to sail out of parking stall completely) magnetic field and background magnetic field (parking stall is without vehicle) differential threshold allow T2 there are the relationship of T1 > T2
=0, T1 and T2 select different empirical values according to usage scenario.Such as open area (around 50 meters do not have pile
Object), and near 10 meters of radius in range number of vehicles be no more than 3 when, be arranged T1=10 and T2=0.8.
If MRBAnd MRCRespectively without vehicle and magnetic field data with vehicles.
The operational process of finite state machine is described as follows:
(1) it powers on, obtain background magnetic field data and saves in three axis geomagnetic sensor of car-free status, be denoted as MRB；
(2) if being currently at the state of no vehicle, three axis geomagnetic sensors can obtain M with certain frequencyRIf: MR> MRBAnd
MR-MRB> T1 then exports vehicle input signal car in, with season MRC=MR。
(3) if being currently at the state of vehicle, three axis geomagnetic sensors can obtain M with certain frequencyRIf: MR< MRCAnd
MR-MRB< T2 then exports vehicle and sails out of signal car departure, with season MRC=0.
The algorithm schematic diagram of the finite state machine is as shown in Fig. 4.
Further, the signal driving camera module that drives into detected is taken pictures, and the camera module is
ATK-OV7725.The picture format comprising license plate taken pictures is the RGB565 pixel square of QVGA (resolution ratio 320*240)
Battle array.
Further, picture pixels matrix conversion is data flow by MCU, using NB-IoT module to send TCP
The mode of (Transmission Control Protocol, transmission control protocol) data packet is transmitted to server.
Further, server end is following steps to the data flow restoring method of picture:
(1) the useless letter of all additional IP (Internet Protocol, Internet protocol) addresses and port numbers is deleted
Breath；
(2) all TCP data packet backwards received are saved；
(3) the TCP data packet that all backwards save is connected as one-dimension array；
(4) one-dimension array is divided into the picture element matrix of 320*240 according to the format of RGB565；
(5) picture element matrix is saved with the format of jpg to server specified folder.
The server end is as shown in Fig. 5 to the restoring method flow chart of image data stream.
The Recognition Algorithm of License Plate of server end includes car plate detection and character recognition two parts.
The step of Detection of License, is as follows:
(1) License Plate
The picture comprising license plate taken pictures is used into following steps:
A. weight is calculated using Gaussian function to the eight neighborhood of each pixel, then further according to the neighborhood of respective weights
Calculated for pixel values goes out Gaussian Blur result of the mean value as current pixel point；
B. gray value is converted by the RGB565 value of each pixel；
C. convolution is carried out using picture element matrix of the Soble operator to picture；
D. the average value for taking all gray values is threshold value, by the picture element matrix binaryzation of picture；
E. the picture element matrix of binaryzation is first expanded into post-etching, obtains connected domain, then take the profile of connected domain；
F. the connected domain for excluding to be unlikely to be license plate obtains candidate license plate connected domain；
G. candidate license plate connected domain is subjected to stretching conversion, obtains the rectangular domain that predetermined size is 136*36 size.
(2) license plate differentiates
Using the machine learning method of SVM (support vector machines), real license plate figure is selected from candidate license plate rectangular domain
Piece includes the following steps:
A. the side parking stall shooting picture by 1000 by algorithm of locating license plate of vehicle processing has sticked license plate and without license plate
Two kinds of labels；
B. labelled picture is obtained into 172 (136+36) a dimensions using Histogram statistics, uses RBF core (Radial
Basis Function, radial basis function) it is trained；
C. the particular model that training obtains is saved as into xml document；
D. use list as output, output result be current image whether there is or not license plate and including blue board, yellow card, green board vehicle
Board color result.
The label application process has used successive iteration automated tag method to select first to reduce the workload of training pattern
The picture training of 1% quantity obtains a naive model out, and the picture for then further taking out 3% is predicted using naive model,
Manual correction is carried out to the result of prediction error, is then trained, is obtained precisely again using above-mentioned 1% and 3% whole pictures
The model promoted is spent, successive iteration obtains the pinpoint accuracy model that can classify automatically to the end.
The step of character recognition algorithm, is as follows:
(1) Character segmentation
The rectangle segment of single character is intercepted from real license plate picture, steps are as follows for specific algorithm:
A. color judgement is carried out using the template matching method in hsv color space；
B. the progress of Otsu threshold method is carried out using different parameters according to blue board, yellow card, green board different colours judging result
Binaryzation；
C. to the independent contouring of Chinese character in license plate, English and numerical character unify contouring, after being divided
Character segment.
(2) character differentiates
Using the machine learning method of ANN (artificial neural network), the character segment after segmentation is identified, list is obtained
A character result, and combine and obtain the final result of Car license recognition.The mind of the input layer of ANN model, hidden layer and output layer
It is respectively 120,40 and 65 through first number.
The algorithm structure figure of the Car license recognition is as shown in Fig. 6.Wherein car plate detection and character two parts collectively constitute
Whole algorithms of Car license recognition.It is divided into License Plate again inside car plate detection and license plate differentiates two parts, character recognition inner part
Two parts are differentiated for Character segmentation and character.All be before every part in all steps by sequential connection, i.e., it is previous
The output of algorithm is the input of the latter algorithm.The input of algorithm entirety is that camera module takes pictures to obtain comprising license plate
The picture of jpg format, export be for form character string license plate content, such as: Zhejiang A12345.
The above shows a kind of side parking detection identifying system proposed by the present invention, can stop for roadside side
Scene realization efficiently stop detection and identification, wherein detection and identification accuracy are high, be conducive to vehicle management system into
The development of one step, total system method have innovative and validity.
The specific embodiment is the preferred embodiments of the present invention, but the present invention is not limited to above-mentioned embodiment party
Formula, without departing from the essence of the present invention, what those skilled in the art can make any obviously changes
It is all belonged to the scope of protection of the present invention into, replacement or modification.
Claims (10)
1. a kind of side parking detection identifying system, which is characterized in that including geomagnetism detecting module, MCU, camera module, NB-
IoT module and server；
The geomagnetism detecting module is used to acquire the magnetic field number of tri- axis direction of X, Y, Z under the rectangular coordinate system in space of side parking stall
According to, and acquisition data are sent to MCU；
The MCU receives the magnetic field data that geomagnetism detecting module is sent, and is had by finite state machine and threshold detection algorithm
The signal that no vehicle drives into, when having detected that vehicle drives into, driving camera module is taken pictures；And receive camera module acquisition
Headstock license plate picture, server is transmitted to stream socket by NB-IoT module；
The camera module is for shooting the headstock license plate picture for driving into vehicle；
The server restores the data flow of headstock license plate picture, and the license plate identified using Recognition Algorithm of License Plate
As a result.
2. parking detection identifying system in a kind of side according to claim 1, which is characterized in that geomagnetism detecting module includes
Three pieces of TMR2012 tunnel resistor magnetic sensor chips, placing direction are tri- axis direction of X, Y, Z under rectangular coordinate system in space, altogether
With three axis geomagnetic sensors of composition.The placement location of the geomagnetism detecting module is the rectangle border vertices on side parking stall
Place, on roadbed stone and tiltedly to the side parking stall for needing to detect.Wherein rectangular coordinate system in space is with earth magnetism sensing module
Position is origin, and X-axis refers to the vehicle commander direction on side parking stall, and Y-axis refers to vertical with X-axis in same ground level and is directed toward vehicle width
Direction, Z axis refer to upwardly direction perpendicular to the ground.The magnetic field data of the geomagnetism detecting module detection placement location, output simulation
Signal is to MCU.
3. a kind of side parking detection identifying system according to claim 2, which is characterized in that the three axis earth magnetism sensing
The output signal of device obtains detecting the presence of the signal that vehicle drives by finite state machine and threshold detection algorithm inside MCU,
Specific step is as follows:
If MRIt is the current magnetic field data result calculated at each moment, calculation formula MR=aMx+bMy+cMz, wherein a, b,
C respectively corresponds the weight of X, Y, Z axis TMR2012 output analog signal, it is desirable that a > b > c.
If T1 is the magnetic field differential threshold for having Che Yuwu vehicle, T2 is the magnetic field and background magnetic field differential threshold when vehicle drives away, and is deposited
In the relationship of T1 > T2, allow T2=0.
If MRBAnd MRCRespectively without vehicle and magnetic field data with vehicles.
The operational process of finite state machine is described as follows:
(1) it powers on, obtain background magnetic field data and saves in three axis geomagnetic sensor of car-free status, be denoted as MRB；
(2) if being currently at the state of no vehicle, three axis geomagnetic sensors can obtain M with certain frequencyRIf: MR> MRBAnd MR-MRB
> T1 then exports vehicle input signal car in, with season MRC=MR。
(3) if being currently at the state of vehicle, three axis geomagnetic sensors can obtain M with certain frequencyRIf: MR< MRCAnd MR-MRB
< T2 then exports vehicle and sails out of signal car departure, with season MRC=0.
4. parking detection identifying system in a kind of side according to claim 1, which is characterized in that the camera module is
ATK-OV7725.The picture format comprising license plate taken pictures is the RGB565 pixel square of QVGA (resolution ratio 320*240)
Battle array.
5. parking detection identifying system in a kind of side according to claim 1, which is characterized in that the MCU is by picture picture
Prime matrix is converted to data flow, is transmitted to server in a manner of sending TCP data packet using NB-IoT module.
6. a kind of side parking detection identifying system according to claim 1, which is characterized in that the server is to picture
Data flow restoring method the following steps are included:
(1) garbage of all additional IP addresses and port numbers is deleted；
(2) all TCP data packet backwards received are saved；
(3) the TCP data packet that all backwards save is connected as one-dimension array；
(4) one-dimension array is divided into the picture element matrix of 320*240 according to the format of RGB565；
(5) picture element matrix is saved with the format of jpg to server specified folder.
7. a kind of side parking detection identifying system according to claim 1, which is characterized in that the license plate of the server
Recognizer includes car plate detection and character recognition two parts.
8. a kind of side parking detection identifying system according to claim 7, which is characterized in that the Detection of License
The step of it is as follows:
(1) License Plate
The picture comprising license plate taken pictures is handled as follows:
A. weight is calculated using Gaussian function to the eight neighborhood of each pixel, then further according to the neighborhood territory pixel of respective weights
Value calculates Gaussian Blur result of the mean value as current pixel point；
B. gray value is converted by the RGB565 value of each pixel；
C. convolution is carried out using picture element matrix of the Soble operator to picture；
D. the average value for taking all gray values is threshold value, by the picture element matrix binaryzation of picture；
E. the picture element matrix of binaryzation is first expanded into post-etching, obtains connected domain, then take the profile of connected domain；
F. the connected domain for excluding to be unlikely to be license plate obtains candidate license plate connected domain；
G. candidate license plate connected domain is subjected to stretching conversion, obtains the rectangular domain that predetermined size is 136*36 size.
(2) license plate differentiates
Using the machine learning method of SVM, real license plate picture is selected from candidate license plate rectangular domain, is included the following steps:
A. by 1000 by algorithm of locating license plate of vehicle processing side parking stall shooting picture stick license plate and without license plate two
Kind label；
B. labelled picture is obtained into 172 dimensions using Histogram statistics, is trained using RBF core；
C. the particular model that training obtains is saved as into xml document；
D. use list as output, output result is current image whether there is or not license plate and the license plate face including blue board, yellow card, green board
Color result.
The label application process has used successive iteration automated tag method, i.e., the picture training for selecting 1% quantity first obtains one
A naive model, the picture for then further taking out 3% are predicted using naive model, are entangled manually to the result of prediction error
Just, it is then trained again using above-mentioned 1% and 3% whole pictures, obtains the model that precision is promoted, successive iteration obtains
The pinpoint accuracy model that can classify automatically to the end.
9. a kind of side parking detection identifying system according to claim 7, which is characterized in that the character recognition algorithm
The step of it is as follows:
(1) Character segmentation
The rectangle segment of single character is intercepted from real license plate picture, steps are as follows for specific algorithm:
A. color judgement is carried out using the template matching method in hsv color space；
B. Otsu threshold method is carried out using different parameters according to blue board, yellow card, green board different colours judging result and carries out two-value
Change；
C. to the independent contouring of Chinese character in license plate, English and numerical character unify contouring, the character after being divided
Segment.
(2) character differentiates
Using the machine learning method of ANN, the character segment after segmentation is identified, obtains single character result, and group
Conjunction obtains the final result of Car license recognition.The neuron number of the input layer of ANN model, hidden layer and output layer is respectively 120,
40 and 65.
10. a kind of carry out side parking detection knowledge method for distinguishing using any one of the claim 1-9 system, feature exists
In, method includes the following steps:
(1) magnetic field data of tri- axis direction of X, Y, Z under the rectangular coordinate system in space of side parking stall is acquired by geomagnetism detecting module,
And acquisition data are sent to MCU；
(2) MCU receive geomagnetism detecting module send magnetic field data, by finite state machine and threshold detection algorithm obtain whether there is or not
The signal that vehicle drives into, when having detected that vehicle drives into, driving camera module work；
(3) after camera module activation, the headstock license plate picture of vehicle is driven into shooting, and is sent to MCU；
(4) MCU receives the headstock license plate picture of camera module acquisition, is transmitted to clothes by NB-IoT module with stream socket
Business device；
(5) license plate that server is restored to the data flow of headstock license plate picture, and identified using Recognition Algorithm of License Plate
As a result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910466245.6A CN110164139B (en) | 2019-05-30 | 2019-05-30 | System and method for detecting and identifying side parking |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910466245.6A CN110164139B (en) | 2019-05-30 | 2019-05-30 | System and method for detecting and identifying side parking |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110164139A true CN110164139A (en) | 2019-08-23 |
| CN110164139B CN110164139B (en) | 2021-08-03 |
Family
ID=67630257
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910466245.6A Active CN110164139B (en) | 2019-05-30 | 2019-05-30 | System and method for detecting and identifying side parking |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110164139B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110969773A (en) * | 2019-12-12 | 2020-04-07 | 中智华清（北京）科技有限公司 | NB-IoT based shared parking stall lock system and method |
| CN111508269A (en) * | 2020-04-23 | 2020-08-07 | 深圳智优停科技有限公司 | Open type parking space vehicle distinguishing method and device based on image recognition |
| CN112037534A (en) * | 2020-09-04 | 2020-12-04 | 深圳华强技术有限公司 | Car detection method and device integrating geomagnetic data and NB (nuclear magnetic resonance) signals |
| CN114155721A (en) * | 2022-02-07 | 2022-03-08 | 北京图盟科技有限公司 | Data conversion method and device for geomagnetic vehicle detector |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20090106201A (en) * | 2008-04-04 | 2009-10-08 | (주)에스엔알 | System and method for managing the admittance to parking place based on geomagnetic sensor |
| CN102722997A (en) * | 2011-12-31 | 2012-10-10 | 北京时代凌宇科技有限公司 | Parking space detection method and system thereof |
| CN106845480A (en) * | 2017-01-13 | 2017-06-13 | 河海大学 | A kind of method that car plate is recognized from picture |
| CN107622672A (en) * | 2017-09-19 | 2018-01-23 | 智慧互通科技有限公司 | Roadside Parking management equipment and method based on earth magnetism and video camera array linkage |
| CN207115756U (en) * | 2017-08-29 | 2018-03-16 | 厦门华方软件科技有限公司 | Parking position monitoring device |
| CN208460200U (en) * | 2018-08-13 | 2019-02-01 | 桂林航天工业学院 | Wireless geomagnetism parking space intelligent awareness apparatus based on narrowband Internet of Things |
-
2019
- 2019-05-30 CN CN201910466245.6A patent/CN110164139B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20090106201A (en) * | 2008-04-04 | 2009-10-08 | (주)에스엔알 | System and method for managing the admittance to parking place based on geomagnetic sensor |
| CN102722997A (en) * | 2011-12-31 | 2012-10-10 | 北京时代凌宇科技有限公司 | Parking space detection method and system thereof |
| CN106845480A (en) * | 2017-01-13 | 2017-06-13 | 河海大学 | A kind of method that car plate is recognized from picture |
| CN207115756U (en) * | 2017-08-29 | 2018-03-16 | 厦门华方软件科技有限公司 | Parking position monitoring device |
| CN107622672A (en) * | 2017-09-19 | 2018-01-23 | 智慧互通科技有限公司 | Roadside Parking management equipment and method based on earth magnetism and video camera array linkage |
| CN208460200U (en) * | 2018-08-13 | 2019-02-01 | 桂林航天工业学院 | Wireless geomagnetism parking space intelligent awareness apparatus based on narrowband Internet of Things |
Non-Patent Citations (1)
| Title |
|---|
| 马兰芳、张红霞等: "基于地磁传感器的车辆检测算法研究", 《自动化仪表》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110969773A (en) * | 2019-12-12 | 2020-04-07 | 中智华清（北京）科技有限公司 | NB-IoT based shared parking stall lock system and method |
| CN111508269A (en) * | 2020-04-23 | 2020-08-07 | 深圳智优停科技有限公司 | Open type parking space vehicle distinguishing method and device based on image recognition |
| CN111508269B (en) * | 2020-04-23 | 2021-05-18 | 深圳智优停科技有限公司 | Open type parking space vehicle distinguishing method and device based on image recognition |
| CN112037534A (en) * | 2020-09-04 | 2020-12-04 | 深圳华强技术有限公司 | Car detection method and device integrating geomagnetic data and NB (nuclear magnetic resonance) signals |
| CN114155721A (en) * | 2022-02-07 | 2022-03-08 | 北京图盟科技有限公司 | Data conversion method and device for geomagnetic vehicle detector |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110164139B (en) | 2021-08-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110164139A (en) | A kind of side parking detection identifying system and method | |
| CN106845890B (en) | Storage monitoring method and device based on video monitoring | |
| CN106548182B (en) | Pavement crack detection method and device based on deep learning and main cause analysis | |
| CN112380952A (en) | Power equipment infrared image real-time detection and identification method based on artificial intelligence | |
| CN102999753B (en) | License plate locating method | |
| CN108121996A (en) | A kind of hard pressing plate state identification method of power screen cabinet based on machine vision | |
| CN107016664B (en) | A kind of bad needle flaw detection method of large circle machine | |
| CN107203990A (en) | A kind of labeling damage testing method based on template matches and image quality measure | |
| CN103605977A (en) | Extracting method of lane line and device thereof | |
| CN105572541A (en) | High-voltage line patrol fault detection method and system based on visual attention mechanism | |
| CN107506765B (en) | License plate inclination correction method based on neural network | |
| CN104766086A (en) | Supervising method and system of way mark | |
| CN108701224A (en) | Visual vehicle parking take sensor | |
| CN109871789A (en) | Vehicle checking method under a kind of complex environment based on lightweight neural network | |
| CN108416774A (en) | A kind of fabric types recognition methods based on fine granularity neural network | |
| CN109191255B (en) | Commodity alignment method based on unsupervised feature point detection | |
| CN102542244A (en) | Face detection method and system and computer program product | |
| CN111242002B (en) | Shared bicycle standardized parking judgment method based on computer vision | |
| CN105550710A (en) | Nonlinear fitting based intelligent detection method for running exception state of contact network | |
| CN107240112A (en) | Individual X Angular Point Extracting Methods under a kind of complex scene | |
| CN101320520B (en) | Vehicle detection method and equipment thereof | |
| CN109271904A (en) | A kind of black smoke vehicle detection method based on pixel adaptivenon-uniform sampling and Bayesian model | |
| CN115797811A (en) | Agricultural product detection method and system based on vision | |
| CN108596244A (en) | A kind of high spectrum image label noise detecting method based on spectrum angle density peaks | |
| CN111079744B (en) | Intelligent vehicle license plate identification method and device suitable for complex illumination environment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |