CN105893953A - Method and system for detecting two license plates of one vehicle - Google Patents
Method and system for detecting two license plates of one vehicle Download PDFInfo
- Publication number
- CN105893953A CN105893953A CN201610191818.5A CN201610191818A CN105893953A CN 105893953 A CN105893953 A CN 105893953A CN 201610191818 A CN201610191818 A CN 201610191818A CN 105893953 A CN105893953 A CN 105893953A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- information
- picture
- car
- board
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/59—Context or environment of the image inside of a vehicle, e.g. relating to seat occupancy, driver state or inner lighting conditions
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
A method for detecting two license plates of one vehicle includes the steps of determining two license plates of one vehicle, supplementing evidences and associating information. Capturing and recording the front and rear license plates of a same vehicle by one camera is realized, thereby preventing disputes. The vehicle head description information and driver information supplementation can be realized by associating and comparing the license plate information of the vehicle head with the vehicle tail, or the vehicle tail description information can be supplemented. More effective clue information can be provided, thereby facilitating arresting work.
Description
Technical field
The invention belongs to field of license plate recognition, particularly relate to an a kind of car two board vehicle checking method
And system.
Background technology
At present, in traffic intelligent management and control, the electric alert or unidirectional management and control of bayonet socket that many dependences are single.When
Occur lawless person by single replacement, block, damage, during the car plate escape of headstock or the tailstock,
By the tracking of interference escape route, bring extreme difficulties for arresting work.
Traditional vehicular traffic recording equipment cannot be with an Intelligent candid camera simultaneously to same
Before and after car, car plate carries out capturing and record, causes the imperfection of law enforcement evidence, produces dispute.
Summary of the invention
Based on this, for above-mentioned technical problem, it is provided that an a kind of car two board vehicle checking method and be
System.
For solving above-mentioned technical problem, the present invention adopts the following technical scheme that
An a kind of car two board vehicle checking method, including:
One car two board vehicle determines step: by being arranged vertically within above passing road and having wide-angle lens
The headstock picture of the first video capture video capture vehicle and tailstock picture, from described headstock picture
And tailstock picture identifies license board information, and the license board information of the headstock identified and the tailstock is carried out
Comparison, determines a car two board vehicle；
Evidence supplements step: grabbed by diagonally disposed the second video capture video camera above passing road
Clap headstock picture or the tailstock picture of vehicle, from described headstock picture, identify that vehicle describes information and drives
The person's of sailing face scratches figure information, or, from described tailstock picture, identify that vehicle describes information, described second
Video capture video camera and ground are 40-50 ° of angle, and described vehicle describes information and includes license board information, car
Mark information, vehicle information and body color information；
Information association step: a described car two board vehicle is determined the described car two board car that step determines
License board information and described evidence supplement the license board information that step obtains and compare, if having identical
From the license board information of headstock picture, then association remaining vehicle corresponding describes information and driver's face is scratched
Figure information is to a described car two board vehicle；If having the identical license board information from tailstock picture, then close
Join remaining vehicle and describe information to a described car two board vehicle.
A described car two board vehicle determines that step farther includes:
First object detects: by described first video capture video camera to the car through predetermined detection area
Detect；
First object is followed the tracks of: the vehicle detected is carried out tracing of the movement；
First object is captured: by the headstock picture of described first video capture video capture vehicle and
Tailstock picture；
First object identification: to described headstock picture and tailstock picture, advanced driving board target detection obtains
Pick up the car the board band of position, in this car plate band of position, then carry out Character segmentation, finally to segmentation after
Character carries out character recognition, obtains headstock and the license board information of the tailstock；
Target comparison: compare the license board information of the headstock identified and the tailstock, determines a car two
Board vehicle.
Described evidence supplements step and farther includes:
Second target detection: by described second video capture video camera to the car through predetermined detection area
Detecting, this second video capture video camera and ground are 45° angle；
Second target following: the vehicle detected is carried out tracing of the movement；
Second target is captured: by headstock picture or the car of described second video capture video capture vehicle
Tail picture；
Second target recognition: described headstock picture or tailstock picture are carried out image procossing, obtains described car
Head picture or tailstock picture include the characteristic information of texture, gray scale and direction gradient, and by described spy
Reference breath mates with the target sample model pre-build, identifies vehicle and describes information；
Face scratches figure information identification: from upper 1/2 region of described headstock picture, enter with the form of sliding window
Row multi-scale sliding window mouth detects, and includes the spy of texture, gray scale and direction gradient in extracting described window
Reference ceases, it is thus achieved that driver's face scratches figure information.
Described first object detecting step and the second target detection step all use the target of Model Matching
Detection algorithm, algorithm of target detection based on context or algorithm of target detection based on background modeling,
Described first object tracking step and the second target following step all use track algorithm based on angle point,
The target tracking algorism of feature based coupling or track algorithm based on multi-tool plate coupling.
This programme further relates to an a kind of car two board vehicle detecting system, including:
One car two board vehicle determines module, for by being arranged vertically within above passing road and having Radix Rumicis
The headstock picture of the first video capture video capture vehicle of camera lens and tailstock picture, from described headstock
Picture and tailstock picture identify license board information, and to the headstock identified and the license board information of the tailstock
Compare, determine a car two board vehicle；
Evidence complementary module, for imaging by diagonally disposed the second video capture above passing road
Machine captures the headstock picture of vehicle or tailstock picture, identify from described headstock picture vehicle describe information with
And driver's face scratches figure information, or, from described tailstock picture, identify that vehicle describes information, described
Second video capture video camera and ground are 40-50 ° of angle, described vehicle describe information include license board information,
Car mark information, vehicle information and body color information；
Information association module, determines the described car two board car that step determines by a described car two board vehicle
License board information and described evidence supplement the license board information that step obtains and compare, if having identical
From the license board information of headstock picture, then association remaining vehicle corresponding describes information and driver's face is scratched
Figure information is to a described car two board vehicle；If having the identical license board information from tailstock picture, then close
Join remaining vehicle and describe information to a described car two board vehicle.
A described car two board vehicle determines that module is built in described first video capture video camera, and it enters one
Step includes:
First object detector unit, for detecting through default by described first video capture video camera
The vehicle in region detects；
First object tracking cell, for carrying out tracing of the movement to the vehicle detected；
First object captures unit, for by the headstock of described first video capture video capture vehicle
Picture and tailstock picture；
First object recognition unit, for described headstock picture and tailstock picture, advanced person's driving board mesh
Mark detection obtains the car plate band of position, then carries out Character segmentation in this car plate band of position, the most right
Character after segmentation carries out character recognition, obtains headstock and the license board information of the tailstock；
Target comparing unit, for comparing, really to the license board information of the headstock identified and the tailstock
A fixed car two board vehicle.
Described evidence complementary module is built in described second video capture video camera, and it farther includes:
Second object detection unit, for detecting through default by described second video capture video camera
The vehicle in region detects, and this second video capture video camera and ground are 45° angle；
Second target tracking unit, for carrying out tracing of the movement to the vehicle detected；
Second target captures unit, for by the headstock of described second video capture video capture vehicle
Picture or tailstock picture；
Second object-recognition unit, carries out image procossing to described headstock picture or tailstock picture, obtains institute
State headstock picture or tailstock picture includes the characteristic information of texture, gray scale and direction gradient, and by institute
The target sample model stating characteristic information and pre-build mates, and identifies vehicle and describes information；
Face scratches figure information identificating unit, for from upper 1/2 region of described headstock picture, with sliding window
Form carry out multi-scale sliding window mouth detection, include texture, gray scale and direction in extracting described window
The characteristic information of gradient, it is thus achieved that driver's face scratches figure information.
Described first object detector unit and the second object detection unit all use the target of Model Matching
Detection algorithm, algorithm of target detection based on context or algorithm of target detection based on background modeling,
Described first object tracking cell and the second target tracking unit all use track algorithm based on angle point,
The target tracking algorism of feature based coupling or track algorithm based on multi-tool plate coupling.
The present invention can realize being grabbed car plate before and after same car by a candid camera simultaneously
Clap and record, it is to avoid the generation of dispute, and by headstock or the association comparison of the vehicle board information of the tailstock,
Achieve and headstock vehicle is described information and driver information supplements, or tailstock vehicle describes the benefit of information
Fill, it is provided that more effective hint information, bring convenience for arresting work.
Accompanying drawing explanation
It is described in detail with the detailed description of the invention present invention below in conjunction with the accompanying drawings:
Fig. 1 is the schematic diagram of a kind of car two board vehicle checking method of the present invention；
Fig. 2 is the flow chart of a kind of car two board vehicle checking method of the present invention；
Fig. 3 is the structural representation of a kind of car two board vehicle detecting system of the present invention；
Fig. 4 is that schematic diagram captured by a kind of camera of the present invention；
Fig. 5 is that schematic diagram captured by the another kind of camera of the present invention.
Detailed description of the invention
As shown in Fig. 1 and 2, an a kind of car two board vehicle checking method, including:
A, a car two board vehicle determine step:
As shown in FIG. 4 and 5, by being arranged vertically within above passing road and there is wide-angle lens
Headstock picture and the tailstock picture of vehicle 5 captured by first video capture video camera 3, from headstock picture with
And tailstock picture identifies license board information, and the license board information of the headstock identified and the tailstock is compared
Right, determine a car two board vehicle.
Wide-angle lens is installed downwards, and the road of vehicle pass-through is carried out round-the-clock monitoring, and wide-angle lens is found a view
In scene, the overall process through vehicle 5 can be clearly apparent, and headstock car plate and tailstock license board information,
Can realize car plate before and after same car being captured and record by a candid camera simultaneously, keep away
Exempt from the generation of dispute.
Specifically, a car two board vehicle determines that step A farther includes:
A01, first object detect: by the first video capture video camera 3 to through predetermined detection area
Vehicle detect.
A02, first object are followed the tracks of: the vehicle detected is carried out tracing of the movement, follow-up to guarantee
Target Photo capture effect.
A03, first object are captured: captured the headstock picture of vehicle by the first video capture video camera 3
And tailstock picture, headstock picture and tailstock picture are close up view, facilitate in succeeding target identification
Feature extraction.
A04, first object identification: to headstock picture and tailstock picture, advanced driving board target detection
Obtain the car plate band of position, in this car plate band of position, then carry out Character segmentation, finally to segmentation after
Character carry out character recognition, obtain headstock and the license board information of the tailstock.
Wherein, the detection of car plate position can be selected for detection algorithm based on edge, or target based on ACF inspection
Method of determining and calculating, character recognition can be selected for machine learning algorithm, such as Adboost, SVM etc..
A05, target comparison: the license board information of the headstock identified and the tailstock is compared, if car
The license board information of head and the tailstock is inconsistent, then can determine that a car two board vehicle.
B, evidence supplement step:
As shown in FIG. 4 and 5, for being grabbed by diagonally disposed the second video above passing road
Shooting camera 4 captures headstock picture or the tailstock picture of vehicle 5, due to headstock may face or facing away from
Second video capture video camera 4, therefore candid photograph is headstock picture or tailstock picture.
As shown in Figure 4, candid photograph be headstock picture time, from headstock picture identify vehicle describe information with
And driver's face scratches figure information.
Or, as it is shown in figure 5, capture be the tailstock picture time, from tailstock picture identify vehicle describe
Information.
Wherein, vehicle describes information and includes license board information, car mark information, vehicle information and vehicle body equally
Colouring information, the second video capture video camera 4 is 40-50 ° of angle with ground, the present embodiment preferably 45 °
Angle.
Specifically, evidence supplements step B and farther includes:
B01, the second target detection: by the second video capture video camera 4 to through predetermined detection area
Vehicle detect；
B02, the second target following: the vehicle detected is carried out tracing of the movement, follow-up to guarantee
Target Photo capture effect.
B03, the second target are captured: captured the headstock picture of vehicle by the second video capture video camera 4
Or tailstock picture, headstock picture and tailstock picture are close up view, facilitate the feature in succeeding target identification
Extract.
B04, the second target recognition: headstock picture and tailstock picture carry out image procossing, obtain car
Head picture or tailstock picture include the characteristic information of texture, gray scale and direction gradient, and feature are believed
Breath mates with the target sample model pre-build, identifies vehicle and describes information, i.e. from headstock figure
Sheet identifying, vehicle describes information, or, from tailstock picture, identify that vehicle describes information；
B05, face scratch figure information identification: from upper 1/2 region of headstock picture, with the form of sliding window
Carry out multi-scale sliding window mouth detection, in extracting window, include the feature of texture, gray scale and direction gradient
Information, it is thus achieved that driver's face scratches figure information.
C, information association step:
One car two board vehicle is determined, and the license board information of the car two board vehicle that step A determines is mended with evidence
The license board information filling the acquisition of step B is compared, if having the identical license board information from headstock picture,
Then association remaining vehicle corresponding describes information and driver's face scratches figure information to a described car two board car
?.
If having the identical license board information from tailstock picture, then association remaining vehicle corresponding describes letter
Breath is to a car two board vehicle.
Remaining vehicle above-mentioned describes information and refers to car mark information, vehicle information and body color information.
Above-mentioned first object detecting step and the second target detection step all can use the mesh of Model Matching
Mark detection algorithm, algorithm of target detection based on context or target detection based on background modeling are calculated
Method.
Algorithm of target detection based on Model Matching, the sample number of detection initial stage collection classification target to be detected
According to, training sample data, extract sample image characteristic information (such as: texture, direction gradient, gray scale etc.),
Obtain sample pattern, detected the video image hit exactly obtaining, and carried out characteristic processing based on image,
Extract the image feature information with Model Matching, and by the information obtained and sample pattern Data Matching, defeated
Go out the result that matching degree is high, finally realize the detection to moving object in video sequences.Former based on this algorithm
The classical detection method of reason has: HOG, SVM, Harr etc..
Target detection based on context, utilizes the relatedness between video frame image, extracts adjacent two frames
Or the relatedness such as geometry between multi-frame video image, position, it is achieved the detection to moving target.
In target detection based on background modeling, i.e. video data processing procedure, first build detection region
Background model, then the video image photographed is done with this background model frame by frame background subtraction, and then detection
Go out target prospect.The method background modeling algorithms most in use has background modeling based on Gaussian modeling, or
Background modeling based on Bayes Modeling.
Above-mentioned first object detector unit 111 and the second object detection unit 121 all can use model
The algorithm of target detection joined, algorithm of target detection based on context or target based on background modeling inspection
Method of determining and calculating.
Track algorithm based on angle point, is on the basis of moving target being detected, extracts target critical special
Levy a little, in video image, realized the tracking of target by target characteristic Point matching.This algorithm is commonly used
Classical way be Mean Shift algorithm.
Track algorithm based on target characteristic coupling, by extracting the rigidity characteristic following the tracks of target, such as target
The features such as gray level image, binary segmentation image, marginal point, angle point, color histogram, carry out target with
Track.
Based on multi-tool plate matched jamming method, the method is based on multi-objective Model, during following the tracks of,
Judge target trajectory by matching error, in conjunction with intensity correlation matching realize target long lasting for
Track.
As it is shown on figure 3, the invention still further relates to an a kind of car two board vehicle detecting system, including:
As shown in FIG. 4 and 5, a car two board vehicle determines module 110, for by being arranged vertically
Above passing road and there is the first video capture video camera 3 of wide-angle lens capture the headstock of vehicle 5
Picture and tailstock picture, identify license board information from headstock picture and tailstock picture, and to identifying
Headstock and the license board information of the tailstock compare, determine a car two board vehicle.
Wide-angle lens is installed downwards, and the road of vehicle pass-through is carried out round-the-clock monitoring, and wide-angle lens is found a view
In scene, the overall process through vehicle 5 can be clearly apparent, and headstock car plate and tailstock license board information,
Can realize car plate before and after same car being captured and record by a candid camera simultaneously, keep away
Exempt from the generation of dispute.
Specifically, a car two board vehicle determines that module 110 is built in the first video capture video camera 3,
It farther includes:
First object detector unit 111, for examining through default by the first video capture video camera 3
The vehicle surveying region detects.
First object tracking cell 112, for carrying out tracing of the movement to the vehicle detected, with really
Protect follow-up Target Photo and capture effect.
First object captures unit 113, for being captured the car of vehicle by the first video capture video camera 3
Head picture and tailstock picture, headstock picture and tailstock picture are close up view, facilitate succeeding target to know
Feature extraction in not.
First object recognition unit 114, for headstock picture and tailstock picture, advanced person's driving board mesh
Mark detection obtains the car plate band of position, then carries out Character segmentation in this car plate band of position, the most right
Character after segmentation carries out character recognition, obtains headstock and the license board information of the tailstock.
Wherein, the detection of car plate position can be selected for detection algorithm based on edge, or target based on ACF inspection
Method of determining and calculating, character recognition can be selected for machine learning algorithm, such as Adboost, SVM etc..
Target comparing unit 115, for the license board information of the headstock identified and the tailstock is compared,
If the license board information of headstock and the tailstock is inconsistent, then can determine that a car two board vehicle.
Evidence complementary module 120, for by diagonally disposed the second video capture above passing road
Headstock picture or the tailstock picture of vehicle 5 captured by video camera 4, owing to headstock may face or facing away from
Two video capture video cameras 4, therefore candid photograph is headstock picture or tailstock picture.
As shown in Figure 4, candid photograph be headstock picture time, from headstock picture identify vehicle describe information with
And driver's face scratches figure information.
Or, as it is shown in figure 5, capture be the tailstock picture time, from tailstock picture identify vehicle describe
Information.
Wherein, vehicle describes information and includes license board information, car mark information, vehicle information and vehicle body equally
Colouring information, the second video capture video camera 4 is 40-50 ° of angle with ground, the present embodiment preferably 45 °
Angle.
Specifically, evidence complementary module 120 is built in the second video capture video camera 4, and it is further
Including:
Second object detection unit 121, for examining through default by the second video capture video camera 4
The vehicle surveying region detects；
Second target tracking unit 122, for carrying out tracing of the movement to the vehicle detected, with really
Protect follow-up Target Photo and capture effect.
Second target captures unit 123, for being captured the car of vehicle by the second video capture video camera 4
Head picture or tailstock picture, headstock picture and tailstock picture are close up view, facilitate in succeeding target identification
Feature extraction.
Second object-recognition unit 124, for headstock picture or tailstock picture are carried out image procossing, obtains
Pick up the car a picture or tailstock picture includes the characteristic information of texture, gray scale and direction gradient, and by spy
Reference breath mates with the target sample model pre-build, identifies vehicle and describes information, i.e. from car
Head picture identifies that vehicle describes information, or, from tailstock picture, identify that vehicle describes information；
Face scratches figure information identificating unit 125, for from upper 1/2 region of headstock picture, with sliding window
Form carry out multi-scale sliding window mouth detection, include texture, gray scale and direction gradient in extracting window
Characteristic information, it is thus achieved that driver's face scratches figure information.
Information association module 130, for determining the car two board car that step A determines by a car two board vehicle
License board information and evidence supplement the license board information that step B obtains and compare, if having identical coming
From the license board information of headstock picture, then association remaining vehicle corresponding describes information and driver's face scratches figure
Information is to a described car two board vehicle.
If having the identical license board information from tailstock picture, then association remaining vehicle corresponding describes letter
Breath is to a car two board vehicle.
Remaining vehicle above-mentioned describes information and refers to car mark information, vehicle information and body color information.
Above-mentioned first object detector unit 111 and the second object detection unit 121 all can use model
The algorithm of target detection joined, algorithm of target detection based on context or target based on background modeling inspection
Method of determining and calculating.
Algorithm of target detection based on Model Matching, the sample number of detection initial stage collection classification target to be detected
According to, training sample data, extract sample image characteristic information (such as: texture, direction gradient, gray scale etc.),
Obtain sample pattern, detected the video image hit exactly obtaining, and carried out characteristic processing based on image,
Extract the image feature information with Model Matching, and by the information obtained and sample pattern Data Matching, defeated
Go out the result that matching degree is high, finally realize the detection to moving object in video sequences.Former based on this algorithm
The classical detection method of reason has: HOG, SVM, Harr etc..
Target detection based on context, utilizes the relatedness between video frame image, extracts adjacent two frames
Or the relatedness such as geometry between multi-frame video image, position, it is achieved the detection to moving target.
In target detection based on background modeling, i.e. video data processing procedure, first build detection region
Background model, then the video image photographed is done with this background model frame by frame background subtraction, and then detection
Go out target prospect.The method background modeling algorithms most in use has background modeling based on Gaussian modeling, or
Background modeling based on Bayes Modeling.
Above-mentioned first object tracking cell 112 and the second target tracking unit 122 all can use based on angle
The target tracking algorism of the track algorithm of point, feature based coupling or tracking based on multi-tool plate coupling
Algorithm.
Track algorithm based on angle point, is on the basis of moving target being detected, extracts target critical special
Levy a little, in video image, realized the tracking of target by target characteristic Point matching.This algorithm is commonly used
Classical way be Mean Shift algorithm.
Track algorithm based on target characteristic coupling, by extracting the rigidity characteristic following the tracks of target, such as target
The features such as gray level image, binary segmentation image, marginal point, angle point, color histogram, carry out target with
Track.
Based on multi-tool plate matched jamming method, the method is based on multi-objective Model, during following the tracks of,
Judge target trajectory by matching error, in conjunction with intensity correlation matching realize target long lasting for
Track.
But, those of ordinary skill in the art is it should be appreciated that above embodiment is only to use
The present invention is described, and is not used as limitation of the invention, as long as at the connotation model of the present invention
In enclosing, change, the modification of embodiment described above all will be fallen in the range of claims of the present invention.
Claims (8)
1. a car two board vehicle checking method, it is characterised in that including:
One car two board vehicle determines step: by being arranged vertically within above passing road and having wide-angle lens
The headstock picture of the first video capture video capture vehicle of head and tailstock picture, from described headstock
Picture and tailstock picture identify license board information, and the car plate of the headstock identified and the tailstock is believed
Breath is compared, and determines a car two board vehicle；
Evidence supplements step: by diagonally disposed the second video capture video camera above passing road
Capture the headstock picture of vehicle or tailstock picture, identify from described headstock picture vehicle describe information with
And driver's face scratches figure information, or, from described tailstock picture, identify that vehicle describes information, institute
Stating the second video capture video camera with ground is 40-50 ° of angle, and described vehicle describes information and includes car plate
Information, car mark information, vehicle information and body color information；
Information association step: a described car two board vehicle is determined described car two board that step determines
The license board information of vehicle and described evidence supplement the license board information of step acquisition and compare, if having
The identical license board information from headstock picture, then association remaining vehicle corresponding describes information and drives
The person's of sailing face scratches figure information to a described car two board vehicle；If having identical from tailstock picture
License board information, then associate remaining vehicle and describe information to a described car two board vehicle.
An a kind of car two board vehicle checking method the most according to claim 1, it is characterised in that
A described car two board vehicle determines that step farther includes:
First object detects: by described first video capture video camera to through predetermined detection area
Vehicle detects；
First object is followed the tracks of: the vehicle detected is carried out tracing of the movement；
First object capture: by the headstock picture of described first video capture video capture vehicle with
And tailstock picture；
First object identification: to described headstock picture and tailstock picture, advanced driving board target detection
Obtain the car plate band of position, then in this car plate band of position, carry out Character segmentation, finally to segmentation
After character carry out character recognition, obtain headstock and the license board information of the tailstock；
Wherein, the detection of car plate position can be selected for detection algorithm based on edge, or target based on ACF
Detection algorithm, character recognition can be selected for machine learning algorithm, such as Adboost, SVM etc..
Target comparison: compare the license board information of the headstock identified and the tailstock, determines a car
Two board vehicles.
An a kind of car two board vehicle checking method the most according to claim 2, it is characterised in that
Described evidence supplements step and farther includes:
Second target detection: by described second video capture video camera to through predetermined detection area
Vehicle detects, and this second video capture video camera and ground are 45° angle；
Second target following: the vehicle detected is carried out tracing of the movement；
Second target is captured: by the headstock picture of described second video capture video capture vehicle or
Tailstock picture；
Second target recognition: described headstock picture or tailstock picture carry out image procossing, obtains described
Headstock picture or tailstock picture include the characteristic information of texture, gray scale and direction gradient, and by institute
The target sample model stating characteristic information and pre-build mates, and identifies vehicle and describes information；
From described headstock picture, i.e. identify that vehicle describes information, or, from described tailstock picture, identify car
Describe information；
Face scratches figure information identification: from upper 1/2 region of described headstock picture, with the form of sliding window
Carry out multi-scale sliding window mouth detection, in extracting described window, include texture, gray scale and direction gradient
Characteristic information, it is thus achieved that driver's face scratches figure information.
4., according to the one one car two board vehicle checking method described in Claims 2 or 3, its feature exists
In, described first object detecting step and the second target detection step all use the target of Model Matching
Detection algorithm, algorithm of target detection based on context or algorithm of target detection based on background modeling,
Described first object tracking step and the second target following step all use tracking based on angle point to calculate
The target tracking algorism of method, feature based coupling or track algorithm based on multi-tool plate coupling.
5. a car two board vehicle detecting system, it is characterised in that including:
One car two board vehicle determines module, for by being arranged vertically within above passing road and having wide
The headstock picture of the first video capture video capture vehicle of angle mirror head and tailstock picture, from described
Headstock picture and tailstock picture identify license board information, and to the headstock identified and the car of the tailstock
Board information is compared, and determines a car two board vehicle；
Evidence complementary module, for taking the photograph by diagonally disposed the second video capture above passing road
Camera captures headstock picture or the tailstock picture of vehicle, identifies that vehicle describes letter from described headstock picture
Breath and driver's face scratch figure information, or, from described tailstock picture, identify that vehicle describes information,
Described second video capture video camera and ground are 40-50 ° of angle, and described vehicle describes information and includes car
Board information, car mark information, vehicle information and body color information；
Information association module, determines the described car two board car that step determines by a described car two board vehicle
License board information and described evidence supplement the license board information that step obtains and compare, if having identical
The license board information from headstock picture, then association remaining vehicle corresponding describes information and driver people
Face scratches figure information to a described car two board vehicle；If having the identical license board information from tailstock picture,
Then associate remaining vehicle and describe information to a described car two board vehicle.
An a kind of car two board vehicle detecting system the most according to claim 5, it is characterised in that
A described car two board vehicle determines that module is built in described first video capture video camera, and it is further
Including:
First object detector unit, for examining through default by described first video capture video camera
The vehicle surveying region detects；
First object tracking cell, for carrying out tracing of the movement to the vehicle detected；
First object captures unit, for by the car of described first video capture video capture vehicle
Head picture and tailstock picture；
First object recognition unit, for described headstock picture and tailstock picture, advanced person's driving board
Target detection obtains the car plate band of position, then carries out Character segmentation in this car plate band of position,
Afterwards the character after segmentation is carried out character recognition, obtain headstock and the license board information of the tailstock；
Target comparing unit, for the license board information of the headstock identified and the tailstock is compared,
Determine a car two board vehicle.
An a kind of car two board vehicle detecting system the most according to claim 6, it is characterised in that
Described evidence complementary module is built in described second video capture video camera, and it farther includes:
Second object detection unit, for examining through default by described second video capture video camera
The vehicle surveying region detects, and this second video capture video camera and ground are 45° angle；
Second target tracking unit, for carrying out tracing of the movement to the vehicle detected；
Second target captures unit, for by the car of described second video capture video capture vehicle
Head picture or tailstock picture；
Second object-recognition unit, carries out image procossing to described headstock picture or tailstock picture, obtains
Described headstock picture or tailstock picture include the characteristic information of texture, gray scale and direction gradient, and
Described characteristic information is mated with the target sample model pre-build, identifies vehicle and describe letter
Breath；
Face scratches figure information identificating unit, for from upper 1/2 region of described headstock picture, to slide
The form of window carries out multi-scale sliding window mouth detection, include in extracting described window texture, gray scale and
The characteristic information of direction gradient, it is thus achieved that driver's face scratches figure information.
8., according to the one one car two board vehicle detecting system described in claim 6 or 7, its feature exists
In, described first object detector unit and the second object detection unit all use the target of Model Matching
Detection algorithm, algorithm of target detection based on context or algorithm of target detection based on background modeling,
Described first object tracking cell and the second target tracking unit all use tracking based on angle point to calculate
The target tracking algorism of method, feature based coupling or track algorithm based on multi-tool plate coupling.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610191818.5A CN105893953A (en) | 2016-03-30 | 2016-03-30 | Method and system for detecting two license plates of one vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610191818.5A CN105893953A (en) | 2016-03-30 | 2016-03-30 | Method and system for detecting two license plates of one vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105893953A true CN105893953A (en) | 2016-08-24 |
Family
ID=57014715
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610191818.5A Pending CN105893953A (en) | 2016-03-30 | 2016-03-30 | Method and system for detecting two license plates of one vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105893953A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106124533A (en) * | 2016-08-30 | 2016-11-16 | 北京华力兴科技发展有限责任公司 | Rapid detection system |
| CN108269404A (en) * | 2016-12-30 | 2018-07-10 | 上海泓鎏智能科技有限公司 | Number plate consistent verification device and verification method before and after vehicle |
| CN108694399A (en) * | 2017-04-07 | 2018-10-23 | 杭州海康威视数字技术股份有限公司 | Licence plate recognition method, apparatus and system |
| CN109243182A (en) * | 2018-10-19 | 2019-01-18 | 天津天地基业科技有限公司 | A kind of car plate detection system based on vertical candid camera |
| CN109285355A (en) * | 2018-10-19 | 2019-01-29 | 天津天地人和企业管理咨询有限公司 | A kind of front and back candid photograph traffic cameras system |
| CN110058991A (en) * | 2018-11-30 | 2019-07-26 | 阿里巴巴集团控股有限公司 | A kind of automatic test approach and system of application software |
| CN111081031A (en) * | 2019-12-26 | 2020-04-28 | 北京万集科技股份有限公司 | Vehicle snapshot method and system |
| CN111767776A (en) * | 2019-12-28 | 2020-10-13 | 西安宇视信息科技有限公司 | Abnormal license plate selection method and device |
| CN112419740A (en) * | 2019-08-23 | 2021-02-26 | 杭州海康威视***技术有限公司 | Vehicle state identification method and device |
| CN112818744A (en) * | 2020-12-30 | 2021-05-18 | 青岛中兴智能交通有限公司 | Method and system for automatically correcting identification inconsistency of entering and exiting license plates |
| CN113343763A (en) * | 2021-05-08 | 2021-09-03 | 山西三友和智慧信息技术股份有限公司 | Vehicle identification system based on big data |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101540105A (en) * | 2009-04-15 | 2009-09-23 | 四川川大智胜软件股份有限公司 | Fake-licensed car detection method based on number-plate identification and gridding supervision |
| KR100991106B1 (en) * | 2010-02-23 | 2010-11-01 | 주식회사 넥스파시스템 | System and method for recognizing license plate using camera module photographing front and back face of vehicle |
| CN103730009A (en) * | 2012-10-12 | 2014-04-16 | 博隆建亚科技(北京)有限公司 | Method and system for detecting vehicle with one number plate missing |
| CN104036640A (en) * | 2014-05-16 | 2014-09-10 | 北京卓视智通科技有限责任公司 | Panoramic image acquisition device and panoramic image acquisition method |
| CN104200671A (en) * | 2014-09-09 | 2014-12-10 | 安徽四创电子股份有限公司 | Method and system for managing virtual gate based on big data platform |
-
2016
- 2016-03-30 CN CN201610191818.5A patent/CN105893953A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101540105A (en) * | 2009-04-15 | 2009-09-23 | 四川川大智胜软件股份有限公司 | Fake-licensed car detection method based on number-plate identification and gridding supervision |
| KR100991106B1 (en) * | 2010-02-23 | 2010-11-01 | 주식회사 넥스파시스템 | System and method for recognizing license plate using camera module photographing front and back face of vehicle |
| CN103730009A (en) * | 2012-10-12 | 2014-04-16 | 博隆建亚科技(北京)有限公司 | Method and system for detecting vehicle with one number plate missing |
| CN104036640A (en) * | 2014-05-16 | 2014-09-10 | 北京卓视智通科技有限责任公司 | Panoramic image acquisition device and panoramic image acquisition method |
| CN104200671A (en) * | 2014-09-09 | 2014-12-10 | 安徽四创电子股份有限公司 | Method and system for managing virtual gate based on big data platform |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106124533A (en) * | 2016-08-30 | 2016-11-16 | 北京华力兴科技发展有限责任公司 | Rapid detection system |
| CN108269404A (en) * | 2016-12-30 | 2018-07-10 | 上海泓鎏智能科技有限公司 | Number plate consistent verification device and verification method before and after vehicle |
| CN108694399B (en) * | 2017-04-07 | 2021-01-26 | 杭州海康威视数字技术股份有限公司 | License plate recognition method, device and system |
| CN108694399A (en) * | 2017-04-07 | 2018-10-23 | 杭州海康威视数字技术股份有限公司 | Licence plate recognition method, apparatus and system |
| CN109243182A (en) * | 2018-10-19 | 2019-01-18 | 天津天地基业科技有限公司 | A kind of car plate detection system based on vertical candid camera |
| CN109285355A (en) * | 2018-10-19 | 2019-01-29 | 天津天地人和企业管理咨询有限公司 | A kind of front and back candid photograph traffic cameras system |
| CN110058991A (en) * | 2018-11-30 | 2019-07-26 | 阿里巴巴集团控股有限公司 | A kind of automatic test approach and system of application software |
| CN112419740A (en) * | 2019-08-23 | 2021-02-26 | 杭州海康威视***技术有限公司 | Vehicle state identification method and device |
| CN112419740B (en) * | 2019-08-23 | 2022-07-08 | 杭州海康威视***技术有限公司 | Vehicle state identification method and device |
| CN111081031A (en) * | 2019-12-26 | 2020-04-28 | 北京万集科技股份有限公司 | Vehicle snapshot method and system |
| CN111081031B (en) * | 2019-12-26 | 2022-06-17 | 北京万集科技股份有限公司 | Vehicle snapshot method and system |
| CN111767776A (en) * | 2019-12-28 | 2020-10-13 | 西安宇视信息科技有限公司 | Abnormal license plate selection method and device |
| CN111767776B (en) * | 2019-12-28 | 2024-03-29 | 西安宇视信息科技有限公司 | Abnormal license plate selecting method and device |
| CN112818744A (en) * | 2020-12-30 | 2021-05-18 | 青岛中兴智能交通有限公司 | Method and system for automatically correcting identification inconsistency of entering and exiting license plates |
| CN113343763A (en) * | 2021-05-08 | 2021-09-03 | 山西三友和智慧信息技术股份有限公司 | Vehicle identification system based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105893953A (en) | Method and system for detecting two license plates of one vehicle | |
| Choudhury et al. | Vehicle detection and counting using haar feature-based classifier | |
| CN105654733B (en) | A kind of front and back licence plate recognition method and device based on video detection | |
| Davis et al. | A two-stage template approach to person detection in thermal imagery | |
| US8798314B2 (en) | Detection of vehicles in images of a night time scene | |
| CN101587544B (en) | Based on the carried on vehicle antitracking device of computer vision | |
| CN108288386A (en) | Road-surface concrete tracking based on video | |
| CN105740855A (en) | Front and rear license plate detection and recognition method based on deep learning | |
| KR101781358B1 (en) | Personal Identification System And Method By Face Recognition In Digital Image | |
| US10255511B2 (en) | Real time traffic sign recognition | |
| CN103310231A (en) | Auto logo locating and identifying method | |
| WO2024037408A1 (en) | Underground coal mine pedestrian detection method based on image fusion and feature enhancement | |
| Zhong et al. | Learning to tell brake lights with convolutional features | |
| CN104978746A (en) | Running vehicle body color identification method | |
| CN114639078A (en) | Vehicle type recognition method, device and system | |
| Ramezani et al. | A new DSWTS algorithm for real-time pedestrian detection in autonomous agricultural tractors as a computer vision system | |
| Wang et al. | Real-time vehicle classification based on eigenface | |
| JP2015046126A (en) | Vehicle detector | |
| CN109034171B (en) | Method and device for detecting unlicensed vehicles in video stream | |
| CN110796099A (en) | Vehicle overrun detection method and device | |
| CN109492513B (en) | Face space duplication eliminating method for light field monitoring | |
| CN110119769A (en) | A kind of detection method for early warning based on multi-modal vehicle characteristics | |
| Yang et al. | Bicyclist detection in large scale naturalistic driving video | |
| Wang et al. | The color identification of automobiles for video surveillance | |
| CN105206060A (en) | Vehicle type recognition device and method based on SIFT characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160824 |