CN109977941A - Licence plate recognition method and device - Google Patents
Licence plate recognition method and device Download PDFInfo
- Publication number
- CN109977941A CN109977941A CN201811573659.0A CN201811573659A CN109977941A CN 109977941 A CN109977941 A CN 109977941A CN 201811573659 A CN201811573659 A CN 201811573659A CN 109977941 A CN109977941 A CN 109977941A
- Authority
- CN
- China
- Prior art keywords
- license plate
- mentioned
- model
- color
- plate area
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/63—Scene text, e.g. street names
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/625—License plates
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of licence plate recognition method and devices.Wherein, this method comprises: receiving target image to be detected；License plate area in above-mentioned target image is positioned using target detection model for the first time, wherein include at least in above-mentioned license plate area: license plate；Secondary positioning is carried out to above-mentioned license plate area based on color of object model and left and right edges regression model；Content recognition is carried out to the license plate area after above-mentioned secondary positioning using target nerve network, obtains the license board information of above-mentioned license plate.The present invention solves the recognition accuracy of traditional licence plate recognition method identification characters on license plate and the technical problem that recognition efficiency is lower.
Description
Technical field
The present invention relates to identification technology fields, in particular to a kind of licence plate recognition method and device.
Background technique
Basic core technology of the Car license recognition as intelligent transportation system, car plate detection positioning are then the passes in Car license recognition
One of link of key.After the nineties, due to the promotion of computer performance and the fast development of image processing techniques, Car license recognition
The recognition efficiency of technology is greatly improved, and then has and completes Vehicle License Plate Recognition System using digital image processing techniques,
And propose the identification technology of template matching, and by optical character recognition technology using border tracing technique to character feature into
Row extracts, and proposes that the profile of multiplicity and testing mechanism carry out License Plate and be then based on connected domain analysis and Euclidean distance again later
To be split to characters on license plate.
But traditional licence plate recognition method is to pass through to summarize feature of image after manually handling image mostly, with
The positioning and identification for realizing license plate, since current identification scene is more complicated, (such as illumination condition deforms license plate, dirt vehicle
Board etc.), cause the recognition accuracy and recognition efficiency to characters on license plate lower.
For above-mentioned problem, currently no effective solution has been proposed.
Summary of the invention
The embodiment of the invention provides a kind of licence plate recognition method and devices, at least to solve traditional licence plate recognition method
Identify the recognition accuracy of characters on license plate and the technical problem that recognition efficiency is lower.
According to an aspect of an embodiment of the present invention, a kind of licence plate recognition method is provided, comprising: receive mesh to be detected
Logo image；License plate area in above-mentioned target image is positioned using target detection model for the first time, wherein in above-mentioned license plate area extremely
It less include: license plate；Secondary positioning is carried out to above-mentioned license plate area based on color of object model and left and right edges regression model；Using
Target nerve network carries out content recognition to the license plate area after above-mentioned secondary positioning, obtains the license board information of above-mentioned license plate.
Further, above-mentioned target detection model includes at least: YOLO V3 target detection model；Above-mentioned color of object mould
Type includes at least: hexagonal pyramid hsv color model；Above-mentioned target nerve network includes at least: based on gating cycle unit GRU's
Target nerve network.
Further, before receiving target image to be detected, the above method further include: multiple sample images are obtained,
Wherein, it is included at least in above-mentioned sample image: sample license plate and sample license plate area；Mark the above-mentioned sample in above-mentioned sample image
This license plate and above-mentioned sample license plate area；Sample image training based on above-mentioned mark obtains above-mentioned target detection model.
Further, before receiving target image to be detected, the above method further include: control picture pick-up device shoots mesh
Mark object obtains image sequence, and above-mentioned target object includes at least: vehicle；It is obtained by the above-mentioned image sequence of analysis processing above-mentioned
Target image, wherein above-mentioned target image includes at least: video image.
Further, secondary fixed to the progress of above-mentioned license plate area using color of object model and left and right edges regression model
Before position, the above method further include: obtain the length and width dimensions data and Aspect Ratio of above-mentioned license plate；Based on above-mentioned length and width dimensions number
According to above-mentioned Aspect Ratio, be modified processing to obtained above-mentioned license plate area is positioned for the first time.
Further, secondary fixed to the progress of above-mentioned license plate area based on color of object model and left and right edges regression model
Before position, the above method further include: whether the color model for detecting above-mentioned target image is RGB color model；In testing result
In the case where indicating that above-mentioned color model is above-mentioned RGB color model, by the color model of above-mentioned target image from above-mentioned RGB face
Color model is converted to above-mentioned color of object model.
Further, it is secondary fixed to be carried out based on color of object model and left and right edges regression model to above-mentioned license plate area
Position, comprising: determine the colouring information of above-mentioned license plate；Based on above-mentioned colouring information and above-mentioned color of object model, to above-mentioned target
Image carries out color constraint processing；To carrying out the constraint of above-mentioned color treated, above-mentioned target image carries out image procossing, wherein
Above-mentioned image procossing includes at least: gray processing processing and binary conversion treatment；Based on above-mentioned left and right edges regression model to carrying out on
Above-mentioned license plate area in target image after stating image procossing carries out secondary positioning.
According to another aspect of an embodiment of the present invention, a kind of license plate recognition device is additionally provided, comprising: receiving module is used
In reception target image to be detected；First locating module, for positioning above-mentioned target image for the first time using target detection model
In license plate area, wherein included at least in above-mentioned license plate area: license plate；Second locating module, for being based on color of object mould
Type and left and right edges regression model carry out secondary positioning to above-mentioned license plate area；Identification module, for using target nerve network
Content recognition is carried out to the license plate area after above-mentioned secondary positioning, obtains the license board information of above-mentioned license plate.
According to another aspect of an embodiment of the present invention, a kind of storage medium is additionally provided, above-mentioned storage medium includes storage
Program, wherein equipment where controlling above-mentioned storage medium in above procedure operation executes the above-mentioned license plate of any one and knows
Other method.
According to another aspect of an embodiment of the present invention, a kind of processor is additionally provided, above-mentioned processor is used to run program,
Wherein, any one above-mentioned licence plate recognition method is executed when above procedure is run.
In embodiments of the present invention, by receiving target image to be detected；On being positioned for the first time using target detection model
State the license plate area in target image, wherein include at least in above-mentioned license plate area: license plate；Based on color of object model and a left side
Right hand edge regression model carries out secondary positioning to above-mentioned license plate area；Using target nerve network to the vehicle after above-mentioned secondary positioning
Board region carries out content recognition, obtains the license board information of above-mentioned license plate, has reached the recognition accuracy for improving identification characters on license plate
And the purpose of recognition efficiency, to realize the fast-developing technical effect for promoting intelligent transportation system, and then solve
The lower technical problem of the recognition accuracy and recognition efficiency of traditional licence plate recognition method identification characters on license plate.
Detailed description of the invention
The drawings described herein are used to provide a further understanding of the present invention, constitutes part of this application, this hair
Bright illustrative embodiments and their description are used to explain the present invention, and are not constituted improper limitations of the present invention.In the accompanying drawings:
Fig. 1 is a kind of flow chart of licence plate recognition method according to an embodiment of the present invention；
Fig. 2 is a kind of flow chart of optional licence plate recognition method according to an embodiment of the present invention；And
Fig. 3 is a kind of structural schematic diagram of license plate recognition device according to an embodiment of the present invention.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
The model that the present invention protects all should belong in member's every other embodiment obtained without making creative work
It encloses.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that cover
Cover it is non-exclusive include, for example, the process, method, system, product or equipment for containing a series of steps or units are not necessarily limited to
Step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, product
Or other step or units that equipment is intrinsic.
Embodiment 1
According to embodiments of the present invention, a kind of embodiment of licence plate recognition method is provided, it should be noted that in attached drawing
The step of process illustrates can execute in a computer system such as a set of computer executable instructions, although also,
Logical order is shown in flow chart, but in some cases, it can be to be different from shown by sequence execution herein or retouch
The step of stating.
Fig. 1 is a kind of flow chart of licence plate recognition method according to an embodiment of the present invention, as shown in Figure 1, this method includes
Following steps:
Step S102 receives target image to be detected；
Step S104 positions the license plate area in above-mentioned target image using target detection model, wherein above-mentioned vehicle for the first time
Board includes at least in region: license plate；
Step S106, it is secondary fixed to be carried out based on color of object model and left and right edges regression model to above-mentioned license plate area
Position；
Step S108 carries out content recognition to the license plate area after above-mentioned secondary positioning using target nerve network, obtains
The license board information of above-mentioned license plate.
In an alternative embodiment, it before receiving target image to be detected, is being pre-positioned by controlling setting
The picture pick-up device shooting motor vehicles for setting (for example, garage, highway, traffic light intersection etc.) obtain image sequence, and pass through analysis
It handles above-mentioned image sequence and obtains above-mentioned target image, wherein above-mentioned target image includes at least: video image.
Optionally, above-mentioned target image can be, but not limited to shoot the front of motor vehicles, or shooting motor vehicles
The obtained video image data in rear portion, include at least in the target image: the license plate area of motor vehicles, that is, license plate setting
Region.
In an alternative embodiment, above-mentioned target detection model includes at least: YOLO V3 target detection model；On
It states color of object model to include at least: hexagonal pyramid hsv color model；Above-mentioned target nerve network includes at least: based on gate
The target nerve network of cycling element GRU.
Wherein, above-mentioned YOLO V3 target detection model is a kind of feature to be learnt using depth convolutional neural networks to examine
Survey the target detection model of object.
It should be noted that the purpose of above-mentioned first positioning and secondary positioning is in order to which determining one most on target image
License plate area that may be small on target image (alternatively, determine the subgraph rectangle as small as possible of an encirclement license plate area
Frame), license board information is obtained in order to carry out content recognition to license plate area.
In the embodiment of the present application, above-mentioned YOLO V3 target detection model can carry out the target image that needs identify
Autonomous learning and autonomous sophisticated model do not need manual intervention in whole process, vehicle in target image can not only be recognized accurately
Board information, and speed can achieve real-time requirement；Above-mentioned YOLO V3 target detection model can also use full figure as in
Hold information, background mistake (background is misdeemed as object) is fewer, and then can effectively improve the stationkeeping ability of license plate, character
The generalization ability of discrimination, recognition speed and identifying system.
Optionally, GRU (Gated Recurrent Unit) is a variant of LSTM, and GRU maintains the effect of LSTM
Keep structure simpler again simultaneously, the neural network model based on GRU can carry out sequence automatic marking to the text information of acquisition
And information extraction, so as to realize content recognition to the license plate area after secondary positioning well.
In the embodiment of the present application, the information extraction algorithm of the neural network model based on GRU is in accuracy rate, recall rate
It is all apparently higher than RNN model, it is identical as LSTM modelling effect, and due to structure optimization, training speed and predetermined speed are all
Higher than LSTM model, therefore, the neural network model based on GRU can be applied to large-scale internet common data information extraction
Scene has the advantages that continue to optimize network structure to obtain higher accuracy rate and stronger versatility.
In embodiments of the present invention, by receiving target image to be detected；On being positioned for the first time using target detection model
State the license plate area in target image, wherein include at least in above-mentioned license plate area: license plate；Based on color of object model and a left side
Right hand edge regression model carries out secondary positioning to above-mentioned license plate area；Using target nerve network to the vehicle after above-mentioned secondary positioning
Board region carries out content recognition, obtains the license board information of above-mentioned license plate, has reached the recognition accuracy for improving identification characters on license plate
And the purpose of recognition efficiency, to realize the fast-developing technical effect for promoting intelligent transportation system, and then solve
The lower technical problem of the recognition accuracy and recognition efficiency of traditional licence plate recognition method identification characters on license plate.
In an alternative embodiment, before receiving target image to be detected, the above method further include:
Step S202 obtains multiple sample images, wherein includes at least in above-mentioned sample image: sample license plate and sample
License plate area；
Step S204 marks above-mentioned sample license plate and above-mentioned sample license plate area in above-mentioned sample image；
Step S206, the sample image training based on above-mentioned mark obtain above-mentioned target detection model.
Optionally, above-mentioned multiple sample images can be, but not limited to as a large amount of license plate picture, in above-mentioned sample image extremely
It less include: sample license plate and sample license plate area；By being labeled place to above-mentioned sample license plate and above-mentioned sample license plate area
Reason, and initial detecting model is trained based on the sample image after mark, obtain above-mentioned target detection model.
In an alternative embodiment, before receiving target image to be detected, the above method further include:
Step S302, control picture pick-up device photographic subjects object obtain image sequence, and above-mentioned target object includes at least: vehicle
?；
Step S304 handles above-mentioned image sequence by analysis and obtains above-mentioned target image, wherein above-mentioned target image is extremely
It less include: video image.
Optionally, above-mentioned picture pick-up device can be, but not limited to as video camera, rotating camera etc., can to target object into
Row video recording, camera shooting etc..
Optionally, above-mentioned target image can be, but not limited to shoot the front of motor vehicles, or shooting motor vehicles
The obtained video image data in rear portion, include at least in the target image: the license plate area of motor vehicles, that is, license plate setting
Region.
In an alternative embodiment, it before receiving target image to be detected, is being pre-positioned by controlling setting
The picture pick-up device shooting motor vehicles for setting (for example, garage, highway, traffic light intersection etc.) obtain image sequence, and pass through analysis
It handles above-mentioned image sequence and obtains above-mentioned target image.
In an alternative embodiment, in use color of object model and left and right edges regression model to above-mentioned license plate area
Before domain carries out secondary positioning, the above method further include:
Step S402 obtains the length and width dimensions data and Aspect Ratio of above-mentioned license plate；
Step S404 is based on above-mentioned length and width dimensions data and above-mentioned Aspect Ratio, to positioning obtained above-mentioned license plate for the first time
Region is modified processing.
Optionally, above-mentioned length and width dimensions data can be, but not limited to for length data, width data, and according to length data
Above-mentioned Aspect Ratio is calculated with width data；Above-mentioned correcting process can be, but not limited to expand correcting process.
In above-mentioned optional embodiment, it is based on above-mentioned length and width dimensions data and above-mentioned Aspect Ratio, it can be to first fixed
The obtained license plate area in position is screened, it is determined whether for rectangular area etc., and to positioning obtained above-mentioned license plate area for the first time
It is normalized and expands correcting process.
In an alternative embodiment, color of object model and left and right edges regression model are being based on to above-mentioned license plate area
Before domain carries out secondary positioning, the above method further include:
Step S502, whether the color model for detecting above-mentioned target image is RGB color model；
Step S504, in the case where testing result indicates that above-mentioned color model is above-mentioned RGB color model, by above-mentioned mesh
The color model of logo image is converted from above-mentioned RGB color model to above-mentioned color of object model.
Optionally, above-mentioned color of object model can be, but not limited to as hsv color model；It should be noted that RGB color
Model is towards hardware, hsv color model user oriented, three dimensional representation from RGB along cube cornerwise White vertex to black
Color vertex observation, it can be seen that the hexagonal external feature of cube, wherein hexagonal boundaries indicate that tone, trunnion axis indicate saturation
Degree, vertical axis indicate lightness.
In the embodiment of the present application, in order to improve the accuracy rate that characters on license plate positions, above-mentioned target image is being detected
In the case that color model is RGB color model, first the color model of target image can be converted from above-mentioned RGB color model
To above-mentioned color of object model, for example, image is transformed into hsv color space from RGB color.
In an alternative embodiment, based on color of object model and left and right edges regression model to above-mentioned license plate area
Carry out secondary positioning, comprising:
Step S602 determines the colouring information of above-mentioned license plate；
Step S604 is based on above-mentioned colouring information and above-mentioned color of object model, carries out color about to above-mentioned target image
Beam processing；
Step S606, to carrying out the constraint of above-mentioned color treated, above-mentioned target image carries out image procossing, wherein above-mentioned
Image procossing includes at least: gray processing processing and binary conversion treatment；
Step S608, based on above-mentioned left and right edges regression model to upper in the target image after carrying out above-mentioned image procossing
It states license plate area and carries out secondary positioning.
Optionally, in the above-described embodiments, based on color of object model and left and right edges regression model to above-mentioned license plate area
Domain carries out secondary positioning, that is, being based on color of object model and left and right edges regression model, determines the edge of above-mentioned license plate area
Position, to realize the above-mentioned license plate area of precise positioning.
In above-mentioned optional embodiment, after the colouring information for determining above-mentioned license plate, it can be believed based on above-mentioned color
Breath and above-mentioned color of object model carry out color constraint processing to above-mentioned target image, and handle above-mentioned color constraint is carried out
Above-mentioned target image afterwards carries out gray processing processing and binary conversion treatment；Also, due to the ginseng when handling image border
There are many burr points at the uneven edge of difference, can be combined with left and right edges regression model to the target after carrying out above-mentioned image procossing
Above-mentioned license plate area in image is modified, to realize that license plate area is accurately positioned.
It should be noted that the variation of image grayscale of Same Scene has on different resolution due in actual environment
Institute is different, even if a big object edge will be degenerated in the lower image of resolution ratio for " texture ", on the other hand, image
In " noise " can also show as mixed and disorderly gray scale mutation, calculate RGB mean value to extract the Pixel Information in target image, and
Simple clustering piecemeal is carried out to target image, noise reduction can also be filtered to the image of processing, color lump boundary is made to become to the greatest extent may be used
It can be soft.
Fig. 2 is a kind of flow chart of optional licence plate recognition method according to an embodiment of the present invention, as shown in Fig. 2, below
By a kind of optional licence plate recognition method embodiment, licence plate recognition method provided by the embodiment of the present application is illustrated
It is bright:
Step S702 receives target image to be detected；
Step S704 positions the license plate area in above-mentioned target image using target detection model, wherein above-mentioned vehicle for the first time
Board includes at least in region: license plate.
In an alternative embodiment, it before receiving target image to be detected, is being pre-positioned by controlling setting
The picture pick-up device shooting motor vehicles for setting (for example, garage, highway, traffic light intersection etc.) obtain image sequence, and pass through analysis
It handles above-mentioned image sequence and obtains above-mentioned target image.
Optionally, above-mentioned target image can be, but not limited to shoot the front of motor vehicles, or shooting motor vehicles
The obtained video image data in rear portion, include at least in the target image: the license plate area of motor vehicles, that is, license plate setting
Region.
Step S706 obtains the length and width dimensions data and Aspect Ratio of above-mentioned license plate；
Step S708 is based on above-mentioned length and width dimensions data and above-mentioned Aspect Ratio, to positioning obtained above-mentioned license plate for the first time
Region is modified processing.
Optionally, above-mentioned length and width dimensions data can be, but not limited to for length data, width data, and according to length data
Above-mentioned Aspect Ratio is calculated with width data；Above-mentioned correcting process can be, but not limited to expand correcting process.
In above-mentioned optional embodiment, it is based on above-mentioned length and width dimensions data and above-mentioned Aspect Ratio, it can be to first fixed
The obtained license plate area in position is screened, it is determined whether for rectangular area etc., and to positioning obtained above-mentioned license plate area for the first time
Domain is normalized and expands correcting process.
Step S710, it is secondary fixed to be carried out based on color of object model and left and right edges regression model to above-mentioned license plate area
Position；
Step S712 carries out content recognition to the license plate area after above-mentioned secondary positioning using target nerve network, obtains
The license board information of above-mentioned license plate.
Optionally, in the above-described embodiments, based on color of object model and left and right edges regression model to above-mentioned license plate area
Domain carries out secondary positioning, that is, being based on color of object model and left and right edges regression model, determines the edge of above-mentioned license plate area
Position, to realize the above-mentioned license plate area of precise positioning.
In an alternative embodiment, above-mentioned target detection model includes at least: YOLO V3 target detection model；On
It states color of object model to include at least: hexagonal pyramid hsv color model；Above-mentioned target nerve network includes at least: based on gate
The target nerve network of cycling element GRU.
It should be noted that the purpose of above-mentioned first positioning and secondary positioning is in order to which determining one most on target image
License plate area that may be small on target image (alternatively, determine the subgraph rectangle as small as possible of an encirclement license plate area
Frame), license board information is obtained in order to carry out content recognition to license plate area.
In the embodiment of the present application, above-mentioned YOLO V3 target detection model can carry out the target image that needs identify
Autonomous learning and autonomous sophisticated model do not need manual intervention in whole process, vehicle in target image can not only be recognized accurately
Board information, and speed can achieve real-time requirement；Above-mentioned YOLO V3 target detection model can also use full figure as in
Hold information, background mistake (background is misdeemed as object) is fewer, and then can effectively improve the stationkeeping ability of license plate, character
The generalization ability of discrimination, recognition speed and identifying system.
It should be noted that GRU (Gated Recurrent Unit) is a variant of LSTM, GRU maintains LSTM
Effect simultaneously again keep structure simpler, the neural network model based on GRU can to the text information of acquisition carry out sequence from
Dynamic mark and information extraction, so as to realize content recognition to the license plate area after secondary positioning well.
In the embodiment of the present application, the information extraction algorithm of the neural network model based on GRU is in accuracy rate, recall rate
It is all apparently higher than RNN model, it is identical as LSTM modelling effect, and due to structure optimization, training speed and predetermined speed are all
Higher than LSTM model, therefore, the neural network model based on GRU can be applied to large-scale internet common data information extraction
Scene has the advantages that continue to optimize network structure to obtain higher accuracy rate and stronger versatility.
Embodiment 2
According to embodiments of the present invention, it additionally provides a kind of for implementing the Installation practice of above-mentioned licence plate recognition method, Fig. 3
It is a kind of structural schematic diagram of license plate recognition device according to an embodiment of the present invention, as shown in figure 3, above-mentioned license plate recognition device,
It include: receiving module 30, the first locating module 32, the second locating module 34 and identification module 36, in which:
Receiving module 30, for receiving target image to be detected；First locating module 32, for using target detection mould
Type positions the license plate area in above-mentioned target image for the first time, wherein includes at least in above-mentioned license plate area: license plate；Second positioning
Module 34, for carrying out secondary positioning to above-mentioned license plate area based on color of object model and left and right edges regression model；Identification
Module 36 obtains above-mentioned vehicle for carrying out content recognition to the license plate area after above-mentioned secondary positioning using target nerve network
The license board information of board.
It should be noted that above-mentioned modules can be realized by software or hardware, for example, for the latter,
Can be accomplished by the following way: above-mentioned modules can be located in same processor；Alternatively, above-mentioned modules are with any
Combined mode is located in different processors.
Herein it should be noted that above-mentioned receiving module 30, the first locating module 32, the second locating module 34 and identification mould
Block 36 corresponds to the step S102 to step S106 in embodiment 1, the example and answer that above-mentioned module and corresponding step are realized
It is identical with scene, but it is not limited to the above embodiments 1 disclosure of that.It should be noted that above-mentioned module as device one
Part may operate in terminal.
It should be noted that the optional or preferred embodiment of the present embodiment may refer to the associated description in embodiment 1,
Details are not described herein again.
Above-mentioned license plate recognition device can also include processor and memory, above-mentioned receiving module 30, the first positioning mould
Block 32, the second locating module 34 and identification module 36 etc. store in memory as program unit, are deposited by processor execution
Above procedure unit in memory is stored up to realize corresponding function.
Include kernel in processor, is gone in memory to transfer corresponding program unit by kernel, above-mentioned kernel can be set
One or more.Memory may include the non-volatile memory in computer-readable medium, random access memory (RAM)
And/or the forms such as Nonvolatile memory, such as read-only memory (ROM) or flash memory (flash RAM), memory includes at least one
Storage chip.
According to the embodiment of the present application, a kind of storage medium embodiment is additionally provided.Optionally, in the present embodiment, above-mentioned
Storage medium includes the program of storage, wherein equipment where controlling above-mentioned storage medium in above procedure operation executes above-mentioned
Any one licence plate recognition method.
Optionally, in the present embodiment, above-mentioned storage medium can be located in computer network in computer terminal group
In any one terminal, or in any one mobile terminal in mobile terminal group, above-mentioned storage medium packet
Include the program of storage.
Optionally, when program is run, equipment where control storage medium executes following functions: receiving target to be detected
Image；License plate area in above-mentioned target image is positioned using target detection model for the first time, wherein in above-mentioned license plate area at least
It include: license plate；Secondary positioning is carried out to above-mentioned license plate area based on color of object model and left and right edges regression model；Using mesh
It marks neural network and content recognition is carried out to the license plate area after above-mentioned secondary positioning, obtain the license board information of above-mentioned license plate.
Optionally, when program is run, equipment where control storage medium executes following functions: multiple sample images are obtained,
Wherein, it is included at least in above-mentioned sample image: sample license plate and sample license plate area；Mark the above-mentioned sample in above-mentioned sample image
This license plate and above-mentioned sample license plate area；Sample image training based on above-mentioned mark obtains above-mentioned target detection model.
Optionally, when program is run, equipment where control storage medium executes following functions: control picture pick-up device shooting
Target object obtains image sequence, and above-mentioned target object includes at least: vehicle；Above-mentioned image sequence is handled by analysis to obtain
State target image, wherein above-mentioned target image includes at least: video image.
Optionally, when program is run, equipment where control storage medium executes following functions: obtaining the length of above-mentioned license plate
Wide dimension data and Aspect Ratio；Based on above-mentioned length and width dimensions data and above-mentioned Aspect Ratio, first positioning is obtained above-mentioned
License plate area is modified processing.
Optionally, when program is run, equipment where control storage medium executes following functions: detecting above-mentioned target image
Color model whether be RGB color model；The case where above-mentioned color model is above-mentioned RGB color model is indicated in testing result
Under, the color model of above-mentioned target image is converted from above-mentioned RGB color model to above-mentioned color of object model.
Optionally, when program is run, equipment where control storage medium executes following functions: determining the face of above-mentioned license plate
Color information；Based on above-mentioned colouring information and above-mentioned color of object model, color constraint processing is carried out to above-mentioned target image；To into
Treated that above-mentioned target image carries out image procossing for the constraint of row above-mentioned color, wherein above-mentioned image procossing includes at least: gray scale
Change processing and binary conversion treatment；Based on above-mentioned left and right edges regression model in the target image after carrying out above-mentioned image procossing
Above-mentioned license plate area carries out secondary positioning.
According to the embodiment of the present application, a kind of processor embodiment is additionally provided.Optionally, in the present embodiment, above-mentioned place
Reason device is for running program, wherein above procedure executes any one of the above licence plate recognition method when running.
The embodiment of the present application provides a kind of equipment, equipment include processor, memory and storage on a memory and can
The program run on a processor, processor performs the steps of when executing program receives target image to be detected；Using mesh
Mark detection model positions the license plate area in above-mentioned target image for the first time, wherein includes at least in above-mentioned license plate area: license plate；
Secondary positioning is carried out to above-mentioned license plate area based on color of object model and left and right edges regression model；Using target nerve network
Content recognition is carried out to the license plate area after above-mentioned secondary positioning, obtains the license board information of above-mentioned license plate.
Optionally, when above-mentioned processor executes program, multiple sample images can also be obtained, wherein above-mentioned sample image
In include at least: sample license plate and sample license plate area；Mark above-mentioned sample license plate and the above-mentioned sample in above-mentioned sample image
License plate area；Sample image training based on above-mentioned mark obtains above-mentioned target detection model.
Optionally, when above-mentioned processor executes program, picture pick-up device photographic subjects object can also be controlled and obtain image sequence
Column, above-mentioned target object include at least: vehicle；Above-mentioned image sequence, which is handled, by analysis obtains above-mentioned target image, wherein on
It states target image to include at least: video image.
Optionally, when above-mentioned processor executes program, the length and width dimensions data and length-width ratio of above-mentioned license plate can also be obtained
Example；Based on above-mentioned length and width dimensions data and above-mentioned Aspect Ratio, place is modified to obtained above-mentioned license plate area is positioned for the first time
Reason.
Optionally, when above-mentioned processor executes program, whether the color model that can also detect above-mentioned target image is RGB
Color model；In the case where testing result indicates that above-mentioned color model is above-mentioned RGB color model, by above-mentioned target image
Color model is converted from above-mentioned RGB color model to above-mentioned color of object model.
Optionally, when above-mentioned processor executes program, the colouring information of above-mentioned license plate can also be determined；Based on above-mentioned color
Information and above-mentioned color of object model carry out color constraint processing to above-mentioned target image；It handles above-mentioned color constraint is carried out
Above-mentioned target image afterwards carries out image procossing, wherein above-mentioned image procossing includes at least: at gray processing processing and binaryzation
Reason；The above-mentioned license plate area in the target image after carrying out above-mentioned image procossing is carried out based on above-mentioned left and right edges regression model
Secondary positioning.
Present invention also provides a kind of computer program products, when executing on data processing equipment, are adapted for carrying out just
The program of beginningization there are as below methods step: target image to be detected is received；Above-mentioned mesh is positioned using target detection model for the first time
License plate area in logo image, wherein included at least in above-mentioned license plate area: license plate；Based on color of object model and left and right side
Edge regression model carries out secondary positioning to above-mentioned license plate area；Using target nerve network to the license plate area after above-mentioned secondary positioning
Domain carries out content recognition, obtains the license board information of above-mentioned license plate.
Optionally, when above-mentioned computer program product executes program, multiple sample images can also be obtained, wherein above-mentioned
It is included at least in sample image: sample license plate and sample license plate area；Mark above-mentioned sample license plate in above-mentioned sample image and
Above-mentioned sample license plate area；Sample image training based on above-mentioned mark obtains above-mentioned target detection model.
Optionally, when above-mentioned computer program product executes program, picture pick-up device photographic subjects object can also be controlled and obtained
To image sequence, above-mentioned target object is included at least: vehicle；Above-mentioned image sequence, which is handled, by analysis obtains above-mentioned target figure
Picture, wherein above-mentioned target image includes at least: video image.
Optionally, when above-mentioned computer program product executes program, the length and width dimensions data of above-mentioned license plate can also be obtained
And Aspect Ratio；Based on above-mentioned length and width dimensions data and above-mentioned Aspect Ratio, to position for the first time obtained above-mentioned license plate area into
Row correcting process.
Optionally, when above-mentioned computer program product executes program, the color model of above-mentioned target image can also be detected
It whether is RGB color model；It, will be above-mentioned in the case where testing result indicates that above-mentioned color model is above-mentioned RGB color model
The color model of target image is converted from above-mentioned RGB color model to above-mentioned color of object model.
Optionally, when above-mentioned computer program product executes program, the colouring information of above-mentioned license plate can also be determined；It is based on
Above-mentioned colouring information and above-mentioned color of object model carry out color constraint processing to above-mentioned target image；To the above-mentioned color of progress
Treated that above-mentioned target image carries out image procossing for constraint, wherein above-mentioned image procossing includes at least: gray processing processing and two
Value processing；Based on above-mentioned left and right edges regression model to the above-mentioned license plate area in the target image after carrying out above-mentioned image procossing
Domain carries out secondary positioning.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
In the above embodiment of the invention, it all emphasizes particularly on different fields to the description of each embodiment, does not have in some embodiment
The part of detailed description, reference can be made to the related descriptions of other embodiments.
In several embodiments provided herein, it should be understood that disclosed technology contents can pass through others
Mode is realized.Wherein, the apparatus embodiments described above are merely exemplary, such as the division of the unit, Ke Yiwei
A kind of logical function partition, there may be another division manner in actual implementation, for example, multiple units or components can combine or
Person is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Between coupling, direct-coupling or communication connection can be through some interfaces, the INDIRECT COUPLING or communication link of unit or module
It connects, can be electrical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple
On unit.It can some or all of the units may be selected to achieve the purpose of the solution of this embodiment according to the actual needs.
It, can also be in addition, the functional units in various embodiments of the present invention may be integrated into one processing unit
It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list
Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can store in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
Equipment (can for personal computer, server or network equipment etc.) execute each embodiment the method for the present invention whole or
Part steps.And storage medium above-mentioned includes: that USB flash disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), mobile hard disk, magnetic or disk etc. be various to can store program code
Medium.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (10)
1. a kind of licence plate recognition method characterized by comprising
Receive target image to be detected；
License plate area in the target image is positioned using target detection model for the first time, wherein in the license plate area at least
It include: license plate；
Secondary positioning is carried out to the license plate area based on color of object model and left and right edges regression model；
Content recognition is carried out to the license plate area after the secondary positioning using target nerve network, obtains the license plate of the license plate
Information.
2. the method according to claim 1, wherein the target detection model includes at least: YOLO V3 target
Detection model；The color of object model includes at least: hexagonal pyramid hsv color model；The target nerve network at least wraps
It includes: the target nerve network based on gating cycle unit GRU.
3. the method according to claim 1, wherein before receiving target image to be detected, the method
Further include:
Obtain multiple sample images, wherein include at least in the sample image: sample license plate and sample license plate area；
Mark the sample license plate and the sample license plate area in the sample image；
Sample image training based on the mark obtains the target detection model.
4. the method according to claim 1, wherein before receiving target image to be detected, the method
Further include:
Control picture pick-up device photographic subjects object obtains image sequence, and the target object includes at least: vehicle；
The target image is obtained by analysis processing described image sequence, wherein the target image includes at least: video figure
Picture.
5. the method according to claim 1, wherein using color of object model and left and right edges regression model
Before carrying out secondary positioning to the license plate area, the method also includes:
Obtain the length and width dimensions data and Aspect Ratio of the license plate；
Based on the length and width dimensions data and the Aspect Ratio, place is modified to the obtained license plate area is positioned for the first time
Reason.
6. the method according to claim 1, wherein being based on color of object model and left and right edges regression model
Before carrying out secondary positioning to the license plate area, the method also includes:
Whether the color model for detecting the target image is RGB color model；
In the case where testing result indicates that the color model is the RGB color model, by the color of the target image
Model is converted from the RGB color model to the color of object model.
7. according to the method described in claim 6, it is characterized in that, being based on color of object model and left and right edges regression model pair
The license plate area carries out secondary positioning, comprising:
Determine the colouring information of the license plate；
Based on the colouring information and the color of object model, color constraint processing is carried out to the target image；
To carrying out color constraint treated, the target image carries out image procossing, wherein described image processing is at least
It include: gray processing processing and binary conversion treatment；
Based on the left and right edges regression model to the license plate area carried out in described image treated target image into
The secondary positioning of row.
8. a kind of license plate recognition device characterized by comprising
Receiving module, for receiving target image to be detected；
First locating module, for positioning the license plate area in the target image for the first time using target detection model, wherein institute
It states in license plate area and includes at least: license plate；
Second locating module, it is secondary for being carried out based on color of object model and left and right edges regression model to the license plate area
Positioning；
Identification module is obtained for carrying out content recognition to the license plate area after the secondary positioning using target nerve network
The license board information of the license plate.
9. a kind of storage medium, which is characterized in that the storage medium includes the program of storage, wherein run in described program
When control the storage medium where equipment perform claim require any one of 1 to 7 described in licence plate recognition method.
10. a kind of processor, which is characterized in that the processor is for running program, wherein right of execution when described program is run
Benefit require any one of 1 to 7 described in licence plate recognition method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811573659.0A CN109977941A (en) | 2018-12-21 | 2018-12-21 | Licence plate recognition method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811573659.0A CN109977941A (en) | 2018-12-21 | 2018-12-21 | Licence plate recognition method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109977941A true CN109977941A (en) | 2019-07-05 |
Family
ID=67076270
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811573659.0A Pending CN109977941A (en) | 2018-12-21 | 2018-12-21 | Licence plate recognition method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109977941A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110807459A (en) * | 2019-10-31 | 2020-02-18 | 深圳市捷顺科技实业股份有限公司 | License plate correction method and device and readable storage medium |
| CN112560728A (en) * | 2020-12-22 | 2021-03-26 | 上海哔哩哔哩科技有限公司 | Target object identification method and device |
| CN112686252A (en) * | 2020-12-28 | 2021-04-20 | 中国联合网络通信集团有限公司 | License plate detection method and device |
| CN112699813A (en) * | 2020-12-31 | 2021-04-23 | 哈尔滨市科佳通用机电股份有限公司 | Multi-country license plate positioning method based on improved MTCNN (multiple terminal communication network) model |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101261722A (en) * | 2008-01-17 | 2008-09-10 | 北京航空航天大学 | Electronic police background intelligent management and automatic implementation system |
| CN101789080A (en) * | 2010-01-21 | 2010-07-28 | 上海交通大学 | Detection method for vehicle license plate real-time positioning character segmentation |
| CN103870803A (en) * | 2013-10-21 | 2014-06-18 | 北京邮电大学 | Vehicle license plate recognition method and system based on coarse positioning and fine positioning fusion |
| EP3182334A1 (en) * | 2015-12-17 | 2017-06-21 | Xerox Corporation | License plate recognition using coarse-to-fine cascade adaptations of convolutional neural networks |
| CN107301385A (en) * | 2017-06-09 | 2017-10-27 | 浙江宇视科技有限公司 | One kind blocks licence plate recognition method and device |
| CN108985214A (en) * | 2018-07-09 | 2018-12-11 | 上海斐讯数据通信技术有限公司 | The mask method and device of image data |
-
2018
- 2018-12-21 CN CN201811573659.0A patent/CN109977941A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101261722A (en) * | 2008-01-17 | 2008-09-10 | 北京航空航天大学 | Electronic police background intelligent management and automatic implementation system |
| CN101789080A (en) * | 2010-01-21 | 2010-07-28 | 上海交通大学 | Detection method for vehicle license plate real-time positioning character segmentation |
| CN103870803A (en) * | 2013-10-21 | 2014-06-18 | 北京邮电大学 | Vehicle license plate recognition method and system based on coarse positioning and fine positioning fusion |
| EP3182334A1 (en) * | 2015-12-17 | 2017-06-21 | Xerox Corporation | License plate recognition using coarse-to-fine cascade adaptations of convolutional neural networks |
| CN107301385A (en) * | 2017-06-09 | 2017-10-27 | 浙江宇视科技有限公司 | One kind blocks licence plate recognition method and device |
| CN108985214A (en) * | 2018-07-09 | 2018-12-11 | 上海斐讯数据通信技术有限公司 | The mask method and device of image data |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110807459A (en) * | 2019-10-31 | 2020-02-18 | 深圳市捷顺科技实业股份有限公司 | License plate correction method and device and readable storage medium |
| CN110807459B (en) * | 2019-10-31 | 2022-06-17 | 深圳市捷顺科技实业股份有限公司 | License plate correction method and device and readable storage medium |
| CN112560728A (en) * | 2020-12-22 | 2021-03-26 | 上海哔哩哔哩科技有限公司 | Target object identification method and device |
| CN112560728B (en) * | 2020-12-22 | 2023-07-11 | 上海幻电信息科技有限公司 | Target object identification method and device |
| CN112686252A (en) * | 2020-12-28 | 2021-04-20 | 中国联合网络通信集团有限公司 | License plate detection method and device |
| CN112699813A (en) * | 2020-12-31 | 2021-04-23 | 哈尔滨市科佳通用机电股份有限公司 | Multi-country license plate positioning method based on improved MTCNN (multiple terminal communication network) model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103824066B (en) | A kind of licence plate recognition method based on video flowing | |
| CN109993056B (en) | Method, server and storage medium for identifying vehicle illegal behaviors | |
| CN108364010B (en) | License plate recognition method, device, equipment and computer readable storage medium | |
| Alvarez et al. | Road detection based on illuminant invariance | |
| CN107545239B (en) | Fake plate detection method based on license plate recognition and vehicle characteristic matching | |
| Alvarez et al. | Road scene segmentation from a single image | |
| Ajmal et al. | A comparison of RGB and HSV colour spaces for visual attention models | |
| CN106599792B (en) | Method for detecting hand driving violation behavior | |
| Rotaru et al. | Color image segmentation in HSI space for automotive applications | |
| CN109711264B (en) | Method and device for detecting occupation of bus lane | |
| CN108154102B (en) | Road traffic sign identification method | |
| CN111767882A (en) | Multi-mode pedestrian detection method based on improved YOLO model | |
| CN106022232A (en) | License plate detection method based on deep learning | |
| CN109977941A (en) | Licence plate recognition method and device | |
| CN111325769B (en) | Target object detection method and device | |
| CN107103317A (en) | Fuzzy license plate image recognition algorithm based on image co-registration and blind deconvolution | |
| CN107480676B (en) | Vehicle color identification method and device and electronic equipment | |
| Ye et al. | A two-stage real-time YOLOv2-based road marking detector with lightweight spatial transformation-invariant classification | |
| CN109543632A (en) | A kind of deep layer network pedestrian detection method based on the guidance of shallow-layer Fusion Features | |
| Türkyılmaz et al. | License plate recognition system using artificial neural networks | |
| CN107665327B (en) | Lane line detection method and device | |
| CN110659550A (en) | Traffic sign recognition method, traffic sign recognition device, computer equipment and storage medium | |
| CN106203237A (en) | The recognition methods of container-trailer numbering and device | |
| CN113128507B (en) | License plate recognition method and device, electronic equipment and storage medium | |
| CN111382658B (en) | Road traffic sign detection method in natural environment based on image gray gradient consistency |
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 | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200602 Address after: 207, 5th floor, 101 / F, 1-14 / F, building 4, yard 59, Chemical Road, Chaoyang District, Beijing 100000 Applicant after: BEIJING MIXISLINK TECHNOLOGY Co.,Ltd. Applicant after: Shenhua hydrogen technology Rugao Co., Ltd Address before: Room 504, 5th floor, Block C, Zhonghui Square, 11 Dongzhimen South Street, Dongcheng District, Beijing, 100010 Applicant before: BEIJING MIXISLINK TECHNOLOGY Co.,Ltd. |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190705 |