CN110414507A - Licence plate recognition method, device, computer equipment and storage medium - Google Patents
Licence plate recognition method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110414507A CN110414507A CN201910626156.3A CN201910626156A CN110414507A CN 110414507 A CN110414507 A CN 110414507A CN 201910626156 A CN201910626156 A CN 201910626156A CN 110414507 A CN110414507 A CN 110414507A
- Authority
- CN
- China
- Prior art keywords
- license plate
- image
- candidate
- detection model
- obtains
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Character Input (AREA)
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The embodiment of the invention discloses a kind of licence plate recognition methods, comprising: detects to the image for including license plate, obtains the candidate license plate region detected；Using the candidate license plate region as the input of license plate vertex detection model, the position on four license plate vertex that the detection that license plate vertex detection model exports obtains is obtained；License plate image is corrected according to the position on four license plate vertex, the target license plate image after being corrected；Characters on license plate in the target license plate image is identified, recognition result is obtained.The wide-angle license plate that the licence plate recognition method can obtain shooting accurately identifies, and substantially increases the accuracy of wide-angle Car license recognition.Furthermore, it is also proposed that a kind of license plate recognition device, computer equipment and storage medium.
Description
Technical field
The present invention relates to field of license plate recognition more particularly to a kind of licence plate recognition method, device, computer equipment and
Storage medium.
Background technique
Car license recognition is all based on greatly standard card mouth at present, the scenes such as parking lot, in the picture that these scenes are captured, general vehicle
Board is all than calibration, no inclination, and than more visible, and existing Recognition Algorithm of License Plate is directed to these generic scenarios, and accuracy rate is all very
It is high.
Now with the increase of domestic automobile ownership, it is badly in need of specification curb parking and needs to do for curb parking scene
License plate differentiation, picture duplicate removal and Car license recognition.In the license plate picture shot due to the front-end camera of roadside scene, license plate angle
Bigger (tilting very big), is easy very big distortion, is also easy to be illuminated by the light influence, so current Recognition Algorithm of License Plate needle
It is very low to this scene Recognition rate.
Summary of the invention
Based on this, it is necessary to which, in view of the above-mentioned problems, proposing that one kind can be big to license plate angle, the license plate being distorted carries out
Licence plate recognition method, device, computer equipment and the storage medium accurately identified.
A kind of licence plate recognition method, which is characterized in that the described method includes:
The image for including license plate is detected, the candidate license plate region detected is obtained；
Using the candidate license plate region as the input of license plate vertex detection model, license plate vertex detection model is obtained
The position on four license plate vertex that the detection of output obtains；
License plate image is corrected according to the position on four license plate vertex, the target license plate figure after being corrected
Picture；
Character in the target license plate image is identified, recognition result is obtained.
A kind of license plate recognition device, described device include:
It obtains module and obtains the candidate license plate region detected for detecting to the image for including license plate；
Detection module, for obtaining the vehicle using the candidate license plate region as the input of license plate vertex detection model
The position on four license plate vertex that the detection of board vertex detection model output obtains；
Rectification module, for being corrected according to the position on four license plate vertex to license plate image, after obtaining correction
Target license plate image；
Identification module obtains recognition result for identifying to the character in the target license plate image.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the calculating
When machine program is executed by the processor, so that the processor executes following steps:
The image for including license plate is detected, the candidate license plate region detected is obtained；
Using the candidate license plate region as the input of license plate vertex detection model, license plate vertex detection model is obtained
The position on four license plate vertex that the detection of output obtains；
License plate image is corrected according to the position on four license plate vertex, the target license plate figure after being corrected
Picture；
Character in the target license plate image is identified, recognition result is obtained.
A kind of computer readable storage medium is stored with computer program, when the computer program is executed by processor,
So that the processor executes following steps:
The image for including license plate is detected, the candidate license plate region detected is obtained；
Using the candidate license plate region as the input of license plate vertex detection model, license plate vertex detection model is obtained
The position on four license plate vertex that the detection of output obtains；
License plate image is corrected according to the position on four license plate vertex, the target license plate figure after being corrected
Picture；
Characters on license plate in the target license plate image is identified, recognition result is obtained.
Above-mentioned licence plate recognition method, device, computer equipment and storage medium, by being carried out to the image for including license plate
Detection, obtains the candidate license plate region detected and obtains then using candidate license plate region as the input of license plate vertex detection model
The position on four license plate vertex that the detection of pick-up board vertex detection model output obtains, then according to the position on four license plate vertex
It sets and license plate image is corrected, the target license plate image after being corrected, later to the characters on license plate in target license plate image
It is identified, obtains recognition result.The licence plate recognition method to shooting obtained wide-angle license plate and can be distorted
License plate is accurately identified, and the accuracy of wide-angle license plate and the Car license recognition that distorts is substantially increased.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Wherein:
Fig. 1 is the flow chart of licence plate recognition method in one embodiment；
Fig. 2 is the effect diagram identified in one embodiment to wide-angle license plate image；
Fig. 3 is the method flow diagram of license plate vertex detection in one embodiment；
Fig. 4 is the method flow diagram detected in one embodiment to candidate license plate region；
Fig. 5 is the process schematic of Car license recognition in one embodiment；
Fig. 6 is the structural block diagram of license plate recognition device in one embodiment；
Fig. 7 is the structural block diagram of detection module in one embodiment；
Fig. 8 is the structural block diagram that module is obtained in one embodiment；
Fig. 9 is the structural block diagram of computer equipment in one embodiment.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
As shown in Figure 1, in one embodiment, provide a kind of licence plate recognition method, which both can be with
Applied to terminal, server also can be applied to, specifically includes the following steps:
Step 102, the image for including license plate is detected, obtains the candidate license plate region detected.
Wherein, candidate license plate region refers to the region where the license plate that detection obtains.The image for including license plate is carried out
Detecting the mode for obtaining candidate license plate region has very much, for example, can carry out license plate area using eight connectivity range searching algorithm
Search.
Step 104, using candidate license plate region as the input of license plate vertex detection model, license plate vertex detection model is obtained
The position on four license plate vertex that the detection of output obtains.
Wherein, the area in candidate license plate region is much larger than the area of license plate image, so in order to which vehicle is accurately positioned out
The position of board, needs to detect the position on four license plate vertex, and license plate vertex refers to the point at four angles of license plate.Detect mould in license plate vertex
Type refers to the model for being used to detect four license plate vertex in candidate license plate region that training obtains.In one embodiment, license plate
Vertex detection model can be obtained using convolutional neural networks model training.The training of license plate vertex detection model, which can use, to be had
The mode training of supervision obtains, and obtains the mark of training sample and sample, training sample refers to candidate license plate area sample, sample
This mark refers to the position coordinates at four angles of license plate marked in candidate license plate area sample.
Step 106, license plate image is corrected according to the position on four license plate vertex, the target license plate after being corrected
Image.
It wherein, may be the image of the image under wide-angle or distortion due to license plate image (for example, being clapped by low level angle
The picture taken the photograph), so needing the position according to four license plate vertex after the position coordinates for orienting four license plate vertex
License plate image is corrected, i.e., license plate is converted into the license plate image than calibration, is convenient for subsequent identification.
As shown in Fig. 2, for the effect diagram in one embodiment, identified to wide-angle license plate image.From figure
As can be seen that the license plate image inclined for the comparison taken has obtained the license plate image than calibration after correcting.
Step 108, the characters on license plate in target license plate image is identified, obtains recognition result.
Wherein, target license plate image refers to the license plate image of the ratio calibration after correction, includes characters on license plate in license plate image,
Characters on license plate includes letter and number.License plate is numbered in order to obtain, is identified to target license plate image, is obtained recognition result.
Since the target license plate image after correction is as the license plate image that normal photographing obtains, so just using traditional recognizer
It can identify to obtain license plate number.For example, can be using the method based on connected region search and upright projection to characters on license plate
It is split, the characters on license plate after segmentation is identified using the character recognition model that deep learning training obtains then, is obtained
To the recognition result of each character, to obtain the license plate number that identification obtains.
Above-mentioned licence plate recognition method obtains the candidate license plate detected by detecting to the image for including license plate
Region obtains the output of license plate vertex detection model then using candidate license plate region as the input of license plate vertex detection model
The position for detecting four obtained license plate vertex, then corrects license plate image according to the position on four license plate vertex, obtains
Target license plate image after to correction, later identifies the characters on license plate in target license plate image, obtains recognition result.It should
Licence plate recognition method can accurately identify the obtained wide-angle license plate of shooting and the license plate being distorted, and greatly improve
The accuracy of wide-angle license plate and the Car license recognition that distorts.
As shown in figure 3, in one embodiment, license plate vertex detection model includes: the first detection model, the second detection mould
Type and third detection model；Using candidate license plate region as the input of license plate vertex detection model, obtains license plate vertex and detect mould
The position on four license plate vertex that the detection of type output obtains, comprising:
Step 104A, using candidate license plate region as the input of the first detection model, the first detection model is used for candidate
License plate area carries out detection and generates candidate license plate window.
Wherein, license plate vertex detection model include three submodels, be respectively the first detection model, the second detection model and
Third detection model.First detection model is for generating candidate window.Candidate window can be selected to the identical window of license plate size
Mouthful.
In one embodiment, the model that the first detection model is made of convolutional layer, comprising: three convolutional layers, input
Image width be 16, a height of 12, the convolution kernel of each convolutional layer uses 3X3.
Step 104B obtains the candidate license plate window of the first detection model output, is corrected place to candidate license plate window
Reason, obtains the first candidate license plate window.
Wherein, correction process refers to that the position to candidate license plate window is corrected.Get the output of the first detection model
Candidate license plate window after, since the candidate license plate window of generation is without so accurate, so also needing to candidate license plate window
It is corrected processing, obtains the first candidate license plate window.It in another embodiment, can also be to the candidate after correction process
License plate window merges the high candidate license plate window of Duplication using NMS (Non-Maximum Suppression, NMS).
Step 104C, using the first candidate license plate window as the input of the second detection model, the second detection model for pair
First candidate license plate window is screened.
Wherein, the second detection model screens obtained multiple first candidate license plate windows, vetos existing a large amount of
Non- license plate window, to obtain accurate candidate license plate window.
In one embodiment, the second detection model is also to be made of convolutional layer, the image of the second detection model input
Width be 36, a height of 24, comprising: three convolutional layers and a full articulamentum, the convolution of the first convolutional layer and the second convolutional layer
That core is selected is 3X3, and third convolutional layer is using 2X2.
Step 104D is corrected processing to the candidate license plate window of the second detection model output, obtains the second candidate vehicle
Board window.
Wherein, similarly, processing is corrected to the candidate license plate window of the second detection model output, obtains the second candidate
License plate window.The mode of correction can be corrected candidate license plate window using license plate frame regression vector.Similarly may be used later
To merge the high candidate license plate window of Duplication using NMS, the second candidate license plate window is finally obtained.
Step 104E, using the second candidate license plate window as the input of third detection model, third detection model is for knowing
Other license plate area, obtains the position on four license plate vertex.
Wherein, third detection model is to use the model for having monitor mode training to obtain, for identification license plate area, then
Determine the position on four license plate vertex.In order to improve the accuracy of identification, third detection model needs to use when training
More training samples are trained, and the sample of the third detection model is labeled as the position coordinates on four license plate vertex, so instruction
The third detection model got can directly obtain the position on four license plate vertex.
In one embodiment, third detection model includes: four convolutional layers and 1 full articulamentum, the first convolutional layer,
The convolution kernel of two convolutional layers and third convolutional layer is 3X3, and the convolution kernel of Volume Four lamination is 2X2, the input of third detection model
Image width be 48, be highly 72.
It is real that above-mentioned first detection model, the second detection model and third detection model all use convolutional neural networks model
It is existing.It is realized jointly to the accurate of four vertex of license plate by the first detection model, the second detection model and third detection model
Positioning, is conducive to the accuracy for improving subsequent identification.
In one embodiment, the candidate license plate window for obtaining the output of the first detection model carries out candidate license plate window
Correction process obtains the first candidate license plate window, comprising: obtains the candidate license plate window and license plate frame of the output of the first detection model
Regression vector；Candidate license plate window is corrected using license plate frame regression vector, obtains the first candidate license plate window.
Wherein, while the first detection model output candidate license plate window, license plate frame regression vector is also outputed, using vehicle
Board frame regression vector is corrected candidate license plate window, obtains the first candidate license plate window.
In one embodiment, license plate image is corrected according to the position on four license plate vertex, after being corrected
Target license plate image, comprising: perspective transformation matrix is calculated according to the position on four license plate vertex；According to perspective transformation matrix
License plate image is corrected, target license plate image is obtained.
It is known that the position coordinates (i.e. source coordinate) on four license plate vertex that detection obtains, and be arranged and need to convert
Four license plate vertex target location coordinate (i.e. coordinates of targets), then can be sat according to four source coordinates and four targets
Perspective transformation matrix is calculated in mark.It later can be to each in license plate image according to the perspective transformation matrix being calculated
Pixel coordinate is converted, the target license plate image after being corrected.
The formula of perspective transform can be expressed as follows:
(u, v, w) refers to the coordinate before transformation, and (x', y', z') refers to transformed coordinate, for two dimensional image, w=1.
Wherein, x=x '/w ', y=y '/w '.
As shown in figure 4, in one embodiment, detecting to the image for including license plate, the candidate detected is obtained
License plate area, comprising:
Step 102A, acquisition include the original image of license plate, are pre-processed to original image, and pretreatment includes: ash
At least one of degree processing, scaling processing, denoising.
Wherein, the process detected to the image of license plate can be put into headend equipment, i.e. terminal, can also be placed on backstage
Equipment.Original image refers to that shooting obtained includes the original image of license plate.Gray proces, which refer to, converts the image into gray scale
Image, the in this way speed convenient for improving detection.Scaling processing, which refers to, scales the images to preset size.Denoising refers to pair
Noise in image is removed.Denoising can be handled using gaussian filtering convolution.
Step 102B carries out vertical edge detection to pretreated image, determines the height and width of image.
Wherein, vertical edge detection is the vertical edge figure of detection image, four sides of image is determined, so that it is determined that image
Height and width.Vertical edge detection can be filtered using filter.
Step 102C divides an image into multiple images block according to the height of image and width.
Wherein, in order to more accurately highlight characteristics of image, before carrying out binaryzation to image, according to image
Height and width divide an image into multiple images block, carry out binaryzation for each image block convenient for subsequent.If it is whole into
Row binaryzation is easily lost many information, influences subsequent Car license recognition.
In one embodiment, it is assumed that the width of image is W, a height of H, wide and height is divided into m parts and n parts respectively, in total
It is divided into m*n block region.
Step 102D determines the corresponding binarization threshold of each image block according to the pixel value in image block.
Wherein, before carrying out binaryzation to each image block, it is first determined the corresponding binarization threshold of each image block.In
In one embodiment, the pixel value of each pixel in image block can be summed up, then average, average value is made
For the binarization threshold of image block.The pixel that will be greater than binarization threshold is set as 255, is otherwise set as 0.
In another embodiment, max pixel value and minimum pixel value in each image block are obtained respectively, by maximum picture
Plain value and minimum pixel value are averagely obtained corresponding comparison value.The calculating of comparison value can be indicated using following formula:
Thri=(max+min)/2
Wherein, max is the max pixel value in the region, and min is the minimum pixel value in the region, then utilizes following public affairs
Formula carries out binarization operation to each point in region:
Wherein, B is the value of each pixel of binary picture, and v is the pixel value at image midpoint, ThriFor belonging to the pixel
Piece comparison value, T be binaryzation threshold value, 5 can be taken here, can also be adjusted according to scene adaptive.Pass through calculating
The absolute value of difference, is then greater than the two-value of setting by the absolute value of the difference of the corresponding pixel value of each pixel and comparison value
The pixel for changing threshold value is set as 255, is otherwise set as 0.
Step 102E carries out at binaryzation image block according to the comparison value of each image block and preset binarization threshold
Reason, obtains binary image.
Wherein, binary conversion treatment is carried out to image block according to the comparison value of each image block and preset binarization threshold,
To obtain binary image.
Step 102F detects the license plate area in binary image, obtains candidate license plate region.
Wherein, the detection of license plate area is carried out for binary image, obtains the candidate license plate region that detection obtains.License plate
The detection in region can be detected using existing mode, for example, can be detected using the retrieval mode of connected region,
Here the method for detection is not limited.
The process in above-mentioned acquisition candidate license plate region can more accurately obtain candidate license plate region, subsequent convenient for improving
The accuracy of Car license recognition.
In one embodiment, the license plate area in binary image is detected, obtains candidate license plate region, wrapped
It includes: corrosion expansion process is carried out to each region detected using Morphology Algorithm to binary image, obtain including to connect
The image in logical region；Connected region search is carried out to the image for including connected region, obtains suspected license plate area；According to each
The textural characteristics of suspected license plate area are detected, and are removed pseudo- license plate area, are obtained candidate license plate region.
Wherein, the region of fracture can be merged using Morphology Algorithms such as corrosion expansions, while eliminates small region
Interference, obtain include connected region image.Then eight connectivity range searching algorithm can be used, to after treatment
Image carries out connected region search, searches suspected license plate area, is then identified using textural characteristics, and pseudo- license plate area is removed
Domain finally obtains candidate license plate region.
Textural characteristics identification specifically can be in the following way: suspected license plate area being carried out binary conversion treatment, is then united
The line number L that jump in the region is greater than given threshold is counted, if line number L removes the suspected license plate area less than preset threshold T,
The half that preset threshold T can take region high.
In one embodiment, the character in target license plate image is identified, obtains recognition result, comprising: to mesh
Character in mark license plate image is detected and is divided, and characters on license plate one by one is obtained；
Using characters on license plate as the input of character recognition model, the recognition result of character recognition model output, character are obtained
Identification model is obtained using the training of deep learning algorithm.
It wherein, can be using connected region search and upright projection method, according to what is detected to the method for character machining
Character is split, and obtains character one by one, is then identified to each character.Training obtains character recognition model in advance,
The characters on license plate that segmentation is obtained obtains the recognition result of character recognition model output as the input of character recognition model.
In another embodiment, the character in target license plate image is identified, obtains recognition result, comprising: is right
Character in target license plate image is detected and is divided, and characters on license plate image one by one is obtained；Characters on license plate image is carried out
Normalized obtains target license plate character picture；Feature extraction is carried out to target license plate character picture, according to the feature of extraction
Carry out character recognition；The confidence level for obtaining character recognition, when confidence level is less than preset threshold, by corresponding target license plate character
Input of the image as character recognition model, obtains the recognition result of character recognition model output, and character recognition model is to use
The training of deep learning algorithm obtains.Characters on license plate image is normalized to obtain target license plate character picture；To mesh
It marks characters on license plate image and carries out feature extraction, character recognition is carried out according to the feature of extraction；The confidence level for obtaining character recognition, when
When confidence level is less than preset threshold, using corresponding target license plate character picture as the input of character recognition model, character is obtained
The recognition result of identification model output, character recognition model are obtained using the training of deep learning algorithm.
Wherein it is possible to which dividing the character to get off uniformly normalizes to width 48,24 pixel sizes of height, character picture is extracted
Gabor characteristic (Gabor characteristic is a kind of feature that can be used to describe image texture information), using SVM (Support
Vector Machine, refers to support vector machines, is a kind of common method of discrimination) character recognition is carried out, work as character recognition
Confidence level be less than given threshold when, the character is recognized using convolutional neural networks (i.e. character recognition model),
Then result is exported.It combines SVM method very high to clearly character recognition accuracy rate it is demonstrated experimentally that Gabor characteristic is extracted, and rolls up
The product neural network character recognition effect low to dirty, contrast is preferable, therefore the two is combined, the whole recognition accuracy of character
99.9% can be greater than.
As shown in figure 5, in one embodiment, giving a kind of process schematic of Car license recognition.It include: three ranks
Section, first stage, car plate detection, second stage, license plate correction, phase III, Car license recognition.
The car plate detection of first stage includes: image preprocessing, vertical edge detection, piecemeal binaryzation, Morphological scale-space,
It searches for license plate area and removes six processes such as pseudo- license plate area.
The license plate correction of second stage includes: that four license plate vertex of identification and license plate correct two processes.
The Car license recognition of phase III includes: two processes of Character segmentation and character recognition.
In one embodiment, car plate detection can be put into headend equipment to go to complete, then by license plate correction and license plate
Identification is placed on cloud (i.e. server) and goes to realize.In the first stage if do not detected comprising license plate, do not have to upload to again
Cloud.
As shown in fig. 6, in one embodiment it is proposed that a kind of license plate recognition device, comprising:
It obtains module 602 and obtains the candidate license plate region detected for detecting to the image for including license plate；
Detection module 604, for using the candidate license plate region as the input of license plate vertex detection model, described in acquisition
The position on four license plate vertex that the detection of license plate vertex detection model output obtains；
Rectification module 606 is corrected for being corrected according to the position on four license plate vertex to license plate image
Target license plate image afterwards；
Identification module 608 obtains recognition result for identifying to the characters on license plate in the target license plate image.
As shown in fig. 7, in one embodiment, license plate vertex detection model includes: the first detection model, the second inspection
Survey model and third detection model；Detection module 604 includes:
First detection module 604A, for using the candidate license plate region as the input of first detection model, institute
It states the first detection model and generates candidate license plate window for carrying out detection to the candidate license plate region；Obtain first detection
The candidate license plate window of model output, is corrected processing to the candidate license plate window, obtains the first candidate license plate window；
Second detection module 604B, for using the first candidate license plate window as the defeated of second detection model
Enter, second detection model is for screening the first candidate license plate window；Second detection model is exported
Candidate license plate window be corrected processing, obtain the second candidate license plate window；
Third detection module 604C, for using the second candidate license plate window as the defeated of the third detection model
Enter, third detection model license plate area for identification obtains the position on four license plate vertex.
In one embodiment, the first detection module 604A is also used to obtain the time of the first detection model output
Select license plate window and license plate frame regression vector；The license plate window is corrected using the license plate frame regression vector, is obtained
The first candidate license plate window.
In one embodiment, rectification module is also used to be calculated perspective according to the position on four license plate vertex and becomes
Change matrix；The license plate image is corrected according to the perspective transformation matrix, obtains target license plate image.
As shown in figure 8, in one embodiment, obtaining module 602 includes:
Preprocessing module 602A, for obtain include license plate original image, the original image is pre-processed,
The pretreatment includes: at least one of gray proces, scaling processing, denoising；
Edge detection module 602B determines described image for carrying out vertical edge detection to pretreated image
Height and width；
Piecemeal binarization block 602C, for described image to be divided into multiple figures according to the height and width of described image
As block；The corresponding comparison value of each image block is determined according to the pixel value in image block；According to the comparison value of each image block and
Preset binarization threshold carries out binary conversion treatment to image block, obtains binary image；
Region detection module 602D obtains candidate vehicle for detecting to the license plate area in the binary image
Board region.
In one embodiment, the region detection module 602D is also used to calculate the binary image using morphology
Method carries out corrosion expansion process to each region detected, obtain include connected region image；To it is described include connect
The image in logical region carries out connected region search, obtains suspected license plate area；According to the textural characteristics of each suspected license plate area
It is detected, removes pseudo- license plate area, obtain candidate license plate region.
In one embodiment, identification module is also used to that the character in the target license plate image is detected and divided
It cuts, obtains characters on license plate image one by one；The characters on license plate image is normalized to obtain target license plate character figure
Picture；Feature extraction is carried out to the target license plate character picture, character recognition is carried out according to the feature of extraction；Obtain character recognition
Confidence level, when the confidence level be less than preset threshold when, using corresponding target license plate character picture as character recognition model
Input, obtain the recognition result of character recognition model output, the character recognition model is using deep learning algorithm
What training obtained.
Fig. 9 shows the internal structure chart of computer equipment in one embodiment.The computer equipment specifically can be clothes
Business device and terminal device, the server include but is not limited to high-performance computer and high-performance computer cluster；The terminal
Equipment includes but is not limited to mobile terminal device and terminal console equipment, the mobile terminal device include but is not limited to mobile phone,
Tablet computer, smartwatch and laptop, the terminal console equipment includes but is not limited to desktop computer and vehicle-mounted computer.
As shown in figure 9, the computer equipment includes processor, memory and the network interface connected by system bus.Wherein, it stores
Device includes non-volatile memory medium and built-in storage.The non-volatile memory medium of the computer equipment is stored with operation system
System, can also be stored with computer program, when which is executed by processor, processor may make to realize Car license recognition side
Method.Computer program can also be stored in the built-in storage, when which is executed by processor, processor may make to hold
Row licence plate recognition method.It will be understood by those skilled in the art that structure shown in Fig. 9, only related to application scheme
Part-structure block diagram, do not constitute the restriction for the computer equipment being applied thereon to application scheme, it is specific to count
Calculating machine equipment may include perhaps combining certain components or with different portions than more or fewer components as shown in the figure
Part arrangement.
In one embodiment, licence plate recognition method provided by the present application can be implemented as a kind of shape of computer program
Formula, computer program can be run in computer equipment as shown in Figure 9.Composition vehicle can be stored in the memory of computer equipment
Each process template of board identification device.For example, obtaining module 602, detection module 604, rectification module 606, identification module
608。
A kind of computer equipment, including memory, processor and storage are in the memory and can be in the processing
The computer program run on device, the processor realize following steps when executing the computer program: to including license plate
Image detected, obtain the candidate license plate region detected；Mould is detected using the candidate license plate region as license plate vertex
The input of type obtains the position on four license plate vertex that the detection that license plate vertex detection model exports obtains；According to described
License plate image is corrected in the position on four license plate vertex, the target license plate image after being corrected；To the target license plate
Characters on license plate in image is identified, recognition result is obtained.
In one embodiment, license plate vertex detection model includes: the first detection model, the second detection model and
Three detection models；It is described using the candidate license plate region as the input of license plate vertex detection model, obtain the license plate vertex
The position on four license plate vertex that the detection of detection model output obtains, comprising: using the candidate license plate region as described the
The input of one detection model, first detection model are used to carry out the candidate license plate region detection to generate candidate license plate window
Mouthful；The candidate license plate window for obtaining the first detection model output, is corrected processing to the candidate license plate window, obtains
First candidate license plate window；Using the first candidate license plate window as the input of second detection model, second inspection
Model is surveyed for screening to the first candidate license plate window；To the candidate license plate window of second detection model output
It is corrected processing, obtains the second candidate license plate window；Using the second candidate license plate window as the third detection model
Input, third detection model license plate area for identification obtains the position on four license plate vertex.
In one embodiment, the candidate license plate window for obtaining the first detection model output, to the candidate
License plate window is corrected processing, obtains the first candidate license plate window, comprising: obtains the candidate of the first detection model output
License plate window and license plate frame regression vector；The license plate window is corrected using the license plate frame regression vector, obtains institute
State the first candidate license plate window.
In one embodiment, license plate image is corrected in the position according to four license plate vertex, obtains
Target license plate image after correction, comprising: perspective transformation matrix is calculated according to the position on four license plate vertex；According to
The perspective transformation matrix corrects the license plate image, obtains target license plate image.
In one embodiment, described pair includes that the image of license plate detects, and obtains the candidate license plate area detected
Domain, comprising: acquisition includes the original image of license plate, is pre-processed to the original image, and the pretreatment includes: gray scale
At least one of processing, scaling processing, denoising；To pretreated image carry out vertical edge detection, determine described in
The height and width of image；Described image is divided into multiple images block according to the height of described image and width；According to image
Pixel value in block determines the corresponding comparison value of each image block；According to the comparison value of each image block and preset binaryzation threshold
Value carries out binary conversion treatment to image block, obtains binary image；License plate area in the binary image is detected,
Obtain candidate license plate region.
In one embodiment, the license plate area in the binary image detects, and obtains candidate license plate
Region, comprising: corrosion expansion process is carried out to each region detected using Morphology Algorithm to the binary image, is obtained
To the image for including connected region；To it is described include connected region image carry out connected region search, obtain doubtful vehicle
Board region；It is detected according to the textural characteristics of each suspected license plate area, removes pseudo- license plate area, obtain candidate license plate area
Domain.
In one embodiment, the character in the target license plate image identifies, obtains recognition result, packet
It includes: the character in the target license plate image being detected and divided, characters on license plate image one by one is obtained；To the license plate
Character picture is normalized to obtain target license plate character picture；Feature is carried out to the target license plate character picture to mention
It takes, character recognition is carried out according to the feature of extraction；The confidence level for obtaining character recognition, when the confidence level is less than preset threshold
When, using corresponding target license plate character picture as the input of character recognition model, obtain the character recognition model output
Recognition result, the character recognition model are obtained using the training of deep learning algorithm.
A kind of computer readable storage medium, the computer-readable recording medium storage have computer program, feature
It is, the computer program realizes following steps when being executed by processor: the image for including license plate is detected, obtains
The candidate license plate region detected；Using the candidate license plate region as the input of license plate vertex detection model, the vehicle is obtained
The position on four license plate vertex that the detection of board vertex detection model output obtains；According to the position pair on four license plate vertex
License plate image is corrected, the target license plate image after being corrected；Characters on license plate in the target license plate image is carried out
Identification, obtains recognition result.
In one embodiment, license plate vertex detection model includes: the first detection model, the second detection model and
Three detection models；It is described using the candidate license plate region as the input of license plate vertex detection model, obtain the license plate vertex
The position on four license plate vertex that the detection of detection model output obtains, comprising: using the candidate license plate region as described the
The input of one detection model, first detection model are used to carry out the candidate license plate region detection to generate candidate license plate window
Mouthful；The candidate license plate window for obtaining the first detection model output, is corrected processing to the candidate license plate window, obtains
First candidate license plate window；Using the first candidate license plate window as the input of second detection model, second inspection
Model is surveyed for screening to the first candidate license plate window；To the candidate license plate window of second detection model output
It is corrected processing, obtains the second candidate license plate window；Using the second candidate license plate window as the third detection model
Input, third detection model license plate area for identification obtains the position on four license plate vertex.
In one embodiment, the candidate license plate window for obtaining the first detection model output, to the candidate
License plate window is corrected processing, obtains the first candidate license plate window, comprising: obtains the candidate of the first detection model output
License plate window and license plate frame regression vector；The license plate window is corrected using the license plate frame regression vector, obtains institute
State the first candidate license plate window.
In one embodiment, license plate image is corrected in the position according to four license plate vertex, obtains
Target license plate image after correction, comprising: perspective transformation matrix is calculated according to the position on four license plate vertex；According to
The perspective transformation matrix corrects the license plate image, obtains target license plate image.
In one embodiment, described pair includes that the image of license plate detects, and obtains the candidate license plate area detected
Domain, comprising: acquisition includes the original image of license plate, is pre-processed to the original image, and the pretreatment includes: gray scale
At least one of processing, scaling processing, denoising；To pretreated image carry out vertical edge detection, determine described in
The height and width of image；Described image is divided into multiple images block according to the height of described image and width；According to image
Pixel value in block determines the corresponding comparison value of each image block；According to the comparison value of each image block and preset binaryzation threshold
Value carries out binary conversion treatment to image block, obtains binary image；License plate area in the binary image is detected,
Obtain candidate license plate region.
In one embodiment, the license plate area in the binary image detects, and obtains candidate license plate
Region, comprising: corrosion expansion process is carried out to each region detected using Morphology Algorithm to the binary image, is obtained
To the image for including connected region；To it is described include connected region image carry out connected region search, obtain doubtful vehicle
Board region；It is detected according to the textural characteristics of each suspected license plate area, removes pseudo- license plate area, obtain candidate license plate area
Domain.
In one embodiment, the character in the target license plate image identifies, obtains recognition result, packet
It includes: the character in the target license plate image being detected and divided, characters on license plate image one by one is obtained；To the license plate
Character picture is normalized to obtain target license plate character picture；Feature is carried out to the target license plate character picture to mention
It takes, character recognition is carried out according to the feature of extraction；The confidence level for obtaining character recognition, when the confidence level is less than preset threshold
When, using corresponding target license plate character picture as the input of character recognition model, obtain the character recognition model output
Recognition result, the character recognition model are obtained using the training of deep learning algorithm.
It should be noted that above-mentioned licence plate recognition method, license plate recognition device, computer equipment and computer-readable storage
Medium belongs to a total inventive concept, licence plate recognition method, license plate recognition device, computer equipment and computer-readable storage
Content in media embodiment can be mutually applicable in.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the program can be stored in a non-volatile computer and can be read
In storage medium, the program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, provided herein
Each embodiment used in any reference to memory, storage, database or other media, may each comprise non-volatile
And/or volatile memory.Nonvolatile memory may include that read-only memory (ROM), programming ROM (PROM), electricity can be compiled
Journey ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include random access memory
(RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms, such as static state RAM
(SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhanced SDRAM
(ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) directly RAM (RDRAM), straight
Connect memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
The limitation to the application the scope of the patents therefore cannot be interpreted as.It should be pointed out that for those of ordinary skill in the art
For, without departing from the concept of this application, various modifications and improvements can be made, these belong to the guarantor of the application
Protect range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.It please input specific implementation content part.
Claims (10)
1. a kind of licence plate recognition method, which is characterized in that the described method includes:
The image for including license plate is detected, the candidate license plate region detected is obtained；
Using the candidate license plate region as the input of license plate vertex detection model, the output of license plate vertex detection model is obtained
The obtained position on four license plate vertex of detection；
License plate image is corrected according to the position on four license plate vertex, the target license plate image after being corrected；
Characters on license plate in the target license plate image is identified, recognition result is obtained.
2. the method according to claim 1, wherein license plate vertex detection model includes: the first detection mould
Type, the second detection model and third detection model；
It is described using the candidate license plate region as the input of license plate vertex detection model, obtain license plate vertex detection model
The position on four license plate vertex that the detection of output obtains, comprising:
Using the candidate license plate region as the input of first detection model, first detection model is used for the time
It selects license plate area to carry out detection and generates candidate license plate window；
The candidate license plate window for obtaining the first detection model output, is corrected processing to the candidate license plate window, obtains
To the first candidate license plate window；
Using the first candidate license plate window as the input of second detection model, second detection model is used for institute
The first candidate license plate window is stated to be screened；
Processing is corrected to the candidate license plate window of second detection model output, obtains the second candidate license plate window；
Using the second candidate license plate window as the input of the third detection model, the third detection model is for identification
License plate area obtains the position on four license plate vertex.
3. according to the method described in claim 2, it is characterized in that, the candidate vehicle for obtaining the first detection model output
Board window is corrected processing to the candidate license plate window, obtains the first candidate license plate window, comprising:
Obtain the candidate license plate window and license plate frame regression vector of the first detection model output；
The license plate window is corrected using the license plate frame regression vector, obtains the first candidate license plate window.
4. the method according to claim 1, wherein the position according to four license plate vertex is to license plate
Image is corrected, the target license plate image after being corrected, comprising:
Perspective transformation matrix is calculated according to the position on four license plate vertex；
The license plate image is corrected according to the perspective transformation matrix, obtains target license plate image.
5. being obtained the method according to claim 1, wherein the image that described pair includes license plate detects
The candidate license plate region detected, comprising:
Acquisition includes the original image of license plate, is pre-processed to the original image, and the pretreatment includes: at gray scale
At least one of reason, scaling processing, denoising；
Vertical edge detection is carried out to pretreated image, determines the height and width of described image；
Described image is divided into multiple images block according to the height of described image and width；
The corresponding comparison value of each image block is determined according to the pixel value in image block；
Binary conversion treatment is carried out to image block according to the comparison value of each image block and preset binarization threshold, obtains binaryzation
Image；
License plate area in the binary image is detected, candidate license plate region is obtained.
6. according to the method described in claim 5, it is characterized in that, the license plate area in the binary image carries out
Detection, obtains candidate license plate region, comprising:
Corrosion expansion process is carried out to each region detected using Morphology Algorithm to the binary image, is included
There is the image of connected region；
To it is described include connected region image carry out connected region search, obtain suspected license plate area；
It is detected according to the textural characteristics of each suspected license plate area, removes pseudo- license plate area, obtain candidate license plate region.
7. the method according to claim 1, wherein the character in the target license plate image is known
Not, recognition result is obtained, comprising:
Character in the target license plate image is detected and divided, characters on license plate image one by one is obtained；
The characters on license plate image is normalized to obtain target license plate character picture；
Feature extraction is carried out to the target license plate character picture, character recognition is carried out according to the feature of extraction；
The confidence level for obtaining character recognition, when the confidence level is less than preset threshold, by corresponding target license plate character picture
As the input of character recognition model, the recognition result of the character recognition model output is obtained, the character recognition model is
It is obtained using the training of deep learning algorithm.
8. a kind of license plate recognition device, which is characterized in that described device includes:
It obtains module and obtains the candidate license plate region detected for detecting to the image for including license plate；
Detection module, for obtaining the license plate top using the candidate license plate region as the input of license plate vertex detection model
The position on four license plate vertex that the detection of point detection model output obtains；
Rectification module, for being corrected according to the position on four license plate vertex to license plate image, the mesh after being corrected
Mark license plate image；
Identification module obtains recognition result for identifying to the characters on license plate in the target license plate image.
9. a kind of computer equipment, which is characterized in that in the memory and can be including memory, processor and storage
The computer program run on the processor, which is characterized in that the processor is realized such as when executing the computer program
The step of any one of claim 1 to 7 licence plate recognition method.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has computer program, and feature exists
In the step of realization licence plate recognition method as described in any one of claim 1 to 7 when the computer program is executed by processor
Suddenly.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910626156.3A CN110414507B (en) | 2019-07-11 | 2019-07-11 | License plate recognition method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910626156.3A CN110414507B (en) | 2019-07-11 | 2019-07-11 | License plate recognition method and device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110414507A true CN110414507A (en) | 2019-11-05 |
| CN110414507B CN110414507B (en) | 2022-07-26 |
Family
ID=68361096
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910626156.3A Active CN110414507B (en) | 2019-07-11 | 2019-07-11 | License plate recognition method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110414507B (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079744A (en) * | 2019-12-06 | 2020-04-28 | 鲁东大学 | Intelligent vehicle license plate identification method and device suitable for complex illumination environment |
| CN111126383A (en) * | 2019-12-06 | 2020-05-08 | 广州烽火众智数字技术有限公司 | License plate detection method, system, device and storage medium |
| CN111191653A (en) * | 2019-12-25 | 2020-05-22 | 北京精英路通科技有限公司 | License plate recognition method and device, computer equipment and storage medium |
| CN111340045A (en) * | 2020-02-12 | 2020-06-26 | 浙江大华技术股份有限公司 | License plate number recognition method and device and storage medium |
| CN111695561A (en) * | 2020-05-25 | 2020-09-22 | 南京博雅集智智能技术有限公司 | License plate detection and correction recognition method and recognition system based on SSD |
| CN111914834A (en) * | 2020-06-18 | 2020-11-10 | 绍兴埃瓦科技有限公司 | Image recognition method and device, computer equipment and storage medium |
| CN112434700A (en) * | 2020-11-25 | 2021-03-02 | 创新奇智(上海)科技有限公司 | License plate recognition method, device, equipment and storage medium |
| CN112560754A (en) * | 2020-12-23 | 2021-03-26 | 北京百度网讯科技有限公司 | Bill information acquisition method, device, equipment and storage medium |
| CN112580648A (en) * | 2020-12-14 | 2021-03-30 | 成都中科大旗软件股份有限公司 | Method for realizing image information identification based on image segmentation technology |
| CN112634141A (en) * | 2020-12-23 | 2021-04-09 | 浙江大华技术股份有限公司 | License plate correction method, device, equipment and medium |
| 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 |
| CN112926583A (en) * | 2021-04-25 | 2021-06-08 | 南京甄视智能科技有限公司 | License plate recognition method and license plate recognition system |
| CN113095320A (en) * | 2021-04-01 | 2021-07-09 | 湖南大学 | License plate recognition method and system and computing device |
| CN113313124A (en) * | 2021-07-29 | 2021-08-27 | 佛山市墨纳森智能科技有限公司 | Method and device for identifying license plate number based on image segmentation algorithm and terminal equipment |
| CN113326836A (en) * | 2020-02-28 | 2021-08-31 | 深圳市丰驰顺行信息技术有限公司 | License plate recognition method and device, server and storage medium |
| CN115019297A (en) * | 2022-08-04 | 2022-09-06 | 之江实验室 | Real-time license plate detection and identification method and device based on color augmentation |
| CN117409376A (en) * | 2023-12-15 | 2024-01-16 | 南京中鑫智电科技有限公司 | Infrared online monitoring method and system for high-voltage sleeve |
Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101408977A (en) * | 2008-11-24 | 2009-04-15 | 东软集团股份有限公司 | Method and apparatus for dividing candidate barrier region |
| CN102156868A (en) * | 2011-03-31 | 2011-08-17 | 汉王科技股份有限公司 | Image binaryzation method and device |
| CN102163284A (en) * | 2011-04-11 | 2011-08-24 | 西安电子科技大学 | Chinese environment-oriented complex scene text positioning method |
| CN102426649A (en) * | 2011-10-13 | 2012-04-25 | 石家庄开发区冀科双实科技有限公司 | Simple steel seal digital automatic identification method with high accuracy rate |
| CN103065138A (en) * | 2012-12-06 | 2013-04-24 | 新疆公众信息产业股份有限公司 | Recognition method of license plate number of motor vehicle |
| CN103093181A (en) * | 2011-11-01 | 2013-05-08 | 青岛海信网络科技股份有限公司 | License plate image locating method and device |
| CN104050450A (en) * | 2014-06-16 | 2014-09-17 | 西安通瑞新材料开发有限公司 | Vehicle license plate recognition method based on video |
| CN104392205A (en) * | 2014-10-24 | 2015-03-04 | 浙江力石科技股份有限公司 | Abnormal vehicle license plate recognition method and system |
| CN105260701A (en) * | 2015-09-14 | 2016-01-20 | 中电海康集团有限公司 | Front vehicle detection method applied to complex scene |
| CN105320953A (en) * | 2015-09-28 | 2016-02-10 | 万永秀 | License plate recognition method |
| US20170177965A1 (en) * | 2015-12-17 | 2017-06-22 | Xerox Corporation | Coarse-to-fine cascade adaptations for license plate recognition with convolutional neural networks |
| CN107229929A (en) * | 2017-04-12 | 2017-10-03 | 西安电子科技大学 | A kind of license plate locating method based on R CNN |
| CN107506763A (en) * | 2017-09-05 | 2017-12-22 | 武汉大学 | A kind of multiple dimensioned car plate precise positioning method based on convolutional neural networks |
| CN107679531A (en) * | 2017-06-23 | 2018-02-09 | 平安科技（深圳）有限公司 | Licence plate recognition method, device, equipment and storage medium based on deep learning |
-
2019
- 2019-07-11 CN CN201910626156.3A patent/CN110414507B/en active Active
Patent Citations (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101408977A (en) * | 2008-11-24 | 2009-04-15 | 东软集团股份有限公司 | Method and apparatus for dividing candidate barrier region |
| CN102156868A (en) * | 2011-03-31 | 2011-08-17 | 汉王科技股份有限公司 | Image binaryzation method and device |
| CN102163284A (en) * | 2011-04-11 | 2011-08-24 | 西安电子科技大学 | Chinese environment-oriented complex scene text positioning method |
| CN102426649A (en) * | 2011-10-13 | 2012-04-25 | 石家庄开发区冀科双实科技有限公司 | Simple steel seal digital automatic identification method with high accuracy rate |
| CN103093181A (en) * | 2011-11-01 | 2013-05-08 | 青岛海信网络科技股份有限公司 | License plate image locating method and device |
| CN103065138A (en) * | 2012-12-06 | 2013-04-24 | 新疆公众信息产业股份有限公司 | Recognition method of license plate number of motor vehicle |
| CN104050450A (en) * | 2014-06-16 | 2014-09-17 | 西安通瑞新材料开发有限公司 | Vehicle license plate recognition method based on video |
| CN104392205A (en) * | 2014-10-24 | 2015-03-04 | 浙江力石科技股份有限公司 | Abnormal vehicle license plate recognition method and system |
| CN105260701A (en) * | 2015-09-14 | 2016-01-20 | 中电海康集团有限公司 | Front vehicle detection method applied to complex scene |
| CN105320953A (en) * | 2015-09-28 | 2016-02-10 | 万永秀 | License plate recognition method |
| US20170177965A1 (en) * | 2015-12-17 | 2017-06-22 | Xerox Corporation | Coarse-to-fine cascade adaptations for license plate recognition with convolutional neural networks |
| CN107229929A (en) * | 2017-04-12 | 2017-10-03 | 西安电子科技大学 | A kind of license plate locating method based on R CNN |
| CN107679531A (en) * | 2017-06-23 | 2018-02-09 | 平安科技（深圳）有限公司 | Licence plate recognition method, device, equipment and storage medium based on deep learning |
| CN107506763A (en) * | 2017-09-05 | 2017-12-22 | 武汉大学 | A kind of multiple dimensioned car plate precise positioning method based on convolutional neural networks |
Non-Patent Citations (6)
| Title |
|---|
| GABRIEL RESENDE GONÇALVES ET.AL: "Real-Time Automatic License Plate Recognition through Deep Multi-Task Networks", 《2018 31ST SIBGRAPI CONFERENCE ON GRAPHICS, PATTERNS AND IMAGES (SIBGRAPI)》 * |
| MEHRARA 等: "Nonlinear Edge Detection Based on Image Density Transform", 《RESEARCHGATE》 * |
| MENG DONG ET.AL: "A CNN-Based Approach for Automatic License Plate Recognition in the Wild", 《BRITISH MACHINE VISION CONFERENCE 2017》 * |
| 崔吉 等: "《工业视觉实用教程》", 31 October 2018, 上海交通大学出版社 * |
| 李松柏: "复杂场景下车牌识别方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 贾志刚: "图像二值化-局部阈值方法汇总", 《HTTPS://CLOUD.TENCENT.COM/DEVELOPER/ARTICLE/1084305》 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111079744A (en) * | 2019-12-06 | 2020-04-28 | 鲁东大学 | Intelligent vehicle license plate identification method and device suitable for complex illumination environment |
| CN111126383A (en) * | 2019-12-06 | 2020-05-08 | 广州烽火众智数字技术有限公司 | License plate detection method, system, device and storage medium |
| CN111191653A (en) * | 2019-12-25 | 2020-05-22 | 北京精英路通科技有限公司 | License plate recognition method and device, computer equipment and storage medium |
| CN111340045A (en) * | 2020-02-12 | 2020-06-26 | 浙江大华技术股份有限公司 | License plate number recognition method and device and storage medium |
| CN111340045B (en) * | 2020-02-12 | 2023-09-01 | 浙江大华技术股份有限公司 | License plate number recognition method, device and storage medium |
| CN113326836A (en) * | 2020-02-28 | 2021-08-31 | 深圳市丰驰顺行信息技术有限公司 | License plate recognition method and device, server and storage medium |
| CN111695561A (en) * | 2020-05-25 | 2020-09-22 | 南京博雅集智智能技术有限公司 | License plate detection and correction recognition method and recognition system based on SSD |
| CN111914834A (en) * | 2020-06-18 | 2020-11-10 | 绍兴埃瓦科技有限公司 | Image recognition method and device, computer equipment and storage medium |
| CN111914834B (en) * | 2020-06-18 | 2024-04-02 | 绍兴埃瓦科技有限公司 | Image recognition method, device, computer equipment and storage medium |
| CN112434700A (en) * | 2020-11-25 | 2021-03-02 | 创新奇智(上海)科技有限公司 | License plate recognition method, device, equipment and storage medium |
| CN112580648A (en) * | 2020-12-14 | 2021-03-30 | 成都中科大旗软件股份有限公司 | Method for realizing image information identification based on image segmentation technology |
| CN112634141A (en) * | 2020-12-23 | 2021-04-09 | 浙江大华技术股份有限公司 | License plate correction method, device, equipment and medium |
| CN112634141B (en) * | 2020-12-23 | 2024-03-29 | 浙江大华技术股份有限公司 | License plate correction method, device, equipment and medium |
| CN112560754A (en) * | 2020-12-23 | 2021-03-26 | 北京百度网讯科技有限公司 | Bill information acquisition method, device, equipment and storage medium |
| 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 |
| CN113095320A (en) * | 2021-04-01 | 2021-07-09 | 湖南大学 | License plate recognition method and system and computing device |
| CN112926583B (en) * | 2021-04-25 | 2022-08-16 | 南京甄视智能科技有限公司 | License plate recognition method and license plate recognition system |
| CN112926583A (en) * | 2021-04-25 | 2021-06-08 | 南京甄视智能科技有限公司 | License plate recognition method and license plate recognition system |
| CN113313124A (en) * | 2021-07-29 | 2021-08-27 | 佛山市墨纳森智能科技有限公司 | Method and device for identifying license plate number based on image segmentation algorithm and terminal equipment |
| CN115019297A (en) * | 2022-08-04 | 2022-09-06 | 之江实验室 | Real-time license plate detection and identification method and device based on color augmentation |
| CN115019297B (en) * | 2022-08-04 | 2022-12-09 | 之江实验室 | Real-time license plate detection and identification method and device based on color augmentation |
| CN117409376A (en) * | 2023-12-15 | 2024-01-16 | 南京中鑫智电科技有限公司 | Infrared online monitoring method and system for high-voltage sleeve |
| CN117409376B (en) * | 2023-12-15 | 2024-05-10 | 南京中鑫智电科技有限公司 | Infrared online monitoring method and system for high-voltage sleeve |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110414507B (en) | 2022-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110414507A (en) | Licence plate recognition method, device, computer equipment and storage medium | |
| CN112686812B (en) | Bank card inclination correction detection method and device, readable storage medium and terminal | |
| CN111640157B (en) | Checkerboard corner detection method based on neural network and application thereof | |
| US20070058856A1 (en) | Character recoginition in video data | |
| CN111461170A (en) | Vehicle image detection method and device, computer equipment and storage medium | |
| CN109447117B (en) | Double-layer license plate recognition method and device, computer equipment and storage medium | |
| CN107704797B (en) | Real-time detection method, system and equipment based on pedestrians and vehicles in security video | |
| CN111191535B (en) | Pedestrian detection model construction method based on deep learning and pedestrian detection method | |
| CN113888461A (en) | Method, system and equipment for detecting defects of hardware parts based on deep learning | |
| CN113592911A (en) | Apparent enhanced depth target tracking method | |
| CN112598922A (en) | Parking space detection method, device, equipment and storage medium | |
| CN116433666A (en) | Board card line defect online identification method, system, electronic equipment and storage medium | |
| CN111523439A (en) | Method, system, device and medium for target detection based on deep learning | |
| CN111027544A (en) | MSER license plate positioning method and system based on visual saliency detection | |
| CN112686248A (en) | Certificate increase and decrease type detection method and device, readable storage medium and terminal | |
| CN111444916A (en) | License plate positioning and identifying method and system under unconstrained condition | |
| CN117495891B (en) | Point cloud edge detection method and device and electronic equipment | |
| CN108876776B (en) | Classification model generation method, fundus image classification method and device | |
| CN113920434A (en) | Image reproduction detection method, device and medium based on target | |
| CN110458234B (en) | Vehicle searching method with map based on deep learning | |
| CN115797970B (en) | Dense pedestrian target detection method and system based on YOLOv5 model | |
| CN104966064A (en) | Pedestrian ahead distance measurement method based on visual sense | |
| CN110334703B (en) | Ship detection and identification method in day and night image | |
| CN105184809A (en) | Moving object detection method and moving object detection device | |
| CN113657378B (en) | Vehicle tracking method, vehicle tracking system and computing device |
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: 20210430 Address after: A1926, building 9, zone 2, Shenzhen Bay science and technology ecological park, 3609 Baishi Road, high tech Zone community, Yuehai street, Nanshan District, Shenzhen, Guangdong 518000 Applicant after: Shenzhen zhiyouting Technology Co.,Ltd. Address before: 518000 2nd floor, No.5 Zhongxing Road, guantian community, Shiyan street, Bao'an District, Shenzhen City, Guangdong Province Applicant before: HECHANG FUTURE TECHNOLOGY (SHENZHEN) Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |