CN109359651A - A kind of License Plate processor and its location processing method - Google Patents
A kind of License Plate processor and its location processing method Download PDFInfo
- Publication number
- CN109359651A CN109359651A CN201811323756.4A CN201811323756A CN109359651A CN 109359651 A CN109359651 A CN 109359651A CN 201811323756 A CN201811323756 A CN 201811323756A CN 109359651 A CN109359651 A CN 109359651A
- Authority
- CN
- China
- Prior art keywords
- license plate
- candidate
- image
- locating module
- 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
- 238000003672 processing method Methods 0.000 title claims description 11
- 238000004040 coloring Methods 0.000 claims abstract description 60
- 230000000877 morphologic effect Effects 0.000 claims abstract description 15
- 238000001914 filtration Methods 0.000 claims abstract description 7
- 238000000034 method Methods 0.000 claims description 35
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 16
- 238000007781 pre-processing Methods 0.000 claims description 10
- 238000012216 screening Methods 0.000 claims description 8
- 238000001514 detection method Methods 0.000 claims description 7
- 230000002708 enhancing effect Effects 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 7
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000003708 edge detection Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 4
- 238000004364 calculation method Methods 0.000 abstract description 4
- 238000013461 design Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 5
- 230000004807 localization Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 230000033228 biological regulation Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
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
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- 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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- 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)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
A kind of License Plate processor, including license plate MTCNN coarse positioning unit, license plate fine positioning unit, license plate MTCNN coarse positioning unit includes candidate frame generation module, candidate frame merges and regression block, candidate frame generation module includes colouring information locating module, morphology locating module, colouring information locating module, the input terminal of morphology locating module is connected with image collecting device, output end is merged by candidate frame and regression block is connected with license plate fine positioning unit, when use, first, colouring information locating module goes out license plate candidate area according to colouring information feature location, morphology locating module orients license plate candidate area according to morphological image operation, candidate frame merges and regression block carries out the merging and recurrence of candidate frame to obtained license plate candidate area again, subsequent license plate fine positioning unit is to candidate Frame carries out SVM filtering.The design reduces calculation amount while improving positioning accuracy.
Description
Technical field
The invention belongs to technical field of automotive electronics, and in particular to a kind of License Plate processor and its localization process side
Method, suitable for improving the precision of positioning, reducing calculation amount.
Background technique
With the arrival of economic globalization, people's lives level is also greatly improved, and automobile quantity sharply increases
Add, situation so at the same time also brings unprecedented pressure to urban transportation and transport service.Traffic jam, friendship at present
Interpreter thus it is increasingly frequent phenomena such as by the concern of people, therefore, be accurately located with detection vehicle so that traffic is more logical
Freely, have become modern city traffic system primary study object.With the fast development of current science and technology, modern intelligent transportation
System is exactly to transport the integrated applications such as Digital Image Processing, pattern-recognition, computer vision processing technique in modern intelligent transportation
In defeated system, so that traffic control system runs more intelligent, scientific and standardization, to solve traffic fortune
A series of problems present in defeated industry.Wherein, license plate number is the unique information that automobile has, on this basis, vehicle
The pith that board positions and detection technique is indispensable at Car license recognition.
License Plate and identification technology handle monitored license plate image with currently advanced computer vision technique,
License plate number is identified using a large amount of digital image processing techniques, to greatly improve the efficiency of management of vehicle, is saved
Human and material resources so that urban traffic control it is scientific with it is intelligent.In modern society, License Plate and identification technology be
It is widely used in inspection station Vehicular real time monitoring, highway electric charge, monitoring, alarming, steals vehicle recognition, parking
Factory's vehicle safety parking management system, vehicular traffic monitoring violating the regulations, traffic police's checking and administration, wagon flow statistics etc. need License Plate to know
Otherwise, especially in terms of modernization highway realizes non-parking charge.License Plate and identifying system are mainly by scheming
As obtaining, license plate image pretreatment, license plate area positions and the parts such as segmentation, Character segmentation, character recognition form.It is domestic at present
Most common vehicle license location technique mainly has the location algorithm based on cromogram, the location algorithm based on grey scale change, based on mind
The shortcomings that location algorithm etc. through network, the above technology, is to need largely to calculate to realize positioning, and positioning accuracy is that have
Limit.
Summary of the invention
The purpose of the present invention is overcoming the problems, such as that positioning accuracy of the existing technology is inadequate, computationally intensive, one kind is provided
It can be improved the precision of positioning and the lesser License Plate processor of calculation amount and its location processing method.
In order to achieve the above object, technical scheme is as follows:
A kind of License Plate processor, including license plate MTCNN coarse positioning unit, license plate fine positioning unit, the license plate
MTCNN coarse positioning unit includes candidate frame generation module, candidate frame merges and regression block, the candidate frame generation module include
Colouring information locating module, morphology locating module, the signal input of the colouring information locating module, morphology locating module
End be connected with image collecting device, colouring information locating module, morphology locating module signal output end pass through time
Frame merging and regression block is selected to be connected with license plate fine positioning unit；
The colouring information locating module is used to go out license plate candidate area according to colouring information feature location；
The morphology locating module is used to orient license plate candidate area according to morphological image operation；
The vehicle that the candidate frame merges and regression block is used to obtain colouring information locating module, morphology locating module
The merging and recurrence of board candidate region progress candidate frame；
The candidate frame that the license plate fine positioning unit is used to merge candidate frame and regression block obtains carries out SVM filtering,
Obtain accurate license plate area.
The processor further includes Retinex image enhancement module, the signal input of the Retinex image enhancement module
End is connected with image collecting device, the signal output end and colouring information locating module, form of Retinex image enhancement module
Locating module is learned to be connected；
The Retinex image enhancement module is used to calculate under greasy weather, cloudy day, sleet sky, night-time scene using Retinex
Method carries out enhancing processing to the image data of image acquisition device.
A kind of location processing method of License Plate processor, successively the following steps are included:
Step 1, the colouring information locating module go out license plate candidate area, the form according to colouring information feature location
It learns locating module and license plate candidate area is oriented according to morphological image operation；
Step 2, the candidate frame merge and regression block obtains colouring information locating module, morphology locating module
The merging and recurrence of license plate candidate area progress candidate frame；
Step 3, the candidate frame that the license plate fine positioning unit merges candidate frame and regression block obtains carry out SVM mistake
Filter, obtains accurate license plate area, at this point, completing the detection positioning of license plate.
In step 1, the colouring information locating module goes out license plate candidate area according to colouring information feature location and successively wraps
Include following steps:
Image preprocessing and color space conversion are first pre-processed received image data using histogram equalization,
The color space of treated image is switched into HSV from RGB again；
Color threshold segmentation binaryzation, all pixels for successively traversing image, by H value be 200 ﹣ 280 and S value and V value is
The pixel that the pixel or H value of 0.35 ﹣ 1.0 is 30 ﹣ 80 and S value and V value is 0.35 ﹣ 1.0 is labeled as white pixel, is otherwise black
Pixel obtains binary image, wherein white area is license plate candidate area, and black is non-license plate area；
Closed operation first carries out closed operation to binary image, then obtains a rectangular area by license plate candidate area
That is candidate license plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle
The threshold value of area screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate regions into corresponding original image by candidate license plate frame
Domain.
In step 1, the morphology locating module is oriented license plate candidate area according to morphological image operation and is successively wrapped
Include following steps:
Image preprocessing, first by received image data carry out greyscale transformation, recycle Gaussian Blur picture filter out image
Then noise carries out Image Edge-Detection using Sobel algorithm, license plate area is made to be distinguished out；
Image binaryzation converts bianry image for pretreated image using thresholding method；
Closed operation first carries out closed operation to bianry image, then obtains a rectangular area i.e. by license plate candidate area
Candidate license plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle
The threshold value of area screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate regions into corresponding original image by candidate license plate frame
Domain.
The processor further includes Retinex image enhancement module, the signal input of the Retinex image enhancement module
End is connected with image collecting device, the signal output end and colouring information locating module, form of Retinex image enhancement module
Locating module is learned to be connected；
The method also includes image processing step, which is located at before step 1；
Described image processing step are as follows: under greasy weather, cloudy day, sleet sky, night-time scene, Retinex image enhancement module
Enhancing processing is carried out using image data from image collecting device of the Retinex algorithm to acquisition.
In step 2, the candidate frame merges and regression block carries out the merging and recurrence of candidate frame using NMS algorithm.
Compared with prior art, the invention has the benefit that
License plate MTCNN coarse positioning unit includes candidate frame generation module, candidate in a kind of License Plate processor of the present invention
Frame merges and regression block, candidate frame generation module include colouring information locating module, morphology locating module, and colouring information is fixed
Position module, morphology locating module signal input part be connected with image collecting device, colouring information locating module, form
The signal output end for learning locating module is merged by candidate frame and regression block is connected with license plate fine positioning unit, is positioning
In treatment process, firstly, colouring information locating module goes out license plate candidate area, morphology positioning according to colouring information feature location
Module orients license plate candidate area according to morphological image operation, and subsequent candidate frame merges and regression block is fixed to colouring information
The license plate candidate area that position module, morphology locating module obtain carries out the merging and recurrence of candidate frame, and the design combines biography
System license plate locating method and deep learning method respectively obtain candidate vehicle using based on colouring information and morphologic localization method
Then board frame is merged and is returned to candidate frame, obtain more accurate license plate candidate frame, efficiently solve MTCNN algorithm
Not high, the computationally intensive problem of location efficiency.Therefore, the present invention not only increases the precision of positioning, and reduces calculation amount.
Detailed description of the invention
Fig. 1 is structure principle chart of the invention.
In figure: license plate MTCNN coarse positioning unit 1, candidate frame generation module 11, colouring information locating module 111, morphology
Locating module 112, candidate frame merges and regression block 12, license plate fine positioning unit 2, Retinex image enhancement module 3, image
Acquisition device 4.
Specific embodiment
The present invention is described in further detail with specific embodiment for explanation with reference to the accompanying drawing.
Referring to Fig. 1, a kind of License Plate processor, including license plate MTCNN coarse positioning unit 1, license plate fine positioning unit 2,
The license plate MTCNN coarse positioning unit 1 includes candidate frame generation module 11, candidate frame merges and regression block 12, the candidate
Frame generation module 11 includes colouring information locating module 111, morphology locating module 112, the colouring information locating module
111, the signal input part of morphology locating module 112 is connected with image collecting device 4, colouring information locating module 111,
The signal output end of morphology locating module 112 passes through candidate frame merging and regression block 12 and 2 phase of license plate fine positioning unit
Connection；
The colouring information locating module 111 is used to go out license plate candidate area according to colouring information feature location；
The morphology locating module 112 is used to orient license plate candidate area according to morphological image operation；
The candidate frame merges and regression block 12 is used for colouring information locating module 111, morphology locating module 112
Obtained license plate candidate area carries out the merging and recurrence of candidate frame；
The candidate frame that the license plate fine positioning unit 2 is used to merge candidate frame and regression block 12 obtains carries out SVM mistake
Filter, obtains accurate license plate area.
The processor further includes Retinex image enhancement module 3, and the signal of the Retinex image enhancement module 3 is defeated
Enter end to be connected with image collecting device 4, the signal output end and colouring information locating module of Retinex image enhancement module 3
111, morphology locating module 112 is connected；
The Retinex image enhancement module 3 is used to use Retinex under greasy weather, cloudy day, sleet sky, night-time scene
The image data that algorithm acquires image collecting device 4 carries out enhancing processing.
A kind of location processing method of License Plate processor, successively the following steps are included:
Step 1, the colouring information locating module 111 go out license plate candidate area according to colouring information feature location, described
Morphology locating module 112 orients license plate candidate area according to morphological image operation；
Step 2, the candidate frame merge and regression block 12 is to colouring information locating module 111, morphology locating module
112 obtained license plate candidate areas carry out the merging and recurrence of candidate frame；
Step 3, the candidate frame that the license plate fine positioning unit 2 merges candidate frame and regression block 12 obtains carry out SVM
Filtering, obtains accurate license plate area, at this point, completing the detection positioning of license plate.
In step 1, the colouring information locating module 111 goes out license plate candidate area successively according to colouring information feature location
The following steps are included:
Image preprocessing and color space conversion are first pre-processed received image data using histogram equalization,
The color space of treated image is switched into HSV from RGB again；
Color threshold segmentation binaryzation, all pixels for successively traversing image, by H value be 200 ﹣ 280 and S value and V value is
The pixel that the pixel or H value of 0.35 ﹣ 1.0 is 30 ﹣ 80 and S value and V value is 0.35 ﹣ 1.0 is labeled as white pixel, is otherwise black
Pixel obtains binary image, wherein white area is license plate candidate area, and black is non-license plate area；
Closed operation first carries out closed operation to binary image, then obtains a rectangular area by license plate candidate area
That is candidate license plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle
The threshold value of area screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate regions into corresponding original image by candidate license plate frame
Domain.
In step 1, the morphology locating module 112 orients license plate candidate area successively according to morphological image operation
The following steps are included:
Image preprocessing, first by received image data carry out greyscale transformation, recycle Gaussian Blur picture filter out image
Then noise carries out Image Edge-Detection using Sobel algorithm, license plate area is made to be distinguished out；
Image binaryzation converts bianry image for pretreated image using thresholding method；
Closed operation first carries out closed operation to bianry image, then obtains a rectangular area i.e. by license plate candidate area
Candidate license plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle
The threshold value of area screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate regions into corresponding original image by candidate license plate frame
Domain.
The processor further includes Retinex image enhancement module 3, and the signal of the Retinex image enhancement module 3 is defeated
Enter end to be connected with image collecting device 4, the signal output end and colouring information locating module of Retinex image enhancement module 3
111, morphology locating module 112 is connected；
The method also includes image processing step, which is located at before step 1；
Described image processing step are as follows: under greasy weather, cloudy day, sleet sky, night-time scene, Retinex image enhancement module 3
Enhancing processing is carried out using image data from image collecting device 4 of the Retinex algorithm to acquisition.
In step 2, the candidate frame merges and regression block 12 carries out the merging and recurrence of candidate frame using NMS algorithm.
The principle of the present invention is described as follows:
Since original MTCNN algorithm location efficiency is not high, the method efficiency for generating candidate frame using sliding window is lower.Example
Such as, the candidate license plate frame generated on 1080*1920 image has nearly 500,000, this results in first layer substantially to filter candidate frame
Work is time-consuming larger, and individually can generate many erroneous detection license plates using the method for color positioning and candidate license plate positioning, positioning accurate
It spends poor.In this regard, using and being based on the invention proposes a kind of method that traditional license plate locating method is combined with deep learning
Colouring information and morphologic localization method respectively obtain candidate license plate frame, then right using non-maxima suppression method (NMS)
All candidate frames such as merge, delete, filtering at the operation, so as to obtain accurate license plate candidate frame.
Image preprocessing: pre-processing image using histogram equalization, eliminates influence of the illumination to image.
Color threshold segmentation binaryzation: license plate color is usually blue or yellow, and the H value of blue pixel is 200 ﹣ 280 and S
Value and V value are 0.35 ﹣ 1.0, and the H value of yellow pixel is 30 ﹣ 80 and S value and V value is 0.35 ﹣ 1.0, and the present invention passes through will be by H value
For 200 ﹣ 280 and pixel region or H value that S value and V value are 0.35 ﹣ 1.0 are 30 ﹣ 80 and S value and V value is the pixel of 0.35 ﹣ 1.0
Zone marker is white area, and other regions are black region, to realize the positioning of license plate candidate area.
Embodiment 1:
Referring to Fig. 1, a kind of License Plate processor, including license plate MTCNN coarse positioning unit 1, license plate fine positioning unit 2,
Retinex image enhancement module 3, the license plate MTCNN coarse positioning unit 1 include candidate frame generation module 11, candidate frame merging
With regression block 12, the candidate frame generation module 11 includes colouring information locating module 111, morphology locating module 112, institute
State colouring information locating module 111, the signal input part of morphology locating module 112 passes through Retinex image enhancement module 3
Be connected with image collecting device 4, colouring information locating module 111, morphology locating module 112 signal output end pass through
Candidate frame merges and regression block 12 is connected with license plate fine positioning unit 2.
The location processing method of above-mentioned License Plate processor, successively follows the steps below:
Step 1, image collecting device 4 obtain current scene image data, if the current scene be the greasy weather, the cloudy day,
Sleet sky or night, the image data that image collecting device 4 will acquire are sent to Retinex image enhancement module 3 and enter step
Rapid 2, if current scene is transmitted directly to color letter without carrying out image enhancement, the image data that image collecting device 4 will acquire
Locating module 111, morphology locating module 112 are ceased, and enters step 3；
Step 2, the Retinex image enhancement module 3 are acquired using Retinex algorithm to from image collecting device 4
Image data carry out enhancing processing.
Step 3, the colouring information locating module 111 go out license plate candidate area according to colouring information feature location, described
Morphology locating module 112 orients license plate candidate area according to morphological image operation；
Step 4, the candidate frame merge and regression block 12 uses NMS algorithm to colouring information locating module 111, form
Learn merging and recurrence that the license plate candidate area that locating module 112 obtains carries out candidate frame；
Step 5, the candidate frame that the license plate fine positioning unit 2 merges candidate frame and regression block 12 obtains carry out SVM
Filtering, obtains accurate license plate area, at this point, completing the detection positioning of license plate；
In step 1, the colouring information locating module 111 goes out license plate candidate area successively according to colouring information feature location
It follows the steps below:
Image preprocessing and color space conversion are first pre-processed received image data using histogram equalization,
The color space of treated image is switched into HSV from RGB again；
Color threshold segmentation binaryzation, all pixels for successively traversing image, by H value be 200 ﹣ 280 and S value and V value is
The pixel that the pixel or H value of 0.35 ﹣ 1.0 is 30 ﹣ 80 and S value and V value is 0.35 ﹣ 1.0 is labeled as white pixel, is otherwise black
Pixel obtains binary image, wherein white area is license plate candidate area, and black is non-license plate area；
Closed operation first carries out closed operation to binary image, then obtains a rectangular area by license plate candidate area
That is candidate license plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle
The threshold value of area screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate regions into corresponding original image by candidate license plate frame
Domain.
In step 1, the morphology locating module 112 orients license plate candidate area successively according to morphological image operation
It follows the steps below:
Image preprocessing, first by received image data carry out greyscale transformation, recycle Gaussian Blur picture filter out image
Then noise carries out Image Edge-Detection using Sobel algorithm, license plate area is made to be distinguished out；
Image binaryzation converts bianry image for pretreated image using thresholding method；
Closed operation first carries out closed operation to bianry image, then obtains a rectangular area i.e. by license plate candidate area
Candidate license plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle
The threshold value of area screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate regions into corresponding original image by candidate license plate frame
Domain.
Claims (7)
1. a kind of License Plate processor, it is characterised in that:
The processor includes license plate MTCNN coarse positioning unit (1), license plate fine positioning unit (2), and the license plate MTCNN is slightly fixed
Bit location (1) includes candidate frame generation module (11), candidate frame merges and regression block (12), the candidate frame generation module
It (11) include colouring information locating module (111), morphology locating module (112), the colouring information locating module (111),
The signal input part of morphology locating module (112) is connected with image collecting device (4), colouring information locating module
(111), the signal output end of morphology locating module (112) is merged by candidate frame and regression block (12) is carefully determined with license plate
Bit location (2) is connected；
The colouring information locating module (111) is used to go out license plate candidate area according to colouring information feature location；
The morphology locating module (112) is used to orient license plate candidate area according to morphological image operation；
The candidate frame merges and regression block (12) are used for colouring information locating module (111), morphology locating module
(112) license plate candidate area obtained carries out the merging and recurrence of candidate frame；
The candidate frame that the license plate fine positioning unit (2) is used to merge candidate frame and regression block (12) obtains carries out SVM mistake
Filter, obtains accurate license plate area.
2. a kind of License Plate processor according to claim 1, it is characterised in that:
The processor further includes Retinex image enhancement module (3), and the signal of the Retinex image enhancement module (3) is defeated
Enter end to be connected with image collecting device (4), the signal output end and colouring information of Retinex image enhancement module (3) position
Module (111), morphology locating module (112) are connected；
The Retinex image enhancement module (3) is used to calculate under greasy weather, cloudy day, sleet sky, night-time scene using Retinex
The image data that method acquires image collecting device (4) carries out enhancing processing.
3. a kind of location processing method of License Plate processor described in claim 1, it is characterised in that:
The method successively the following steps are included:
Step 1, the colouring information locating module (111) go out license plate candidate area, the shape according to colouring information feature location
State locating module (112) orients license plate candidate area according to morphological image operation；
Step 2, the candidate frame merge and regression block (12) is to colouring information locating module (111), morphology locating module
(112) license plate candidate area obtained carries out the merging and recurrence of candidate frame；
Step 3, the candidate frame that the license plate fine positioning unit (2) merges candidate frame and regression block (12) obtains carry out SVM
Filtering, obtains accurate license plate area, at this point, completing the detection positioning of license plate.
4. a kind of location processing method of License Plate processor according to claim 3, it is characterised in that:
In step 1, the colouring information locating module (111) goes out license plate candidate area according to colouring information feature location and successively wraps
Include following steps:
Image preprocessing and color space conversion are first pre-processed received image data using histogram equalization, then will
The color space of treated image switchs to HSV from RGB；
Color threshold segmentation binaryzation, all pixels for successively traversing image, by H value be 200 ﹣ 280 and S value and V value is 0.35 ﹣
The pixel that 1.0 pixel or H value is 30 ﹣ 80 and S value and V value is 0.35 ﹣ 1.0 is labeled as white pixel, is otherwise black picture element,
Obtain binary image, wherein white area is license plate candidate area, and black is non-license plate area；
Closed operation first carries out closed operation to binary image, then obtains a rectangular area by license plate candidate area and waits
Select license plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle area
The threshold value of screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate area into corresponding original image by candidate license plate frame.
5. a kind of location processing method of License Plate processor according to claim 3, it is characterised in that:
In step 1, the morphology locating module (112) is oriented license plate candidate area according to morphological image operation and is successively wrapped
Include following steps:
Image preprocessing, first by received image data carry out greyscale transformation, recycle Gaussian Blur picture filtering image noise,
Then Image Edge-Detection is carried out using Sobel algorithm, license plate area is made to be distinguished out；
Image binaryzation converts bianry image for pretreated image using thresholding method；
Closed operation first carries out closed operation to bianry image, and it is i.e. candidate then to obtain a rectangular area by license plate candidate area
License plate outline region；
It screens profile, calculate the size in all candidate license plate outline regions in image first with profile calculating method, recycle area
The threshold value of screening and the setting of profile length-width ratio filters out candidate contours region；
Obtained candidate contours area maps are obtained license plate candidate area into corresponding original image by candidate license plate frame.
6. a kind of location processing method of License Plate processor according to claim 3, it is characterised in that:
The processor further includes Retinex image enhancement module (3), and the signal of the Retinex image enhancement module (3) is defeated
Enter end to be connected with image collecting device (4), the signal output end and colouring information of Retinex image enhancement module (3) position
Module (111), morphology locating module (112) are connected；
The method also includes image processing step, which is located at before step 1；
Described image processing step are as follows: under greasy weather, cloudy day, sleet sky, night-time scene, Retinex image enhancement module (3) is adopted
Enhancing processing is carried out with image data from image collecting device (4) of the Retinex algorithm to acquisition.
7. a kind of location processing method of License Plate processor according to claim 3, it is characterised in that: step (2)
In, the candidate frame merges and regression block (12) carry out the merging and recurrence of candidate frame using NMS algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811323756.4A CN109359651A (en) | 2018-11-08 | 2018-11-08 | A kind of License Plate processor and its location processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811323756.4A CN109359651A (en) | 2018-11-08 | 2018-11-08 | A kind of License Plate processor and its location processing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109359651A true CN109359651A (en) | 2019-02-19 |
Family
ID=65344634
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811323756.4A Pending CN109359651A (en) | 2018-11-08 | 2018-11-08 | A kind of License Plate processor and its location processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109359651A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110321969A (en) * | 2019-07-11 | 2019-10-11 | 山东领能电子科技有限公司 | A kind of vehicle face alignment schemes based on MTCNN |
| CN111429727A (en) * | 2020-04-23 | 2020-07-17 | 深圳智优停科技有限公司 | License plate identification method and system in open type parking space |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355180A (en) * | 2016-09-07 | 2017-01-25 | 武汉安可威视科技有限公司 | Method for positioning license plates on basis of combination of color and edge features |
-
2018
- 2018-11-08 CN CN201811323756.4A patent/CN109359651A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106355180A (en) * | 2016-09-07 | 2017-01-25 | 武汉安可威视科技有限公司 | Method for positioning license plates on basis of combination of color and edge features |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110321969A (en) * | 2019-07-11 | 2019-10-11 | 山东领能电子科技有限公司 | A kind of vehicle face alignment schemes based on MTCNN |
| CN110321969B (en) * | 2019-07-11 | 2023-06-30 | 山东领能电子科技有限公司 | MTCNN-based face alignment method |
| CN111429727A (en) * | 2020-04-23 | 2020-07-17 | 深圳智优停科技有限公司 | License plate identification method and system in open type parking space |
| CN111429727B (en) * | 2020-04-23 | 2021-04-02 | 深圳智优停科技有限公司 | License plate identification method and system in open type parking space |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109145798B (en) | Driving scene target identification and travelable region segmentation integration method | |
| Li et al. | Nighttime lane markings recognition based on Canny detection and Hough transform | |
| CN104463241A (en) | Vehicle type recognition method in intelligent transportation monitoring system | |
| Deb et al. | HSI color based vehicle license plate detection | |
| CN103440503B (en) | The recognition methods of a kind of automobile body color detection | |
| Cao et al. | An adaptive approach to vehicle license plate localization | |
| Ingole et al. | Characters feature based Indian vehicle license plate detection and recognition | |
| CN111046741A (en) | Method and device for identifying lane line | |
| Agbeyangi et al. | Automatic identification of vehicle plate number using raspberry pi | |
| CN105512662A (en) | Detection method and apparatus for unlicensed vehicle | |
| CN109359651A (en) | A kind of License Plate processor and its location processing method | |
| Wen et al. | A rear-vehicle detection system for static images based on monocular vision | |
| FAN et al. | Robust lane detection and tracking based on machine vision | |
| Nguwi et al. | Number plate recognition in noisy image | |
| Rahaman et al. | Lane detection for autonomous vehicle management: PHT approach | |
| Mori et al. | Visibility estimation in foggy conditions by in-vehicle camera and radar | |
| Vaishnav et al. | An integrated automatic number plate recognition for recognizing multi language fonts | |
| Khoshki et al. | Improved Automatic License Plate Recognition (ALPR) system based on single pass Connected Component Labeling (CCL) and reign property function | |
| Kaur et al. | An Efficient Method of Number Plate Extraction from Indian Vehicles Image | |
| CN108470175A (en) | Vehicle License Plate Recognition System based on neural network and recognition methods | |
| CN114926456A (en) | Rail foreign matter detection method based on semi-automatic labeling and improved deep learning | |
| Deb et al. | Vehicle license plate detection algorithm based on color space and geometrical properties | |
| Mir et al. | Automated speed limit identification for efficient driving system | |
| Seifert et al. | Visual object detection for mobile road sign inventory | |
| Mohammad et al. | Arabic license plate recognition system |
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 |