CN106548170B - A kind of violation vehicle image processing method and system - Google Patents
A kind of violation vehicle image processing method and system Download PDFInfo
- Publication number
- CN106548170B CN106548170B CN201610970752.XA CN201610970752A CN106548170B CN 106548170 B CN106548170 B CN 106548170B CN 201610970752 A CN201610970752 A CN 201610970752A CN 106548170 B CN106548170 B CN 106548170B
- Authority
- CN
- China
- Prior art keywords
- image
- license plate
- characters
- violation vehicle
- module
- 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.)
- Active
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
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
-
- 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
- G06V10/24—Aligning, centring, orientation detection or correction of the image
- G06V10/245—Aligning, centring, orientation detection or correction of the image by locating a pattern; Special marks for positioning
-
- 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)
- Traffic Control Systems (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to technical field of image processing, providing a kind of violation vehicle image processing method and system, method includes: to carry out video frame extraction to video image, and extraction includes the violation vehicle image of complete characters on license plate;The processing of gray scale haze is carried out to single width violation vehicle image, obtains and saves fogless violation vehicle image;License plate area positioning is carried out to fogless violation vehicle image, obtains character zone image corresponding to each characters on license plate;Character zone image corresponding to each characters on license plate is parsed, the corresponding character zone of each character zone image is obtained;Each character zone is successively compared with pre-stored character pattern according to putting in order, it obtains and saves to obtain the character for being included in license plate area, to realize the acquisition to violation vehicle information, clearly record vehicle illegal image, the license board information for accurately judging violation vehicle simultaneously, provides convenience for subsequent procedure of punishment or program of calling to account.
Description
Technical field
The present invention relates to technical field of image processing, specially a kind of violation vehicle image processing method and system.
Background technique
With being constantly progressive for society, automobile gradually replaces original traditional trip mode, becomes people's trip and rides instead of walk
First choice, especially in recent years, vehicle price gradually decreases, and gradually caters to consumer's purchasing power.The increase of vehicle, virtually
Larger pressure, such as traffic jam and traffic accident are brought to traffic, wherein traffic accident is caused by running red light for vehicle mostly.
Currently, for running red light for vehicle recording mode generally using photographic device shooting record by the way of, camera will
The vehicle violation image taken is saved, as in down-stream to the punishment evidence of violator and warning.But in mist
Under haze weather, photographic device is more fuzzy to collected vehicle image, more difficult to the confirmation of the license board information of vehicle, very
It to being to be difficult to confirm, is difficult to call to account to illegal vehicle, is also difficult to provide to have simultaneously for due to bring traffic accident violating the regulations
The evidence of convincingness.
Summary of the invention
In order to overcome the defect of the prior art as indicated above, the present inventor has made intensive studies this, is paying
After a large amount of creative works, so as to complete the present invention.
Specifically, the technical problems to be solved by the present invention are: a kind of violation vehicle image processing method is provided, with solution
Under certainly above-mentioned haze weather, photographic device is more fuzzy to collected vehicle image, to the confirmation of the license board information of vehicle compared with
It for difficulty, is even difficult to confirm, is difficult to call to account to illegal vehicle, be also difficult to simultaneously for due to bring traffic accident violating the regulations
It provides and has the technical issues of convictive evidence.
In order to solve the above technical problems, the technical scheme is that
A kind of violation vehicle image processing method, the method includes the following steps:
Video frame extraction is carried out to the video image that image acquisition device arrives, extraction includes complete characters on license plate
Violation vehicle image, the characters on license plate include Chinese character, number and letter;
The processing of gray scale haze is carried out to the single width violation vehicle image extracted, obtains fogless violation vehicle image,
And it saves;
License plate area positioning is carried out to the fogless violation vehicle image acquired, obtains each license plate in license plate area
Character zone image corresponding to character;
Character zone image corresponding to each characters on license plate got is parsed, each character zone figure is obtained
As corresponding character zone;
The each character zone that will acquire successively is compared with pre-stored character pattern according to putting in order, and obtains
The character for being included in license plate area is obtained, and is saved.
As an improvement scheme, the video image that arrives to image acquisition device carries out video frame extraction,
The step of extraction includes the violation vehicle image of complete characters on license plate specifically include the following steps:
Video frame segmentation is carried out to the video image that image acquisition device arrives, obtains N number of video frame;
Eigenvalues analysis is carried out to N number of video frame, judges in video frame whether to include pre-set characteristic value
Parameter, the feature value parameter are rectangular character zone;
It will include the video frame of the feature value parameter as the violation vehicle image.
As an improvement scheme, the violation vehicle image described in the single width extracted carries out at gray scale haze
Reason, the step of obtaining fogless violation vehicle image specifically include the following steps:
Three color filtering are carried out to violation vehicle image, initial pictures are calculated;
According to the initial pictures being calculated, atmospheric transmissivity function and air light value are calculated;
According to atmospherical scattering model, air light value and atmospheric transmissivity function, the fogless violation vehicle image is generated;
Wherein, the step of initial pictures that the basis is calculated, calculating atmospheric transmissivity function, specifically includes
Following step:
Average filter processing is carried out to initial pictures, obtains smoothed image;
Grey level compensation is carried out to smoothed image, obtains dark primary image;
Dark primary image is modified, and air light value is combined to calculate initial atmosphere transmittance function;
The initial pictures that the basis is calculated, calculate air light value the step of specifically include the following steps:
The gray-level projection of horizontal direction is carried out to the first image got;
Summation operation is carried out to the horizontal direction gray-level projection value of initial pictures, is acquired under floor projection most
Big area image;
Vertical direction gray-level projection is carried out to the maximum region image under floor projection;
Summation operation is carried out to vertical direction gray-level projection value, acquires the maximum region figure under upright projection
Picture;
The average value that the maximum pixel of a number of brightness value is chosen in the maximum region image under upright projection is made
For the air light value.
As an improvement scheme, it is fixed that the described pair of fogless violation vehicle image acquired carries out license plate area
Position, the step of obtaining character zone image corresponding to each characters on license plate in license plate area specifically include the following steps:
Smothing filtering is carried out to the license plate area image;
According to the orientation of several characters on license plate, two gray-scale edges between adjacent characters on license plate are obtained, with two
Any one line vertical with the orientation picture of the characters on license plate in region between a gray-scale edges is boundary
Line carries out region division to adjacent characters on license plate, generates several individual characters on license plate area images.
As an improvement scheme, character zone image corresponding to the described pair of each characters on license plate got carries out
Parsing, the step of obtaining each character zone image corresponding character zone specifically include the following steps:
Arrangement by several characters on license plate area images of generation according to characters on license plate on the license plate area image
Sequence is ranked up;
Image segmentation is carried out to each characters on license plate area image after sequence, several image-regions are obtained, each
Seed growth point is obtained in image-region;
Based on the seed growth point, growth obtains the character zone and background area in characters on license plate area image
Domain.
Another object of the present invention is to provide at a kind of violation vehicle image based on violation vehicle image processing method
Reason system, the system comprises:
Video frame extraction module, the video image for arriving to image acquisition device carry out video frame extraction, extract
It include the violation vehicle image of complete characters on license plate, the characters on license plate includes Chinese character, number and letter;
Gray scale haze processing module is obtained for carrying out the processing of gray scale haze to the single width violation vehicle image extracted
Take fogless violation vehicle image;
Character zone image collection module, it is fixed for carrying out license plate area to the fogless violation vehicle image acquired
Position obtains character zone image corresponding to each characters on license plate in license plate area;
Character zone obtains module, for solving to character zone image corresponding to each characters on license plate got
Analysis, obtains the corresponding character zone of each character zone image;
Character contrast module, each character zone for will acquire according to put in order successively with pre-stored word
Symbol pattern is compared, and acquires the character for being included in license plate area;
Database, for storing the character for being included in fogless the violation vehicle image and license plate area.
As an improvement scheme, the video frame extraction module specifically includes:
Video frame divides module, and the video image for arriving to image acquisition device carries out video frame segmentation, obtains N
A video frame;
Eigenvalues analysis module, for carrying out Eigenvalues analysis to N number of video frame, judge in video frame whether include
There is pre-set feature value parameter, the feature value parameter is rectangular character zone;
Violation vehicle image confirmation module, for that will include the video frame of the feature value parameter as the vehicle violating the regulations
Image.
As an improvement scheme, the gray scale haze processing module specifically includes:
Initial pictures are calculated for carrying out three color filtering to violation vehicle image in three color filter modules;
Computing module, for calculating atmospheric transmissivity function and atmosphere light according to the initial pictures being calculated
Value;
Fog free images generation module, for generating institute according to atmospherical scattering model, air light value and atmospheric transmissivity function
State fogless violation vehicle image;
Wherein, the computing module specifically includes:
Smoothed image obtains module, for carrying out average filter processing to initial pictures, obtains smoothed image;
Grey level compensation module obtains dark primary image for carrying out grey level compensation to smoothed image;
Atmospheric transmissivity function computation module for being modified to dark primary image, and combines air light value to calculate just
Beginning atmospheric transmissivity function;
The computing module further include:
Floor projection module, for carrying out the gray-level projection of horizontal direction to the first image got;
First summation operation module carries out summation operation for the horizontal direction gray-level projection value to initial pictures,
Acquire the maximum region image under floor projection;
Upright projection module, for carrying out vertical direction gray-level projection to the maximum region image under floor projection;
Second summation operation module acquires vertical for carrying out summation operation to vertical direction gray-level projection value
Deliver directly the maximum region image under shadow;
Air light value computing module, for choosing a number of brightness value in the maximum region image under upright projection
The average value of maximum pixel is as the air light value.
As an improvement scheme, the character zone image collection module specifically includes:
Smothing filtering module, for carrying out smothing filtering to the license plate area image;
Gray-scale edges obtain module, for the orientation according to several characters on license plate, obtain adjacent characters on license plate it
Between two gray-scale edges;
Division module, for any one in the region between two gray-scale edges and the characters on license plate
The vertical line of orientation picture is line of demarcation, carries out region division to adjacent characters on license plate, generates several individual license plates
Character zone image.
As an improvement scheme, the character zone obtains module and specifically includes:
Sorting module, several characters on license plate area images for that will generate are according to characters on license plate in the license plate area
Putting in order on image is ranked up;
Image segmentation module obtains several for carrying out image segmentation to each characters on license plate area image after sequence
A image-region obtains seed growth point in each image-region;
Character zone pop-in upgrades, for based on the seed growth point, growth to obtain characters on license plate area image
In character zone and background area.
After above-mentioned technical proposal, the beneficial effects of the present invention are:
Video frame extraction is carried out to the video image that image acquisition device arrives, extraction includes complete characters on license plate
Violation vehicle image;The processing of gray scale haze is carried out to the single width violation vehicle image extracted, obtain and saves fogless disobey
Chapter vehicle image;License plate area positioning is carried out to the fogless violation vehicle image acquired, is obtained each in license plate area
Character zone image corresponding to characters on license plate;Character zone image corresponding to each characters on license plate got is solved
Analysis, obtains the corresponding character zone of each character zone image;The each character zone that will acquire is according to putting in order successively
It is compared with pre-stored character pattern, obtains and saves to obtain the character for being included in license plate area, thus realization pair
The acquisition of violation vehicle information clearly records vehicle illegal image, while accurately judging the license board information of violation vehicle, is
Subsequent procedure of punishment or program of calling to account provide convenience.
Detailed description of the invention
Fig. 1 is the implementation flow chart of violation vehicle image processing method provided by the invention;
Fig. 2 is that the video image provided by the invention arrived to image acquisition device carries out video frame extraction, extracts packet
The implementation flow chart of violation vehicle image containing complete characters on license plate;
Fig. 3 is that the violation vehicle image provided by the invention described in the single width extracted carries out the processing of gray scale haze, is obtained
Take the implementation flow chart of fogless violation vehicle image;
Fig. 4 is the initial pictures that basis provided by the invention is calculated, and calculates the realization of atmospheric transmissivity function
Flow chart;
Fig. 5 is the initial pictures that basis provided by the invention is calculated, and calculates the implementation flow chart of air light value;
Fig. 6 is the fogless violation vehicle image progress license plate area positioning provided by the invention to acquiring, and is obtained
The implementation flow chart of character zone image corresponding to each characters on license plate in license plate area;
Fig. 7 be it is provided by the invention character zone image corresponding to each characters on license plate got is parsed,
Obtain the implementation flow chart of the corresponding character zone of each character zone image;
Fig. 8 is the structural block diagram of violation vehicle image processing system provided by the invention;
Fig. 9 is the structural block diagram of gray scale haze processing module provided by the invention.
Specific embodiment
Below with reference to specific embodiment, the present invention is further described.But the purposes and mesh of these exemplary embodiments
Be only used to enumerate the present invention, any type of any restriction not is constituted to real protection scope of the invention, it is more non-to incite somebody to action this
The protection scope of invention is confined to this.
Fig. 1 shows the implementation flow chart of violation vehicle image processing method provided by the invention, specifically includes following
Step:
In step s101, video frame extraction is carried out to the video image that image acquisition device arrives, extraction includes
The violation vehicle image of complete characters on license plate, the characters on license plate include Chinese character, number and letter.
In this step, which is set to the photographic device being arranged on each traffic intersection, such as sets at present
The high-definition camera set, by corresponding communication interface and communications protocol, by the transmission of video images taken to server
Host on, host saves video image, and to have suspicion violating the regulations video image carry out video frame extraction, so as to point
Analyse vehicle violation situation.
Wherein, the extracting mode of the video frame is following is discussed in detail, and details are not described herein, which is common vehicle
Board, such as blue bottom wrongly written or mispronounced character, black matrix wrongly written or mispronounced character, white gravoply, with black engraved characters, yellow bottom black word etc., details are not described herein.
In step s 102, the processing of gray scale haze is carried out to the single width violation vehicle image extracted, obtained fogless
Violation vehicle image, and save.
Defogging processing is carried out to single width violation vehicle image, i.e., under conditions of bad weather, such as haze, rain and snow weather
Under conditions of lower, the disturbing factor on the single width violation vehicle image extracted is removed, pure image is acquired, so as to
Clearly violation vehicle information is confirmed, also provides evidence for subsequent punishment on contravention of regulation.
Meanwhile saving defogging treated violation vehicle image in the database, so as to subsequent query calling.
In step s 103, license plate area positioning is carried out to the fogless violation vehicle image acquired, obtains license plate
Character zone image corresponding to each characters on license plate in region.
In this step, region segmentation is carried out to violation vehicle image, obtains the corresponding character area of each character on license plate
Area image.
In step S104, character zone image corresponding to each characters on license plate got is parsed, is obtained
The corresponding character zone of each character zone image.
The pattern of character included in each character zone image is separated, so as to pre-stored pattern into
Row compares, and confirms characters on license plate, finally confirms the license board information of violation vehicle.
In step s105, each character zone that will acquire according to put in order successively with pre-stored character figure
Case is compared, and acquires the character for being included in license plate area, and save.
In this step, all character patterns for being included on license plate are saved in advance, the word that then will acquire
Symbol region is compared with character pattern, determines that the arrangement of the character for being included on the license plate of violation vehicle and character is suitable
Sequence.
Fig. 2 shows the video images provided by the invention arrived to image acquisition device to carry out video frame extraction, mentions
Take include the violation vehicle image of complete characters on license plate implementation flow chart, specifically include the following steps:
In step s 201, video frame segmentation is carried out to the video image that image acquisition device arrives, obtains N number of video
Frame.
In this step, to the segmentation of video image, existing software realization, such as premiere software can be used,
According to the size of video image, several video frames are divided into, N is the integer more than or equal to 1.
In step S202, Eigenvalues analysis is carried out to N number of video frame, judges in video frame whether to include to preset
Feature value parameter, the feature value parameter be rectangular character zone.
In this step, can use ergodic algorithm, N number of video frame is analyzed, judge be in the image of video frame
No includes pre-set feature value parameter, i.e., found out from N number of video frame include complete license plate image, this includes
The image of complete license plate may have several, can be from more one of them is chosen in image completely and clearly, i.e. step
S203。
It will include the video frame of the feature value parameter as the violation vehicle image in step S203.
When the image for including complete license plate there are several, then wherein one can be chosen according to the actual needs.
Fig. 3 shows the violation vehicle image provided by the invention described in the single width extracted and carries out at gray scale haze
Reason, obtain the implementation flow chart of fogless violation vehicle image, step specifically include the following steps:
In step S301, three color filtering are carried out to violation vehicle image, initial pictures are calculated.
In this step, three color filtering refer to the triple channel (RGB) of any pixel in violation vehicle image into
Row mini-value filtering, no longer imposes herein.
In step s 302, according to the initial pictures being calculated, atmospheric transmissivity function and atmosphere light are calculated
Value.
In this step, atmospherical scattering model are as follows:
According to atmospheric scattering theory, the scattering of atmospheric particles includes two classes:
(1) luminous energy that body surface reflects in scene is decayed during reaching sensor by the scattering of atmospheric particles
Process;
(2) solar energy is reached the process of sensor after the suspended particles scattering in atmosphere.
When haze sky is imaged, attenuation model and atmosphere light imaging model are existed simultaneously and are risen scenery in atmospherical scattering model
The theoretical basis for the features such as leading role is fuzzy Misty Image presentation, low contrast;
And air light value and atmospheric transmissivity function are the core parameter of entire atmospherical scattering model, as long as this is calculated
The defogging processing to foggy image can be realized in two parameters.
In step S303, according to atmospherical scattering model, air light value and atmospheric transmissivity function, generate described fogless
Violation vehicle image.
It can be concluded that, after the air light value and atmospheric transmissivity function being calculated, nothing can be obtained from step S302
The violation vehicle image of mist, details are not described herein.
Wherein, on the basis of embodiment shown in Fig. 3, Fig. 4 shows the institute that basis provided by the invention is calculated
Initial pictures are stated, the implementation flow chart of atmospheric transmissivity function is calculated, specifically include the following steps:
In step S401, average filter processing is carried out to initial pictures, obtains smoothed image.
In this step, it in order to avoid the gray scale bounce between adjacent pixel, needs to carry out average filter processing.
In step S402, grey level compensation is carried out to smoothed image, obtains dark primary image.
In this step, after average filter, the variation tendency of dark primary in image, but itself and true value be can reflect
It has a certain gap, needs to carry out grey level compensation.
In step S403, dark primary image is modified, and air light value is combined to calculate initial atmosphere transmissivity letter
Number.
In this step, adjusting for depth is carried out to dark primary image, makes image that there is certain sense of depth, while if big
Gas light value is it is known that atmospheric transmissivity function can be calculated.
On the basis of embodiment shown in Fig. 3, it is described initial to show that basis provided by the invention is calculated by Fig. 5
Image calculates the implementation flow chart of air light value, specifically include the following steps:
In step S501, the gray-level projection of horizontal direction is carried out to the first image got.
In step S502, summation operation is carried out to the horizontal direction gray-level projection value of initial pictures, is acquired
Maximum region image under floor projection.
In step S503, vertical direction gray-level projection is carried out to the maximum region image under floor projection.
In step S504, summation operation is carried out to vertical direction gray-level projection value, is acquired under upright projection
Maximum region image.
In step S505, the maximum picture of a number of brightness value is chosen in the maximum region image under upright projection
The average value of element is as the air light value.
It in embodiments of the present invention, can also be in the following way for the calculating of air light value:
(1) all pixels point in violation vehicle image is extracted;
(2) descending arrangement is carried out to the pixel value of all pixels point;
(3) choose violation vehicle image in pixel pixel value sequence preceding 5% pixel average gray value
For air light value.
Details are not described herein, but not to limit the present invention.
In embodiments of the present invention, Fig. 6 shows provided by the invention to the fogless violation vehicle image acquired
License plate area positioning is carried out, the implementation flow chart of character zone image corresponding to each characters on license plate in license plate area is obtained,
Itself specifically include the following steps:
In step s 601, smothing filtering is carried out to the license plate area image.
In step S602, according to the orientation of several characters on license plate, two between adjacent characters on license plate are obtained
Gray-scale edges.
In step S603, with any one in the region between two gray-scale edges and the characters on license plate
The vertical line of orientation picture is line of demarcation, carries out region division to adjacent characters on license plate, generates several individual license plates
Character zone image.
It is single characters on license plate area image that the embodiment, which is by license plate area image segmentation, as subsequent confirmation analysis vehicle
The foundation of the trade mark.
Fig. 7, which is shown, provided by the invention to be solved character zone image corresponding to each characters on license plate got
Analysis, obtains the implementation flow chart of the corresponding character zone of each character zone image, specifically include the following steps:
In step s 701, by several characters on license plate area images of generation according to characters on license plate in the license plate area
Putting in order on image is ranked up.
In step S702, image segmentation is carried out to each characters on license plate area image after sequence, obtains several figures
As region, seed growth point is obtained in each image-region.
In step S703, based on the seed growth point, growth obtains the character in characters on license plate area image
Region and background area.
The character zone that single characters on license plate area image analysis is obtained accurately is known as the foundation of confirmation license plate number
The license plate number of violation vehicle, details are not described herein.
Fig. 8 shows the structural block diagram of violation vehicle image processing system provided by the invention, for ease of description, in figure
Only give part related to the embodiment of the present invention.
Video frame extraction module 11 carries out video frame extraction to the video image that image acquisition device arrives, and extraction includes
There is the violation vehicle image of complete characters on license plate, the characters on license plate includes Chinese character, number and letter;Gray scale haze processing module
12 pairs of single width violation vehicle images extracted carry out the processing of gray scale haze, obtain fogless violation vehicle image;Character area
Area image obtains module 13 and carries out license plate area positioning to the fogless violation vehicle image acquired, obtains in license plate area
Character zone image corresponding to each characters on license plate;It is right to each characters on license plate institute got that character zone obtains module 14
The character zone image answered is parsed, and the corresponding character zone of each character zone image is obtained;Character contrast module 15 will
The each character zone got is successively compared with pre-stored character pattern according to putting in order, and acquires license plate
The character for being included in region;Database 16 stores the word for being included in the fogless violation vehicle image and license plate area
Symbol.
Wherein, video frame extraction module 11 specifically includes:
Video frame divides module 17 and carries out video frame segmentation to the video image that image acquisition device arrives, and obtains N number of
Video frame;Whether Eigenvalues analysis module 18 carries out Eigenvalues analysis to N number of video frame, judge in video frame to include pre-
The feature value parameter being first arranged, the feature value parameter are rectangular character zone;Violation vehicle image confirmation module 19 will include
There is the video frame of the characteristic value parameter as the violation vehicle image.
In this embodiment, character zone image collection module 13 specifically includes:
Smothing filtering module 20 carries out smothing filtering to the license plate area image;If gray-scale edges obtain module 21 according to
The orientation of dry characters on license plate, obtains two gray-scale edges between adjacent characters on license plate;Division module 22 is with two institutes
Stating any one line vertical with the orientation picture of the characters on license plate in the region between gray-scale edges is line of demarcation, right
Adjacent characters on license plate carries out region division, generates several individual characters on license plate area images.
In this embodiment, character zone obtains module 14 and specifically includes:
Sorting module 23 is by several characters on license plate area images of generation according to characters on license plate in the license plate area figure
As upper putting in order is ranked up;Each characters on license plate area image after 24 pairs of image segmentation module sequences carries out image point
It cuts, obtains several image-regions, seed growth point is obtained in each image-region;Character zone pop-in upgrades 25 is with described
Based on seed growth point, growth obtains the character zone and background area in characters on license plate area image.
In embodiments of the present invention, as shown in figure 9, gray scale haze processing module 12 specifically includes:
Three color filter modules 26 carry out three color filtering to violation vehicle image, and initial pictures are calculated;Computing module 27
According to the initial pictures being calculated, atmospheric transmissivity function and air light value are calculated;Fog free images generation module 28
According to atmospherical scattering model, air light value and atmospheric transmissivity function, the fogless violation vehicle image is generated;
Wherein, the computing module 27 specifically includes:
Smoothed image obtains module 29 and carries out average filter processing to initial pictures, obtains smoothed image;Grey level compensation mould
Block 30 carries out grey level compensation to smoothed image, obtains dark primary image;Atmospheric transmissivity function computation module 31 is to dark primary figure
As being modified, and air light value is combined to calculate initial atmosphere transmittance function;
The computing module 27 further include:
Floor projection module 32 carries out the gray-level projection of horizontal direction to the first image got;First summation operation
Module 33 carries out summation operation to the horizontal direction gray-level projection value of initial pictures, acquires the maximum under floor projection
Area image;Upright projection module 34 carries out vertical direction gray-level projection to the maximum region image under floor projection;The
Two summation operation modules 35 carry out summation operation to vertical direction gray-level projection value, acquire the maximum under upright projection
Area image;Air light value computing module 36 chooses a number of brightness value most in the maximum region image under upright projection
The average value of big pixel is as the air light value.
Wherein, the function of above-mentioned modules is as recorded in above method embodiment, and details are not described herein.
In embodiments of the present invention, video frame extraction is carried out to the video image that image acquisition device arrives, extracts packet
Violation vehicle image containing complete characters on license plate;The processing of gray scale haze is carried out to the single width violation vehicle image extracted,
It obtains and saves fogless violation vehicle image;License plate area positioning is carried out to the fogless violation vehicle image acquired,
Obtain character zone image corresponding to each characters on license plate in license plate area;To corresponding to each characters on license plate got
Character zone image is parsed, and the corresponding character zone of each character zone image is obtained;The each character area that will acquire
Domain is successively compared with pre-stored character pattern according to putting in order, and obtains and saves to obtain in license plate area and is included
Character clearly record vehicle illegal image, while accurately judgement is violating the regulations to realize acquisition to violation vehicle information
The license board information of vehicle provides convenience for subsequent procedure of punishment or program of calling to account.
It should be appreciated that the purposes of these embodiments is merely to illustrate the present invention and is not intended to limit protection model of the invention
It encloses.In addition, it should also be understood that, after reading the technical contents of the present invention, those skilled in the art can make the present invention each
Kind change, modification and/or variation, all these equivalent forms equally fall within guarantor defined by the application the appended claims
Within the scope of shield.
Claims (4)
1. a kind of violation vehicle image processing method, which is characterized in that the method includes the following steps:
Video frame extraction is carried out to the video image that image acquisition device arrives, extraction includes the violating the regulations of complete characters on license plate
Vehicle image, the characters on license plate include Chinese character, number and letter;
The processing of gray scale haze is carried out to the single width violation vehicle image extracted, obtains fogless violation vehicle image, and protect
It deposits;
License plate area positioning is carried out to the fogless violation vehicle image acquired, obtains each characters on license plate in license plate area
Corresponding character zone image;
Character zone image corresponding to each characters on license plate got is parsed, each character zone image pair is obtained
The character zone answered;
The each character zone that will acquire successively is compared with pre-stored character pattern according to putting in order, and obtains
The character for being included in license plate area, and save;
The video image arrived to image acquisition device carries out video frame extraction, and extraction includes complete characters on license plate
The step of violation vehicle image specifically include the following steps:
Video frame segmentation is carried out to the video image that image acquisition device arrives, obtains N number of video frame;
Eigenvalues analysis is carried out to N number of video frame, judges in video frame whether to include pre-set feature value parameter,
The feature value parameter is rectangular character zone;
It will include the video frame of the feature value parameter as the violation vehicle image;
The violation vehicle image described in the single width extracted carries out the processing of gray scale haze, obtains fogless violation vehicle figure
The step of picture specifically include the following steps:
Three color filtering are carried out to violation vehicle image, initial pictures are calculated;
According to the initial pictures being calculated, atmospheric transmissivity function and air light value are calculated;
According to atmospherical scattering model, air light value and atmospheric transmissivity function, the fogless violation vehicle image is generated;
Wherein, the step of initial pictures that the basis is calculated, calculating atmospheric transmissivity function, specifically includes following
Step:
Average filter processing is carried out to initial pictures, obtains smoothed image;
Grey level compensation is carried out to smoothed image, obtains dark primary image;
Dark primary image is modified, and air light value is combined to calculate initial atmosphere transmittance function;
The initial pictures that the basis is calculated, calculate air light value the step of specifically include the following steps:
The initial pictures got are carried out with the gray-level projection of horizontal direction;
Summation operation is carried out to the horizontal direction gray-level projection value of initial pictures, acquires the maximum area under floor projection
Area image;
Vertical direction gray-level projection is carried out to the maximum region image under floor projection;
Summation operation is carried out to vertical direction gray-level projection value, acquires the maximum region image under upright projection;
The brightness value of the maximum pixel of a number of brightness value is chosen in the maximum region image under upright projection as institute
Air light value is stated,
The described pair of fogless violation vehicle image acquired carries out license plate area positioning, obtains each license plate in license plate area
Corresponding to character the step of character zone image specifically include the following steps:
Smothing filtering is carried out to the license plate area image;
According to the orientation of several characters on license plate, two gray-scale edges between adjacent characters on license plate are obtained, with two institutes
Any one line vertical with the orientation of the characters on license plate stated in the region between gray-scale edges is line of demarcation, to phase
Adjacent characters on license plate carries out region division, generates several individual characters on license plate area images.
2. violation vehicle image processing method according to claim 1, which is characterized in that the described pair of each vehicle got
The step of character zone image corresponding to board character is parsed, and each character zone image corresponding character zone is obtained tool
Body includes the following steps:
By several characters on license plate area images of generation according to characters on license plate putting in order on the license plate area image
It is ranked up;
Image segmentation is carried out to each characters on license plate area image after sequence, several image-regions are obtained, in each image
Seed growth point is obtained in region;
Based on the seed growth point, growth obtains the character zone and background area in characters on license plate area image.
3. a kind of violation vehicle image processing system based on violation vehicle image processing method described in claim 1, special
Sign is, the system comprises:
Video frame extraction module, the video image for arriving to image acquisition device carry out video frame extraction, and extraction includes
There is the violation vehicle image of complete characters on license plate, the characters on license plate includes Chinese character, number and letter;
Gray scale haze processing module obtains nothing for carrying out the processing of gray scale haze to the single width violation vehicle image extracted
The violation vehicle image of mist;
Character zone image collection module, for carrying out license plate area positioning to the fogless violation vehicle image acquired,
Obtain character zone image corresponding to each characters on license plate in license plate area;
Character zone obtains module, for being parsed to character zone image corresponding to each characters on license plate got,
Obtain the corresponding character zone of each character zone image;
Character contrast module, each character zone for will acquire according to put in order successively with pre-stored character figure
Case is compared, and acquires the character for being included in license plate area;
Database, for storing the character for being included in fogless the violation vehicle image and license plate area;
The video frame extraction module specifically includes:
Video frame divides module, and the video image for arriving to image acquisition device carries out video frame segmentation, obtains N number of view
Frequency frame;
Whether Eigenvalues analysis module judges in video frame to include pre- for carrying out Eigenvalues analysis to N number of video frame
The feature value parameter being first arranged, the feature value parameter are rectangular character zone;
Violation vehicle image confirmation module, for that will include the video frame of the feature value parameter as the violation vehicle figure
Picture;
The gray scale haze processing module specifically includes:
Initial pictures are calculated for carrying out three color filtering to violation vehicle image in three color filter modules;
Computing module, for calculating atmospheric transmissivity function and air light value according to the initial pictures being calculated;
Fog free images generation module, for generating the nothing according to atmospherical scattering model, air light value and atmospheric transmissivity function
The violation vehicle image of mist;
Wherein, the computing module specifically includes:
Smoothed image obtains module, for carrying out average filter processing to initial pictures, obtains smoothed image;
Grey level compensation module obtains dark primary image for carrying out grey level compensation to smoothed image;
Atmospheric transmissivity function computation module, for being modified to dark primary image, and it is initial big to combine air light value to calculate
Gas transmittance function;
The computing module further include:
Floor projection module, for the initial pictures got to be carried out with the gray-level projection of horizontal direction;
First summation operation module carries out summation operation for the horizontal direction gray-level projection value to initial pictures, obtains
Obtain the maximum region image under floor projection;
Upright projection module, for carrying out vertical direction gray-level projection to the maximum region image under floor projection;
Second summation operation module acquires vertical throwing for carrying out summation operation to vertical direction gray-level projection value
Maximum region image under shadow;
Air light value computing module, it is maximum for choosing a number of brightness value in the maximum region image under upright projection
Pixel brightness value as the air light value;
The character zone image collection module specifically includes:
Smothing filtering module, for carrying out smothing filtering to the license plate area image;
Gray-scale edges obtain module and obtain between adjacent characters on license plate for the orientation according to several characters on license plate
Two gray-scale edges;
Division module, for the arrangement with any one in the region between two gray-scale edges with the characters on license plate
The vertical line in direction is line of demarcation, carries out region division to adjacent characters on license plate, generates several individual characters on license plate areas
Area image.
4. violation vehicle image processing system according to claim 3, which is characterized in that the character zone obtains module
It specifically includes:
Sorting module, several characters on license plate area images for that will generate are according to characters on license plate in the license plate area image
On put in order and be ranked up;
Image segmentation module obtains several figures for carrying out image segmentation to each characters on license plate area image after sequence
As region, seed growth point is obtained in each image-region;
Character zone pop-in upgrades, for based on the seed growth point, growth to be obtained in characters on license plate area image
Character zone and background area.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610970752.XA CN106548170B (en) | 2016-11-04 | 2016-11-04 | A kind of violation vehicle image processing method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610970752.XA CN106548170B (en) | 2016-11-04 | 2016-11-04 | A kind of violation vehicle image processing method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106548170A CN106548170A (en) | 2017-03-29 |
CN106548170B true CN106548170B (en) | 2019-11-19 |
Family
ID=58394031
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610970752.XA Active CN106548170B (en) | 2016-11-04 | 2016-11-04 | A kind of violation vehicle image processing method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106548170B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110135418A (en) * | 2019-04-15 | 2019-08-16 | 深圳壹账通智能科技有限公司 | Traffic accident fix duty method, apparatus, equipment and storage medium based on picture |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7857882B1 (en) * | 2007-12-13 | 2010-12-28 | Johnson Welded Products Inc. | Water and oil mist elimination apparatus for a vehicular compressed air storage system |
CN102142090A (en) * | 2011-03-15 | 2011-08-03 | 中国科学技术大学 | Vehicle detection method and system |
CN103903234A (en) * | 2014-03-12 | 2014-07-02 | 南京第五十五所技术开发有限公司 | Real-time image defogging method based on image field depth |
CN104820998A (en) * | 2015-05-27 | 2015-08-05 | 成都通甲优博科技有限责任公司 | Human body detection and tracking method and device based on unmanned aerial vehicle mobile platform |
CN106023118A (en) * | 2016-06-13 | 2016-10-12 | 凌云光技术集团有限责任公司 | Image defogging method and realization method on FPGA |
-
2016
- 2016-11-04 CN CN201610970752.XA patent/CN106548170B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7857882B1 (en) * | 2007-12-13 | 2010-12-28 | Johnson Welded Products Inc. | Water and oil mist elimination apparatus for a vehicular compressed air storage system |
CN102142090A (en) * | 2011-03-15 | 2011-08-03 | 中国科学技术大学 | Vehicle detection method and system |
CN103903234A (en) * | 2014-03-12 | 2014-07-02 | 南京第五十五所技术开发有限公司 | Real-time image defogging method based on image field depth |
CN104820998A (en) * | 2015-05-27 | 2015-08-05 | 成都通甲优博科技有限责任公司 | Human body detection and tracking method and device based on unmanned aerial vehicle mobile platform |
CN106023118A (en) * | 2016-06-13 | 2016-10-12 | 凌云光技术集团有限责任公司 | Image defogging method and realization method on FPGA |
Non-Patent Citations (2)
Title |
---|
"基于阈值分割和区域生长的车牌识别方法";胡伟平,王日凤,;《广西科学院学报》;20160228;第32卷(第1期);期刊第1-4节 * |
"无人侦察图像快速去雾算法";陆士猛,刘昌锦;《红外技术》;20151031;第37卷(第10期);期刊引言,第1-3节 * |
Also Published As
Publication number | Publication date |
---|---|
CN106548170A (en) | 2017-03-29 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103235938B (en) | The method and system of car plate detection and indentification | |
CN107506760B (en) | Traffic signal detection method and system based on GPS positioning and visual image processing | |
CN105374013B (en) | Image visibility restorative procedure and image processing apparatus based on double dark channel priors | |
WO2017028587A1 (en) | Vehicle monitoring method and apparatus, processor, and image acquisition device | |
CN103268470B (en) | Object video real-time statistical method based on any scene | |
CN104616021B (en) | Traffic sign image processing method and device | |
CN102831591B (en) | Gaussian filter-based real-time defogging method for single image | |
CN102867417B (en) | Taxi anti-forgery system and taxi anti-forgery method | |
KR101246120B1 (en) | A system for recognizing license plate using both images taken from front and back faces of vehicle | |
CN103714325A (en) | Left object and lost object real-time detection method based on embedded system | |
CN106991707B (en) | Traffic signal lamp image strengthening method and device based on day and night imaging characteristics | |
CN103824081A (en) | Method for detecting rapid robustness traffic signs on outdoor bad illumination condition | |
CN109147351A (en) | A kind of traffic light control system | |
CN107507142A (en) | Signal lamp image processing method | |
CN107038691A (en) | The dark primary remote sensing image defogging method aided in based on cloud detection | |
CN103530640A (en) | Unlicensed vehicle detection method based on AdaBoost and SVM (support vector machine) | |
CN103034983A (en) | Defogging method based on anisotropic filtering | |
CN107346547A (en) | Real-time foreground extracting method and device based on monocular platform | |
CN112017445A (en) | Pedestrian violation prediction and motion trail tracking system and method | |
CN112528917A (en) | Zebra crossing region identification method and device, electronic equipment and storage medium | |
CN111931726A (en) | Traffic light detection method and device, computer storage medium and road side equipment | |
CN113128507A (en) | License plate recognition method and device, electronic equipment and storage medium | |
CN106548170B (en) | A kind of violation vehicle image processing method and system | |
CN107480683A (en) | A kind of image processing method and device | |
CN106340031A (en) | Method and device for detecting moving object |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |