It is positioned at the electronic recognition platform violating the regulations of traffic intersection
The present invention is Application No. 201510162839.X, filing date April 8 in 2015 day,
The divisional application of the patent of bright entitled " being positioned at the electronic recognition platform violating the regulations of traffic intersection ".
Technical field
The present invention relates to detection of electrons field, particularly relate to a kind of electronics violating the regulations being positioned at traffic intersection and know
Other platform.
Background technology
Making a dash across the red light is one of red route traffic safety hidden danger.He has following hazardness: 1,
When making a dash across the red light, the attention of driver is all on traffic light, it is impossible to inside and outside comprehensive and correct perception car
Change, reaction and measure to driver bring difficulty;When 2, making a dash across the red light, generally pass through crossing
Speed all ratios comparatively fast, reduce driver's cognitive competence to space;When 3, making a dash across the red light, drive
The judgement of all kinds of traffic environments is declined by member, makes the thinking of driver judge or operate out sometimes
Now slip up;4, crossing is point and the conflict point of traffic flow, makes a dash across the red light and increases point and punching
The insecurity of bump itself, causes safety coefficient to decline;5, make a dash across the red light affect driver in time,
Operating and controlling vehicle accurately and efficiently;6, make a dash across the red light and affect the safety performance of vehicle, also subtract simultaneously
The strain reflection ability of other traffic participants weak and strain disposing capacity;7, facts have proved percentage
90 particularly serious vehicle accident all occur in intersection, therefore by caused accident of making a dash across the red light
Consequence generally all ratios are more serious;8, under traffic lights, as unconsciousization of driver occurs, there is pole
End hazardness.
Accident is brought to road traffic in order to avoid making a dash across the red light, the normal driving custom of specification driver,
The severeest punishment rule have all been formulated in the behavior of making a dash across the red light by the control of traffic and road department of every country
Then.And the correct detection of behavior of wherein making a dash across the red light is to ensure control of traffic and road department law enforcement fairness
The only resource.The existing detection device that makes a dash across the red light is still to be passed through red signal interval by common camera
The vehicle at crossing carries out image taking, image recognition and Car license recognition, and for control of traffic and road, department carries
For the information of vehicles of illegal running, this means precision is the highest, it is judged that rule is single, easily causes by mistake
Sentencing, thus bring economic loss or have inconvenient traffic to the driver that abides by the law, meanwhile, this means are easily by mist
The impact of haze weather.
Accordingly, it would be desirable to the technical scheme of a kind of new vehicles peccancy detection of electrons, high accuracy is taked to shoot
Instrument, and come whether the ruling vehicle that passes through makes a dash across the red light by more than one criterion, it is provided that vehicle is rushed
The accuracy that red light judges, it is to avoid the situation of erroneous judgement occurs, and haze weather can be overcome image
Impact, improve further the reliability of detection.
Summary of the invention
In order to solve the problems referred to above, the invention provides a kind of electronic recognition violating the regulations being positioned at traffic intersection
Platform, introduces the CMOS vision sensor vehicle to front end, the crossing presumptive area of predetermined red signal interval
Carry out high accuracy to shoot, and use associated picture processing equipment that the high-definition image of shooting is carried out target knowledge
Not, it is simultaneously introduced two position sensors and is also used for performing the detection of running red light for vehicle, according to two inspections
Survey whether result comprehensive ruling vehicle makes a dash across the red light, it is to avoid normally travel or go out due to driver's erroneous judgement
Front end, crossing presumptive area but the vehicle after the centrage of crossing of stopping are mistaken for making a dash across the red light, and increase
Added sharpening processing apparatus to remove the haze weather adverse effect to image, improve comprehensively identify flat
The accuracy of platform.
According to an aspect of the present invention, it is provided that a kind of electronic recognition violating the regulations being positioned at traffic intersection is put down
Platform, described Identification platform includes CMOS vision sensor, detection band image segmentating device, vehicle image
Dispenser, enumerator and flush bonding processor, described CMOS vision sensor is for traffic intersection
Interior region, Through Lane direction shoots, to obtain craspedodrome directional image, and described detection band
Image segmentating device connects CMOS vision sensor, within the predetermined red light moment in Through Lane direction,
Isolating detection band image from described craspedodrome directional image, described vehicle image dispenser connects described
Detection band image segmentating device, isolates the moving vehicle image in detection band image, and described enumerator is even
Connect described vehicle image dispenser, for calculating the frame number of moving vehicle image in the predetermined red light moment,
Described flush bonding processor is connected with described enumerator, within the predetermined red light moment, based on described counting
The count results of device determines whether there is running red light for vehicle.
It is positioned at described in more specifically, in the electronic recognition platform violating the regulations of traffic intersection, farther includes:
User's entering apparatus, for the input according to user, sets predetermined automobile upper limit gray threshold, makes a reservation for
Automobile lower limit gray threshold and preset count threshold value, always according to the input of user, in traffic intersection
The detection band in Through Lane direction is configured, the craspedodrome that described detection band is arranged in traffic intersection
The footpath front end in direction, track, is a rectangular area, and described detection band is in described craspedodrome directional image
In occupy 120 × 80 picture element matrixs;Power supply, including solar powered device, accumulator, cuts
Change switch and electric pressure converter, described switching switch and described solar powered device and described accumulator
Connect respectively, according to accumulator dump energy decide whether to be switched to described solar powered device with by
Described solar powered device is powered, and described electric pressure converter is connected with described switching switch, leading to
The 5V voltage crossing switching switch input is converted to 3.3V voltage;Position sensor, including primary importance
Sensing unit and second position sensing unit, described primary importance sensing unit is arranged on detection band front end
Position, for having detected whether that within the predetermined red light moment vehicle passes through detection band front position, if
Having, then export the first vehicle through signal, described second position sensing unit is arranged in traffic intersection
Heart line position, for having detected whether that within the predetermined red light moment vehicle is through traffic intersection centrage position
Put, pass through signal, primary importance sensing unit and the second position if it has, then export the second vehicle
Sense unit is all contact-sensing unit, and the object contact that be there is more than predetermined pressure by inspection is come really
Surely having vehicle process, described traffic intersection position of center line is before described detection band front position;With
Step dynamic RAM SDRAM, respectively connect user's entering apparatus, detection band image segmentating device,
Vehicle image dispenser and flush bonding processor, receive and store predetermined automobile upper limit gray threshold, pre-
Determine automobile lower limit gray threshold, preset count threshold value and described detection band configuration information;Sharpening image
Processor, between described CMOS vision sensor and described detection band image segmentating device, is used for
Receive described craspedodrome directional image, described craspedodrome directional image is carried out mist elimination and processes to obtain mist elimination straight
Line direction image, replaces described craspedodrome directional image, by described for the input of described mist elimination craspedodrome directional image
Detection band image segmentating device is to isolate detection band image;Described sharpening image processor includes: mist
Haze Concentration Testing subset, is positioned in air, for the haze of detection traffic intersection position in real time
Concentration, and determine that intensity removed by haze according to haze concentration, described haze is removed intensity value and is arrived 0
Between 1;Overall air light value obtains subset, is connected with described CMOS vision sensor to obtain
Described craspedodrome directional image, calculates the gray value of each pixel in described craspedodrome directional image, by gray scale
The gray value of the pixel that value is maximum is as overall air light value;Atmospheric scattering light value obtains subset, with
Described CMOS vision sensor and described haze Concentration Testing subset connect, respectively to described craspedodrome
Each pixel of directional image, extracts its R, G, B tri-in Color Channel pixel value minima make
For target pixel value, use the Gaussian filter EPGF (edge-preserving keeping edge
Gaussian filter) it is filtered processing to obtain filtered target pixel value to described target pixel value,
Target pixel value is deducted filtered target pixel value to obtain object pixel difference, use EPGF to mesh
Mark pixel value difference is filtered processing to obtain filtered target pixel value difference, is subtracted by filtered target pixel value
Go filtered target pixel value difference to remove reference value to obtain haze, haze is removed intensity and is multiplied by haze
Except reference value removes threshold value to obtain haze, take haze and remove the minima in threshold value and target pixel value
As comparison reference, take the maximum in comparison reference and 0 scattered as the air of each pixel
Penetrate light value;Medium transmission rate obtains subset, obtains subset and described with described overall air light value
Atmospheric scattering light value obtains subset and connects respectively, by the atmospheric scattering light value of each pixel divided by whole
Body atmosphere light value, to obtain except value, deducts described except value is to obtain the medium transmission of each pixel by 1
Rate;Sharpening Image Acquisition subset, with described CMOS vision sensor, described overall atmosphere light
Value obtains subset and described medium transmission rate obtains subset and connects respectively, deducts each picture by 1
Described first difference, to obtain the first difference, is multiplied by overall air light value to obtain by the medium transmission rate of element
Product value, the pixel value of each pixel in described craspedodrome directional image is deducted described product value with
Obtain the second difference, described second difference is each to obtain divided by the medium transmission rate of each pixel
The sharpening pixel value of individual pixel, in described craspedodrome directional image, the pixel value of each pixel includes institute
State the R of each pixel in craspedodrome directional image, G, B tri-Color Channel pixel value, correspondingly,
The sharpening pixel value of each pixel obtained includes the R of each pixel, and G, B tri-colors are logical
Road sharpening pixel value, the sharpening pixel value composition mist elimination craspedodrome directional image of all pixels;Described
Gray value in detection band image at predetermined automobile upper limit gray threshold and is made a reservation for by vehicle image dispenser
Pixel identification between automobile lower limit gray threshold also forms vehicle image, when existing in detection band image
During vehicle image, based on front and back two frame vehicle image, determine whether vehicle moves, if motion, sentence
Before and after Duan, in two frame vehicle images, a frame below is moving vehicle image, and described Counter Value is automatic
Add 1 to calculate the frame number of moving vehicle image, make with the frame after in front and back two frame vehicle image
For new former frame, continue, based on two frame vehicle images before and after new, to perform above-mentioned moving vehicle image
Judge and above-mentioned counter operation, until detection band image does not exist vehicle image;Described embedded
Processor is connected respectively with position sensor, enumerator and vehicle image dispenser, when at vehicle image
It is sequentially received the first vehicle after the time that the first frame moving vehicle image in dispenser is corresponding to pass through
Signal and the second vehicle are through signal, and when Counter Value exceedes preset count threshold value, send vehicle and rush
Modulating signal, and the first frame moving vehicle image is sent to Car license recognition device;Car license recognition device,
Be connected with described flush bonding processor, based on the first frame moving vehicle image, according to algorithm of locating license plate of vehicle,
Character Segmentation of License Plate and the car plate data of optical character recognition algorithms identification vehicle;Wireless data leads to
Letter equipment, connects described Car license recognition device and described flush bonding processor, by defeated for Car license recognition device
The car plate data that go out and rush Word message corresponding to modulating signal with vehicle by cordless communication network respectively
It is sent to mobile terminal and the traffic control center of car plate data correspondence car owner.
Being positioned at described in more specifically, in the electronic recognition platform violating the regulations of traffic intersection, described user inputs
Device, described synchronous DRAM, described vehicle image dispenser, described embedded processing
The instrument case being positioned at traffic intersection is all located at by device and described Car license recognition device.
Being positioned at described in more specifically, in the electronic recognition platform violating the regulations of traffic intersection, described CMOS regards
Sense sensor, described sharpening image processor and described radio data communication equipment are all disposed within traffic
On fixing bar above crossing.
Being positioned at described in more specifically, in the electronic recognition platform violating the regulations of traffic intersection, described CMOS regards
The resolution of the craspedodrome directional image of sense sensor shooting is 1920 × 1080.
It is positioned at described in more specifically, in the electronic recognition platform violating the regulations of traffic intersection, described embedded place
Reason device is Freescale IMX6 processor.
Accompanying drawing explanation
Below with reference to accompanying drawing, embodiment of the present invention are described, wherein:
Fig. 1 is according to the electronic recognition platform violating the regulations being positioned at traffic intersection shown in embodiment of the present invention
Block diagram.
Fig. 2 is according to the electronic recognition platform violating the regulations being positioned at traffic intersection shown in embodiment of the present invention
The block diagram of position sensor.
Detailed description of the invention
Below with reference to accompanying drawings the present invention is positioned at the reality of the electronic recognition platform violating the regulations of traffic intersection
The scheme of executing is described in detail.
Road traffic signal lamp is the kind in traffic safety product, is to strengthen road traffic
Management, reduces the generation of vehicle accident, improves road occupation efficiency, improves a kind of weight of traffic
Want instrument.It is applicable to the intersection such as cross, T-shaped, by road traffic signal controller control, refers to
Guide-car and pedestrains safety are passed through in an orderly manner.The kind of traffic light mainly has: car lane signal
Lamp, pedestrian crosswalk signal lamp (i.e. traffic lights), bicycle lane signal lights, direction instruction signal lamp,
Mobile traffic light, solar energy flashing warning light, charge station's ceiling signal lights.Road is handed over
Ventilating signal lamp is the important component part in traffic signal commander, is the basic language of road traffic.
Road traffic signal lamp by red light (represent no through traffic), green light (represent allow current),
Amber light (representing warning) composition.It is divided into: motor vehicle signal lamp, bicycle signal lights, People's Bank of China's horizontal stroke
Road signal lights, lane signal lamp, direction instruction signal lamp, flashing, alarming signal lights, road and railway
Level-crossing grade crossing signal lamp.Being widely used in cross-road, it is hidden that bend, bridge etc. have safety
The Dangerous Area suffered from, conductor driver or pedestrian traffic, promote to have a good transport and communication network, it is to avoid vehicle accident and meaning
Outer accident occurs.
Along with development and the growth of private car quantity in city, have and drive driver's quantity of qualification more
Come the most.Owing to the driving habits of driver is different, or due to driver because urgent thing is made up for lost time,
When road traffic signal lamp is red, still adhere to, by crossing, i.e. there is situation about making a dash across the red light, and
The time of other direction vehicles and pedestrians the most just, make a dash across the red light and easily cause vehicle collision and mistake
The accident hindering pedestrian occurs, to other people with oneself bring economic loss and personal injury, harm relatively greatly,
It it is one of the target of control of traffic and road department key monitoring.
The technical scheme judged of making a dash across the red light of the prior art, is by the location near traffic intersection
Camera head is set on bar, is identified by the vehicle of front end, crossing presumptive area in red time
And judgement, the license board information determining the vehicle made a dash across the red light is sent to control of traffic and road department so that
Follow-up law enforcement.Although this mode can interpolate that out the vehicle made a dash across the red light to a certain extent, but
Stop after the centrage of crossing for going out front end, crossing presumptive area due to driver's erroneous judgement
Vehicle is also judged as making a dash across the red light, simultaneously as use single judgment rule and perform shooting be common
Camera head, the unavoidable situation that erroneous judgement occurs, thus judge that precision is the highest, further, since city holds
It is vulnerable to the invasion and attack of haze weather, under various haze weather in various degree, the shadow to electronic image
The degree of sound is different, objectively causes the possibility failed to judge.
To this end, the present invention proposes a kind of electronic recognition platform violating the regulations being positioned at traffic intersection, it is possible to carry
For the resolution of shooting image, carry out image taking, image recognition and Car license recognition more accurately, adopt
Combine ruling by two kinds of decision rules to make a dash across the red light behavior, improve the electronic recognition violating the regulations being positioned at traffic intersection
The accuracy of detection of platform, further, increases and goes hazeization to process electronic image, it is to avoid fail to judge
Situation occurs.
Fig. 1 is according to the electronic recognition platform violating the regulations being positioned at traffic intersection shown in embodiment of the present invention
Block diagram, described Identification platform includes CMOS vision sensor 1, sharpening image procossing
Device 2, detection band image segmentating device 3, vehicle image dispenser 4, enumerator 5 and flush bonding processor
6, described CMOS vision sensor 1 is for the region, Through Lane direction in traffic intersection
Shooting, to obtain craspedodrome directional image, described sharpening image processor 2 is used for receiving described
Craspedodrome directional image, carries out mist elimination and processes to obtain mist elimination craspedodrome directional diagram described craspedodrome directional image
Picture, described detection band image segmentating device 3 isolates detection band figure from described mist elimination craspedodrome directional image
Picture, described vehicle image dispenser 4 isolates the moving vehicle image in detection band image, described meter
Number device connects described vehicle image dispenser 4, for calculating moving vehicle image in the predetermined red light moment
Frame number, described flush bonding processor 6 and CMOS vision sensor 1, sharpening image processor
2, detection band image segmentating device 3, vehicle image dispenser 4, enumerator 5 connect respectively, predetermined
In the red light moment, count results based on described enumerator 5 determines whether there is running red light for vehicle.
Then, the structure of the described electronic recognition platform violating the regulations being positioned at traffic intersection is carried out more specifically
Describe.
Described Identification platform also includes: user's entering apparatus, for the input according to user, sets pre-
Determine automobile upper limit gray threshold, predetermined automobile lower limit gray threshold and preset count threshold value, always according to
The input at family, is configured the detection band in the Through Lane direction in traffic intersection, described detection band
The footpath front end in the Through Lane direction being arranged in traffic intersection, is a rectangular area, described
Detection band occupies 120 × 80 picture element matrixs in described craspedodrome directional image.
Described Identification platform also includes: power supply, including solar powered device, accumulator, cuts
Change switch and electric pressure converter, described switching switch and described solar powered device and described accumulator
Connect respectively, according to accumulator dump energy decide whether to be switched to described solar powered device with by
Described solar powered device is powered, and described electric pressure converter is connected with described switching switch, leading to
The 5V voltage crossing switching switch input is converted to 3.3V voltage.
As in figure 2 it is shown, described Identification platform also includes: position sensor 7, pass including primary importance
Sense unit 71 and second position sensing unit 72, described primary importance sensing unit 71 is arranged on detection
Band front position, for having detected whether that within the predetermined red light moment vehicle is through detection position, band front end
Putting, if it has, then export the first vehicle through signal, described second position sensing unit 72 is arranged
At traffic intersection position of center line, for having detected whether that vehicle is through traffic within the predetermined red light moment
Crossing position of center line, if it has, then export the second vehicle through signal, primary importance sensing unit
71 and second position sensing unit 72 be all contact-sensing unit, there is predetermined pressure by inspection
Above object contact determines there is vehicle process, and described traffic intersection position of center line is in described detection
Before band front position.
Described Identification platform also includes: synchronous DRAM SDRAM, connects user respectively
Entering apparatus, detection band image segmentating device 3, vehicle image dispenser 4 and flush bonding processor 6,
Receive and store predetermined automobile upper limit gray threshold, predetermined automobile lower limit gray threshold, preset count threshold
Value and described detection band configuration information.
Described sharpening image processor 2 is positioned at described CMOS vision sensor 1 and described detection band
Between image segmentating device 3, it is used for receiving described craspedodrome directional image, described craspedodrome directional image is entered
Row mist elimination processes to obtain mist elimination craspedodrome directional image, replaces described craspedodrome directional image, goes described
Mist craspedodrome directional image inputs described detection band image segmentating device 3 to isolate detection band image.
Described sharpening image processor 3 includes: haze Concentration Testing subset, is positioned in air,
For the haze concentration of detection traffic intersection position in real time, and determine that haze goes according to haze concentration
Except intensity, described haze removes intensity value between 0 to 1.
Described sharpening image processor 3 includes: overall air light value obtains subset, with described
CMOS vision sensor 1 connects to obtain described craspedodrome directional image, calculates described craspedodrome directional diagram
The gray value of each pixel in Xiang, using the gray value of pixel maximum for gray value as overall atmosphere light
Value.
Described sharpening image processor 3 includes: atmospheric scattering light value obtains subset, with described
CMOS vision sensor 1 and described haze Concentration Testing subset connect, respectively to described craspedodrome side
To each pixel of image, extract its R, G, B tri-minima conduct in Color Channel pixel value
Target pixel value, uses the Gaussian filter EPGF (edge-preserving keeping edge
Gaussian filter) it is filtered processing to obtain filtered target pixel value to described target pixel value,
Target pixel value is deducted filtered target pixel value to obtain object pixel difference, use EPGF to mesh
Mark pixel value difference is filtered processing to obtain filtered target pixel value difference, is subtracted by filtered target pixel value
Go filtered target pixel value difference to remove reference value to obtain haze, haze is removed intensity and is multiplied by haze
Except reference value removes threshold value to obtain haze, take haze and remove the minima in threshold value and target pixel value
As comparison reference, take the maximum in comparison reference and 0 scattered as the air of each pixel
Penetrate light value.
Described sharpening image processor 3 includes: medium transmission rate obtains subset, with described entirety
Air light value obtains subset and described atmospheric scattering light value obtains subset and connects respectively, by each
The atmospheric scattering light value of pixel divided by overall air light value to obtain except value, deduct 1 described except value with
Obtain the medium transmission rate of each pixel.
Described sharpening image processor 3 includes: sharpening Image Acquisition subset, with described CMOS
Vision sensor 1, described overall air light value obtain subset and described medium transmission rate obtains son and sets
Back-up does not connect, and deducts the medium transmission rate of each pixel by 1 to obtain the first difference, by described
First difference is multiplied by overall air light value to obtain product value, by described craspedodrome directional image each
The pixel value of pixel deducts described product value to obtain the second difference, by described second difference divided by each
The medium transmission rate of individual pixel to obtain the sharpening pixel value of each pixel, described craspedodrome directional diagram
The R of each pixel, G, B during in Xiang, the pixel value of each pixel includes described craspedodrome directional image
Three Color Channel pixel values, correspondingly, it is thus achieved that the sharpening pixel value of each pixel include each
The R of individual pixel, G, B tri-Color Channel sharpening pixel value, the sharpening pixel value of all pixels
Composition mist elimination craspedodrome directional image.
Described vehicle image dispenser 4 by gray value in detection band image at predetermined automobile upper limit gray scale threshold
Pixel identification between value and predetermined automobile lower limit gray threshold also forms vehicle image, when detection band figure
When there is vehicle image in Xiang, based on front and back two frame vehicle image, determine whether vehicle moves, if
Motion, it is judged that front and back in two frame vehicle images, a frame below is moving vehicle image, described counting
Device value adds 1 automatically to calculate the frame number of moving vehicle image, after in front and back two frame vehicle image
A frame as new former frame, continue, based on two frame vehicle images before and after new, to perform above-mentioned motion
Vehicle image judges and above-mentioned counter operation, until there is not vehicle image in detection band image.
Described flush bonding processor 6 and position sensor 7, enumerator 5 and vehicle image dispenser 4
Connect respectively, when after the time that the first frame moving vehicle image in vehicle image dispenser 4 is corresponding
It is sequentially received the first vehicle through signal and the second vehicle through signal, and the value of enumerator 5 exceedes
During preset count threshold value, send vehicle and rush modulating signal, and the first frame moving vehicle image is sent to car
Board recognition means.
Described Identification platform also includes: Car license recognition device, is connected with described flush bonding processor 6,
Based on the first frame moving vehicle image, according to algorithm of locating license plate of vehicle, Character Segmentation of License Plate and optics
The car plate data of character recognition algorithm identification vehicle.
Described Identification platform also includes: radio data communication equipment, connect described Car license recognition device and
Described flush bonding processor 6, by Car license recognition device export car plate data and rush modulating signal with vehicle
Corresponding Word message is sent respectively to the movement of car plate data correspondence car owner by cordless communication network
Terminal and traffic control center.
Alternatively, in the described electronic recognition platform violating the regulations being positioned at traffic intersection, described user inputs
Device, described synchronous DRAM, described vehicle image dispenser 4, described embedded place
The instrument case being positioned at traffic intersection is all located at by reason device 6 and described Car license recognition device;Described CMOS
Vision sensor 1, described sharpening image processor 2 and described radio data communication equipment are all arranged
On fixing bar above traffic intersection;The craspedodrome directional diagram of described CMOS vision sensor 1 shooting
The resolution of picture is 1920 × 1080;Described flush bonding processor 6 is Freescale IMX6 processor.
It addition, haze image can remove haze by what a series of images processing equipment realized image,
To obtain the image of sharpening, improve the visibility of image.These image processing equipments perform not respectively
Same image processing function, the principle formed based on haze, reach to remove the effect of haze.Haze figure
The sharpening of picture processes all has great using value, military domain bag for dual-use field
Including military and national defense, remote sensing navigation etc., civil area includes road monitoring, target following and automatic Pilot
Deng.
The process that haze image is formed can be described by atmospheric attenuation process, in haze image and reality
The medium of the available overall air light value of the relation between image i.e. sharpening image and each pixel passes
Defeated rate is stated, i.e. in the case of known haze image, according to overall air light value and each picture
The medium transmission rate of element, can solve sharpening image.
There are some and have in the solving of medium transmission rate for overall air light value and each pixel
Effect and through the means of checking, such as, for the medium transmission rate of each pixel, need to obtain whole
Body atmosphere light value and the atmospheric scattering light value of each pixel, and the atmospheric scattering light value of each pixel
Can be at the Gaussian smoothing that each pixel pixel value in haze image is carried out twice holding edge
Filtering and obtain, therebetween, the intensity that haze is removed is adjustable;And the acquisition pattern of entirety air light value has
Two kinds, a kind of mode is, (i.e. can make in haze image by obtaining the black channel of haze image
The black channel value of some pixels is the lowest, black channel is R, and G, B tri-is in Color Channel
A kind of), in haze image, the multiple pixels bigger than normal by finding black channel pixel value are found
The pixel of gray value maximum obtains, and gray value that will search out, gray value is maximum pixel is made
For overall air light value, participate in the sharpening of each pixel in haze image and process;It addition, it is overall
Air light value also can obtain in the following manner: calculates the gray value of each pixel in haze image, will
The gray value of the pixel that gray value is maximum is as overall air light value.
Relation between concrete haze image and real image i.e. sharpening image, and parameters
Between relation can be found in above content.
By the discussion to haze image formation basic theory, build between haze image and sharpening image
Relation, represent this relation by multiple parameters, subsequently by the multiple parameter values obtained and haze figure
The image that picture the most reducible acquisition definition is higher, owing to some statistical means have been used in the acquisition of parameter
And empirical means, the image that the most described definition is higher can not be fully equivalent to real image, but
Have and considerable degree of gone haze effect, provide effectively for the every field operation under haze weather
Ensure.
Use the electronic recognition platform violating the regulations being positioned at traffic intersection of the present invention, for existing knowledge of making a dash across the red light
The erroneous judgement the highest, easy of other platform precision and the technical problem failed to judge, use high-definition camera to carry out rushing red
Lamp image taking, combines position sensor and two kinds of results of image recognition, is used for whether determining vehicle
In red signal interval by crossing, wherein have employed image recognition, License Plate, License Plate Character Segmentation and
The image processing techniquess such as optical character recognition, whole Identification platform all obtains in terms of accuracy rate and efficiency
Bigger raising, it is often more important that, go the use of haze processing equipment to decrease the mist in red light running image
Haze composition, it is to avoid the generation of situation of failing to judge.
Although it is understood that the present invention discloses as above with preferred embodiment, but above-mentioned enforcement
Example is not limited to the present invention.For any those of ordinary skill in the art, without departing from
Under technical solution of the present invention ambit, all may utilize the technology contents of the disclosure above to the technology of the present invention
Scheme makes many possible variations and modification, or is revised as the Equivalent embodiments of equivalent variations.Therefore,
Every content without departing from technical solution of the present invention, the technical spirit of the foundation present invention is to above example
Any simple modification, equivalent variations and the modification done, all still falls within technical solution of the present invention protection
In the range of.