CN1448885A - Method and device for picking license plate area and correcting skew license plate from automobile image - Google Patents
Method and device for picking license plate area and correcting skew license plate from automobile image Download PDFInfo
- Publication number
- CN1448885A CN1448885A CN 02108154 CN02108154A CN1448885A CN 1448885 A CN1448885 A CN 1448885A CN 02108154 CN02108154 CN 02108154 CN 02108154 A CN02108154 A CN 02108154A CN 1448885 A CN1448885 A CN 1448885A
- Authority
- CN
- China
- Prior art keywords
- image
- license plate
- arithmetic element
- pixels
- plate area
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Traffic Control Systems (AREA)
Abstract
The present invention includes taking image of vehicle in lane with video camera, reading out the taken image, calculating logarithmic grey scale value, decomposing the image into coarse image, horizontal different image, vertical different image and diagonal different image, converting the real numbers of grey scale in the pixels of the horizontal different image into binary value in a binarizing operation unit, seeking the area with highest total binary value in the vehicle image with vehicle number plate cutting unit and defining the said area as the vehicle number plate area, correcting the vehicle number plate image with the correcting unit, and cutting off the non-vehicle number plate parts in the coarse area with the fine vehicle number plate area cutting unit.
Description
Technical field
The invention relates to a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, particularly can be applicable to car plate automatic identification system and traffic monitoring, vehicle gate inhibition's the cutting and rectification that the car plate image is crooked of license plate area image.
Background technology
Press, the location of the license plate area in the automobile image and to cut be the pre-process program of the automatic identification of car plate, so its accuracy influences the overall performance of the automatic identification of car plate; In addition, and even in vehicle entrance guard management on the traffic monitoring, managerial personnel need see through the number-plate number that video camera and display are watched car, if can be automatically cut, correct the part image of license plate area crooked and show, can make things convenient for managerial personnel to watch.Therefore, how automatically and effectively from automobile image, find out car plate the place, position, cut out and correct the crooked of car plate, be the important topic that car plate automatic identification system and traffic monitoring, vehicle gate inhibition etc. use.
Prior art is for the location of the license plate area in the automobile image and cut, generally be directly to use pixel gray level value to come computing to obtain the result, and change because of light easily or shade makes the operation result generation error of GTG value cause the location mistake of license plate area.The GTG value of each pixel is to be multiplied each other (Multiplication) and got by two compositions such as lightness and object reflection strength (to please refer to the Digital Image Processing that RafaelC.Gonzalez and Richard E.Woods are collaborateed in the image, 1993 editions, the 28th page to the 31st page); Lightness is to determine by external light is strong and weak, and the object reflection strength just really is the image characteristics of reflection object itself.
This shows that above-mentioned usual way still has many defectives, and imperfect, and demands urgently being improved.
Summary of the invention
The object of the present invention is to provide a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, can be applicable to automatic identification of car plate and traffic monitoring, vehicle gate inhibition.
Another object of the present invention is to provide a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, pixel gray level value is imposed logarithm operation (Logarithmic operation) to obtain logarithm GTG value, utilize logarithm GTG value to remove the influence of lightness and the place, position that compute location goes out car plate again, so can be because of the location mistake that light changes or shade makes the GTG value change to cause license plate area.
For achieving the above object, provided by the invention a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, main CCD camera, Frame Grabber by the collocation camera lens, and cooperate automobile image reading unit, logarithm GTG value arithmetic element, wavelet decomposition arithmetic element, image binaryzation arithmetic element, license plate area rough lumber to draw unit, skew license plate correcting unit, license plate area to fritter the computing or the processing of cutting the unit; Wherein:
The CCD camera and the Frame Grabber of collocation camera lens absorb automobile image to the track, and read to drive the unit by automobile image Frame Grabber institute picked image is read out, now is come each pixel in the automobile image is calculated its logarithm GTG value by logarithm GTG value arithmetic element, wavelet decomposition arithmetic element then now resolves into thick image with logarithm GTG value image, the level difference image, the vertical differentiation image, the diagonal angle difference image, now by image binaryzation arithmetic element will+, the logarithm GTG value that each pixel of arithmetic element image output is separated in wavelength-division is transferred to 0 or 1 bi-values by real number value; Cutting the unit by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to the default rough value of car plate length and width the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes the skew license plate correcting unit to correct the crooked of license plate area image, and what make is not crooked; Fritter at last and cut the unit and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area, with the place of license plate area to the end.
Wherein this logarithm GTG value arithmetic element adds 0.5 to the GTG value of each pixel, and making the pixel gray level value scope is from 0.5 to 255.5, and now is done the logarithm computing to produce logarithm GTG value to each pixel gray level value again.
Wherein this logarithm GTG value arithmetic element adds one to the GTG value of each pixel greater than 0 numerical value, makes the arithmetic error of each pixel gray level value being made logarithm computing unlikely generation logarithm 0 when producing logarithm GTG value.
Wherein the employed wavelet function of this wavelet decomposition arithmetic element is the Haar function.
Wherein the employed wavelet function of this wavelet decomposition arithmetic element is suitable for the function of wavelet conversion for all.
The level difference image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
The vertical differentiation image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
The diagonal angle difference image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
The thick image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting level difference image, as the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting vertical differentiation image, as the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting diagonal angle difference image, imitatively be the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting thick image, as the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting thick image, again by resulting image after the gradient computing, as the process object of image binaryzation arithmetic element.
Wherein this image binaryzation arithmetic element the bigger pixel number of absolute value of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, the numerical value of other pixels changes and is made as 0.
Wherein this image binaryzation arithmetic element the bigger pixel number of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, the numerical value of other pixels changes and is made as 0.
Wherein this license plate area ancestral cutter unit approximately be worth each twice with the car plate length and width scope as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole binaryzation image; The scope window frame that calculates the bi-values summation moves horizontally the length of a car plate or the width of a car plate of vertical moving at every turn.
Wherein this license plate area rough lumber cut the unit with greater than the scope of car plate length and width as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, just pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels, X-axle bed mark then remains unchanged.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigenvector, to produce the new Y-axle bed of these pixels mark.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigenvector, to produce the new X-axle bed of these pixels mark.
Provided by the invention a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, can be:, and cooperate automobile image reading unit, wavelet decomposition arithmetic element, image binaryzation arithmetic element, license plate area rough lumber to cut unit, skew license plate correcting unit, license plate area to fritter the computing or the processing of cutting the unit mainly by CCD camera, the Frame Grabber of collocation camera lens; Wherein:
The CCD camera and the Frame Grabber of collocation camera lens absorb automobile image to the track, and Frame Grabber institute picked image is read out by the automobile image reading unit, wavelet decomposition arithmetic element then now resolves into thick image, level difference image, vertical differentiation image with image, the diagonal angle difference image, now transfers the GTG value of each pixel of wavelet decomposition arithmetic element institute image output to 0 or 1 bi-values by image binaryzation arithmetic element; Cutting the unit by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to default car plate length and width value the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes the skew license plate correcting unit to correct the crooked of license plate area image, makes it not crooked; Fritter at last and cut the unit and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area, with the place of license plate area to the end.
Wherein the employed wavelet function of this wavelet decomposition arithmetic element is the Haar function.
The employed wavelet function of this wavelet decomposition arithmetic element wherein is for all are suitable for the function of wavelet conversion.
Wherein the level difference image that produced of this wavelet decomposition arithmetic element is the process object of image binaryzation arithmetic element.
The vertical differentiation image that produced of this wavelet decomposition arithmetic element wherein imitatively is the process object of image binaryzation arithmetic element.
Wherein the diagonal angle difference image that produced of this wavelet decomposition arithmetic element is the process object of image binaryzation arithmetic element.
Wherein the thick image that produced of this wavelet decomposition arithmetic element is the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element does one to the resulting level difference image of several wavelet decomposition, is the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element does one to the resulting vertical differentiation image of several wavelet decomposition, is the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element does one to the resulting diagonal angle of several wavelet decomposition difference image, is the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element does one to the resulting thick image of several wavelet decomposition, is the process object of image binaryzation arithmetic element.
Wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting thick image, make the resulting image of gradient computing again, be the process object of image binaryzation arithmetic element.
Wherein this image binaryzation arithmetic element the bigger pixel number of absolute value of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, other pixel numbers change and are made as 0.
Wherein this image binaryzation arithmetic element the bigger pixel number of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, other pixel numbers change and are made as 0.
Wherein this license plate area rough lumber cut the unit with the scope of each twice of car plate length and width value as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image; The scope window frame that calculates the bi-values summation moves horizontally the length of a car plate or the width of a car plate of vertical moving at every turn.
Wherein this license plate area rough lumber cut the unit with greater than the scope of car plate length and width as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels, X-axle bed mark then remains unchanged.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, just pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed of these pixels mark.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigenvector, to produce the new X-axle bed of these pixels mark.
Provided by the invention a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, also can be:, and cooperate automobile image reading unit, image binaryzation arithmetic element, license plate area rough lumber to cut unit, skew license plate correcting unit, license plate area to fritter the computing or the processing of cutting the unit mainly by CCD camera, the Frame Grabber of collocation camera lens; Wherein:
The CCD camera and the Frame Grabber of collocation camera lens absorb automobile image to the track, and by the automobile image reading unit Frame Grabber institute picked image is read out, now transfers the GTG value of each pixel of image to 0 or 1 bi-values by image binaryzation arithmetic element; Cutting the unit by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to default car plate length and width value the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes the skew license plate correcting unit to correct the crooked of license plate area image, makes it not crooked; Fritter at last and cut the unit and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area, with the place of license plate area to the end.
Wherein this license plate area rough lumber cut the unit with the scope of each twice of car plate length and width value as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole binaryzation image; The scope window frame that calculates the bi-values summation moves horizontally the length of a car plate or the width of a car plate of vertical moving at every turn.
Wherein this license plate area rough lumber cut the unit with greater than the scope of car plate length and width as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels, X-axle bed mark then remains unchanged.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigenvector, to produce the new Y-axle bed of these pixels mark.
Wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, also just fixed from the input image of binaryzation pixel number be that the X-axle of all pixels of 1 is marked the intension analysis of deciding with the Y-axle bed, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed of these pixels mark.
Description of drawings
See also the detailed description and the accompanying drawing thereof of following relevant a preferred embodiment of the present invention, can further understand technology contents of the present invention and purpose effect thereof, wherein:
Fig. 1 drags routine calcspar for the present invention picking license plate area and the method for correcting skew license plate and reality of device from automobile image;
Fig. 2 produces the synoptic diagram of thick image, level difference image, vertical differentiation image, diagonal angle difference image with wavelet decomposition for image;
Fig. 3 is the synoptic diagram of wavelet function with the Haar function representation;
Fig. 4 cuts the embodiment synoptic diagram of unit for the license plate area rough lumber;
Fig. 5 is the embodiment process flow diagram of skew license plate correcting unit;
Fig. 6 cuts the embodiment process flow diagram of unit for car plate fritters; And
Fig. 7 cuts the embodiment synoptic diagram of unit for car plate fritters.
Embodiment
The present invention is a kind of image pick up the car of clapping with video camera, and change into logarithm GTG value image, decomposite the level difference image and search the position of license plate area whole automobile image, location, method and the device correcting the crooked of car plate and license plate area is cut out from level difference gradation distribution situation with the wavelet decomposition method from raw video.CCD camera, Frame Grabber (Frame grabber) and the automobile image reading unit, logarithm GTG value (Logarithmicgray-level) arithmetic element, wavelet decomposition (Wavelet decomposition) arithmetic element, image binaryzation (Binarization) arithmetic element, the license plate area rough lumber that the present invention includes the collocation camera lens are cut unit, skew license plate correcting unit, license plate area and are frittered modules such as cutting the unit.
The GTG value of each pixel is to be multiplied each other and got by two compositions such as lightness and object reflection strength (to please refer to the Digital ImageProcessing that Rafael C.Gonzalez and Richard E.Woods are collaborateed in the image, 1993 editions, the 28th page to the 31st page); Lightness is to determine by external light is strong and weak, and the object reflection strength just really is the image characteristics of reflection object itself.The GTG value through logarithm operation resulting logarithm GTG value be logarithm lightness (Logarithmic illumination) with logarithm reflection strength (Logarithmic reflectance) and.
The resolution (Resolution) of supposing image frame is M*N pixel, and coordinate position (x, the GTG value of pixel y) be g (x, y), 0＜x＜M+1,0＜y＜N+1; (x, value y) can be directly reads from the output valve of CCD camera g.If coordinate position (x, pixel lightness y) be i (x, y), reflection strength be r (x, y), then g (x, y)=(x, y) (x, y), (x, (x, value y) can't read from the output valve of CCD camera i * r i for value y) and r.Because the pixel gray level value scope of the digitized video that the Frame Grabber acquisition produces is from 0 to 255 integer, in the digitized process its on average to cast out value (Truncated value) be 0.5, therefore we add 0.5 to the GTG value of each pixel, and making the pixel gray level value scope is from 0.5 to 255.5; After passing through logarithm operation again, logarithm GTG value be g ' (x, y), the logarithm lightness be I ' (x, y), the logarithm reflection strength be r ' (x, y), promptly
g’(x，y)＝1n?g(x，y)
＝1n?i(x，y)+1n?r(x，y)
＝i’(x，y)+r’(x，y)，0＜x＜M+1，0＜y＜N+1；
The relation that so just lightness, reflection strength is transferred to the logarithm addition just can be offseted the logarithm shading value through the process of wavelet decomposition generation difference image again, and difference image is almost presented by the logarithm reflection strength.
Wavelet decomposition (Wavelet decomposition) resolves into thick image, level difference image, vertical differentiation image, diagonal angle difference image with image; Whenever do wavelet decomposition one time, resulting thick image, level difference image, vertical differentiation image, diagonal angle difference image are all decomposes 1/4th of preceding image size.The present invention comes object as the cutting car plate with the level difference image: on the one hand, the size of level difference image is 1/4th of original image, therefore cut out the required operand of car plate just only surplus original 1/4th; On the other hand, the logarithm GTG value of each pixel of level difference image reflects the horizontal variable quantity of logarithm GTG value of former image, is for searching the special efficacy numerical value of car plate position institute foundation.
The present invention utilizes main intension analysis (Principal component analysis with being, be called Hotelling transform again, please refer to the Digital lmage Processing that Rafael C.Gonzalez and Richard E.Woods are collaborateed, 1993 editions, the 148th page to the 156th page) in find out the X-axle bed mark that the maximum pairing Eigen vector of Eigenvalue revises pixel; With pixel (X, Y) coordinate values projection (Projection) is to this Eigen vector, to produce the new X-axle bed mark of this pixel, Y-axle bed mark then remains unchanged; Situation with rectifiable skew license plate like this.
The CCD camera and the Frame Grabber of collocation camera lens absorb automobile image to the track, and Frame Grabber institute picked image is read out by the automobile image reading unit, now is come each pixel in the automobile image is calculated its logarithm GTG value by logarithm GTG value arithmetic element, wavelet decomposition arithmetic element then now resolves into thick image with logarithm GTG value image, the level difference image, the vertical differentiation image, the diagonal angle difference image, now transfers the logarithm GTG value of each pixel of level difference image to 0 or 1 bi-values by real number value (Real number) by image binaryzation arithmetic element.Cutting the unit by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to the default rough value of car plate length and width the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes the skew license plate correcting unit to correct the crooked of license plate area image, makes it not crooked as far as possible; Fritter at last and cut the unit and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area.
See also shown in Figure 1, for a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, mainly be by the CCD camera 1 that comprises the camera lens of arranging in pairs or groups, Frame Grabber (Framegrabber) 2, and cooperate automobile image reading unit 3, logarithm GTG value (Logarithmic gray-level) arithmetic element 4, wavelet decomposition (Wavelet decomposition) arithmetic element 5, image binaryzation (Binarization) arithmetic element 6, unit 7 is cut in the license plate area rough lumber, skew license plate correcting unit 8, license plate area fritters modules such as cutting unit 9 and forms.In Fig. 1, the no housing object that dotted arrow top is indicated is an input signal, and the no housing object that the dotted arrow terminal is indicated is an output signal.2 pairs of track picked-ups of the CCD camera 1 of collocation camera lens and Frame Grabber automobile image, and 2 picked image of Frame Grabber are read out by automobile image reading unit 3, now is come each pixel in the automobile image is calculated its logarithm GTG value by logarithm GTG value arithmetic element 4,5 nows of wavelet decomposition arithmetic element resolve into thick image with logarithm GTG value image, the level difference image, the vertical differentiation image, the diagonal angle difference image, now transfers the logarithm GTG value of each pixel of level difference image to 0 or 1 bi-values by real number value (Real number) by image binaryzation arithmetic element 6.Cutting unit 7 by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to the default rough value of car plate length and width the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes skew license plate correcting unit 8 to correct the crooked of license plate area image, makes it not crooked; Fritter at last and cut unit 9 and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area, with the place of license plate area to the end.
Wherein Da Pei Lens CCD video camera 1, Frame Grabber 2 are known products; Frame Grabber 2 produces raw video, and 3 of automobile image reading units connect with Frame Grabber Jie and image is read out with temporary; Automobile image reading unit 3 can be made up of elements such as processor (Processor) wafer, temporary storage (Temporary memory) wafer, permanent storage (Permanent memory) wafer, time sequential pulse (Sequential Pulse) wafer and power supply units.
Because the pixel gray level scope of the digitized video that Frame Grabber 2 acquisitions produce is from 0 to 255 integer, it is 0.5 that value (Truncated value) is cast out in its average digitizing, therefore we add 0.5 to the GTG value of each pixel in logarithm GTG value arithmetic element 4, make pixel gray level value scope from 0.5 to 255.5, then again each pixel gray level value is done the logarithm computing to produce logarithm GTG value.Logarithm GTG value arithmetic element 4 changes into logarithm GTG value image with raw video, and logarithm GTG value arithmetic element 4 can be made up of elements such as processor (Processor) wafer, temporary storage (Temporary memory) wafer, permanent storage (Permanent memory) wafer, time sequential pulse (Sequential Pulse) wafer and power supply units.
Wavelet decomposition (Wavelet decomposition) arithmetic element 5 resolves into thick image, level difference image, vertical differentiation image, diagonal angle difference image with image; Shown in Figure 2 is image produces thick image, level difference image, vertical differentiation image, diagonal angle difference image with wavelet decomposition synoptic diagram; The spendable wavelet function of wavelet decomposition has a variety of, and that shown in Figure 3 is a kind of of wavelet function: the synoptic diagram of Haar function.We come object as the cutting car plate with the level difference scene: on the one hand, the size of level difference scene is 1/4th of original image, therefore cut out the required operand of car plate just only surplus original 1/4th; On the other hand, the logarithm GTG value of each pixel of level difference image reflects the horizontal variable quantity of logarithm GTG value of former image, is for searching the character numerical value of car plate position institute foundation.Wavelet decomposition arithmetic element 5 utilizes logarithm GTG value image to produce the level difference image, and the wavelet decomposition arithmetic element can be levied and element such as power supply unit is formed by processor wafer, temporary storage wafer, permanent storage wafer, time sequential pulse crystalline substance.
The numerical value of the pixel that image binaryzation arithmetic element 6 is bigger with the absolute value of all pixel numbers of level difference image change be made as 1, the numerical value of other pixels changes and is made as 0; The ratio that the pixel that the absolute value of pixel number is bigger accounts for all sum of all pixels can be 1/100,1/50 or other are less than 1 suitable ratio.Image binaryzation arithmetic element 6 utilizes the level difference image to produce level difference binaryzation image, and image binaryzation arithmetic element 6 can be made up of elements such as processor wafer, temporary storage wafer, permanent storage wafer, time sequential pulse wafer and power supply units.
Shown in Figure 4 is the embodiment synoptic diagram that unit 7 is cut in the license plate area rough lumber, the scope that approximately is worth each twice with the car plate length and width is as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image; The scope window frame that calculates the bi-values summation moves horizontally the length of a car plate or the width of a car plate of vertical moving at every turn.Find out a scope of bi-values summation maximum from each moving, this rectangle scope is the thick zone level difference of car plate binaryzation image; The coordinate values at four angles of the thick zone level difference of car plate binaryzation image be multiply by 2, promptly obtain the coordinate values at four angles, the thick zone of car plate in the raw video, this zone is cut out promptly get the thick area image of car plate.The license plate area rough lumber is cut unit 7 and is utilized the about length and width value of car plate, level difference binaryzation image, raw video to produce the thick zone level difference of car plate binaryzation image, the thick area image of car plate, and the license plate area rough lumber is cut the unit and can be made up of elements such as processor wafer, temporary storage wafer, permanent storage wafer, time sequential pulse wafer and power supply units.
Skew license plate correcting unit 8 utilizes the thick zone level difference of car plate binaryzation image, the thick area image of car plate to produce the thick zone level difference of car plate binaryzation and corrects image, the thick zone rectification of car plate image.Shown in Figure 5 is the embodiment process flow diagram of skew license plate correcting unit; At first to the intension analysis (Principal component analysis) of deciding of the thick zone level difference of car plate binaryzation image, just from the thick zone level difference of car plate binaryzation image pixel number be in the X-axle of all pixels of 1 and Y-axle bed mark are done main the culvert analyze, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, then with the (X of all pixels of the thick zone level difference of car plate binaryzation image, Y) coordinate values projection (Projection) is to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged, and therefore can produce the thick zone level difference of car plate binaryzation and correct image; Same (X to thick all pixels of area image of car plate, Y) coordinate values projection (Projection) is to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged, therefore can produce the thick zone of car plate and correct image, the situation of skew license plate is to solve with this.Skew license plate correcting unit 8 can be made up of elements such as processor wafer, temporary storage wafer, permanent storage wafer, time sequential pulse wafer and power supply units.
Last step is that frittering of license plate area cut, and shown in Figure 6 is the embodiment process flow diagram of license plate area cutter unit 9, and license plate area shown in Figure 7 fritters the operating principle that the embodiment synoptic diagram that cuts unit 9 then helps to understand Fig. 6.At first calculate the center of gravity of the thick zone level difference of car plate binaryzation image, then do following processing: the bi-values summation of every row (Row) that calculating center of gravity Y-axle bed is marked with at the thick zone level difference of car plate binaryzation image, finding out more preceding up bi-values summation from top to bottom increases maximum delegation, and the Y-axle bed that writes down this row is designated as Y1; The bi-values summation of the every row (Row) under calculating center of gravity Y-axle bed is marked with, the bi-values summation of finding out from lower to upper than previous row increases maximum delegation, and the Y-axle bed that writes down this row is designated as Y1; Calculate the bi-values summation that center of gravity X-axle bed is marked with every row (Column) on a left side, increase maximum row by the bi-values summation of finding out from left to right than previous column, the X-axle bed that writes down these row is designated as X1; Calculate the bi-values summation that center of gravity X-axle bed is marked with right every row (Column), increase maximum row by the right side bi-values summation of finding out than previous column of turning left, the X-axle bed that writes down these row is designated as X2.Then the thin zone level difference of car plate binaryzation correct the coordinate of correcting image in the thick zone level difference of car plate binaryzation in four angles of image be (X1, Y1), (X1, Y2), (X2, Y1), (X2, Y2); Because the thin zone level difference of car plate binaryzation is corrected the thick zone level difference of length and width, the car plate binaryzation of image and is corrected the length and width that the length and width of image are respectively the thin zone of car plate, / 2nd of thick regional length and width of car plate, therefore four coordinates of angle in the thick zone of car plate in the thin zone of car plate are (2*X1,2*Y1), (1*X1,2*Y2), (2*X2,2*Y1), (2*X2,2*Y2); Having corrected the thin area image of crooked car plate therefore just can cut out.License plate area fritters and cuts the unit and can be made up of elements such as processor wafer, temporary storage wafer, permanent storage wafer, time sequential pulse wafer and power supply units.
Provided by the present invention from vehicle shadow Lisu Nationality the method and the device of picking license plate area and correcting skew license plate, with aforementioned case and other common technologies quoted as proof mutually relatively the time, have following advantage:
1, the invention provides a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, can be applicable to automatic identification of car plate and traffic monitoring, vehicle gate inhibition.
2, the invention provides a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, be that pixel gray level value is imposed logarithm operation (Logarithmic operation) to obtain logarithm GTG value, utilize logarithm GTG value to remove the influence of lightness and the place, position that compute location goes out car plate again, so can be because of the location mistake that light changes or shade makes the GTG value change to cause license plate area.
3, the invention provides a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, to cutting out the required operand of car plate, have only 1/4th of general common method.
4, the invention provides a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, the crooked situation of rectifiable car plate image.
Above-listed detailed description is specifying at a possible embodiments of the present invention, but this embodiment is not in order to limit claim of the present invention, allly do not break away from the equivalence that the technology of the present invention spirit does and implement or change, for example: wait the equivalence embodiment of variation, all should be contained in the claim of the present invention.
Claims (49)
1, a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, main CCD camera, Frame Grabber by the collocation camera lens, and cooperate automobile image reading unit, logarithm GTG value arithmetic element, wavelet decomposition arithmetic element, image binaryzation arithmetic element, license plate area rough lumber to draw unit, skew license plate correcting unit, license plate area to fritter the computing or the processing of cutting the unit; Wherein:
The CCD camera and the Frame Grabber of collocation camera lens absorb automobile image to the track, and read to drive the unit by automobile image Frame Grabber institute picked image is read out, now is come each pixel in the automobile image is calculated its logarithm GTG value by logarithm GTG value arithmetic element, wavelet decomposition arithmetic element then now resolves into thick image with logarithm GTG value image, the level difference image, the vertical differentiation image, the diagonal angle difference image, now by image binaryzation arithmetic element will+, the logarithm GTG value that each pixel of arithmetic element image output is separated in wavelength-division is transferred to 0 or 1 bi-values by real number value; Cutting the unit by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to the default rough value of car plate length and width the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes the skew license plate correcting unit to correct the crooked of license plate area image, and what make is not crooked; Fritter at last and cut the unit and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area, with the place of license plate area to the end.
2, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this logarithm GTG value arithmetic element adds 0.5 to the GTG value of each pixel, making the pixel gray level value scope is from 0.5 to 255.5, and now is done the logarithm computing to produce logarithm GTG value to each pixel gray level value again.
3, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this logarithm GTG value arithmetic element adds one to the GTG value of each pixel greater than 0 numerical value, makes the arithmetic error of each pixel gray level value being made logarithm computing unlikely generation logarithm 0 when producing logarithm GTG value.
4, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that wherein the employed wavelet function of this wavelet decomposition arithmetic element is the Haar function.
5, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that wherein the employed wavelet function of this wavelet decomposition arithmetic element is suitable for the function of wavelet conversion for all.
6, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that the level difference image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
7, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that the vertical differentiation image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
By described method and the device of driving license plate area and correcting skew license plate of from automobile image, picking of claim 1, it is characterized in that 8, the diagonal angle difference image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
9, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that the thick image that produced of this wavelet decomposition arithmetic element wherein is as the process object of image binaryzation arithmetic element.
10, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting level difference image, as the process object of image binaryzation arithmetic element.
11, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting vertical differentiation image, as the process object of image binaryzation arithmetic element.
12, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting diagonal angle difference image, imitatively be the process object of image binaryzation arithmetic element.
13, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting thick image, as the process object of image binaryzation arithmetic element.
14, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting thick image, again by resulting image after the gradient computing, as the process object of image binaryzation arithmetic element.
15, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this image binaryzation arithmetic element the bigger pixel number of absolute value of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, the numerical value of other pixels changes and is made as 0.
16, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this image binaryzation arithmetic element the bigger pixel number of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, the numerical value of other pixels changes and is made as 0.
17, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this license plate area ancestral cutter unit approximately be worth each twice with the car plate length and width scope as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole binaryzation image; The scope window frame that calculates the bi-values summation moves horizontally the length of a car plate or the width of a car plate of vertical moving at every turn.
18, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this license plate area rough lumber cut the unit with greater than the scope of car plate length and width as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image.
19, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, just pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged.
20, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels, X-axle bed mark then remains unchanged.
21, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed of these pixels mark.
22, by claim 1 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed of these pixels mark.
23, a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, mainly by CCD camera, the Frame Grabber of collocation camera lens, and cooperate automobile image reading unit, wavelet decomposition arithmetic element, image binaryzation arithmetic element, license plate area rough lumber to cut unit, skew license plate correcting unit, license plate area to fritter the computing or the processing of cutting the unit; Wherein:
The CCD camera and the Frame Grabber of collocation camera lens absorb automobile image to the track, and Frame Grabber institute picked image is read out by the automobile image reading unit, wavelet decomposition arithmetic element then now resolves into thick image, level difference image, vertical differentiation image with image, the diagonal angle difference image, now transfers the GTG value of each pixel of wavelet decomposition arithmetic element institute image output to 0 or 1 bi-values by image binaryzation arithmetic element; Cutting the unit by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to default car plate length and width value the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes the skew license plate correcting unit to correct the crooked of license plate area image, makes it not crooked; Fritter at last and cut the unit and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area, with the place of license plate area to the end.
24, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that wherein the employed wavelet function of this wavelet decomposition arithmetic element is the Haar function.
25, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that the employed wavelet function of this wavelet decomposition arithmetic element wherein is for all are suitable for the function of wavelet conversion.
26, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that wherein the level difference image that produced of this wavelet decomposition arithmetic element is the process object of image binaryzation arithmetic element.
27, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, the vertical differentiation image that produced of this wavelet decomposition arithmetic element wherein imitatively is the process object of image binaryzation arithmetic element.
28, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that wherein the diagonal angle difference image that produced of this wavelet decomposition arithmetic element is the process object of image binaryzation arithmetic element.
29, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that wherein the thick image that produced of this wavelet decomposition arithmetic element is the process object of image binaryzation arithmetic element.
30, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element does one to the resulting level difference image of several wavelet decomposition, is the process object of image binaryzation arithmetic element.
31, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element does one to the resulting vertical differentiation image of several wavelet decomposition, is the process object of image binaryzation arithmetic element.
32, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element does one to the resulting diagonal angle of several wavelet decomposition difference image, is the process object of image binaryzation arithmetic element.
33, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element does one to the resulting thick image of several wavelet decomposition, is the process object of image binaryzation arithmetic element.
34, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this wavelet decomposition arithmetic element do one to the several wavelet decomposition resulting thick image, make the resulting image of gradient computing again, be the process object of image binaryzation arithmetic element.
35, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this image binaryzation arithmetic element the bigger pixel number of absolute value of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, other pixel numbers change and are made as 0.
36, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this image binaryzation arithmetic element the bigger pixel number of all pixel numbers of wavelet decomposition arithmetic element institute image output is changed be made as 1, other pixel numbers change and are made as 0.
37, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this license plate area rough lumber cut the unit with the scope of each twice of car plate length and width value as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image; The scope window frame that calculates the bi-values summation moves horizontally the length of a car plate or the width of a car plate of vertical moving at every turn.
38, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this license plate area rough lumber cut the unit with greater than the scope of car plate length and width as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image.
39, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged.
40, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels, X-axle bed mark then remains unchanged.
41, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, just pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed of these pixels mark.
42, by claim 23 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed of these pixels mark.
43, a kind of from automobile image the method and the device of picking license plate area and correcting skew license plate, mainly by CCD camera, the Frame Grabber of collocation camera lens, and cooperate automobile image reading unit, image binaryzation arithmetic element, license plate area rough lumber to cut unit, skew license plate correcting unit, license plate area to fritter the computing or the processing of cutting the unit; Wherein:
The CCD camera and the Frame Grabber of collocation camera lens absorb automobile image to the track, and by the automobile image reading unit Frame Grabber institute picked image is read out, now transfers the GTG value of each pixel of image to 0 or 1 bi-values by image binaryzation arithmetic element; Cutting the unit by the license plate area rough lumber then, to seek in whole the automobile image that regional bi-values summation according to default car plate length and width value the highest, and should the zone tentatively cut out the place into license plate area; Now utilizes the skew license plate correcting unit to correct the crooked of license plate area image, makes it not crooked; Fritter at last and cut the unit and excise the non-part that belongs to car plate in the thick zone of car plate by license plate area, with the place of license plate area to the end.
44, by claim 43 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this license plate area rough lumber cut the unit with the scope of each twice of car plate length and width value as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole binaryzation image; The scope window frame that calculates the bi-values summation moves horizontally the length of a car plate or the width of a car plate of vertical moving at every turn.
45, by claim 43 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this license plate area rough lumber cut the unit with greater than the scope of car plate length and width as the scope window frame that calculates the bi-values summation, the scope window frame of mobile computing bi-values summation and calculate the summation of all bi-values in each scope in whole level difference binaryzation image.
46, by claim 43 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels, Y-axle bed mark then remains unchanged.
47, by claim 43 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels, X-axle bed mark then remains unchanged.
48, by claim 43 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, promptly pixel number is the intension analysis of deciding of the X-axle of all pixels of 1 and Y-axle bed mark from the input image of binaryzation, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed of these pixels mark.
49, by claim 43 described from automobile image the method and the device of picking license plate area and correcting skew license plate, it is characterized in that, wherein this skew license plate correcting unit is at first to the intension analysis of deciding of input image, also just fixed from the input image of binaryzation pixel number be that the X-axle of all pixels of 1 is marked the intension analysis of deciding with the Y-axle bed, and from main intension analysis, find out the maximum pairing Eigen vector of Eigen value, to import (the X of all pixels of image then, Y) coordinate values is projected to this Eigen vector, to produce the new Y-axle bed mark of these pixels; From main intension analysis, find out the pairing Eigen vector of second largest Eigen value simultaneously, will import then all pixels of image (X, Y) coordinate values is projected to this Eigen vector, to produce the new X-axle bed of these pixels mark.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 02108154 CN1198235C (en) | 2002-03-28 | 2002-03-28 | Method and device for picking license plate area and correcting skew license plate from automobile image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 02108154 CN1198235C (en) | 2002-03-28 | 2002-03-28 | Method and device for picking license plate area and correcting skew license plate from automobile image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1448885A true CN1448885A (en) | 2003-10-15 |
| CN1198235C CN1198235C (en) | 2005-04-20 |
Family
ID=28680212
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 02108154 Expired - Lifetime CN1198235C (en) | 2002-03-28 | 2002-03-28 | Method and device for picking license plate area and correcting skew license plate from automobile image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1198235C (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100349175C (en) * | 2004-05-14 | 2007-11-14 | 中华电信股份有限公司 | Method for plucking license-plate region from vehicle image |
| CN101630360B (en) * | 2008-07-14 | 2012-12-19 | 上海分维智能科技有限公司 | Method for identifying license plate in high-definition image |
| TWI715184B (en) * | 2019-09-04 | 2021-01-01 | 新煒科技有限公司 | Method and device for positioning target object in image, computer device and storage medium |
-
2002
- 2002-03-28 CN CN 02108154 patent/CN1198235C/en not_active Expired - Lifetime
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100349175C (en) * | 2004-05-14 | 2007-11-14 | 中华电信股份有限公司 | Method for plucking license-plate region from vehicle image |
| CN101630360B (en) * | 2008-07-14 | 2012-12-19 | 上海分维智能科技有限公司 | Method for identifying license plate in high-definition image |
| TWI715184B (en) * | 2019-09-04 | 2021-01-01 | 新煒科技有限公司 | Method and device for positioning target object in image, computer device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1198235C (en) | 2005-04-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1252986C (en) | Information processing method and apparatus | |
| CN1177298C (en) | Multiple focussing image fusion method based on block dividing | |
| CN1297943C (en) | Image defect inspecting apparatus and image defect inspecting method | |
| CN1885311A (en) | Two-dimensional code, encoding and decoding method thereof | |
| CN1851555A (en) | Method for realizing two-dimensional panoramic true imaging | |
| CN1691050A (en) | 2D rectangular code symbol scanning device and 2D rectangular code symbol scanning method | |
| CN1908955A (en) | Trilateral poly-dimensional bar code easy for omnibearing recognition and reading method thereof | |
| CN1525204A (en) | Scanning device calibration system and method | |
| CN1402551A (en) | Apparatus and method for automatically tracking mobile object | |
| CN101029824A (en) | Method and apparatus for positioning vehicle based on characteristics | |
| CN101039430A (en) | Method for scanning quickly residual matrix in video coding | |
| CN1761309A (en) | Signal processing apparatus and signal processing method for image data | |
| CN1459189A (en) | Screen correcting method and imaging device | |
| CN1856028A (en) | Camera and method for taking picture compensation | |
| CN1835547A (en) | Image processing device and registration data generation method in image processing | |
| CN1531709A (en) | Image characteristic identification signal generation apparatus and image characteristic identification signal generation method | |
| CN101079108A (en) | DSP based multiple channel mechanical digital display digital gas meter automatic detection device | |
| CN1886758A (en) | Method and apparatus for downscaling a digital matrix image | |
| CN1198235C (en) | Method and device for picking license plate area and correcting skew license plate from automobile image | |
| CN101042734A (en) | Method for rapid making image connection element | |
| CN1508751A (en) | Photoelectric mouse and method for preventing it from anormal operation | |
| CN1232926C (en) | Apparatus for conducting binary coding to image and using method thereof | |
| CN1828631A (en) | Method and apparatus for acquiring image of internal structure, and computer product | |
| CN1612155A (en) | Computer pixel moving method and position tracing sensor using same | |
| CN101035196A (en) | Apparatus for detecting a varied area and method of detecting a varied area |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |