CN104966047A - Method and device for identifying vehicle license - Google Patents

Abstract

The invention provides a method and device for identifying vehicle license. The method comprises: predetermining characters and presetting template images of each character, wherein each character corresponds to at least one template image and each template image corresponds to one character; acquiring an image of a vehicle license to be identified; according to characters in the image of the vehicle license to be identified, segmenting the image of the vehicle license to be identified in order to obtain segmented images; matching each current segmented image with a template image in order to determine a matched template image most matching the current segmented image and acquiring the matched template image of each segmented image; and using the character corresponding to the matched template image of the current segmented image as a character corresponding to the current segmented image. The method and device for identifying vehicle license improve a correct identification rate.

Description

A kind of method of Car license recognition and device

Technical field

The present invention relates to technical field of image processing, particularly a kind of method of Car license recognition and device.

Background technology

Along with City ITS develops rapidly, automatic Recognition of License Plate, as its core technology, have also been obtained the development of leap.In China, structure, the form of car plate and be abroad very different, such as, contain Chinese character in the car plate of China.

In prior art, the identification of car plate is mainly identified the profile of character in license plate image, determines character concrete in license plate image by identification character profile.In existing recognition methods, have higher requirement to the sharpness of license plate image to be identified, if the character in license plate image to be identified had a stain, cause recognition result mistake possibly, the correct recognition rata of prior art is lower.

Summary of the invention

In view of this, the invention provides a kind of method and device of Car license recognition, can correct recognition rata be improved.

On the one hand, the invention provides a kind of method of Car license recognition, comprising: pre-determine character, pre-set the template image of each character, wherein, each character at least one template image corresponding, the corresponding character of each template image, also comprises:

S1: obtain license plate image to be identified;

S2: according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains segmentation image;

S3: mated with each template image by current segmentation image, determines and the matching template image that current segmentation image mates most, obtains the matching template image of each segmentation image;

S4: using character corresponding for the matching template image of current segmentation image as character corresponding to current segmentation image.

Further, described predefined character, pre-sets the template image of each character, wherein, each character at least one template image corresponding, the corresponding character of each template image, comprising:

Described character comprises: Chinese character, alphabetic character, numerical character;

Described template image comprises: Chinese character template image, alphabetical template image, digital template image;

Pre-determine Chinese character, alphabetic character, numerical character, pre-set the Chinese character template image of each Chinese character, pre-set the Chinese character template image of each alphabetic character, pre-set the Chinese character template image of each numerical character, wherein, each Chinese character at least one Chinese character template image corresponding, the corresponding Chinese character of each Chinese character template image, each alphabetic character at least one alphabetical template image corresponding, the corresponding alphabetic character of each alphabetical template image, corresponding at least one the digital template image of each numerical character, the corresponding numerical character of each digital template image,

Described segmentation image comprises: Chinese character segmentation image, grapheme segmentation image, digital segmentation image;

Obtain segmentation image described in described S2, comprising: obtain Chinese character segmentation image, obtain grapheme segmentation image, obtain digital segmentation image;

Described matching template image comprises: mate Chinese character template image, mate alphabetical template image, coupling digital template image;

Described S3 comprises:

A1: mated with each Chinese character template image by current Chinese character segmentation image, determines and the coupling Chinese character template image that current Chinese character segmentation image mates most, obtains the coupling Chinese character template image of each Chinese character segmentation image;

A2: current letter is split image and mate with each alphabetical template image, determines the coupling letter template image split image with current letter and mate most, obtains the coupling letter template image of each grapheme segmentation image;

A3: Contemporary Digital is split image and mate with each digital template image, determines the coupling digital template image split image with Contemporary Digital and mate most, obtains the coupling digital template image of each digital segmentation image.

Further, described A1, comprising:

Current Chinese character template image is moved on current Chinese character segmentation image, travel through all pixels of current Chinese character segmentation image, and calculate current Chinese character template image by formula one and moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image, wherein, described formula one is:

$\mathrm{\γ}(u,v)=\frac{{\mathrm{\Σ}}_{x,y}[f(x,y)-{\stackrel{\‾}{f}}_{u,v}][t(x-u,y-v)-\stackrel{\‾}{t}]}{\{{\mathrm{\Σ}}_{x,y}{[f(x,y)-{\stackrel{\‾}{f}}_{u,v}]}^2{\mathrm{\Σ}}_{x,y}{[t(x-u,y-v)-\stackrel{\‾}{t}]}^2{\}}^{0.5}}$

Wherein, u is the distance of current Chinese character template image movement in the x direction, v is the distance of current Chinese character template image movement in y-direction, γ (u, v) for current Chinese character template image is moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image after current Chinese character template movement (u, v) distance, the average gray of the current Chinese character segmentation image under current Chinese character template image covers, for the average gray of current Chinese character template image, t (x-u, y-v) is for current Chinese character template image is at point (x-u, y-v) gray-scale value, f (x, y) is for current Chinese character segmentation image is at the gray-scale value of point (x, y);

By normalized-cross-correlation function maximum in all normalized-cross-correlation function of current Chinese character template image and current Chinese character segmentation image, as the first matching degree of current Chinese character segmentation image and current Chinese character template image;

Determine the first matching degree of current Chinese character segmentation image and each Chinese character template image;

By the Chinese character template image maximum with the first matching degree of current Chinese character segmentation image, as the coupling Chinese character template image of current Chinese character segmentation image.

Further, described A2, comprising:

Sample to current letter segmentation image and obtain the first Division Sampling point of the first predetermined number, sampling to current letter template image obtains the first palette sample point of described first predetermined number;

Determine the shape histogram of each first Division Sampling point under polar coordinate system in each default characteristic area, determine the shape histogram under polar coordinate system in each default characteristic area of each first palette sample point;

Calculate the Matching power flow between current first Division Sampling point and current first palette sample point according to formula two, wherein, formula two is:

$C(p_i,q_j)=\frac{1}{2}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{{[h_i\left(k\right)-h_j\left(k\right)]}^2}{h_i\left(k\right)+h_j\left(k\right)}$

Wherein, p _ifor current first Division Sampling point, q _jfor current first palette sample point, C (p _i, q _j) be a p _iwith a q _jbetween Matching power flow, h _ik () is a kth characteristic area mid point p in K the characteristic area preset under polar coordinate system _ishape histogram, h _jk () is a kth characteristic area mid point q in K the characteristic area preset under polar coordinate system _jshape histogram;

Determine the Matching power flow between arbitrary first Division Sampling point and arbitrary first palette sample point;

All first Division Sampling points are mated with all first palette sample points, obtains alphabetical matching scheme;

Determine the total Matching power flow of letter of each alphabetical matching scheme according to formula three, wherein, formula three is:

$C_{\mathrm{zm}}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i$

Wherein, C _zmfor the total Matching power flow of letter, C _ifor the i-th Matching power flow to the first Division Sampling point matched and the first palette sample point in current letter matching scheme, n is described first predetermined number;

Determine the best letter Matching power flow that the total Matching power flow of letter is minimum, alphabetical for described the best Matching power flow is split the second matching degree of image and current letter template image as current letter;

Determine the second matching degree of current letter segmentation image and each alphabetical template image;

The minimum alphabetical template image of the second matching degree of image is split, as the coupling letter template image of current letter segmentation image by with current letter.

Further, described A3, comprising:

Sample to Contemporary Digital segmentation image and obtain the second Division Sampling point of the second predetermined number, sampling to Contemporary Digital template image obtains the second palette sample point of described second predetermined number;

Determine the shape histogram of each second Division Sampling point under polar coordinate system in each default characteristic area, determine the shape histogram under polar coordinate system in each default characteristic area of each second palette sample point;

Calculate the Matching power flow between current second Division Sampling point and current second palette sample point according to formula four, wherein, formula four is:

$C(r_x,s_y)=\frac{1}{2}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\frac{{[h_x\left(m\right)-h_y\left(m\right)]}^2}{h_x\left(m\right)+h_y\left(m\right)}$

Wherein, r _xfor current first Division Sampling point, s _yfor current first palette sample point, C (r _x, s _y) be a r _xwith a r _xbetween Matching power flow, h _xm () is m characteristic area mid point r in M the characteristic area preset under polar coordinate system _xshape histogram, h _ym () is m characteristic area mid point s in M the characteristic area preset under polar coordinate system _yshape histogram;

Determine the Matching power flow between arbitrary second Division Sampling point and arbitrary second palette sample point;

All second Division Sampling points are mated with all second palette sample points, obtains digital matching scheme;

Determine the total Matching power flow of numeral of each digital matching scheme according to formula five, wherein, formula five is:

$C_{\mathrm{sz}}=\underset{s=1}{\overset{f}{\mathrm{\Σ}}}{C}_s$

Wherein, C _szfor the total Matching power flow of numeral, C _sfor s in Contemporary Digital matching scheme is to the Matching power flow of the second Division Sampling point matched and the second palette sample point, f is described second predetermined number;

Determine the optimal digital Matching power flow that the total Matching power flow of numeral is minimum, described optimal digital Matching power flow is split the 3rd matching degree of image and Contemporary Digital template image as Contemporary Digital;

Determine the 3rd matching degree of Contemporary Digital segmentation image and each digital template image;

The minimum digital template image of the 3rd matching degree of image is split, as the coupling digital template image of Contemporary Digital segmentation image by with Contemporary Digital.

On the other hand, the invention provides a kind of device of Car license recognition, comprising:

Setting unit, for determining character, arranges the template image of each character, wherein, and each character at least one template image corresponding, the corresponding character of each template image;

Acquiring unit, for obtaining license plate image to be identified;

Cutting unit, for according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains segmentation image;

Matching unit, for being mated with each template image by current segmentation image, determining and the matching template image that current segmentation image mates most, obtaining the matching template image of each segmentation image;

Character cell, for using character corresponding for the matching template image of current segmentation image as character corresponding to current segmentation image.

Further, described character comprises: Chinese character, alphabetic character, numerical character;

Described template image comprises: Chinese character template image, alphabetical template image, digital template image;

Described setting unit, for determining Chinese character, alphabetic character, numerical character, the Chinese character template image of each Chinese character is set, pre-set the Chinese character template image of each alphabetic character, pre-set the Chinese character template image of each numerical character, wherein, each Chinese character at least one Chinese character template image corresponding, the corresponding Chinese character of each Chinese character template image, each alphabetic character at least one alphabetical template image corresponding, the corresponding alphabetic character of each alphabetical template image, corresponding at least one the digital template image of each numerical character, the corresponding numerical character of each digital template image,

Described segmentation image comprises: Chinese character segmentation image, grapheme segmentation image, digital segmentation image;

Described cutting unit, for when performing described acquisition segmentation image, specifically performing: obtain Chinese character segmentation image, obtaining grapheme segmentation image, obtaining digital segmentation image;

Described matching template image comprises: mate Chinese character template image, mate alphabetical template image, coupling digital template image;

Described matching unit, comprising: Chinese character matching unit, alphabetical matching unit, digital matching unit;

Described Chinese character matching unit, for being mated with each Chinese character template image by current Chinese character segmentation image, determining and the coupling Chinese character template image that current Chinese character segmentation image mates most, obtaining the coupling Chinese character template image of each Chinese character segmentation image;

Described alphabetical matching unit, mates with each alphabetical template image for current letter being split image, determines the coupling letter template image split image with current letter and mate most, obtains the coupling letter template image of each grapheme segmentation image;

Described digital matching unit, mates with each digital template image for Contemporary Digital being split image, determines the coupling digital template image split image with Contemporary Digital and mate most, obtains the coupling digital template image of each digital segmentation image.

Further, described Chinese character matching unit, comprising:

Cross-correlation unit, for current Chinese character template image is moved on current Chinese character segmentation image, travel through all pixels of current Chinese character segmentation image, and calculate current Chinese character template image by formula one and moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image, wherein, described formula one is:

$\mathrm{\γ}(u,v)=\frac{{\mathrm{\Σ}}_{x,y}[f(x,y)-{\stackrel{\‾}{f}}_{u,v}][t(x-u,y-v)-\stackrel{\‾}{t}]}{\{{\mathrm{\Σ}}_{x,y}{[f(x,y)-{\stackrel{\‾}{f}}_{u,v}]}^2{\mathrm{\Σ}}_{x,y}{[t(x-u,y-v)-\stackrel{\‾}{t}]}^2{\}}^{0.5}}$

Wherein, u is the distance of current Chinese character template image movement in the x direction, v is the distance of current Chinese character template image movement in y-direction, γ (u, v) for current Chinese character template image is moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image after current Chinese character template movement (u, v) distance, the average gray of the current Chinese character segmentation image under current Chinese character template image covers, for the average gray of current Chinese character template image, t (x-u, y-v) is for current Chinese character template image is at point (x-u, y-v) gray-scale value, f (x, y) is for current Chinese character segmentation image is at the gray-scale value of point (x, y);

First Chinese character matching degree unit, for by normalized-cross-correlation function maximum in all normalized-cross-correlation function of current Chinese character template image and current Chinese character segmentation image, as the first matching degree of current Chinese character segmentation image and current Chinese character template image;

Second Chinese character matching degree unit, for determining the first matching degree of current Chinese character segmentation image and each Chinese character template image;

Chinese character matching image determining unit, for by the Chinese character template image maximum with the first matching degree of current Chinese character segmentation image, as the coupling Chinese character template image of current Chinese character segmentation image.

Further, described alphabetical matching unit, comprising:

Letter sampling unit, obtain the first Division Sampling point of the first predetermined number for sampling to current letter segmentation image, sampling to current letter template image obtains the first palette sample point of described first predetermined number;

Letter histogram unit, for determining the shape histogram of each first Division Sampling point under polar coordinate system in each default characteristic area, determines the shape histogram under polar coordinate system in each default characteristic area of each first palette sample point;

First alphabetical Matching power flow unit, for calculating the Matching power flow between current first Division Sampling point and current first palette sample point according to formula two, wherein, formula two is:

$C(p_i,q_j)=\frac{1}{2}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{{[h_i\left(k\right)-h_j\left(k\right)]}^2}{h_i\left(k\right)+h_j\left(k\right)}$

Wherein, p _ifor current first Division Sampling point, q _jfor current first palette sample point, C (p _i, q _j) be a p _iwith a q _jbetween Matching power flow, h _ik () is a kth characteristic area mid point p in K the characteristic area preset under polar coordinate system _ishape histogram, h _jk () is a kth characteristic area mid point q in K the characteristic area preset under polar coordinate system _jshape histogram;

Second alphabetical Matching power flow unit, for determining the Matching power flow between arbitrary first Division Sampling point and arbitrary first palette sample point;

Letter matching scheme unit, for being mated with all first palette sample points by all first Division Sampling points, obtains alphabetical matching scheme;

The total Matching power flow unit of letter, for determining the total Matching power flow of letter of each alphabetical matching scheme according to formula three, wherein, formula three is:

$C_{\mathrm{zm}}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i$

Wherein, C _zmfor the total Matching power flow of letter, C _ifor the i-th Matching power flow to the first Division Sampling point matched and the first palette sample point in current letter matching scheme, n is described first predetermined number;

First alphabetical matching degree unit, for determining the best letter Matching power flow that the total Matching power flow of letter is minimum, splits the second matching degree of image and current letter template image as current letter using alphabetical for described the best Matching power flow;

Second alphabetical matching degree unit, for determining the second matching degree of current letter segmentation image and each alphabetical template image;

Letter matching image determining unit, for split the minimum alphabetical template image of the second matching degree of image, as the alphabetical template image of coupling of current letter segmentation image by with current letter.

Further, described digital matching unit, comprising:

Digital acquisition unit, obtain the second Division Sampling point of the second predetermined number for sampling to Contemporary Digital segmentation image, sampling to Contemporary Digital template image obtains the second palette sample point of described second predetermined number;

Numeral histogram unit, for determining the shape histogram of each second Division Sampling point under polar coordinate system in each default characteristic area, determines the shape histogram under polar coordinate system in each default characteristic area of each second palette sample point;

First digital Matching power flow unit, for calculating the Matching power flow between current second Division Sampling point and current second palette sample point according to formula four, wherein, formula four is:

$C(r_x,s_y)=\frac{1}{2}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\frac{{[h_x\left(m\right)-h_y\left(m\right)]}^2}{h_x\left(m\right)+h_y\left(m\right)}$

Wherein, r _xfor current first Division Sampling point, s _yfor current first palette sample point, C (r _x, s _y) be a r _xwith a r _xbetween Matching power flow, h _xm () is m characteristic area mid point r in M the characteristic area preset under polar coordinate system _xshape histogram, h _ym () is m characteristic area mid point s in M the characteristic area preset under polar coordinate system _yshape histogram;

Second digital Matching power flow unit, determines the Matching power flow between arbitrary second Division Sampling point and arbitrary second palette sample point;

Numeral matching scheme unit, for being mated with all second palette sample points by all second Division Sampling points, obtains digital matching scheme;

The total Matching power flow unit of numeral, for determining the total Matching power flow of numeral of each digital matching scheme according to formula five, wherein, formula five is:

$C_{\mathrm{sz}}=\underset{s=1}{\overset{f}{\mathrm{\Σ}}}{C}_s$

Wherein, C _szfor the total Matching power flow of numeral, C _sfor s in Contemporary Digital matching scheme is to the Matching power flow of the second Division Sampling point matched and the second palette sample point, f is described second predetermined number;

First digital matching degree unit, for determining the optimal digital Matching power flow that the total Matching power flow of numeral is minimum, splits the 3rd matching degree of image and Contemporary Digital template image as Contemporary Digital using described optimal digital Matching power flow;

Second digital matching degree unit, for determining the 3rd matching degree of Contemporary Digital segmentation image and each digital template image;

Numeral matching image determining unit, for split the minimum digital template image of the 3rd matching degree of image by with Contemporary Digital, splits the coupling digital template image of image as Contemporary Digital.

The method of a kind of Car license recognition provided by the invention and device, pre-set template image, go out to split image according to the Character segmentation in license plate image from license plate image to be identified, each segmentation image is mated with template image, determine the template image that each segmentation image is corresponding, the character that this template image is corresponding is exactly the character on the segmentation image of its correspondence, that segmentation image is mated with template image in the method, even if segmentation image is unintelligible owing to there is the reasons such as stain, in the matching process, this segmentation image is also generally the highest with the matching degree of the template image mated most in theory, and then the corresponding character obtained also is correct, improve the correct recognition rata of the character identified on license plate image.

Accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.

Fig. 1 is the process flow diagram of the method for a kind of Car license recognition that one embodiment of the invention provides;

Fig. 2 is the process flow diagram of the method for the another kind of Car license recognition that one embodiment of the invention provides;

Fig. 3 is the process flow diagram of the method for another Car license recognition that one embodiment of the invention provides;

Fig. 4 is the schematic diagram of the device of a kind of Car license recognition that one embodiment of the invention provides.

Embodiment

For making the object of the embodiment of the present invention, technical scheme and advantage clearly; below in conjunction with the accompanying drawing in the embodiment of the present invention; technical scheme in the embodiment of the present invention is clearly and completely described; obviously; described embodiment is the present invention's part embodiment, instead of whole embodiments, based on the embodiment in the present invention; the every other embodiment that those of ordinary skill in the art obtain under the prerequisite not making creative work, all belongs to the scope of protection of the invention.

As shown in Figure 1, embodiments provide a kind of method of Car license recognition, the method can comprise the following steps:

S0: pre-determine character, pre-set the template image of each character, wherein, each character at least one template image corresponding, the corresponding character of each template image;

S1: obtain license plate image to be identified;

S2: according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains segmentation image;

S3: mated with each template image by current segmentation image, determines and the matching template image that current segmentation image mates most, obtains the matching template image of each segmentation image;

S4: using character corresponding for the matching template image of current segmentation image as character corresponding to current segmentation image.

The method of a kind of Car license recognition that the embodiment of the present invention provides, pre-set template image, go out to split image according to the Character segmentation in license plate image from license plate image to be identified, each segmentation image is mated with template image, determine the template image that each segmentation image is corresponding, the character that this template image is corresponding is exactly the character on the segmentation image of its correspondence, that segmentation image is mated with template image in the method, even if segmentation image is unintelligible owing to there is the reasons such as stain, in the matching process, this segmentation image is also generally the highest with the matching degree of the template image mated most in theory, and then the corresponding character obtained also is correct, improve the correct recognition rata of the character identified on license plate image.

General license plate image having three kinds of characters, is Chinese character, alphabetic character, numerical character respectively, in order to reduce the matching range of segmentation image, improves recognition speed, can by three kinds of characters separately coupling.

In a kind of possible implementation, described predefined character, pre-sets the template image of each character, wherein, each character at least one template image corresponding, the corresponding character of each template image, comprising: described character comprises: Chinese character, alphabetic character, numerical character;

Described template image comprises: Chinese character template image, alphabetical template image, digital template image;

Pre-determine Chinese character, alphabetic character, numerical character, pre-set the Chinese character template image of each Chinese character, pre-set the Chinese character template image of each alphabetic character, pre-set the Chinese character template image of each numerical character, wherein, each Chinese character at least one Chinese character template image corresponding, the corresponding Chinese character of each Chinese character template image, each alphabetic character at least one alphabetical template image corresponding, the corresponding alphabetic character of each alphabetical template image, corresponding at least one the digital template image of each numerical character, the corresponding numerical character of each digital template image,

Described segmentation image comprises: Chinese character segmentation image, grapheme segmentation image, digital segmentation image;

Obtain segmentation image described in described S2, comprising: obtain Chinese character segmentation image, obtain grapheme segmentation image, obtain digital segmentation image;

Described matching template image comprises: mate Chinese character template image, mate alphabetical template image, coupling digital template image;

Described S3 comprises:

A1: mated with each Chinese character template image by current Chinese character segmentation image, determines and the coupling Chinese character template image that current Chinese character segmentation image mates most, obtains the coupling Chinese character template image of each Chinese character segmentation image.

Particularly, multiple Chinese character may appear in car plate, the corresponding Chinese character segmentation image of each Chinese character, corresponding arbitrary Chinese character segmentation image all will obtain the coupling Chinese character template image mated most in all Chinese character template images.

A2: current letter is split image and mate with each alphabetical template image, determines the coupling letter template image split image with current letter and mate most, obtains the coupling letter template image of each grapheme segmentation image.

A3: Contemporary Digital is split image and mate with each digital template image, determines the coupling digital template image split image with Contemporary Digital and mate most, obtains the coupling digital template image of each digital segmentation image.

In a kind of possible implementation, described A1, comprising:

Steps A 11: current Chinese character template image is moved on current Chinese character segmentation image, travel through all pixels of current Chinese character segmentation image, and calculate current Chinese character template image by formula one and moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image, wherein, described formula one is:

$\mathrm{\γ}(u,v)=\frac{{\mathrm{\Σ}}_{x,y}[f(x,y)-{\stackrel{\‾}{f}}_{u,v}][t(x-u,y-v)-\stackrel{\‾}{t}]}{\{{\mathrm{\Σ}}_{x,y}{[f(x,y)-{\stackrel{\‾}{f}}_{u,v}]}^2{\mathrm{\Σ}}_{x,y}{[t(x-u,y-v)-\stackrel{\‾}{t}]}^2{\}}^{0.5}}$

Wherein, u is the distance of current Chinese character template image movement in the x direction, v is the distance of current Chinese character template image movement in y-direction, γ (u, v) for current Chinese character template image is moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image after current Chinese character template movement (u, v) distance, the average gray of the current Chinese character segmentation image under current Chinese character template image covers, for the average gray of current Chinese character template image, t (x-u, y-v) is for current Chinese character template image is at point (x-u, y-v) gray-scale value, f (x, y) is for current Chinese character segmentation image is at the gray-scale value of point (x, y).

Particularly, current Chinese character template image often mobile one (u, v) distance all can obtain a γ (u, v), and the position at γ (u, v) larger explanation Chinese character template image place is more mated with current Chinese character segmentation image.

Steps A 12: by normalized-cross-correlation function maximum in all normalized-cross-correlation function of current Chinese character template image and current Chinese character segmentation image, as the first matching degree of current Chinese character segmentation image and current Chinese character template image.

Steps A 13: the first matching degree determining current Chinese character segmentation image and each Chinese character template image.

Steps A 14: by the Chinese character template image maximum with the first matching degree of current Chinese character segmentation image, as the coupling Chinese character template image of current Chinese character segmentation image.

In a kind of possible implementation, described A2, comprising:

Steps A 21: sample to current letter segmentation image and obtain the first Division Sampling point of the first predetermined number, sampling to current letter template image obtains the first palette sample point of described first predetermined number.

Particularly, sample to current letter segmentation image and current letter template image respectively, obtain the sampled point of equal number, the method for sampling can be identical method.The quantity of sampled point is more, and the matching result obtained is more accurate.

Steps A 22: determine the shape histogram of each first Division Sampling point under polar coordinate system in each default characteristic area, determine the shape histogram under polar coordinate system in each default characteristic area of each first palette sample point.

Particularly, can calculate the shape histogram of current first Division Sampling point under polar coordinate system in a kth characteristic area by following histogram formula, this histogram formula is:

h(k)＝#{a≠d:(a-d)∈bin(k)}

Wherein, h (k) is the shape histogram of current first Division Sampling point under polar coordinate system in a kth characteristic area, and d is current first Division Sampling point, and a is other the first Division Sampling points.

Steps A 23: calculate the Matching power flow between current first Division Sampling point and current first palette sample point according to formula two, wherein, formula two is:

$C(p_i,q_j)=\frac{1}{2}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{{[h_i\left(k\right)-h_j\left(k\right)]}^2}{h_i\left(k\right)+h_j\left(k\right)}$

Wherein, p _ifor current first Division Sampling point, q _jfor current first palette sample point, C (p _i, q _j) be a p _iwith a q _jbetween Matching power flow, h _ik () is a kth characteristic area mid point p in K the characteristic area preset under polar coordinate system _ishape histogram, h _jk () is a kth characteristic area mid point q in K the characteristic area preset under polar coordinate system _jshape histogram.

Steps A 24: determine the Matching power flow between arbitrary first Division Sampling point and arbitrary first palette sample point.

Steps A 25: all first Division Sampling points are mated with all first palette sample points, obtains alphabetical matching scheme.

Steps A 26: the total Matching power flow of letter determining each alphabetical matching scheme according to formula three, wherein, formula three is:

$C_{\mathrm{zm}}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i$

Wherein, C _zmfor the total Matching power flow of letter, C _ifor the i-th Matching power flow to the first Division Sampling point matched and the first palette sample point in current letter matching scheme, n is described first predetermined number.

Steps A 27: determine the best letter Matching power flow that the total Matching power flow of letter is minimum, alphabetical for described the best Matching power flow is split the second matching degree of image and current letter template image as current letter.

Steps A 28: the second matching degree determining current letter segmentation image and each alphabetical template image.

Steps A 29: split the minimum alphabetical template image of the second matching degree of image by with current letter, as the coupling letter template image of current letter segmentation image.

In a kind of possible implementation, described A3, comprising:

Steps A 31: sample to Contemporary Digital segmentation image and obtain the second Division Sampling point of the second predetermined number, sampling to Contemporary Digital template image obtains the second palette sample point of described second predetermined number.

Steps A 32: determine the shape histogram of each second Division Sampling point under polar coordinate system in each default characteristic area, determine the shape histogram under polar coordinate system in each default characteristic area of each second palette sample point.

Steps A 33: calculate the Matching power flow between current second Division Sampling point and current second palette sample point according to formula four, wherein, formula four is:

$C(r_x,s_y)=\frac{1}{2}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\frac{{[h_x\left(m\right)-h_y\left(m\right)]}^2}{h_x\left(m\right)+h_y\left(m\right)}$

Wherein, r _xfor current first Division Sampling point, s _yfor current first palette sample point, C (r _x, s _y) be a r _xwith a r _xbetween Matching power flow, h _xm () is m characteristic area mid point r in M the characteristic area preset under polar coordinate system _xshape histogram, h _ym () is m characteristic area mid point s in M the characteristic area preset under polar coordinate system _yshape histogram.

Steps A 34: determine the Matching power flow between arbitrary second Division Sampling point and arbitrary second palette sample point.

Steps A 35: all second Division Sampling points are mated with all second palette sample points, obtains digital matching scheme.

Steps A 36: the total Matching power flow of numeral determining each digital matching scheme according to formula five, wherein, formula five is:

$C_{\mathrm{sz}}=\underset{s=1}{\overset{f}{\mathrm{\Σ}}}{C}_s$

Wherein, C _szfor the total Matching power flow of numeral, C _sfor s in Contemporary Digital matching scheme is to the Matching power flow of the second Division Sampling point matched and the second palette sample point, f is described second predetermined number.

Steps A 37: determine the optimal digital Matching power flow that the total Matching power flow of numeral is minimum, described optimal digital Matching power flow is split the 3rd matching degree of image and Contemporary Digital template image as Contemporary Digital.

Steps A 38: the 3rd matching degree determining Contemporary Digital segmentation image and each digital template image.

Steps A 39: split the minimum digital template image of the 3rd matching degree of image by with Contemporary Digital, as the coupling digital template image of Contemporary Digital segmentation image.

For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with drawings and the specific embodiments, the present invention is described in further detail.

As shown in Figure 2, for the Chinese character part in license plate image, embodiments provide a kind of method of Car license recognition, the method can comprise the following steps:

Step 201: pre-determine Chinese character, pre-set the Chinese character template image of each Chinese character, wherein, each Chinese character at least one Chinese character template image corresponding, the corresponding Chinese character of each Chinese character template image.

For example, the Chinese character on car plate generally has the abbreviation etc. of place province ,city and area, such as: Shandong, capital etc., can arrange the Chinese character template image of the Chinese character likely appeared on car plate, has the profile of corresponding Chinese character in this Chinese character template image.

In order to improve correct recognition rata, can by corresponding for Chinese character multiple Chinese character template image, the Chinese character template image mated most is like this that the probability of correct Chinese character analog image is higher.

Step 202: obtain license plate image to be identified.

Step 203: according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains Chinese character segmentation image.

For example, the corresponding segmentation image of each character in license plate image.

Step 204: current Chinese character template image is moved on current Chinese character segmentation image, travel through all pixels of current Chinese character segmentation image, and calculate current Chinese character template image by formula one and moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image, wherein, described formula one is:

$\mathrm{\γ}(u,v)=\frac{{\mathrm{\Σ}}_{x,y}[f(x,y)-{\stackrel{\‾}{f}}_{u,v}][t(x-u,y-v)-\stackrel{\‾}{t}]}{\{{\mathrm{\Σ}}_{x,y}{[f(x,y)-{\stackrel{\‾}{f}}_{u,v}]}^2{\mathrm{\Σ}}_{x,y}{[t(x-u,y-v)-\stackrel{\‾}{t}]}^2{\}}^{0.5}}$

Wherein, u is the distance of current Chinese character template image movement in the x direction, v is the distance of current Chinese character template image movement in y-direction, γ (u, v) for current Chinese character template image is moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image after current Chinese character template movement (u, v) distance, the average gray of the current Chinese character segmentation image under current Chinese character template image covers, for the average gray of current Chinese character template image, t (x-u, y-v) is for current Chinese character template image is at point (x-u, y-v) gray-scale value, f (x, y) is for current Chinese character segmentation image is at the gray-scale value of point (x, y).

For example, the size of current Chinese character template image is 2 × 2, and the size of current Chinese character segmentation image is 4 × 4.Current Chinese character template image is moved on current Chinese character segmentation image, the position of current Chinese character template image on current Chinese character segmentation image can have 4 kinds, mobile (0 respectively, 0), mobile (1,0), mobile (0,1), (1,1) is moved.For this in 4 move mode have 4 corresponding γ (u, v), be γ (0,0), γ (0,1), γ (1,0), γ (1,1) respectively.

Step 205: by normalized-cross-correlation function maximum in all normalized-cross-correlation function of current Chinese character template image and current Chinese character segmentation image, as the first matching degree of current Chinese character segmentation image and current Chinese character template image.

For example, γ (1,0) is maximum, then the size of the first matching degree of current Chinese character segmentation image and current Chinese character template image is equal with γ (1,0).

Step 206: the first matching degree determining current Chinese character segmentation image and each Chinese character template image.

Step 207: by the Chinese character template image maximum with the first matching degree of current Chinese character segmentation image, as the coupling Chinese character template image of current Chinese character segmentation image.

Step 208: using character corresponding for the coupling Chinese character template image of current Chinese character segmentation image as character corresponding to current Chinese character segmentation image.

For example, the character that the coupling Chinese character template image of current Chinese character segmentation image is corresponding is " capital ", then the character that current Chinese character segmentation image is corresponding is " capital ", and namely the recognition result of current Chinese character segmentation image is " capital ".

As shown in Figure 3, for the letter part in license plate image, embodiments provide a kind of method of Car license recognition, the method can comprise the following steps:

Step 301: pre-determine alphabetic character, pre-set the alphabetical template image of each alphabetic character, wherein, each alphabetic character at least one alphabetical template image corresponding, the corresponding alphabetic character of each alphabetical template image.

For example, the alphabetical template image of the alphabetic character likely appeared on car plate can be set, in this alphabetical template image, have the profile of corresponding alphabetic character.Whole 26 alphabetical alphabetical template images generally can be set.

In order to improve correct recognition rata, can by corresponding for alphabetic character multiple alphabetical template image, the alphabetical template image mated most is like this that the probability of correct alphabetical analog image is higher.

Step 302: obtain license plate image to be identified.

Step 303: according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains grapheme segmentation image.

Step 304: sample to current letter segmentation image and obtain the first Division Sampling point of the first predetermined number, sampling to current letter template image obtains the first palette sample point of described first predetermined number.

For example, the first predetermined number is 100.

Step 305: determine the shape histogram of each first Division Sampling point under polar coordinate system in each default characteristic area, determine the shape histogram under polar coordinate system in each default characteristic area of each first palette sample point.

Step 306: calculate the Matching power flow between current first Division Sampling point and current first palette sample point according to formula two, wherein, formula two is:

$C(p_i,q_j)=\frac{1}{2}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{{[h_i\left(k\right)-h_j\left(k\right)]}^2}{h_i\left(k\right)+h_j\left(k\right)}$

Wherein, p _ifor current first Division Sampling point, q _jfor current first palette sample point, C (p _i, q _j) be a p _iwith a q _jbetween Matching power flow, h _ik () is a kth characteristic area mid point p in K the characteristic area preset under polar coordinate system _ishape histogram, h _jk () is a kth characteristic area mid point q in K the characteristic area preset under polar coordinate system _jshape histogram.

Step 307: determine the Matching power flow between arbitrary first Division Sampling point and arbitrary first palette sample point.

For example, two the first Division Sampling points are had to be d1, d2 respectively, two the first palette sample points are had to be e1, e2 respectively, to calculate in steps A 24, the Matching power flow of d1 and e1, the Matching power flow of d1 and e2, the Matching power flow of d2 and e1, the Matching power flow of d2 and e2, for the number of the first different Division Sampling points, the like.

Step 308: all first Division Sampling points are mated with all first palette sample points, obtains alphabetical matching scheme.

For example, there are two the first Division Sampling points to be d1, d2 respectively, have two the first palette sample points to be e1, e2 respectively.D1 and e1 is corresponding, and d2 and e2 is corresponding, obtains first letter matching scheme, and d1 and e2 is corresponding, and d2 and e1 is corresponding, obtains second letter matching scheme.This step will obtain all possible alphabetical matching scheme.

Step 309: the total Matching power flow of letter determining each alphabetical matching scheme according to formula three, wherein, formula three is:

$C_{\mathrm{zm}}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i$

Wherein, C _zmfor the total Matching power flow of letter, C _ifor the i-th Matching power flow to the first Division Sampling point matched and the first palette sample point in current letter matching scheme, n is described first predetermined number.

For example, in above-mentioned first letter matching scheme, the Matching power flow of d1 and e1 is the Matching power flow of C1, d2 and e2 is C2, then the total Matching power flow C of the letter of first letter matching scheme _zmfor C1+C2.

Step 310: determine the best letter Matching power flow that the total Matching power flow of letter is minimum, alphabetical for described the best Matching power flow is split the second matching degree of image and current letter template image as current letter.

Step 311: the second matching degree determining current letter segmentation image and each alphabetical template image.

Step 312: split the minimum alphabetical template image of the second matching degree of image by with current letter, as the coupling letter template image of current letter segmentation image.

Step 313: character corresponding to coupling letter template image current letter being split image is as character corresponding to current letter segmentation image.

For example, the character that the coupling letter template image of current letter segmentation image is corresponding is " A ", then the character that current letter segmentation image is corresponding is " A ", and namely the recognition result of current letter segmentation image is " A ".

It should be noted that: the recognition methods of digital segmentation image and the recognition methods of grapheme segmentation image similar, specific embodiment can reference letter segmentation image embodiment.

It should be noted that: obtaining license plate image to be identified can realize in the following manner:

Under the actual environment of complexity, the key of whole identifying is how from uneven illumination, accurately locates license plate area in the automobile image of background complexity.

Image semantic classification is the basis positioned car plate; concrete steps comprise carries out gray processing process by the car plate coloured image of input; next gray scale stretching is carried out to image; to strengthen picture contrast; then carry out medium filtering, the method also can Protect edge information profile information preferably while restraint speckle.

Have employed the method that the multiple processing modes such as rim detection, morphology processing, characteristic area analysis and sciagraphy combine car plate is separated.Particularly, after carrying out pre-service to initial license plate image, adopt prewitt operator to make rim detection to it, it can detect the edges of regions that grey scale change is violent; The Morphological scale-space of following utilization several functions, the image after edge detects carries out opening and closing operation, and corrosion expansion process, obtains several candidate regions of car plate connected domain; Finally utilize regional characteristics analysis, by the estimation of the area to candidate region, the ratio of width to height and " dutycycle " etc., filter out rectangle license plate area; Last then realized the fine positioning of car plate by sciagraphy, cut out, obtained license plate image to be identified.

In order to make recognition result more accurate, slant correction can be carried out to the above-mentioned license plate image split.Particularly, from image, locate the car plate obtained may exist tilt phenomenon, affects follow-up Character segmentation, is thus necessary to consider this problem of slant correction.The Radon converter technique based on spatial alternation can be have employed.Before carrying out slant correction to license plate image, canny operator is first used to make rim detection.Radon conversion deals with based on gray level image, its mathematical principle on every straight line, does line integral to pixel, the distance of air line distance initial point is, the anglec of rotation of straight line is, so the line integral value of luv space is mapped as the transform value in space, is finally converted into the detection to point to the detection of straight line.

After obtaining license plate image to be identified, according to the character in described license plate image to be identified, split described license plate image to be identified, obtain segmentation image, can realize in the following manner:

Whole car plate comprises altogether 7 characters, for the ease of identifying, will carry out the segmentation of character.Seven characters are divided into two groups to process by the thinking that this implementation combines based on priori and sciagraphy.

Step D1: utilize level, the upper right corner coordinate of top left co-ordinate that vertical projection determines first character and the 7th character, can obtain character height h, according to priori, the width of character is about h/2 simultaneously.

Step D2: to 2, left side character according to priori, calculates its segmentation starting point coordinate.

Step D3: to 5, the right character, carry out from right to left " jumping characteristic scanning ", obtains 5 characters segmentation starting point coordinate separately.

Step D4: according to 7 characters segmentation starting point coordinate and character duration, highly, be partitioned into single character respectively, obtain split image.

In addition, can nearest neighbor classifier be utilized to the identification of character in license plate image to be identified, after feature extraction is carried out to object to be identified, make similarity or Diversity measure, just according to measurement results, object to be identified can be classified.Adopt nearest neighbor method to treat identification character and carry out Classification and Identification.

See Fig. 4, the device of a kind of Car license recognition that the present embodiment provides, comprising:

Setting unit 401, for determining character, arranges the template image of each character, wherein, and each character at least one template image corresponding, the corresponding character of each template image.

Acquiring unit 402, for obtaining license plate image to be identified.

Cutting unit 403, for according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains segmentation image.

Matching unit 404, for being mated with each template image by current segmentation image, determining and the matching template image that current segmentation image mates most, obtaining the matching template image of each segmentation image.

Character cell 405, for using character corresponding for the matching template image of current segmentation image as character corresponding to current segmentation image.

In a kind of possible implementation, described character comprises: Chinese character, alphabetic character, numerical character;

Described template image comprises: Chinese character template image, alphabetical template image, digital template image;

Described setting unit 401, for determining Chinese character, alphabetic character, numerical character, the Chinese character template image of each Chinese character is set, pre-set the Chinese character template image of each alphabetic character, pre-set the Chinese character template image of each numerical character, wherein, each Chinese character at least one Chinese character template image corresponding, the corresponding Chinese character of each Chinese character template image, each alphabetic character at least one alphabetical template image corresponding, the corresponding alphabetic character of each alphabetical template image, corresponding at least one the digital template image of each numerical character, the corresponding numerical character of each digital template image,

Described segmentation image comprises: Chinese character segmentation image, grapheme segmentation image, digital segmentation image;

Described cutting unit 403, for when performing described acquisition segmentation image, specifically performing: obtain Chinese character segmentation image, obtaining grapheme segmentation image, obtaining digital segmentation image;

Described matching template image comprises: mate Chinese character template image, mate alphabetical template image, coupling digital template image;

Described matching unit 404, comprising: Chinese character matching unit, alphabetical matching unit, digital matching unit;

Described Chinese character matching unit, for being mated with each Chinese character template image by current Chinese character segmentation image, determining and the coupling Chinese character template image that current Chinese character segmentation image mates most, obtaining the coupling Chinese character template image of each Chinese character segmentation image;

Described alphabetical matching unit, mates with each alphabetical template image for current letter being split image, determines the coupling letter template image split image with current letter and mate most, obtains the coupling letter template image of each grapheme segmentation image;

Described digital matching unit, mates with each digital template image for Contemporary Digital being split image, determines the coupling digital template image split image with Contemporary Digital and mate most, obtains the coupling digital template image of each digital segmentation image.

In a kind of possible implementation, described Chinese character matching unit, comprising:

Cross-correlation unit, for current Chinese character template image is moved on current Chinese character segmentation image, travel through all pixels of current Chinese character segmentation image, and calculate current Chinese character template image by formula one and moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image, wherein, described formula one is:

$\mathrm{\γ}(u,v)=\frac{{\mathrm{\Σ}}_{x,y}[f(x,y)-{\stackrel{\‾}{f}}_{u,v}][t(x-u,y-v)-\stackrel{\‾}{t}]}{\{{\mathrm{\Σ}}_{x,y}{[f(x,y)-{\stackrel{\‾}{f}}_{u,v}]}^2{\mathrm{\Σ}}_{x,y}{[t(x-u,y-v)-\stackrel{\‾}{t}]}^2{\}}^{0.5}}$

Wherein, u is the distance of current Chinese character template image movement in the x direction, v is the distance of current Chinese character template image movement in y-direction, γ (u, v) for current Chinese character template image is moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image after current Chinese character template movement (u, v) distance, the average gray of the current Chinese character segmentation image under current Chinese character template image covers, for the average gray of current Chinese character template image, t (x-u, y-v) is for current Chinese character template image is at point (x-u, y-v) gray-scale value, f (x, y) is for current Chinese character segmentation image is at the gray-scale value of point (x, y);

First Chinese character matching degree unit, for by normalized-cross-correlation function maximum in all normalized-cross-correlation function of current Chinese character template image and current Chinese character segmentation image, as the first matching degree of current Chinese character segmentation image and current Chinese character template image;

Second Chinese character matching degree unit, for determining the first matching degree of current Chinese character segmentation image and each Chinese character template image;

Chinese character matching image determining unit, for by the Chinese character template image maximum with the first matching degree of current Chinese character segmentation image, as the coupling Chinese character template image of current Chinese character segmentation image.

In a kind of possible implementation, described alphabetical matching unit, comprising:

Letter sampling unit, obtain the first Division Sampling point of the first predetermined number for sampling to current letter segmentation image, sampling to current letter template image obtains the first palette sample point of described first predetermined number;

Letter histogram unit, for determining the shape histogram of each first Division Sampling point under polar coordinate system in each default characteristic area, determines the shape histogram under polar coordinate system in each default characteristic area of each first palette sample point;

First alphabetical Matching power flow unit, for calculating the Matching power flow between current first Division Sampling point and current first palette sample point according to formula two, wherein, formula two is:

$C(p_i,q_j)=\frac{1}{2}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{{[h_i\left(k\right)-h_j\left(k\right)]}^2}{h_i\left(k\right)+h_j\left(k\right)}$

Wherein, p _ifor current first Division Sampling point, q _jfor current first palette sample point, C (p _i, q _j) be a p _iwith a q _jbetween Matching power flow, h _ik () is a kth characteristic area mid point p in K the characteristic area preset under polar coordinate system _ishape histogram, h _jk () is a kth characteristic area mid point q in K the characteristic area preset under polar coordinate system _jshape histogram;

Second alphabetical Matching power flow unit, for determining the Matching power flow between arbitrary first Division Sampling point and arbitrary first palette sample point;

Letter matching scheme unit, for being mated with all first palette sample points by all first Division Sampling points, obtains alphabetical matching scheme;

The total Matching power flow unit of letter, for determining the total Matching power flow of letter of each alphabetical matching scheme according to formula three, wherein, formula three is:

$C_{\mathrm{zm}}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i$

Wherein, C _zmfor the total Matching power flow of letter, C _ifor the i-th Matching power flow to the first Division Sampling point matched and the first palette sample point in current letter matching scheme, n is described first predetermined number;

First alphabetical matching degree unit, for determining the best letter Matching power flow that the total Matching power flow of letter is minimum, splits the second matching degree of image and current letter template image as current letter using alphabetical for described the best Matching power flow;

Second alphabetical matching degree unit, for determining the second matching degree of current letter segmentation image and each alphabetical template image;

Letter matching image determining unit, for split the minimum alphabetical template image of the second matching degree of image, as the alphabetical template image of coupling of current letter segmentation image by with current letter.

In a kind of possible implementation, described digital matching unit, comprising:

Digital acquisition unit, obtain the second Division Sampling point of the second predetermined number for sampling to Contemporary Digital segmentation image, sampling to Contemporary Digital template image obtains the second palette sample point of described second predetermined number;

Numeral histogram unit, for determining the shape histogram of each second Division Sampling point under polar coordinate system in each default characteristic area, determines the shape histogram under polar coordinate system in each default characteristic area of each second palette sample point;

First digital Matching power flow unit, for calculating the Matching power flow between current second Division Sampling point and current second palette sample point according to formula four, wherein, formula four is:

$C(r_x,s_y)=\frac{1}{2}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\frac{{[h_x\left(m\right)-h_y\left(m\right)]}^2}{h_x\left(m\right)+h_y\left(m\right)}$

Wherein, r _xfor current first Division Sampling point, s _yfor current first palette sample point, C (r _x, s _y) be a r _xwith a r _xbetween Matching power flow, h _xm () is m characteristic area mid point r in M the characteristic area preset under polar coordinate system _xshape histogram, h _ym () is m characteristic area mid point s in M the characteristic area preset under polar coordinate system _yshape histogram;

Second digital Matching power flow unit, determines the Matching power flow between arbitrary second Division Sampling point and arbitrary second palette sample point;

Numeral matching scheme unit, for being mated with all second palette sample points by all second Division Sampling points, obtains digital matching scheme;

The total Matching power flow unit of numeral, for determining the total Matching power flow of numeral of each digital matching scheme according to formula five, wherein, formula five is:

$C_{\mathrm{sz}}=\underset{s=1}{\overset{f}{\mathrm{\Σ}}}{C}_s$

Wherein, C _szfor the total Matching power flow of numeral, C _sfor s in Contemporary Digital matching scheme is to the Matching power flow of the second Division Sampling point matched and the second palette sample point, f is described second predetermined number;

First digital matching degree unit, for determining the optimal digital Matching power flow that the total Matching power flow of numeral is minimum, splits the 3rd matching degree of image and Contemporary Digital template image as Contemporary Digital using described optimal digital Matching power flow;

Second digital matching degree unit, for determining the 3rd matching degree of Contemporary Digital segmentation image and each digital template image;

Numeral matching image determining unit, for split the minimum digital template image of the 3rd matching degree of image by with Contemporary Digital, splits the coupling digital template image of image as Contemporary Digital.

The content such as information interaction, implementation between each unit in said apparatus, due to the inventive method embodiment based on same design, particular content can see in the inventive method embodiment describe, repeat no more herein.

The method of a kind of Car license recognition that the embodiment of the present invention provides and device, have following beneficial effect:

The method of a kind of Car license recognition that the embodiment of the present invention provides and device, pre-set template image, go out to split image according to the Character segmentation in license plate image from license plate image to be identified, each segmentation image is mated with template image, determine the template image that each segmentation image is corresponding, the character that this template image is corresponding is exactly the character on the segmentation image of its correspondence, that segmentation image is mated with template image in the method, even if segmentation image is unintelligible owing to there is the reasons such as stain, in the matching process, this segmentation image is also generally the highest with the matching degree of the template image mated most in theory, and then the corresponding character obtained also is correct, improve the correct recognition rata of the character identified on license plate image.

It should be noted that, in this article, the relational terms of such as first and second and so on is only used for an entity or operation to separate with another entity or operational zone, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or equipment and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or equipment.When not more restrictions, the key element " being comprised " limited by statement, and be not precluded within process, method, article or the equipment comprising described key element and also there is other same factor.

One of ordinary skill in the art will appreciate that: all or part of step realizing said method embodiment can have been come by the hardware that programmed instruction is relevant, aforesaid program can be stored in the storage medium of embodied on computer readable, this program, when performing, performs the step comprising said method embodiment; And aforesaid storage medium comprises: ROM, RAM, magnetic disc or CD etc. various can be program code stored medium in.

Finally it should be noted that: the foregoing is only preferred embodiment of the present invention, only for illustration of technical scheme of the present invention, be not intended to limit protection scope of the present invention.All any amendments done within the spirit and principles in the present invention, equivalent replacement, improvement etc., be all included in protection scope of the present invention.

Claims (10)

1. a method for Car license recognition, is characterized in that, comprising: pre-determine character, pre-set the template image of each character, wherein, each character at least one template image corresponding, the corresponding character of each template image, also comprises:

S1: obtain license plate image to be identified;

S2: according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains segmentation image;

S3: mated with each template image by current segmentation image, determines and the matching template image that current segmentation image mates most, obtains the matching template image of each segmentation image;

S4: using character corresponding for the matching template image of current segmentation image as character corresponding to current segmentation image.

2. method according to claim 1, is characterized in that, described predefined character, pre-sets the template image of each character, wherein, each character at least one template image corresponding, the corresponding character of each template image, comprising:

Described character comprises: Chinese character, alphabetic character, numerical character;

Described template image comprises: Chinese character template image, alphabetical template image, digital template image;

Pre-determine Chinese character, alphabetic character, numerical character, pre-set the Chinese character template image of each Chinese character, pre-set the Chinese character template image of each alphabetic character, pre-set the Chinese character template image of each numerical character, wherein, each Chinese character at least one Chinese character template image corresponding, the corresponding Chinese character of each Chinese character template image, each alphabetic character at least one alphabetical template image corresponding, the corresponding alphabetic character of each alphabetical template image, corresponding at least one the digital template image of each numerical character, the corresponding numerical character of each digital template image,

Described segmentation image comprises: Chinese character segmentation image, grapheme segmentation image, digital segmentation image;

Obtain segmentation image described in described S2, comprising: obtain Chinese character segmentation image, obtain grapheme segmentation image, obtain digital segmentation image;

Described matching template image comprises: mate Chinese character template image, mate alphabetical template image, coupling digital template image;

Described S3 comprises:

A1: mated with each Chinese character template image by current Chinese character segmentation image, determines and the coupling Chinese character template image that current Chinese character segmentation image mates most, obtains the coupling Chinese character template image of each Chinese character segmentation image;

A2: current letter is split image and mate with each alphabetical template image, determines the coupling letter template image split image with current letter and mate most, obtains the coupling letter template image of each grapheme segmentation image;

A3: Contemporary Digital is split image and mate with each digital template image, determines the coupling digital template image split image with Contemporary Digital and mate most, obtains the coupling digital template image of each digital segmentation image.

3. method according to claim 2, is characterized in that, described A1, comprising:

Current Chinese character template image is moved on current Chinese character segmentation image, travel through all pixels of current Chinese character segmentation image, and calculate current Chinese character template image by formula one and moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image, wherein, described formula one is:

$\mathrm{\γ}(u,v)=\frac{{\mathrm{\Σ}}_{x,y}[f(x,y)-{\stackrel{\‾}{f}}_{u,v}][t(x-u,y-v)-\stackrel{\‾}{t}]}{\left\{{\mathrm{\Σ}}_{x,y}\right[f(x,y)-{\stackrel{\‾}{f}}_{u,v}{]}^2{\mathrm{\Σ}}_{x,y}{[t(x-u,y-v)-\stackrel{\‾}{t}]}^2{\}}^{0.5}}$

Wherein, u is the distance of current Chinese character template image movement in the x direction, v is the distance of current Chinese character template image movement in y-direction, γ (u, v) for current Chinese character template image is moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image after current Chinese character template movement (u, v) distance, the average gray of the current Chinese character segmentation image under current Chinese character template image covers, for the average gray of current Chinese character template image, t (x-u, y-v) is for current Chinese character template image is at point (x-u, y-v) gray-scale value, f (x, y) is for current Chinese character segmentation image is at the gray-scale value of point (x, y);

By normalized-cross-correlation function maximum in all normalized-cross-correlation function of current Chinese character template image and current Chinese character segmentation image, as the first matching degree of current Chinese character segmentation image and current Chinese character template image;

Determine the first matching degree of current Chinese character segmentation image and each Chinese character template image;

By the Chinese character template image maximum with the first matching degree of current Chinese character segmentation image, as the coupling Chinese character template image of current Chinese character segmentation image.

4. method according to claim 2, is characterized in that, described A2, comprising:

Sample to current letter segmentation image and obtain the first Division Sampling point of the first predetermined number, sampling to current letter template image obtains the first palette sample point of described first predetermined number;

Determine the shape histogram of each first Division Sampling point under polar coordinate system in each default characteristic area, determine the shape histogram under polar coordinate system in each default characteristic area of each first palette sample point;

Calculate the Matching power flow between current first Division Sampling point and current first palette sample point according to formula two, wherein, formula two is:

$C(p_i,q_j)=\frac{1}{2}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{{[h_i\left(k\right)-h_j\left(k\right)]}^2}{h_i\left(k\right)+h_j\left(k\right)}$

Wherein, p _ifor current first Division Sampling point, q _jfor current first palette sample point, C (p _i, q _j) be a p _iwith a q _jbetween Matching power flow, h _ik () is a kth characteristic area mid point p in K the characteristic area preset under polar coordinate system _ishape histogram, h _jk () is a kth characteristic area mid point q in K the characteristic area preset under polar coordinate system _jshape histogram;

Determine the Matching power flow between arbitrary first Division Sampling point and arbitrary first palette sample point;

All first Division Sampling points are mated with all first palette sample points, obtains alphabetical matching scheme;

Determine the total Matching power flow of letter of each alphabetical matching scheme according to formula three, wherein, formula three is:

$C_{\mathrm{zm}}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i$

Wherein, C _zmfor the total Matching power flow of letter, C _ifor the i-th Matching power flow to the first Division Sampling point matched and the first palette sample point in current letter matching scheme, n is described first predetermined number;

Determine the best letter Matching power flow that the total Matching power flow of letter is minimum, alphabetical for described the best Matching power flow is split the second matching degree of image and current letter template image as current letter;

Determine the second matching degree of current letter segmentation image and each alphabetical template image;

The minimum alphabetical template image of the second matching degree of image is split, as the coupling letter template image of current letter segmentation image by with current letter.

5. method according to claim 2, is characterized in that, described A3, comprising:

Sample to Contemporary Digital segmentation image and obtain the second Division Sampling point of the second predetermined number, sampling to Contemporary Digital template image obtains the second palette sample point of described second predetermined number;

Determine the shape histogram of each second Division Sampling point under polar coordinate system in each default characteristic area, determine the shape histogram under polar coordinate system in each default characteristic area of each second palette sample point;

Calculate the Matching power flow between current second Division Sampling point and current second palette sample point according to formula four, wherein, formula four is:

$C(r_x,s_y)=\frac{1}{2}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\frac{{[h_x\left(m\right)-h_y\left(m\right)]}^2}{h_x\left(m\right)+h_y\left(m\right)}$

Wherein, r _xfor current first Division Sampling point, s _yfor current first palette sample point, C (r _x, s _y) be a r _xwith a r _xbetween Matching power flow, h _xm () is m characteristic area mid point r in M the characteristic area preset under polar coordinate system _xshape histogram, h _ym () is m characteristic area mid point s in M the characteristic area preset under polar coordinate system _yshape histogram;

Determine the Matching power flow between arbitrary second Division Sampling point and arbitrary second palette sample point;

All second Division Sampling points are mated with all second palette sample points, obtains digital matching scheme;

Determine the total Matching power flow of numeral of each digital matching scheme according to formula five, wherein, formula five is:

$C_{\mathrm{sz}}=\underset{s=1}{\overset{f}{\mathrm{\Σ}}}{C}_s$

Wherein, C _szfor the total Matching power flow of numeral, C _sfor s in Contemporary Digital matching scheme is to the Matching power flow of the second Division Sampling point matched and the second palette sample point, f is described second predetermined number;

Determine the optimal digital Matching power flow that the total Matching power flow of numeral is minimum, described optimal digital Matching power flow is split the 3rd matching degree of image and Contemporary Digital template image as Contemporary Digital;

Determine the 3rd matching degree of Contemporary Digital segmentation image and each digital template image;

The minimum digital template image of the 3rd matching degree of image is split, as the coupling digital template image of Contemporary Digital segmentation image by with Contemporary Digital.

6. a device for Car license recognition, is characterized in that, comprising:

Setting unit, for determining character, arranges the template image of each character, wherein, and each character at least one template image corresponding, the corresponding character of each template image;

Acquiring unit, for obtaining license plate image to be identified;

Cutting unit, for according to the character in described license plate image to be identified, splits described license plate image to be identified, obtains segmentation image;

Matching unit, for being mated with each template image by current segmentation image, determining and the matching template image that current segmentation image mates most, obtaining the matching template image of each segmentation image;

Character cell, for using character corresponding for the matching template image of current segmentation image as character corresponding to current segmentation image.

7. device according to claim 6, is characterized in that,

Described character comprises: Chinese character, alphabetic character, numerical character;

Described template image comprises: Chinese character template image, alphabetical template image, digital template image;

Described setting unit, for determining Chinese character, alphabetic character, numerical character, the Chinese character template image of each Chinese character is set, pre-set the Chinese character template image of each alphabetic character, pre-set the Chinese character template image of each numerical character, wherein, each Chinese character at least one Chinese character template image corresponding, the corresponding Chinese character of each Chinese character template image, each alphabetic character at least one alphabetical template image corresponding, the corresponding alphabetic character of each alphabetical template image, corresponding at least one the digital template image of each numerical character, the corresponding numerical character of each digital template image,

Described segmentation image comprises: Chinese character segmentation image, grapheme segmentation image, digital segmentation image;

Described cutting unit, for when performing described acquisition segmentation image, specifically performing: obtain Chinese character segmentation image, obtaining grapheme segmentation image, obtaining digital segmentation image;

Described matching template image comprises: mate Chinese character template image, mate alphabetical template image, coupling digital template image;

Described matching unit, comprising: Chinese character matching unit, alphabetical matching unit, digital matching unit;

Described Chinese character matching unit, for being mated with each Chinese character template image by current Chinese character segmentation image, determining and the coupling Chinese character template image that current Chinese character segmentation image mates most, obtaining the coupling Chinese character template image of each Chinese character segmentation image;

Described alphabetical matching unit, mates with each alphabetical template image for current letter being split image, determines the coupling letter template image split image with current letter and mate most, obtains the coupling letter template image of each grapheme segmentation image;

Described digital matching unit, mates with each digital template image for Contemporary Digital being split image, determines the coupling digital template image split image with Contemporary Digital and mate most, obtains the coupling digital template image of each digital segmentation image.

8. device according to claim 7, is characterized in that, described Chinese character matching unit, comprising:

Cross-correlation unit, for current Chinese character template image is moved on current Chinese character segmentation image, travel through all pixels of current Chinese character segmentation image, and calculate current Chinese character template image by formula one and moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image, wherein, described formula one is:

$\mathrm{\γ}(u,v)=\frac{{\mathrm{\Σ}}_{x,y}[f(x,y)-{\stackrel{\‾}{f}}_{u,v}][t(x-u,y-v)-\stackrel{\‾}{t}]}{\left\{{\mathrm{\Σ}}_{x,y}\right[f(x,y)-{\stackrel{\‾}{f}}_{u,v}{]}^2{\mathrm{\Σ}}_{x,y}{[t(x-u,y-v)-\stackrel{\‾}{t}]}^2{\}}^{0.5}}$

Wherein, u is the distance of current Chinese character template image movement in the x direction, v is the distance of current Chinese character template image movement in y-direction, γ (u, v) for current Chinese character template image is moving (u, v) apart from the normalized-cross-correlation function of rear and current Chinese character segmentation image after current Chinese character template movement (u, v) distance, the average gray of the current Chinese character segmentation image under current Chinese character template image covers, for the average gray of current Chinese character template image, t (x-u, y-v) is for current Chinese character template image is at point (x-u, y-v) gray-scale value, f (x, y) is for current Chinese character segmentation image is at the gray-scale value of point (x, y);

First Chinese character matching degree unit, for by normalized-cross-correlation function maximum in all normalized-cross-correlation function of current Chinese character template image and current Chinese character segmentation image, as the first matching degree of current Chinese character segmentation image and current Chinese character template image;

Second Chinese character matching degree unit, for determining the first matching degree of current Chinese character segmentation image and each Chinese character template image;

Chinese character matching image determining unit, for by the Chinese character template image maximum with the first matching degree of current Chinese character segmentation image, as the coupling Chinese character template image of current Chinese character segmentation image.

9. device according to claim 7, is characterized in that, described alphabetical matching unit, comprising:

Letter sampling unit, obtain the first Division Sampling point of the first predetermined number for sampling to current letter segmentation image, sampling to current letter template image obtains the first palette sample point of described first predetermined number;

Letter histogram unit, for determining the shape histogram of each first Division Sampling point under polar coordinate system in each default characteristic area, determines the shape histogram under polar coordinate system in each default characteristic area of each first palette sample point;

First alphabetical Matching power flow unit, for calculating the Matching power flow between current first Division Sampling point and current first palette sample point according to formula two, wherein, formula two is:

$C(p_i,q_j)=\frac{1}{2}\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\frac{{[h_i\left(k\right)-h_j\left(k\right)]}^2}{h_i\left(k\right)+h_j\left(k\right)}$

Wherein, p _ifor current first Division Sampling point, q _jfor current first palette sample point, C (p _i, q _j) be a p _iwith a q _jbetween Matching power flow, h _ik () is a kth characteristic area mid point p in K the characteristic area preset under polar coordinate system _ishape histogram, h _jk () is a kth characteristic area mid point q in K the characteristic area preset under polar coordinate system _jshape histogram;

Second alphabetical Matching power flow unit, for determining the Matching power flow between arbitrary first Division Sampling point and arbitrary first palette sample point;

Letter matching scheme unit, for being mated with all first palette sample points by all first Division Sampling points, obtains alphabetical matching scheme;

The total Matching power flow unit of letter, for determining the total Matching power flow of letter of each alphabetical matching scheme according to formula three, wherein, formula three is:

$C_{\mathrm{zm}}=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{C}_i$

Wherein, C _zmfor the total Matching power flow of letter, C _ifor the i-th Matching power flow to the first Division Sampling point matched and the first palette sample point in current letter matching scheme, n is described first predetermined number;

First alphabetical matching degree unit, for determining the best letter Matching power flow that the total Matching power flow of letter is minimum, splits the second matching degree of image and current letter template image as current letter using alphabetical for described the best Matching power flow;

Second alphabetical matching degree unit, for determining the second matching degree of current letter segmentation image and each alphabetical template image;

Letter matching image determining unit, for split the minimum alphabetical template image of the second matching degree of image, as the alphabetical template image of coupling of current letter segmentation image by with current letter.

10. device according to claim 7, is characterized in that, described digital matching unit, comprising:

Digital acquisition unit, obtain the second Division Sampling point of the second predetermined number for sampling to Contemporary Digital segmentation image, sampling to Contemporary Digital template image obtains the second palette sample point of described second predetermined number;

Numeral histogram unit, for determining the shape histogram of each second Division Sampling point under polar coordinate system in each default characteristic area, determines the shape histogram under polar coordinate system in each default characteristic area of each second palette sample point;

First digital Matching power flow unit, for calculating the Matching power flow between current second Division Sampling point and current second palette sample point according to formula four, wherein, formula four is:

$C(r_x,s_y)=\frac{1}{2}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}\frac{{[h_x\left(m\right)-h_y\left(m\right)]}^2}{h_x\left(m\right)+h_y\left(m\right)}$

Wherein, r _xfor current first Division Sampling point, s _yfor current first palette sample point, C (r _x, s _y) be a r _xwith a r _xbetween Matching power flow, h _xm () is m characteristic area mid point r in M the characteristic area preset under polar coordinate system _xshape histogram, h _ym () is m characteristic area mid point s in M the characteristic area preset under polar coordinate system _yshape histogram;

Second digital Matching power flow unit, determines the Matching power flow between arbitrary second Division Sampling point and arbitrary second palette sample point;

Numeral matching scheme unit, for being mated with all second palette sample points by all second Division Sampling points, obtains digital matching scheme;

The total Matching power flow unit of numeral, for determining the total Matching power flow of numeral of each digital matching scheme according to formula five, wherein, formula five is:

$C_{\mathrm{sz}}=\underset{s=1}{\overset{f}{\mathrm{\Σ}}}{C}_s$

Wherein, C _szfor the total Matching power flow of numeral, C _sfor s in Contemporary Digital matching scheme is to the Matching power flow of the second Division Sampling point matched and the second palette sample point, f is described second predetermined number;

First digital matching degree unit, for determining the optimal digital Matching power flow that the total Matching power flow of numeral is minimum, splits the 3rd matching degree of image and Contemporary Digital template image as Contemporary Digital using described optimal digital Matching power flow;

Second digital matching degree unit, for determining the 3rd matching degree of Contemporary Digital segmentation image and each digital template image;

Numeral matching image determining unit, for split the minimum digital template image of the 3rd matching degree of image by with Contemporary Digital, splits the coupling digital template image of image as Contemporary Digital.