CN102693431B - Method and device for identifying type of white number plate - Google Patents
Method and device for identifying type of white number plate Download PDFInfo
- Publication number
- CN102693431B CN102693431B CN201210177788.4A CN201210177788A CN102693431B CN 102693431 B CN102693431 B CN 102693431B CN 201210177788 A CN201210177788 A CN 201210177788A CN 102693431 B CN102693431 B CN 102693431B
- Authority
- CN
- China
- Prior art keywords
- white
- license plate
- car plate
- plate area
- distance
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Character Input (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a method and a device for identifying the type of a white number plate. The problems of low precision and accuracy during the identification of the type of the white number plate in the prior art are solved. The method for identifying the type of the white number plate comprises the following steps of: performing character segmentation on a preprocessed white number plate area, identifying a character sequence in line with each white number plate type in the white number plate according to a distance between every two segmented characters, and determining the type of the white number plate according to a matching distance between a template corresponding to each white number plate type and the character sequence. The number plate sequence in line with each type is determined according to the distances between the characters, and then the type of the number plate is finally determined according to the distance between the template corresponding to the number plate type and the character sequence, so that the accuracy and precision of an identification result can be ensured under twice identification condition.
Description
Technical field
The present invention relates to computer image processing technology field, relate in particular to a kind of recognition methods and device of white car plate type.
Background technology
Along with socioeconomic development, the quantity of vehicle constantly increases, and therefore by computer information, intelligentized project management vehicle, becomes inevitable.In prior art, car plate is identified in the important role of intelligent transportation field performer, and it has a wide range of applications in traffic flow monitoring, the charge of highway bayonet socket, red light violation vehicle monitoring and community's automatic fare collection system.
But existing conventional vehicle license type is many, from background color, be divided into wrongly written or mispronounced character car plate of the blue end, yellow end surplus car plate, black matrix wrongly written or mispronounced character car plate and white gravoply, with black engraved characters car plate; Type, can be divided into again normal blue board, common individual layer yellow card, double-deck yellow card, coach with board, list temporarily, black licence plate, white alert board, white army board and white People's Armed Police's car plate etc.To the car plate adularescent of white gravoply, with black engraved characters alert board, white army board and white People's Armed Police's car plate.
In prior art, the car plate of white gravoply, with black engraved characters is when identifying, the general method adopting comprises: utilize the color of car plate to orient white car plate, three kind car plate types corresponding according to white gravoply, with black engraved characters car plate afterwards, the white car plate of orienting is carried out to Character segmentation, finally, by identification, select the last recognition result of conduct of degree of confidence maximum.In the identifying of above-mentioned car plate type, only according to degree of confidence, determine car plate type, the standard of judgement is more single, can not effectively to car plate type, distinguish, and therefore the recognition methods precision of white car plate type of the prior art is low, poor accuracy.
Summary of the invention
In view of this, the embodiment of the present invention provides a kind of recognition methods and device of white car plate type, low in order to solve prior art precision when the type of white car plate is identified, the problem of poor accuracy.
The invention provides a kind of recognition methods of white car plate type, described method comprises:
In the image of input, carry out license plate area location, in the license plate area of location, identify white license plate area;
The white license plate area identifying is carried out to pre-service, and pretreated white license plate area is carried out to Character segmentation;
According to the distance between every two characters after Character segmentation, identify the character string that meets every kind of white car plate type in this white car plate;
By each character string, mate with the template of the corresponding car plate type of this character string, according to average matching distance, determine the type of each white car plate.
The invention provides a kind of recognition device of white car plate type, described device comprises:
Identification module, carries out license plate area location for the image in input, in the license plate area of location, identifies white license plate area;
Module is cut apart in pre-service, for the white license plate area identifying is carried out to pre-service, and pretreated white license plate area is carried out to Character segmentation;
Character string is selected module, for according to the distance between every two characters after Character segmentation, identifies the character string that meets every kind of white car plate type in this white car plate;
Matching module, for by each character string, mates with the template of the corresponding car plate type of this character string, determines the type of each white car plate according to average matching distance.
The invention provides a kind of recognition methods and device of white car plate type, the method is when carrying out white car plate type identification, pretreated white license plate area has been carried out to Character segmentation, according to the distance of cutting apart between rear every two characters, identify the character string that meets every kind of white car plate type in this white car plate, according to the template corresponding with every kind of white car plate type and the matching distance of this character string, determine the car plate type of this white car plate.Due in the present invention according to the distance between character string, determine and meet the car plate sequence of every type, according to the distance of template corresponding to the type car plate and this character string, finally determine the type of this car plate again afterwards, therefore the in the situation that of twice identification, can guarantee accuracy and the degree of accuracy of recognition result.
Accompanying drawing explanation
Accompanying drawing described herein is used to provide a further understanding of the present invention, forms a part of the present invention, and schematic description and description of the present invention is used for explaining the present invention, does not form inappropriate limitation of the present invention.In the accompanying drawings:
Fig. 1 is the identifying schematic diagram of this white car plate type provided by the invention;
Fig. 2 is the process schematic diagram that this white license plate area is carried out to rectification provided by the invention;
Fig. 3 is the process schematic diagram of this white license plate area being removed to up-and-down boundary processing provided by the invention;
Fig. 4 is the process schematic diagram of identifying white car plate type by distance provided by the invention;
Fig. 5 is the detailed identifying schematic diagram of this white car plate type provided by the invention;
Fig. 6 is the structural representation of this white car plate type identification device provided by the invention.
Embodiment
The embodiment of the present invention, in order to improve the accuracy of identification of the type of white car plate and accuracy, provides a kind of recognition methods and device of white car plate type.
Below in conjunction with Figure of description, the present invention is described in detail.
Fig. 1 is the identifying schematic diagram of this white car plate type provided by the invention, and this identifying comprises the following steps:
S101: carry out license plate area location in the image of input, identify white license plate area in the license plate area of location.
Recognition methods due to this white car plate type, can be applied in a plurality of systems such as traffic flow monitoring, the charge of highway bayonet socket, red light violation vehicle monitoring and community's automatic charging, concrete identifies white car plate type by white car plate type identification device in the present invention.
After the image of the vehicle that comprises license board information is input in this device, this device carries out license plate area location in the image of input, concrete can adopt several different methods of the prior art when carrying out license plate area location, for example can adopt edge detection method or machine learning scheduling algorithm to carry out car plate location in the image of input, concrete for example can carry out sobel rim detection, according to the result detecting, using the larger region of marginal density as car plate region.
Owing to carrying out license plate area location in the image in input, identifying white license plate area, be the common method of this area, just this content is not repeated in the present invention.
S102: the white license plate area identifying is carried out to pre-service, and pretreated white license plate area is carried out to Character segmentation.
Concrete when the white license plate area identifying is carried out to pre-service, can adopt multiple preprocess method, for example can directly to license plate area, carry out binary conversion treatment.
In order further to improve the accuracy of white car plate type identification, in the present invention the white license plate area identifying being carried out to pre-service comprises: adopt following at least one preprocess method to carry out pre-service to the white license plate area identifying, rectification preprocess method and removal up-and-down boundary preprocess method.
Rectification preprocess method and removal up-and-down boundary preprocess method should be used in combination, and first carry out horizontal tilt correction, and then remove boundary treatment.Because pretreated process has been carried out rectification to white license plate area, and removed the interference to the up-and-down boundary in car plate type identification, therefore can in this pretreated car plate white portion, carry out Character segmentation.Concrete when carrying out Character segmentation,, in the region at white car plate place, determine the border of each character, thereby determine the position of each character.
S103: according to the distance between every two characters after Character segmentation, identify the character string that meets every kind of white car plate type.
White license plate area is carried out, after Character segmentation, can determining the position at each character place, according to the position at each character place, can determine the distance between any two adjacent characters.In the present invention because the distance between any two characters can be determined, and the white car plate of every type, there is certain relation in the distance between one fixed number object character string, for example, for the alert board of white, between its 7 character strings, distance between the first two character is greater than the distance between other any two characters, and for white army board, between its 7 character strings, distance between second character and the 3rd character is greater than the distance between other any two characters, and for white People's Armed Police's board, in its 5 character strings, distance between any two characters is basic identical.
Therefore according to the distance feature between the character of the white car plate of above-mentioned every type, can, for each white license plate area, identify the character string that meets every kind of white car plate type.
S104: by each character string, mate with the template of the corresponding car plate type of this character string, determine the type of each white car plate according to average matching distance.
Owing to being each character comprising from white license plate area in the present invention, selection meets the character string of every kind of white car plate type, and for the character comprising in same white license plate area, its different character string may meet different white car plate types.Therefore when carrying out follow-up identification, for the character comprising in same white license plate area, the white car plate type that its different character string meets, selects the template corresponding with the type, this template is mated with this character string, according to matching distance, determine the type of each white car plate.
For example, the character comprising for a certain white license plate area, its 3 character strings are respectively the first character string, the second character string and three-character doctrine sequence, wherein the first character string is the character string that meets white army board, the second character string is for meeting the character string of the alert board of white, three-character doctrine sequence is the character string that meets white People's Armed Police's board, adopt the first template that white army board is corresponding to mate with the region at the first character string place, adopt the second template corresponding to the alert board of white to mate with the region at the second character string place, adopt the 3rd template that white People's Armed Police's board is corresponding to mate with the region at three-character doctrine sequence place, when the Region Matching distance at the second template and the second character string place the most in short-term, determine that this white car plate is for the alert board of white.
The present invention is when carrying out white car plate type identification, pretreated white license plate area has been carried out to Character segmentation, according to the distance of cutting apart between rear every two characters, identify the character string that meets every kind of white car plate type in this white car plate, according to the template corresponding with every kind of white car plate type and the matching distance of this character string, determine the car plate type of this white car plate.Due in the present invention according to the distance between character string, determine and meet the car plate sequence of every type, according to the distance of template corresponding to the type car plate and this character string, finally determine the type of this car plate again afterwards, therefore the in the situation that of twice identification, can guarantee accuracy and the degree of accuracy of recognition result.
Below by specific embodiment, car plate type identification process of the present invention is elaborated.
In the image in input, be generally gray level image in the present invention, carry out, behind license plate area location, can determining the region at car plate place in this image, and by corresponding method, by the white car plate identifying processing in these car plates.
For the region at each white car plate place of identifying, pre-service is carried out in this region, concrete preprocessing process comprises rectification and/or removes up-and-down boundary.When this preprocessing process comprises rectification and removes up-and-down boundary, the better effects if of identification.
Fig. 2 is the process schematic diagram that this white license plate area is carried out to rectification provided by the invention, and this process comprises the following steps:
S201: adopt corresponding method to carry out binary conversion treatment to the described white license plate area identifying.
S202: according to the saltus step of pixel value in the white license plate area after the binary conversion treatment identifying, the position assignment that white pixel is put corresponding pixel during with this saltus step in horizontal edge hum pattern is 1, and the position assignment of other pixels is 0.
S203: with the horizontal marginal information figure of step-length rotation setting, for postrotational each angle, determine that in the angular range of setting every row comprises the number of white pixel point in the horizontal direction.
S204: for postrotational each angle, select the more setting line number of number of white pixel point, add up this setting line number white pixel point number with.
S205: according to the number of white pixel point and the anglec of rotation corresponding to maximal value, determine and the RA of this white license plate area this white license plate area carried out to rectification.
Concrete when this white license plate area is carried out to rectification, for this white license plate area, carry out corresponding binary conversion treatment, for example can adopt otsu Binarization methods.After binary conversion treatment, character is black, and background is white.In order to identify after this binary conversion treatment the state of each pixel in this white license plate area, adopted in the present invention horizontal edge hum pattern.
In a certain order, search for each pixel in the white license plate area after this binary conversion treatment, when the pixel value of this pixel and the next pixel being adjacent searching is identical, in this horizontal edge hum pattern, be 0 with the position assignment of the corresponding pixel of this pixel; When the pixel value of this pixel and the next pixel being adjacent searching is different, be that the pixel value of pixel is while occurring saltus step, the pixel value that is pixel is 1 by 0 saltus step, or it is 0 o'clock by 1 saltus step, in this horizontal edge hum pattern, the pixel position assignment corresponding with this white pixel point position is 1, by the position assignment of the pixel corresponding with black pixel point position, is 0.By the above-mentioned search to each pixel in the white license plate area of binaryzation, finally determine the assignment of each pixel in horizontal edge hum pattern.
For final definite horizontal edge hum pattern, the pixel in the upper left corner of this horizontal edge hum pattern of take is the center of circle, and definite transverse axis, the step-length of setting in the present invention can be 1 degree, 1 degree of take rotates this horizontal edge hum pattern as step-length, in the scope of spending at transverse axis positive and negative 20 after rotation, determine every row projection value in the horizontal direction in horizontal edge hum pattern, determine the pixel that in this horizontal edge hum pattern, every row pixel value is 1, the number in positive and negative 20 scopes of spending of this transverse axis.
According to after each number of degrees of rotation, every row projection value in the horizontal direction in horizontal edge hum pattern, determine the number of the white pixel point that in horizontal edge hum pattern, every row comprises in this angular range, select the white pixel more n that counts out capable, preferably, this n value is 4 in the present invention,, after these number of degrees of rotation, selects 4 row of projection value maximum.For this angle, add up this 4 row projection value in the horizontal direction and, the number of selecting the white pixel point that this 4 row comprises in this angular range with.
According to rotation each angle, 4 row that projection value is larger, the number of the white pixel point comprising in this angular range with, determine this number and maximal value.When determined this number and maximal value after, angle corresponding to this maximal value also can be determined, according to this angle, can carry out rectification to this white license plate area.Concrete when this white license plate area is carried out to rectification, when this angle is A, utilize bilinear interpolation to carry out horizontal tilt rectification.
In order to improve the accuracy of white car plate type identification, the white license plate area obtaining after locating is carried out after rectification, can to the white license plate area after this rectification, remove again the processing of up-and-down boundary in the present invention.
Fig. 3 is the process schematic diagram of this white license plate area being removed to up-and-down boundary processing provided by the invention, and this process comprises the following steps:
S301: the white license plate area after described rectification is carried out to binary conversion treatment.
S302: in the white license plate area after binary conversion treatment, identify the transition times of every row pixel value, and determine the longest white line length and the longest black line length that every row comprises.
S303: according at least one in described transition times, the longest white line length and the longest black line length, with the respective distances threshold value of setting, determine the upper and lower border of this white license plate area.
Concrete when this white license plate area being removed to up-and-down boundary and process, first carrying out the white license plate area of rectification, adopting otsu Binarization methods, this white license plate area is being carried out to binary conversion treatment.This white license plate area is carried out after binary conversion treatment, from the middle line position of this white license plate area, search for up and down respectively every a line.
Often search a line, this row carried out to following treatment step:
Identify the hop value of this row, in this row, the pixel value of two adjacent pixels becomes 0 from 1, and becomes 1 number of times, the hop value SumLineJump using this number of times as this row from 0.In addition, when this is advanced to line search, also to identify every black line comprising in this row and the length of white line, identify the number of the pixel that every black line and white line comprise, and therefrom determine the length L ineWhiteLength of long white line that this row comprises, and the length L ineBlackLength of long black line.
In the identifying in car plate type, the situation that may occur Characters Stuck, now when carrying out the interpretation of white line length, just there will be error, in the present invention in order further to improve the accuracy of white car plate type identification, during white line in identification in white license plate area, first need to judge whether this white line is real white line, because may be the white line of Adhesion formation between two characters, rather than the white line of real character representative, when being judged as real white line, this white line is joined in the selection set of the longest white line length.
Concrete when determining whether real white line, first determine the scope that forms this white line in this row, in this row initial pixel to the scope that stops pixel.In this row forms the scope of this white line, from pixel corresponding to the starting point of this white line, judge adjacent with this pixel, whether the pixel that is arranged in two row up and down of same row there is black pixel point, when there is a black pixel point, the identification parameter iSumBlack that is initialized as 0 is added to 1, when one other pixel point being detected also for black pixel point, this identification parameter is added to 1 again; Afterwards, for second pixel in this white line, judge adjacent with this pixel, whether the pixel that is arranged in two row up and down of same row there is black pixel point, this identification parameter is added to 1 again, loop said process, until each pixel in this white line has been carried out to above-mentioned judgement.
The numerical value of this identification parameter after completing according to judgement, the length iWhiteLineLen of this white line, and the first proportion threshold value α arranging, determine whether this white line is real white line, specifically when judging, judge whether this parameter is greater than the long-pending of this first proportion threshold value α and this white line length, judge whether iSumBlack> α * iWhiteLineLen, when this identification parameter is greater than the first proportion threshold value α and this white line length iWhiteLineLen long-pending, determine that this white line is real white line, this white line is joined in the selection set of the longest white line length, otherwise, be defined as disturbing, as white line, do not process, the general value in 0.5 to 0.8 scope of the first proportion threshold value α wherein.
Or, when whether every the white line identifying in judgement is real white line, also can be according to the length iWhiteLineLen of this white line, the the second proportion threshold value β arranging and the width W of car plate are determined, concrete is whether the length that judges this white line is greater than the long-pending of the second proportion threshold value β and this car plate width W, judge whether iWhiteLineLen> β * W, when the length iWhiteLineLen of white line is greater than the second proportion threshold value β and this car plate width W long-pending, determine that this white line is real white line, this white line is joined in the selection set of the longest white line length, otherwise, be defined as disturbing, as white line, do not process, the general value in 0.3 to 0.5 scope of the second proportion threshold value β wherein.
When having determined the hop value SumLineJump of every row, and after the longest black line length L ineBlackLength and the longest white line length L ineWhiteLength of every row, can judge whether this row is border, specifically when judging: can judge whether the hop value SumLineJump of this row is less than the saltus step threshold value Thr1 of setting, judge whether SumLineJump<Thr1, when the hop value SumLineJump of this row is less than the saltus step threshold value Thr1 of setting, think behavior coboundary or lower boundary;
Or, also can judge whether the longest black line length L ineBlackLength of this row is greater than the black line length threshold Thr2 of setting, judge whether LineBlackLength>Thr2, when the longest black line length L ineBlackLength of this row is greater than the black line length threshold Thr2 of setting, think behavior coboundary or lower boundary;
Or, also can judge whether the longest white line length L ineWhiteLength of this row is greater than the white line length threshold Thr3 of setting, wherein the value of Thr1, Thr2 and Thr3 1/4th of the board width of generally picking up the car.Judge whether LineWhiteLength>Thr3, when the longest white line length L ineWhiteLength of this row is greater than the white line length threshold Thr3 of setting, think behavior coboundary or lower boundary.
Determined behind the coboundary and lower boundary of white license plate area, the region that surpasses upper and lower border can have been removed, thereby reduce when follow-up white car plate type identification the interference of borderline region to identifying.
White license plate area is carried out after pre-service, in this white license plate area, comprise character zone and background area, can in this white license plate area, carry out having cut apart of character.Concrete while carrying out Character segmentation in this white license plate area, this pretreated white license plate area is carried out after binary conversion treatment, then carry out inverse and blow, making character in this white license plate area is partly white, background parts is black.
In white license plate area after this binaryzation inverse are processed, carry out the projection of vertical direction and calculate, calculate the number sum of every row white pixel point, concrete when calculating, can be according to following formula:
Wherein, Sum
inumber sum for the white pixel point of i row in the white license plate area after binaryzation inverse processing, WhitePlate (j, i) white pixel point array, represent the white pixel point of the capable i row of j,, iHeight is the width of this white license plate area, the line number comprising, after carrying out inverse processing, in this white license plate area, the pixel value of each white pixel point is 1, and the pixel value of black pixel point is 0.
According to the projection that is often listed in vertical direction of determining, determine character zone.Determine when definite character region, the order according to setting, is generally order from left to right, searches the projection value that is often listed in vertical direction, when i is listed in the projection value Sum of vertical direction
iwhile being greater than zero, determine that this i classifies the left margin of character as, continue afterwards to search next column, as the projection value Sum of m row
mequal at 0 o'clock, determine that this m classifies the right margin of character as, the region between left margin and right margin is character region.According to said method, in a certain order, can in white license plate area, determine the position at each character place, thereby realize cutting apart of character.
Owing in white license plate area, character being cut apart, determined the position at each character place, therefore the distance between every two characters can be determined, the region at each character place is rectangular area in the present invention, by the center of rectangle, can determine the distance between character.
Three kinds of its structural formulas of dissimilar licence plate of white car plate are different, therefore can be according to the structures of every kind of car plate type, and identification car plate type.For example, for alert board, in 7 continuous character strings, the distance between the first each and every one and the second character is greater than the distance between all the other any two adjacent characters, that is:
d
12>max(d
23,d
34,d
45,d
56,d
67)
D wherein
12, d
23, d
34, d
45, d
56, d
67, be respectively in these 7 continuous character strings the distance between any two adjacent characters.For army's board, in 7 continuous character strings, the distance between second and the 3rd character is greater than the distance between all the other any two adjacent characters, that is:
d
23>max(d
12,d
34,d
45,d
56,d
67)
And for People's Armed Police's car plate, in 5 continuous character strings, the distance between every two characters equates, when the spacing of each adjacent character is less than 3 pixel values, can think that the distance between two characters equates in the present invention.
According to the distance feature of character string corresponding to above-mentioned every kind of car plate type, after each white is cut apart for the character of license plate area, identify corresponding character string in the present invention, judge whether each character string meets corresponding distance condition.Fig. 4 is the process schematic diagram of identifying white car plate type by distance provided by the invention, and this process comprises the following steps:
S401: according to the distance between every two characters after Character segmentation, whether the distance between the adjacent character sequence of judgement setting quantity, meet the first distance condition, second distance conditioned disjunction the 3rd distance condition of setting.
S402: according to the satisfied distance condition of this character string, determine the car plate type of this character string ownership.
S403: adopt template corresponding to this car plate type, determine the distance between this template and this character string, selection and template are apart from minimum sequence as car plate, and type corresponding to this template is this car plate type.
Concrete according to the distance between the character in white car plate, while determining car plate type, for each car plate, take out any 7 continuous character strings and 5 continuous character strings in this car plate, may in each white license plate area, there be a plurality of character strings that meet above-mentioned condition, for each character string of taking out, according to the length of this character string, select corresponding distance condition, judge whether this character string meets this corresponding distance condition, when for example this character string is the character string of 5 characters, its tool distance condition is whether distance between the adjacent character of every two equates, when this character string is the character string of 7 characters, judge in the character string of this 7 character, whether the distance between the first two character is greater than the distance between all the other any two adjacent characters, or judge whether the distance between second and the 3rd character is greater than the distance between all the other any two adjacent characters.The character string that meets above-mentioned distance condition is taken out, and record distance condition corresponding to each character string.
According to each that select, meet the character string of above-mentioned distance condition, adopt the template that this distance condition is corresponding, mate with this character string, determine the distance of this character string and this template.According to each character string of taking out from this white car plate sequence and the average matching distance between its template, selecting the car plate type corresponding to template of mean distance minimum value is the type of this white car plate.Adopt following formula:
Wherein, d
avgfor average matching distance, d
ifor the distance of i character in character string and template correspondence position, the length that N is this character string.
For example, in this white license plate area, have 3 character strings, wherein, wherein the first character string is the character string of 7 characters, it meets the first distance condition, meet the distance condition of alert board, the second character string is the character string of 7 characters, and it meets second distance condition, meet the distance condition of army's board, three-character doctrine sequence is the character string of 5 characters, and it meets the 3rd distance condition, meets the distance condition of People's Armed Police's board, these 3 character strings are taken out respectively, the template that the army's of employing board is corresponding is respectively mated with the first character string, the template that army's board is corresponding meets the template of army's board distance condition, location positioning corresponding to each character in this template, the center of each character is also determined, and the center of each character is also determined in character string, therefore according to the center of each character in template, and the center of corresponding each character, can determine template and character string matching distance and, according to the length of this character string, can determine the average matching distance of character string and template.Owing to having 3 character strings all to meet the distance condition of respective type car plate in this white car plate sequence, therefore each character string and corresponding template can be determined corresponding average matching distance, select the car plate type corresponding to template of average matching distance minimum value to be the type that this white car plate is corresponding.
Fig. 5 is the detailed identifying schematic diagram of this white car plate type provided by the invention, and this process comprises the following steps:
S501: carry out license plate area location in the image of input, identify white license plate area in the license plate area of location.
S502: this white license plate area is carried out to binary conversion treatment, this white license plate area is carried out to rectification.
S503: in the binaryzation white license plate area figure after rectification, remove up-and-down boundary.
S504: carry out binary conversion treatment and carry out inverse processing having removed the white license plate area of up-and-down boundary, determining the projection of the white license plate area vertical direction after this inverse is processed, the character in white license plate area is cut apart.
S505: after Character segmentation, according to the distance between every two adjacent characters in the character string comprising in this white license plate area, and corresponding distance condition, selection meets the character string of every kind of distance condition.
S506: according to the character string of selecting, adopt this character string to mate with corresponding template, determine average matching distance.
S507: the character string of selecting average matching distance minimum is this car plate sequence, car plate type corresponding to this template is the type of this white car plate.
Due in the present invention according to the distance between character string, determine and meet the car plate sequence of every type, according to the distance of template corresponding to the type car plate and this character string, finally determine the type of this car plate again afterwards, therefore the in the situation that of twice identification, can guarantee accuracy and the degree of accuracy of recognition result.
Fig. 6 is the structural representation of this white car plate type identification device provided by the invention, and this device comprises:
Identification module 61, carries out license plate area location for the image in input, in the license plate area of location, identifies white license plate area;
Module 62 is cut apart in pre-service, for the white license plate area identifying is carried out to pre-service, and pretreated white license plate area is carried out to Character segmentation;
Character string is selected module 63, for according to the distance between every two characters after Character segmentation, identifies the character string that meets every kind of white car plate type in this white car plate;
Matching module 64, for by each character string, mates with the template of the corresponding car plate type of this character string, determines the type of each white car plate according to average matching distance.
Module 62 is cut apart in described pre-service, specifically for adopting corresponding method to carry out binary conversion treatment to the described white license plate area identifying; According to the saltus step of pixel value in the white license plate area after the binary conversion treatment identifying, the position assignment that white pixel is put corresponding pixel during with this saltus step in horizontal edge hum pattern is 1, and the position assignment of other pixels is 0; With the step-length of setting, rotate horizontal marginal information figure, for postrotational each angle, in the angular range of setting, determine that every row comprises the number of white pixel point in the horizontal direction; For postrotational each angle, select the more setting line number of number of white pixel point, add up this setting line number white pixel point number with; According to the number of white pixel point and the anglec of rotation corresponding to maximal value, determine and the RA of this white license plate area this white license plate area carried out to rectification.
Module 62 is cut apart in described pre-service, specifically for described white license plate area is carried out to binary conversion treatment; In white license plate area after binary conversion treatment, identify the transition times of every row pixel value, and determine the longest white line length and the longest black line length that every row comprises; According at least one in described transition times, the longest white line length and the longest black line length, with the respective distances threshold value of setting, determine the upper and lower border of this white license plate area.
Module 62 is cut apart in described pre-service, and specifically for before determining the longest white line length that every row comprises, whether every white line that judgement is determined is real white line; When determining that this white line is real white line, this white line is joined in the selection set of the longest white line length.
Module 62 is cut apart in described pre-service, length specifically for the number of pixel that this white line is comprised as this white line, according to the length of this white line, whether be greater than the long-pending of the second proportion threshold value of setting and the car plate width of setting, judge whether this white line is real white line; Or
For the white line of every row, the number of the pixel comprising according to this white line, and in this white line, whether the neighbor pixel of pixel same column is black pixel point, determines the value of identification parameter; Length long-pending that whether is greater than the first proportion threshold value and white line according to the value of identification parameter, judges whether this white line is real white line.
Module 62 is cut apart in described pre-service, described pretreated white license plate area is carried out to binary conversion treatment, and carry out inverse processing; Add up every row white pixel point number and; According to the number of every row white pixel point and, judge whether every row are the left and right border of character.
Described matching module 64, specifically for according to the distance between every two characters after Character segmentation, judges the distance between the adjacent character sequence of setting quantity, whether meets the first distance condition, second distance conditioned disjunction the 3rd distance condition of setting; According to the satisfied distance condition of this character string, determine the car plate type of this character string ownership; Adopt template corresponding to this car plate type, determine the matching distance between this template and this character string, select the sequence of average matching distance minimum as car plate, type corresponding to this template is this car plate type.
The invention provides a kind of recognition methods and device of white car plate type, the method is when carrying out white car plate type identification, pretreated white license plate area has been carried out to Character segmentation, according to the distance of cutting apart between rear every two characters, identify the character string that meets every kind of white car plate type in this white car plate, according to the template corresponding with every kind of white car plate type and the matching distance of this character string, determine the car plate type of this white car plate.Due in the present invention according to the distance between character string, determine and meet the car plate sequence of every type, according to the distance of template corresponding to the type car plate and this character string, finally determine the type of this car plate again afterwards, therefore the in the situation that of twice identification, can guarantee accuracy and the degree of accuracy of recognition result.
Above-mentioned explanation illustrates and has described a preferred embodiment of the present invention, but as previously mentioned, be to be understood that the present invention is not limited to disclosed form herein, should not regard the eliminating to other embodiment as, and can be used for various other combinations, modification and environment, and can, in invention contemplated scope described herein, by technology or the knowledge of above-mentioned design or association area, improve.And the change that those skilled in the art carry out and variation do not depart from the spirit and scope of the present invention, all should be in the protection domain of claims of the present invention.
Claims (6)
1. a recognition methods for white car plate type, is characterized in that, described method comprises:
In the image of input, carry out license plate area location, in the license plate area of location, identify white license plate area;
The white license plate area identifying is carried out to pre-service, and pretreated white license plate area is carried out to Character segmentation;
Wherein, described Character segmentation comprises:
Described pretreated white license plate area is carried out to binary conversion treatment, and carry out inverse processing;
Add up every row white pixel point number and;
According to the number of every row white pixel point and, judge whether every row are the left and right border of character;
According to the distance between every two characters after Character segmentation, identify the character string that meets every kind of white car plate type in this white car plate;
By each character string, mate with the template of the corresponding car plate type of this character string, according to average matching distance, determine the type of each white car plate;
Wherein, described according to the distance between every two characters after Character segmentation, identify the character string that meets every kind of white car plate type in this white car plate and comprise:
According to the distance between every two characters after Character segmentation, whether the distance between the adjacent character sequence of judgement setting quantity, meet the first distance condition, second distance conditioned disjunction the 3rd distance condition of setting;
Described by each character string, mate with the template of the corresponding car plate type of this character string, according to average matching distance, determine that the type of each white car plate comprises:
According to the satisfied distance condition of this character string, determine the car plate type of this character string ownership;
Adopt template corresponding to this car plate type, determine the matching distance between this template and this character string, select the sequence of average matching distance minimum as car plate, type corresponding to this template is this car plate type.
2. recognition methods as claimed in claim 1, is characterized in that, describedly the white license plate area identifying is carried out to pre-service comprises:
Rectification and removal up-and-down boundary;
Adopt described rectification preprocess method, the white license plate area identifying carried out to pre-service and comprise:
The described white license plate area identifying is carried out to binary conversion treatment;
According to the saltus step of pixel value in the white license plate area after the binary conversion treatment identifying, the position assignment that white pixel is put corresponding pixel during with this saltus step in horizontal edge hum pattern is 1, and the position assignment of other pixels is 0;
With the step-length of setting, rotate horizontal marginal information figure, for postrotational each angle, in the angular range of setting, determine that every row comprises the number of white pixel point in the horizontal direction;
For postrotational each angle, select the more setting line number of number of white pixel point, add up this setting line number white pixel point number with;
According to the number of white pixel point and the anglec of rotation corresponding to maximal value, determine and the RA of this white license plate area this white license plate area carried out to rectification;
Adopt described removal up-and-down boundary preprocess method, the white license plate area identifying carried out to pre-service and comprise:
Described white license plate area is carried out to binary conversion treatment;
In white license plate area after binary conversion treatment, identify the transition times of every row pixel value, and determine the longest white line length and the longest black line length that every row comprises;
According at least one in described transition times, the longest white line length and the longest black line length, with the respective distances threshold value of setting, determine the upper and lower border of this white license plate area.
3. method as claimed in claim 2, is characterized in that, described before determining the longest white line length that every row comprises, described method also comprises:
Whether the number of the pixel that this white line is comprised, as the length of this white line, is greater than the long-pending of the second proportion threshold value of setting and the car plate width of setting according to the length of this white line, judges whether this white line is real white line; Or
For the white line of every row, the number of the pixel comprising according to this white line, and in this white line, whether the neighbor pixel of pixel same column is black pixel point, determines the value of identification parameter;
Length long-pending that whether is greater than the first proportion threshold value and white line according to the value of identification parameter, judges whether this white line is real white line;
When determining that this white line is real white line, this white line is joined in the selection set of the longest white line length.
4. a recognition device for white car plate type, is characterized in that, described device comprises:
Identification module, carries out license plate area location for the image in input, in the license plate area of location, identifies white license plate area;
Module is cut apart in pre-service, for the white license plate area identifying is carried out to pre-service, and pretreated white license plate area is carried out to Character segmentation; Specifically for described pretreated white license plate area is carried out to binary conversion treatment, and carry out inverse processing; Add up every row white pixel point number and; According to the number of every row white pixel point and, judge whether every row are the left and right border of character;
Character string is selected module, for according to the distance between every two characters after Character segmentation, identifies the character string that meets every kind of white car plate type in this white car plate; Specifically for according to the distance between every two characters after Character segmentation, judge the distance between the adjacent character sequence of setting quantity, whether meet the first distance condition, second distance conditioned disjunction the 3rd distance condition of setting;
Matching module, for by each character string, mates with the template of the corresponding car plate type of this character string, determines the type of each white car plate according to average matching distance; Specifically for according to the satisfied distance condition of this character string, determine the car plate type of this character string ownership; Adopt template corresponding to this car plate type, determine the matching distance between this template and this character string, select the sequence of average matching distance minimum as car plate, type corresponding to this template is this car plate type.
5. device as claimed in claim 4, is characterized in that, module is cut apart in described pre-service, specifically for the described white license plate area identifying is carried out to binary conversion treatment; According to the saltus step of pixel value in the white license plate area after the binary conversion treatment identifying, the position assignment that white pixel is put corresponding pixel during with this saltus step in horizontal edge hum pattern is 1, and the position assignment of other pixels is 0;
With the step-length of setting, rotate horizontal marginal information figure, for postrotational each angle, in the angular range of setting, determine that every row comprises the number of white pixel point in the horizontal direction; For postrotational each angle, select the more setting line number of number of white pixel point, add up this setting line number white pixel point number with; According to the number of white pixel point and the anglec of rotation corresponding to maximal value, determine and the RA of this white license plate area this white license plate area carried out to rectification; Described white license plate area is carried out to binary conversion treatment; In white license plate area after binary conversion treatment, identify the transition times of every row pixel value, and determine the longest white line length and the longest black line length that every row comprises; According at least one in described transition times, the longest white line length and the longest black line length, with the respective distances threshold value of setting, determine the upper and lower border of this white license plate area.
6. device as claimed in claim 5, it is characterized in that, module is cut apart in described pre-service, length specifically for the number of pixel that this white line is comprised as this white line, according to the length of this white line, whether be greater than the long-pending of the second proportion threshold value of setting and the car plate width of setting, judge whether this white line is real white line; Or
For the white line of every row, the number of the pixel comprising according to this white line, and in this white line, whether the neighbor pixel of pixel same column is black pixel point, determines the value of identification parameter; Length long-pending that whether is greater than the first proportion threshold value and white line according to the value of identification parameter, judges whether this white line is real white line;
When determining that this white line is real white line, this white line is joined in the selection set of the longest white line length.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210177788.4A CN102693431B (en) | 2012-05-31 | 2012-05-31 | Method and device for identifying type of white number plate |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210177788.4A CN102693431B (en) | 2012-05-31 | 2012-05-31 | Method and device for identifying type of white number plate |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102693431A CN102693431A (en) | 2012-09-26 |
| CN102693431B true CN102693431B (en) | 2014-09-24 |
Family
ID=46858848
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210177788.4A Expired - Fee Related CN102693431B (en) | 2012-05-31 | 2012-05-31 | Method and device for identifying type of white number plate |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102693431B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103093245B (en) * | 2013-01-21 | 2016-01-20 | 信帧电子技术（北京）有限公司 | The method of marker lamp in video image |
| CN104200210B (en) * | 2014-08-12 | 2018-11-06 | 合肥工业大学 | A kind of registration number character dividing method based on component |
| CN104268509B (en) * | 2014-09-17 | 2018-05-29 | 深圳市捷顺科技实业股份有限公司 | The method and system of dump truck car plate detection |
| CN107368771B (en) * | 2016-05-12 | 2020-10-09 | 北京君正集成电路股份有限公司 | Method and device for positioning license plate |
| CN108108734B (en) * | 2016-11-24 | 2021-09-24 | 杭州海康威视数字技术股份有限公司 | License plate recognition method and device |
| CN108319951B (en) * | 2017-01-16 | 2020-10-02 | 杭州海康威视数字技术股份有限公司 | Method and device for recognizing characters in license plate |
| CN109657664B (en) * | 2017-10-12 | 2021-02-05 | 杭州海康威视数字技术股份有限公司 | License plate type identification method and device and electronic equipment |
| CN110135416B (en) * | 2018-02-09 | 2021-06-04 | 杭州海康威视数字技术股份有限公司 | License plate recognition method and device |
| CN110533019B (en) * | 2018-05-23 | 2022-08-12 | 杭州海康威视数字技术股份有限公司 | License plate positioning method and device and storage medium |
| CN109063702B (en) * | 2018-08-08 | 2022-05-17 | 北京旷视科技有限公司 | License plate recognition method, device, equipment and storage medium |
| CN110956169A (en) * | 2018-09-27 | 2020-04-03 | 杭州海康威视数字技术股份有限公司 | License plate recognition method and device and electronic equipment |
| CN109271993A (en) * | 2018-09-28 | 2019-01-25 | 徐常柏 | A kind of document scanning system, computer equipment and readable storage medium storing program for executing |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1825381A (en) * | 2005-02-24 | 2006-08-30 | 株式会社日立制作所 | Number plate recognition device, adjusting method of electric lens of number plate recognition device |
| CN101383003A (en) * | 2008-10-31 | 2009-03-11 | 江西赣粤高速公路股份有限公司 | Real-time precise recognition method for vehicle number board |
| CN101630361A (en) * | 2008-12-30 | 2010-01-20 | 北京邮电大学 | Plate number, body color and mark identification-based equipment and plate number, body color and mark identification-based method for identifying fake plate vehicles |
-
2012
- 2012-05-31 CN CN201210177788.4A patent/CN102693431B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1825381A (en) * | 2005-02-24 | 2006-08-30 | 株式会社日立制作所 | Number plate recognition device, adjusting method of electric lens of number plate recognition device |
| CN101383003A (en) * | 2008-10-31 | 2009-03-11 | 江西赣粤高速公路股份有限公司 | Real-time precise recognition method for vehicle number board |
| CN101630361A (en) * | 2008-12-30 | 2010-01-20 | 北京邮电大学 | Plate number, body color and mark identification-based equipment and plate number, body color and mark identification-based method for identifying fake plate vehicles |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102693431A (en) | 2012-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102693431B (en) | Method and device for identifying type of white number plate | |
| CN103116751B (en) | A kind of Method of Automatic Recognition for Character of Lcecse Plate | |
| CN102708356B (en) | Automatic license plate positioning and recognition method based on complex background | |
| CN104616009A (en) | Character cutting and recognizing method | |
| Moutarde et al. | Robust on-vehicle real-time visual detection of American and European speed limit signs, with a modular Traffic Signs Recognition system | |
| CN100414561C (en) | Vehicle plate extracting method based on skiagraphy and mathematical morphology | |
| CN101877050B (en) | Automatic extracting method for characters on license plate | |
| CN103093181A (en) | License plate image locating method and device | |
| CN102722711A (en) | License plate character sequence identification method and device in license plate type identification | |
| CN107301385A (en) | One kind blocks licence plate recognition method and device | |
| CN103136528B (en) | A kind of licence plate recognition method based on dual edge detection | |
| CN102915433B (en) | Character combination-based license plate positioning and identifying method | |
| CN101334836A (en) | License plate positioning method incorporating color, size and texture characteristic | |
| CN102722733A (en) | Identification method and device of license plate types | |
| CN103824452A (en) | Lightweight peccancy parking detection device based on full view vision | |
| CN102364496A (en) | Method and system for identifying automobile license plates automatically based on image analysis | |
| CN103345766A (en) | Method and device for identifying signal light | |
| CN102750530B (en) | Character recognition method and device | |
| CN111666805A (en) | Category tagging system for autonomous driving | |
| CN100385452C (en) | Registration number character dividing method | |
| CN102254152A (en) | License plate location method based on color change points and color density | |
| CN106845482A (en) | A kind of license plate locating method | |
| CN103544480A (en) | Vehicle color recognition method | |
| CN111295666A (en) | Lane line detection method, device, control equipment and storage medium | |
| CN105574485A (en) | Vehicle information identification method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| ASS | Succession or assignment of patent right |
Owner name: HE JIANGTAO Free format text: FORMER OWNER: XINZHENG ELECTRONIC TECHNOLOGY (BEIJING) CO., LTD. Effective date: 20131122 Owner name: LIU ZHONGXUAN TONG HUIYONG Effective date: 20131122 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20131122 Address after: 100085, A, block 6A, block 28, information road, Beijing, Haidian District Applicant after: He Jiangtao Applicant after: Liu Zhongxuan Applicant after: Tong Huiyong Address before: 100085, room 228, Section 2, building 1, building C, 3rd Street, Haidian District, Beijing Applicant before: Xinzheng Electronic Technology (Beijing) Co., Ltd. |
|
| ASS | Succession or assignment of patent right |
Free format text: FORMER OWNER: LIU ZHONGXUAN TONG HUIYONG Effective date: 20140418 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20140418 Address after: 100085, A, block 6A, block 28, information road, Beijing, Haidian District Applicant after: He Jiangtao Address before: 100085, A, block 6A, block 28, information road, Beijing, Haidian District Applicant before: He Jiangtao Applicant before: Liu Zhongxuan Applicant before: Tong Huiyong |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140924 Termination date: 20190531 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |