CN110135416A - A kind of licence plate recognition method and device - Google Patents

A kind of licence plate recognition method and device Download PDF

Info

Publication number
CN110135416A
CN110135416A CN201810133583.3A CN201810133583A CN110135416A CN 110135416 A CN110135416 A CN 110135416A CN 201810133583 A CN201810133583 A CN 201810133583A CN 110135416 A CN110135416 A CN 110135416A
Authority
CN
China
Prior art keywords
license plate
character
recognition result
default
plate structure
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201810133583.3A
Other languages
Chinese (zh)
Other versions
CN110135416B (en
Inventor
高增辉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Hikvision Digital Technology Co Ltd
Original Assignee
Hangzhou Hikvision Digital Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hangzhou Hikvision Digital Technology Co Ltd filed Critical Hangzhou Hikvision Digital Technology Co Ltd
Priority to CN201810133583.3A priority Critical patent/CN110135416B/en
Publication of CN110135416A publication Critical patent/CN110135416A/en
Application granted granted Critical
Publication of CN110135416B publication Critical patent/CN110135416B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/63—Scene text, e.g. street names
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
    • G06V20/625—License plates

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Traffic Control Systems (AREA)
  • Character Discrimination (AREA)
  • Character Input (AREA)

Abstract

The embodiment of the invention provides a kind of licence plate recognition method and device, licence plate recognition method includes: to carry out Car license recognition to images to be recognized, obtains the recognition result of license board information；It by the location information of character each in recognition result, is matched with the location information of each preset rectangle frame in default license plate structure model, determines the corresponding target license plate structural model of recognition result；According to the matching result of recognition result and target license plate structural model, judge whether the number of characters in recognition result is consistent with the number of characters of target license plate structural model, if NO, it is determined that lack character in recognition result or there are redundant characters.It can judge whether lose character or multiword symbol in recognition result using the embodiment of the present invention, improve Car license recognition accuracy rate.

Description

A kind of licence plate recognition method and device
Technical field
The present invention relates to intelligent transportation fields, more particularly to a kind of licence plate recognition method and device.
Background technique
License plate is the identification information of vehicle, is an important information for being different from other motor vehicles.License plate recognition technology It has been widely used in the scenes such as bayonet, parking lot and electronic police, to obtain the license board information of vehicle in scene.
The detailed process of existing licence plate recognition method are as follows: firstly, being positioned in the images to be recognized comprising license board information License plate area out；Then, by the image in analysis license plate area, the character in license plate area is split；Finally, to dividing The character cut is identified, license plate recognition result is obtained.
But that the background image of license plate is complicated, the image quality of images to be recognized is bad, tilting occurs in license plate etc. is severe Car license recognition environment in, the stability of above-mentioned licence plate recognition method is poor, susceptible to various factors, may cause vehicle Occur multiword symbol in board recognition result, lack the problems such as character, so that the accuracy rate of Car license recognition is lower.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of licence plate recognition method and device, with judge in recognition result whether Character or multiword symbol are lost, Car license recognition accuracy rate is improved.Specific technical solution is as follows:
In a first aspect, the embodiment of the invention provides a kind of licence plate recognition methods, which comprises
Car license recognition is carried out to images to be recognized, obtains the recognition result of license board information；
By preset rectangle frame each in the location information of character each in the recognition result, with default license plate structure model Location information is matched, and determines the corresponding target license plate structural model of the recognition result；
According to the matching result of the recognition result and the target license plate structural model, judge in the recognition result Whether number of characters is consistent with the number of characters of the target license plate structural model；
If NO, it is determined that lack character in the recognition result or there are redundant characters.
Second aspect, the embodiment of the invention provides a kind of license plate recognition device, described device includes:
Identification module obtains the recognition result of license board information for carrying out Car license recognition to images to be recognized；
Matching module, for will in the location information of character each in the recognition result, with default license plate structure model it is each The location information of preset rectangle frame is matched, and determines the corresponding target license plate structural model of the recognition result；
First judgment module is sentenced for the matching result according to the recognition result and the target license plate structural model Whether the number of characters to break in the recognition result is consistent with the number of characters of the target license plate structural model；
Determining module, if for the first judgment module judging result in the recognition result number of characters and institute The number of characters for stating target license plate structural model is inconsistent, it is determined that lacks character in the recognition result or there are redundant characters.
The third aspect, the embodiment of the invention provides a kind of electronic equipment, including processor and memory, wherein
The memory, for storing computer program；
The processor when for executing the program stored on the memory, realizes first party of the embodiment of the present invention Method and step described in face.
Licence plate recognition method and device provided in an embodiment of the present invention obtain firstly, carrying out Car license recognition to images to be recognized To the recognition result of license board information；It then, will be each in the location information of character each in recognition result, with default license plate structure model The location information of preset rectangle frame is matched, and determines the corresponding target license plate structural model of recognition result；Finally, according to knowledge The matching result of other result and target license plate structural model, judge number of characters in recognition result whether with target license plate structure mould The number of characters of type is consistent, if NO, it is determined that lacks character in recognition result or there are redundant characters.
In such manner, it is possible to judge with the presence or absence of character or multicharacter problem is lost in recognition result, further to identification As a result it is adjusted, keeps recognition result more accurate and reliable.Certainly, it implements any of the products of the present invention or method must be needed not necessarily To reach above all advantages simultaneously.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with It obtains other drawings based on these drawings.
Fig. 1 is a kind of flow diagram of the licence plate recognition method of the embodiment of the present invention；
Fig. 2 is a kind of schematic diagram of the license plate structure model of the embodiment of the present invention；
Fig. 3 is another schematic diagram of the license plate structure model of the embodiment of the present invention；
Fig. 4 is the recognition result of the embodiment of the present invention and a kind of flow diagram of default license plate structure Model Matching；
Fig. 5 is another flow diagram of the licence plate recognition method of the embodiment of the present invention；
Fig. 6 is that license plate structure model and a kind of matched schematic diagram of recognition result progress are preset in the embodiment of the present invention；
Fig. 7 is that license plate structure model and another matched schematic diagram of recognition result progress are preset in the embodiment of the present invention；
Fig. 8 is that license plate structure model and the matched another schematic diagram of recognition result progress are preset in the embodiment of the present invention；
Fig. 9 be the embodiment of the present invention in single layer license plate rectangle frame center away from schematic diagram；
Figure 10 be the double-deck license plate rectangle frame in the embodiment of the present invention center away from schematic diagram；
Figure 11 is another flow diagram of the licence plate recognition method of the embodiment of the present invention；
Figure 12 is the structural schematic diagram of the license plate recognition device of the embodiment of the present invention；
Figure 13 is the structural schematic diagram of the electronic equipment of the embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other Embodiment shall fall within the protection scope of the present invention.
The embodiment of the invention provides a kind of licence plate recognition methods, know referring to the license plate that Fig. 1, Fig. 1 are the embodiment of the present invention A kind of flow diagram of other method, includes the following steps:
Step 101, Car license recognition is carried out to images to be recognized, obtains the recognition result of license board information.
In this step, by carrying out Car license recognition to images to be recognized, the recognition result of license board information is got.The knowledge Character distribution line number in location information and license plate in other result comprising each character, each character that recognize in license plate etc. The detailed process of information, this step can refer to the relevant technologies, and details are not described herein.
It, can be first according to the confidence level and character of character each in recognition result after obtaining the recognition result of license board information The confidence level of position judges whether to meet output condition, if not satisfied, subsequent Matching and modification process is executed again, if satisfied, pressing Output form requires output license plate recognition result.It certainly, can also direct basis after obtaining the recognition result of license board information Recognition result executes subsequent Matching and modification process, without the confidence level and character position to character each in recognition result Confidence level is made whether to meet the judgement of output condition.
Step 102, by preset rectangle each in the location information of character each in recognition result, with default license plate structure model The location information of frame is matched, and determines the corresponding target license plate structural model of recognition result.
In this step, the location information of each character in the recognition result of license board information step 101 obtained, and it is default The location information of each preset rectangle frame is matched in license plate structure model, to determine and know from default license plate structure model The corresponding target license plate structural model of other result.
Wherein, presetting license plate structure model is at least one, specifically multiple license plate structures can be arranged according to application scenarios Model.Referring to figs. 2 and 3, Fig. 2 is a kind of schematic diagram of the license plate structure model of the embodiment of the present invention, and Fig. 3 is that the present invention is implemented Another schematic diagram of the license plate structure model of example.From Fig. 2 and Fig. 3 as it can be seen that in default license plate structure model, a rectangle frame Indicate a character, multiple rectangle frame embodied in combination character distribution situation of license plate.
Specifically, Fig. 2 and it is shown in Fig. 3 be Italian Civil license plate structure.License plate structure shown in Fig. 2 is 2-5 Type, the i.e. character string of license plate include two sections of character strings, and first segment character string is continuous two characters, and second segment character string is Continuous five characters.License plate structure shown in Fig. 3 be 2-3-2 type, i.e., the character string of license plate include three sections of character strings, first Section character string is continuous two characters, and second segment character string is continuous three characters, and third section character string is continuous two A character.
In this way, by the location information of character each in recognition result, with each preset rectangle frame in default license plate structure model Location information matched, wherein location information specifically includes the width of character or the width of rectangle frame and intercharacter Center away from or rectangle frame between center away to determine the matched target license plate structural model of recognition result, to further use mesh Mark license plate structure model is come the problem of judgement in recognition result with the presence or absence of multiword symbol or few character.
The frequency and probability difference, such as the actual traffic in Italy occurred due to different default license plate structure models In scene, utilization rate making for single layer 2-3-2 type license plate higher than the utilization rate of single layer 2-3-2 type license plate of single layer 2-5 type license plate It is higher than bilayer 2-3-2 type license plate with rate, i.e., the corresponding target license plate structural model of recognition result is single layer 2-5 under normal circumstances The probability of type license plate is higher than single layer 2-3-2 type license plate, the probability of single layer 2-3-2 type license plate is higher than the double-deck 2-3-2 type license plate, Therefore, in order to reduce matching workload, the matching priority of each type of default license plate structure model can be set, match excellent The setting of first grade can be configured according to actual utilization rate, i.e. the matching priority of single layer 2-5 type license plate structure model is high It is higher than in the matching priority of single layer 2-3-2 type license plate structure model, the matching priority of single layer 2-3-2 type license plate structure model The matching priority of the double-deck 2-3-2 type license plate structure model.Matching priority can also be configured according to scene, such as in city The utilization rate of single layer 2-3-2 type license plate is higher than bilayer 2-3-2 type license plate in city, then single layer 2-3-2 type license plate structure can be set The matching priority of model is higher than bilayer 2-3-2 type license plate structure model.
Optionally, step 102, specifically:
By each preset rectangle frame in the location information of character each in recognition result, with the first default license plate structure model Location information is matched, and the first default license plate structure model is to match priority most in default license plate structure model to be matched High default license plate structure model；
If successful match, it is determined that the first default license plate structure model is the corresponding target license plate structure mould of recognition result Type；
If matching is unsuccessful, the default vehicle of matching highest priority is rejected from default license plate structure model to be matched Board structural model, and returning to execution will be each in the location information of character each in recognition result, with the first default license plate structure model The location information of preset rectangle frame is matched.
For example, default license plate structure model include: the license plate of single layer 2-5 type structure, single layer 2-3-2 type structure license plate, The matching priority of the license plate of single layer 2-5 type structure is arranged most according to the utilization rate of license plate in the license plate of the double-deck 2-3-2 type structure High, the preferential high license plate for single layer 2-3-2 type structure of level of matching, matching priority are minimum for the double-deck 2-3-2 type structure License plate.Then recognition result and the process of default license plate structure Model Matching are as shown in Figure 4, comprising:
It step 401, will be each default in the location information of character each in recognition result, with single layer 2-5 type license plate structure model The location information of rectangle frame matched, judge whether successful match, it is no to then follow the steps if so, then follow the steps 402 403。
Step 402, determine that the corresponding target license plate structural model of recognition result is single layer 2-5 type license plate structure model.
It step 403, will be each pre- in the location information of character each in recognition result, with single layer 2-3-2 type license plate structure model If the location information of rectangle frame matched, judge whether successful match, it is no to then follow the steps if so, then follow the steps 404 405。
Step 404, determine that the corresponding target license plate structural model of recognition result is single layer 2-3-2 type license plate structure model.
It step 405, will be each pre- in the location information of character each in recognition result, with the double-deck 2-3-2 type license plate structure model If the location information of rectangle frame matched, judge whether successful match, it is no to then follow the steps if so, then follow the steps 406 407。
Step 406, determine that the corresponding target license plate structural model of recognition result is bilayer 2-3-2 type license plate structure model.
Step 407, it is not present and the matched default license plate structure model of recognition result.
Such as above-mentioned matching process, due to the utilization rate highest of single layer 2-5 type license plate, the target license plate structural model that recognizes A possibility that it is also maximum, if carrying out primary identification can identify that target license plate structural model is single layer 2-5 type license plate, Can no longer be matched with the license plate structure model of other structures, effectively reduce matching workload, also, be arranged condition in addition to Other than the utilization rate of the license plate of different structure, it can also be configured according to the probability for the license plate for occurring different structure in scene, Matched efficiency can be improved.
Certainly, for the license plate of the single layer 2-5 type structure mentioned in examples detailed above, single layer 2-3-2 type structure license plate, The default license plate structure model such as the license plate of the double-deck 2-3-2 type structure presets the location information of character each in recognition result with each License plate structure model synchronizes matching；Which matching result based on recognition result Yu each default license plate structure model again determines One default license plate structure model can be used as target license plate structural model.
As above-mentioned, the line number of character can also be double in multirow, such as examples detailed above other than for a line in license plate Layer 2-3-2 type license plate may include character distribution line number when carrying out Car license recognition to images to be recognized, in recognition result, The character line number that can exactly identify license plate is a line or multirow, and being distributed line number according to character in this way can be first to default vehicle Board structural model carries out primary screening, such as identifies that the character distribution line number of license plate is 2 rows, then when being matched, only needs Match character line number be 2 default license plate structure model, without again to the default license plate structure model of other line numbers into Row matching, so as to reduce matching workload, improve matching efficiency.
Optionally, before step 102, licence plate recognition method can also include:
It is distributed line number according to the character in recognition result, the character in screening character distribution line number and recognition result, which is distributed, goes The identical default license plate structure model of number.
Due in the recognition result that is identified to license plate, it is understood that there may be the case where multiword symbol or scarce character, therefore It when carrying out the matching of default license plate structure model, needs to be matched character by character, and then which default license plate structure mould is judged Type and license plate to be identified are most close, this is preset license plate structure model as target license plate structural model, carries out license plate more Accurately identification.
Optionally, the width that license plate structure model includes preset rectangle frame is preset；
Therefore, step 102, can specifically include:
For each default license plate structure model, every a line internal moment shape frame is preset in license plate structure model from head to tail according to this Put in order, calculate the overlapping widths between the width of rectangle frame and the width of recognition result middle finger location character；
According to the default confidence level of overlapping widths and specific bit character, the corresponding matching value of specific bit character is calculated；
The corresponding matching value of characters all in recognition result is added, recognition result and the default license plate structure model are calculated Matching total value；
It determines and matches the maximum default license plate structure model of total value in each default license plate structure model as recognition result matching Target license plate structural model.
Actual matching process can be the feelings that each rectangle frame is overlapping with specific bit character in default license plate structure model Condition is matched, and specific bit character corresponding to multiword symbol and scarce character can be different, but for same situation, most It is bigger that similar default license plate structure model matches the matching total value come, therefore, can be with according to the size of matching total value Determine that the matching maximum default license plate structure model of total value is target license plate structural model.If the character distribution line number of license plate is A line then matches the character of the row according to above-mentioned matching step one by one；If the character distribution line number of license plate is more Row, then can match the character of every a line according to above-mentioned matching step one by one respectively.
Step 103, according to the matching result of recognition result and target license plate structural model, judge the character in recognition result Whether number is consistent with the number of characters of target license plate structural model, if it has not, thening follow the steps 104.
In this step, according to the determining target license plate structural model of step 102 and recognition result and target license plate knot The matching result of structure model judges whether the number of characters in recognition result is consistent with the number of characters of target license plate structural model, if Unanimously, then illustrate the problem of there is no multiword symbol or few characters in recognition result, if inconsistent, illustrate to deposit in recognition result The problem of multiword accords with or lacks character.
Step 104, it determines in recognition result and lacks character or there are redundant characters.
In this step, when the number of characters for determining the number of characters in recognition result and target license plate structural model is inconsistent When, it can determine in recognition result and lack character or there are redundant characters, further recognition result can be adjusted.
Lack character in determining recognition result or there are redundant characters, it can be true according to the process for matching comparison character by character Determine which character lacked in recognition result, or there are which redundant characters, therefore, in order to which correct license plate can be restored, Optionally, after step 104, licence plate recognition method can also include:
If lacking character in recognition result, character lacking in addition；If there are redundant characters in recognition result, go Except the redundant character.
Due to lacking character and multiword symbol there may be after the character of tail position, before the first character or other positions, because This, correspondingly, lacking in addition the step of character, it can be with are as follows:
If lacking character after the tail position character of recognition result, identified from the position after the character of recognition result tail position Character, and the character that will identify that is added in recognition result；
If lacking character before the first character in recognition result, from the first character before position from identify character, And the character that will identify that is added in recognition result.
The step of removing redundant character, can be with are as follows:
If the tail position character of recognition result is redundant character, tail position character is removed；
If the first character in recognition result is redundant character, the first character is removed.
Similarly, if identifying that lacking character or redundant character appears in other intermediate positions, can be used corresponding knowledge After other method identifies, corresponding addition or removal.
As it can be seen that licence plate recognition method provided in an embodiment of the present invention, firstly, according to the matching pair of the recognition result of license board information The target license plate structural model answered, so by comparing the number of characters in recognition result whether the word with target license plate structural model Symbol number unanimously judges with the presence or absence of character or multicharacter problem is lost in recognition result, if number of characters and mesh in recognition result The number of characters for marking license plate structure model is inconsistent, then illustrates to exist in recognition result and lose character or multicharacter problem, in this way, just Further recognition result can be adjusted, keep recognition result more accurate and reliable.
In addition, after step 104, licence plate recognition method can also include: in another implementation
According to the alphabetical distribution rule of target license plate structural model and digital distribution rule, judge whether recognition result has Accidentally；
If it has, then modifying according to preset alteration ruler to recognition result.
In practical applications, since the shape of certain letter and numbers is more similar, such as number 1 and capital I, number Word 0 and upper case character O, number 8 and capitalization B etc., are easy to appear identification mistake, therefore, can be according to the corresponding mesh of recognition result The regularity of distribution of character in license plate results model is marked to solve the problems, such as this.
For example, first segment character string and third section character string are letter for the 2-3-2 type license plate of Italian Civil, Second segment character string is number, then, when appearance 1,0,8 etc. in the first segment character string and third section character string in recognition result When digital, which can be changed to the capitalizations such as I, O, B, equally, when the second segment character in recognition result When occurring the capitalizations such as I, O, B in string, the capitalizations such as I, O, B can be changed to the numbers such as 1,0,8.
In a kind of implementation, according to the alphabetical distribution rule of target license plate structural model and digital distribution rule, judgement The whether wrong step of recognition result may include:
According to the arrangement position of rectangle frame where presetting letter in target license plate structural model, judge in recognition result and pre- If whether the character that rectangle frame where alphabetical is in same arrangement position is number；
If it has, then will be in recognition result and rectangle frame where default letter according to preset character similarity table The character change of same arrangement position is the corresponding capitalization of number；
According to the arrangement position of rectangle frame where preset number in target license plate structural model, judge in recognition result and pre- If rectangle frame where digital is in whether the character of same arrangement position is capitalization；
If it has, then will be in recognition result and rectangle frame where preset number according to preset character similarity table The character change of same arrangement position is the corresponding number of capitalization.
It in specific implementation, can be alphabetical and default according to presetting in the matched target license plate structural model of recognition result The arrangement position of rectangle frame where data, it is whether wrong to judge the character in recognition result, for example, judge in recognition result and Rectangle frame where default letter is in whether the character of same arrangement position is the numbers such as 1,0,8, if it has, then the character is repaired The capitalizations such as I, O, B are changed to, for another example, judge to be in same arrangement position with rectangle frame where preset number in recognition result Character whether be the capitalizations such as I, O, B, if it has, then by the character change be the numbers such as 1,0,8, further improve knowledge The accuracy of other result.
Due to that in general, before the matching for carrying out default license plate structure model, can not know that license plate is to lack word Symbol or multiword accord with, and therefore, it is necessary to carry out the matching of character one by one according to different situations, can determine target license plate structure Model.With reference to Fig. 5, Fig. 5 is another flow diagram of the licence plate recognition method of the embodiment of the present invention, is included the following steps:
Step 501, Car license recognition is carried out to images to be recognized, obtains the recognition result of license board information.
The detailed process and technical effect of this step can refer to the step 101 in licence plate recognition method shown in FIG. 1, This is repeated no more.
Step 502, for each default license plate structure model, every a line internal moment shape frame in license plate structure model is preset according to this Putting in order from head to tail calculates the first overlapping widths and rectangle frame of the width of rectangle frame and the width of the first character Width and the second character width the second overlapping widths.
Default license plate structure model includes the width of preset rectangle frame.First character is to be in and preset in recognition result The character of the same arrangement position of rectangle frame in license plate structure model, for example, being located at primary square in default license plate structure model Shape frame, the first corresponding character are to be located at primary character in recognition result, preset and are located at the in license plate structure model Two rectangle frames, the first corresponding character are to be located at deputy character in recognition result, and so on, each rectangle Frame is corresponding with first character.Second character is to be in recognition result with rectangle frame in default license plate structure model with arrangement A character is moved to right at position, for example, being located at primary rectangle frame, the second corresponding character in default license plate structure model It is located at deputy character as in recognition result, presets and be located at deputy rectangle frame in license plate structure model, it is corresponding Second character is the character for being located at third position in recognition result, and so on, in addition to being located at last rectangle frame, other Each rectangle frame is corresponding with second character.
In a kind of implementation, the width of each preset rectangle frame can be identical in license plate structure model, specifically can root According to the width of character in actual license plate, the average value of character width is calculated, and using the average value as in license plate structure model The width of preset rectangle frame.
It should be noted that since the width of alphabetical I and number 1 are significantly less than other characters, in order to avoid influencing character The applicability of the average value of width can reject the alphabetical I in license plate with number 1 and then according to the width of character in license plate Degree calculates the average value of character width.
In this step, by the width of each preset rectangle frame in preset license plate model, one by one with license board information The width of the first character is compared in recognition result, the first overlapping widths is calculated, with further true according to the first overlapping widths Determine the corresponding target license plate structural model of recognition result.
Specifically, Fig. 6 can be referred to, Fig. 6 is that license plate structure model and recognition result progress are preset in the embodiment of the present invention A kind of matched schematic diagram.In Fig. 6, by the width of preset rectangle frame each in default license plate structure model, with recognition result The width of the first character, is compared one by one respectively in (DD 12345), corresponding one first overlapping of final each rectangle frame Width.
It should be noted that license plate structure model further include center between rectangle frame away from.In this way, in license plate structure model The first rectangle frame and recognition result in the first alignment, license plate structure model " translation cover " is arrived into recognition result On, make each character in recognition result and each rectangle frame " Chong Die " in license plate structure model, obtains the first overlapping widths.
In this step, by the width of each rectangle frame in preset license plate model, one by one with the recognition result of license board information In the width of the second character be compared, the second overlapping widths are calculated, to further determine that identification knot according to the second overlapping widths The corresponding target license plate structural model of fruit.
Specifically, Fig. 7 can be referred to, Fig. 7 is that license plate structure model and recognition result progress are preset in the embodiment of the present invention Another matched schematic diagram.In Fig. 7, first row is default license plate structure model, and second row is wrong recognition result (XDD 12345), third row are correct recognition result (DD 12345).Specifically, by being preset in default license plate structure model Each rectangle frame width, be compared one by one with the width of the second character in wrong recognition result, so that each rectangle frame is equal Corresponding second overlapping widths.
It should be noted that with the first rectangle frame in license plate structure model and the second character in recognition result Alignment, by each character in addition to initial character on license plate structure model " translation cover " to recognition result, made in recognition result with Each rectangle frame " overlapping " in license plate structure model, obtains the second overlapping widths of each second character and rectangle frame.
Step 503, according to the default confidence level of the first overlapping widths and the first character, the first character corresponding the is calculated One matching value；According to the default confidence level of the second overlapping widths and the second character, corresponding second matching of the second character is calculated Value.
In this step, the first overlapping widths step 502 being calculated, the ratio between with the width of the rectangle frame, multiplied by The default confidence level of the character can calculate the corresponding matching value of the character, to further determine that identification knot according to the matching value The corresponding target license plate structural model of fruit；Wherein, in recognition result the default confidence level of each character be it is known, can specifically join The relevant technologies are examined, details are not described herein.
Specifically, shown in the calculation formula of matching value such as formula (1).
In formula (1), crd is the corresponding matching value of a character；Match_len is character overlap width,For the width matching result for corresponding to rectangle frame in the character and preset license plate model, wherein char_crd is The default confidence level of the character；Char_w is the width that preset license plate model corresponds to character rectangle frame.
For example, the width for being located at the rectangle frame of license plate structure model first place is 1 centimetre, the character positioned at recognition result first place It is matched with the rectangle frame, and character width is 0.8 centimetre, meanwhile, which spatially includes the character of the first place completely, So, character overlap width is 0.8 centimetre, and the corresponding width matching result of the character of the first place is 0.8/1=0.8.
In this step, the second overlapping widths step 502 being calculated, the ratio between with the width of the rectangle frame, multiplied by The default confidence level of the character can calculate the corresponding matching value of the character, to further determine that identification knot according to the matching value The corresponding target license plate structural model of fruit.
Step 504, corresponding first matching value of characters all in recognition result is added, calculates recognition result and this is default First matching total value of license plate structure model；Corresponding second matching value of characters all in recognition result is added, identification is calculated As a result the second of license plate structure model is preset with this match total value.
In this step, the corresponding matching value of characters all in recognition result is added, obtains recognition result and license plate knot The matching total value of structure model, same recognition result matching total value more than one corresponding with same license plate structure model.
Step 505, it for each default license plate structure model, is preset according to this and removes first place in license plate structure model in every a line Rectangle frame putting in order from head to tail other than rectangle frame, calculates the third weight of the width of rectangle frame and the width of third character Folded width.
Third character be in recognition result in rectangle frame in default license plate structure model with moving to left one at arrangement position The character of position, for example, being located at deputy rectangle frame in default license plate structure model, corresponding third character is to identify knot It is located at primary character in fruit, presets the rectangle frame for being located at third position in license plate structure model, corresponding third character is To be located at deputy character in recognition result, and so on, in addition to being located at primary rectangle frame, other each rectangle frames pair There should be a third character.
In this step, by the width of each rectangle frame in addition to the first rectangle frame in preset license plate model, one by one It is compared with the width of third character in the recognition result of license board information, third overlapping widths is calculated, to be overlapped according to third Width further determines that the corresponding target license plate structural model of recognition result.
Specifically, Fig. 8 can be referred to, Fig. 8 is that license plate structure model and recognition result progress are preset in the embodiment of the present invention Matched another kind schematic diagram.In fig. 8, first row is default license plate structure model, and second row is wrong recognition result (D 12345), third row is correct recognition result (DD 12345).Specifically, the first rectangle will be removed in default license plate structure model The width of each rectangle frame other than frame, the width with third character in wrong recognition result, is compared one by one, final except head Each rectangle frame other than the rectangle frame of position corresponds to a third overlapping widths.
It should be noted that with the second rectangle frame in license plate structure model and the first character in recognition result Alignment makes each character and license plate structure model in recognition result on license plate structure model " translation covers " to recognition result In except the first outer rectangular frame each rectangle frame " overlapping ", obtain the third overlapping widths of each third character and rectangle frame.
Step 506, according to the default confidence level of third overlapping widths and third character, third character corresponding the is calculated Three matching values.
In this step, third overlapping widths step 505 being calculated, the ratio between with the width of the rectangle frame, multiplied by The default confidence level of the character can calculate the corresponding matching value of the character, to further determine that identification knot according to the matching value The corresponding target license plate structural model of fruit.
Step 507, the corresponding third matching value of characters all in recognition result is added, calculates recognition result and this is default The third of license plate structure model matches total value.
In this step, the corresponding matching value of characters all in recognition result is added, obtains recognition result and license plate knot The matching total value of structure model.
It should be noted that step 502 is to step 504, the execution sequence of step 505 to step 507 can be according to reality Situation is arranged, and certainly, step 502 to step 504, step 505 to step 507 also may be performed simultaneously.
Step 508, the first matching total value, the second matching total value and the third matching of each default license plate structure model are determined In total value, the maximum value of all values.
Step 509, by the corresponding default license plate structure model of maximum value, it is determined as the matched target license plate knot of recognition result Structure model.If target license plate structural model is the corresponding default license plate structure model of the first matching total value, step 510 is executed；If Target license plate structural model is that the second matching total value or third match the corresponding default license plate structure model of total value, determines identification knot The number of characters of number of characters and target license plate structural model in fruit is inconsistent, if target license plate structural model is the second matching total value Corresponding default license plate structure model, thens follow the steps 513；If target license plate structural model is third matching, total value is corresponding pre- If license plate structure model, thens follow the steps 514.
In this step, by the first of calculated each default license plate structure model the matching total value, the second matching total value with And the corresponding default license plate structure model of maximum value in all values composed by third matching total value, it is determined as the recognition result Target license plate structural model the most matched；Also, according to the type of maximum value, to determine first place or the tail in recognition result Position is with the presence or absence of redundant character or lacks character.Specifically, if maximum value is the first matching total value, the tail in recognition result There may be lack character or tail position character for redundant character after the character of position；If maximum value is the second matching total value, The first character in recognition result is redundant character；If maximum value is that third matches total value, the first word in recognition result Lack character before symbol.
For example, a recognition result is matched with three default license plate structure models, the recognition result and first it is pre- If the corresponding first, second, and third matching total value of license plate structure model is respectively 4,5 and 3, with second default license plate structure The corresponding first, second, and third matching total value of model is respectively 3,5 and 6, corresponding with the default license plate structure model of third First, second, and third matching total value is respectively 18,16 and 17, then, in all matching total values, the first matching total value 18 is right The default license plate structure model of third answered is the matched target license plate structural model of the recognition result.
In the specific implementation process, the center between the rectangle frame in license plate structure model is away from including phase in same section of character string The first center between adjacent two rectangle frames away from and two sections of character strings between two neighboring rectangle frame the second center away from such as Shown in Fig. 9, Fig. 9 be the embodiment of the present invention in single layer license plate rectangle frame center away from schematic diagram, in Fig. 9, char_dis is First center away from, seg_dis be the second center away from, the second center away from namely character field spacing.It should be understood that different license plates The second center for including in type away from number it is different, for example, including 2 the second centers in 2-3-2 type license plate away from 2-5 type vehicle Include 1 the second center in board away from.Again for example shown in Figure 10, Figure 10 is the center of the double-deck license plate rectangle frame in the embodiment of the present invention Away from schematic diagram, in Figure 10, char_dis is that the first center of bottom away from, seg_dis is the second center of bottom away from hori_ Seg_dis is distance of the top layer left side edge away from bottom left side edge, and vert_seg_dis is top layer lower edge and bottom top The distance of edge.
In a kind of implementation, the width of rectangle frame can also be set, the first center between rectangle frame away from rectangle frame Proportional region of second center away from the width with rectangle frame between the proportional region and rectangle frame of width, in this way, can basis The width of rectangle frame, calculate the first center between rectangle frame away from the second center away from.
Although structural model is identical, such as be 2-5 type license plate due to the license plate of country variant, the second center away from It is variant, in order to reduce storage default license plate structure model number, such as only store a 2-5 type license plate structure model, It in this way can be by adjusting the second different centers away from reaching the matched effect of multi-model.Specifically, with 2-5 type shown in Fig. 2 For license plate, can preset the second center away from numberical range, executing step 502, before step 505, gradually adjusting Second center away from value, adjust every time, the first matching value, the second matching value and the can be generated in step 503, step 506 Three matching values, corresponding, new the first matching total value, the second matching total value and third also can be generated in step 504, step 507 Total value is matched, in this way, step 508, step 509 are just needed from multiple first matchings total values, multiple second matching total values and multiple Third, which matches, determines maximum value in total value, and the corresponding license plate structure model of maximum value is determined as the matched target of recognition result License plate structure model.Include multiple second centers for other away from license plate type, can according to actual needs, successively or together When adjust multiple second centers away from value.
For example, default second center away from numberical range be 20~30 pixels, the second center of setting away from initial value be 20 pixels, at this point, one first matching total value can be calculated by executing step 502 to step 507, one second Total value is matched with total value and a third；Next, being 21 pixels by the second center distance regulation, at this point, by executing step 502 can be calculated one first matching total value to step 507 again, and one second matching total value and a third matching are total Value；It similarly, then by the second center distance regulation is 22 pixels, 23 pixels, 24 pixels, 25 pixels, 26 pixels, 27 In the case where pixel, 28 pixels, 29 pixels and 30 pixels, step 502 is executed respectively to step 507.In this way, in total may be used To obtain 11 first matching total values, 11 second matching total values and 11 thirds match total value.Finally, can be from upper 11 first matching total values are stated, are maximized in 11 second matching total values and 11 third matching total values, and will most It is worth corresponding license plate structure model greatly and is determined as the matched target license plate structural model of recognition result.
It should be noted that step 502 to step 509 can be in filtering out character distribution line number and recognition result After the identical default license plate structure model of character distribution function, target license plate structural model therefrom is determined by matching again, this Sample can reduce matched workload, improve matching efficiency.
Step 510, by comparing the word in the quantity and recognition result of preset rectangle frame in target license plate structural model The quantity of symbol determines whether the number of characters in recognition result is consistent with the number of characters of target license plate structural model；If target license plate The quantity of preset rectangle frame is greater than the quantity of the character in recognition result in structural model, thens follow the steps 511；If target carriage The quantity of preset rectangle frame is less than the quantity of the character in recognition result in board structural model, thens follow the steps 512.
It in this step, can if target license plate structural model is the corresponding default license plate structure model of the first matching total value Come with the quantity by comparing the character in the quantity and recognition result of preset rectangle frame in target license plate structural model into one Step judges that the tail position character in recognition result whether there is later and lacks whether character or tail position character are redundant character；If The quantity of character in target license plate structural model in the quantity and recognition result of preset rectangle frame is inconsistent, then recognition result In tail position character after have that lack character or tail position character be redundant character.
Step 511, it determines and lacks character in recognition result.
In this step, when the number that maximum value is preset rectangle frame in the first matching total value and target license plate structural model When amount is greater than the quantity of the character in recognition result, it can determine and lack character in recognition result, for example, after the character of tail position Lack character or lacks character in intermediate position.With specific reference to each matching value, determination is which position has lacked character.This Sample, so that it may supplement the character lacked in the corresponding position of recognition result, improve the accuracy of recognition result.
Step 512, determine that there are redundant characters in recognition result.
In this step, when the number that maximum value is preset rectangle frame in the first matching total value and target license plate structural model Amount be less than recognition result in character quantity when, can determine that there are redundant characters in recognition result, for example, at tail position it is more Several characters.In this manner it is possible to remove the character of corresponding position in recognition result, the accuracy of recognition result is improved.
Step 513, determine that the first character in recognition result is redundant character.
In this step, when maximum value is the second matching total value, it can determine that the first character of recognition result is extra Character improves the accuracy of recognition result in this manner it is possible to remove the first character of recognition result.
Step 514, character is lacked before determining the first character in recognition result.
In this step, when maximum value is that third matches total value, before can determining the first character in recognition result Lack character, in this manner it is possible to supplement the character lacked before the first character, improves the preparatory of recognition result.
In the specific implementation process, when the number of characters and target license plate by step 509 into the determining recognition result of step 514 The number of characters of structural model is inconsistent, and has determined and lack in recognition result or there are when the specific location of redundant character, Following steps can be executed:
If lacking character in recognition result, character lacking in addition；If there are redundant characters in recognition result, go Except redundant character.
Specifically, being needed pair when the number of characters of number of characters and target license plate structural model in recognition result is inconsistent Recognition result is adjusted, and removes redundant character or addition lacks character.
In one implementation, above-mentioned steps can specifically include as follows:
If lacking character after the tail position character of recognition result, identified from the position after the character of recognition result tail position Character, and the character that will identify that is added in recognition result；
If lacking character before the first character in recognition result, from the first character before position from identify character, And the character that will identify that is added in recognition result；
If the tail position character of recognition result is redundant character, tail position character is removed；
If the first character in recognition result is redundant character, the first character is removed.
Specifically, when lacking character after the tail position character of recognition result, it can be from the position after the character of tail position Start to identify character, and the character that will identify that is added to after the tail position character in recognition result；When the tail in recognition result When position character is redundant character, tail position character can be removed, the further adjustment to recognition result is realized, improves recognition result Accuracy.When lacking character before the first character of recognition result, character can be identified from the position before the first character, And the character that will identify that is added to before the first character in recognition result；When the first character in recognition result is extra word Fu Shi can remove the first character, realize the further adjustment to recognition result, improve the accuracy of recognition result.Certainly, such as The middle position of fruit recognition result lacks character can be adjusted when perhaps multiword accords with by the processing added or removed.
When carrying out the Matching and modification of recognition result and default license plate structure model, in order to improve matched efficiency, simplify Process is matched, position first can be corresponded to recognition result and default license plate structure model and carry out overlapping matching, if meeting matching Condition can then directly determine target license plate structural model, if being unsatisfactory for matching condition, then carry out moving to left one or move to right One overlapping matching.It is another flow diagram of the licence plate recognition method of the embodiment of the present invention, packet with reference to Figure 11, Figure 11 Include following steps:
Step 1101, Car license recognition is carried out to images to be recognized, obtains the recognition result of license board information.
Step 1102, for each default license plate structure model, every a line internal moment shape in license plate structure model is preset according to this Frame putting in order from head to tail calculates the first overlapping widths of the width of rectangle frame and the width of the first character.
Step 1103, according to the default confidence level of the first overlapping widths and the first character, it is corresponding to calculate the first character First matching value.
Step 1104, corresponding first matching value of characters all in recognition result is added, it is pre- with this calculates recognition result If the first matching total value of license plate structure model.
Step 1105, judge whether maximum first matching total value is greater than the first default threshold in each default license plate structure model Value, it is no to then follow the steps 1107 if so then execute step 1106.
After obtaining the first matching total value, if the first matching total value is sufficiently large, illustrate corresponding default license plate structure Model is that the possibility of target license plate structural model is very big, therefore this directly can be preset license plate structure model and be determined as target carriage Otherwise board structural model illustrates that a possibility that corresponding default license plate structure model is target license plate structural model is smaller, then needs Calculate the second matching total value.
For example, above-mentioned default license plate structure model includes: the license plate of single layer 2-5 type structure, single layer 2-3-2 type structure The license plate of license plate, bilayer 2-3-2 type structure is obtained by matching the calculating of total value to first with each default license plate structure model The first matching total value to the license plate of single layer 2-3-2 type structure is maximum, and is greater than the first preset threshold, it is determined that single layer 2-3-2 The license plate of type structure is target license plate structural model.If the first matching total value of the license plate of single layer 2-3-2 type structure is maximum, but Be the license plate of single layer 2-3-2 type structure the first matching total value less than the first preset threshold, then carry out the second matching total value or The calculating of third matching total value.
Step 1106, determine that the first matching maximum default license plate structure model of total value is in each default license plate structure model The matched target license plate structural model of recognition result.
Step 1107, every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, counted Calculate the second overlapping widths of the width of rectangle frame and the width of the second character.
Step 1108, according to the default confidence level of the second overlapping widths and the second character, it is corresponding to calculate the second character Second matching value.
Step 1109, corresponding second matching value of characters all in recognition result is added, it is pre- with this calculates recognition result If the second matching total value of license plate structure model.
Step 1110, judge whether maximum second matching total value is greater than the second default threshold in each default license plate structure model Value, it is no to then follow the steps 1112 if so then execute step 1111.
After obtaining the second matching total value, if the second matching total value is sufficiently large, illustrate corresponding default license plate structure Model is that the possibility of target license plate structural model is very big, therefore this directly can be preset license plate structure model and be determined as target carriage Otherwise board structural model illustrates that a possibility that corresponding default license plate structure model is target license plate structural model is smaller, then needs Calculate third matching total value.
For example, above-mentioned default license plate structure model includes: the license plate of single layer 2-5 type structure, single layer 2-3-2 type structure The license plate of license plate, bilayer 2-3-2 type structure is obtained by matching the calculating of total value to second with each default license plate structure model The second matching total value to the license plate of single layer 2-5 type structure is maximum, and is greater than the second preset threshold, it is determined that single layer 2-5 type knot The license plate of structure is target license plate structural model.If the second matching total value of the license plate of single layer 2-5 type structure is maximum, but single layer Second matching total value of the license plate of 2-5 type structure then carries out the calculating of third matching total value less than the second preset threshold.
Step 1111, determine that the second matching maximum default license plate structure model of total value is in each default license plate structure model The matched target license plate structural model of recognition result.
Step 1112, the rectangle frame in license plate structure model in every a line in addition to the first rectangle frame is preset from head according to this To putting in order for tail, the third overlapping widths of the width of rectangle frame and the width of third character are calculated.
Step 1113, according to the default confidence level of third overlapping widths and third character, it is corresponding to calculate third character Third matching value.
Step 1114, the corresponding third matching value of characters all in recognition result is added, it is pre- with this calculates recognition result If the third of license plate structure model matches total value.
Step 1115, judge whether maximum third matching total value is greater than the default threshold of third in each default license plate structure model Value, if so, determining that the third matching maximum default license plate structure model of total value is identification knot in each default license plate structure model The matched target license plate structural model of fruit.
After obtaining third matching total value, if third matching total value is sufficiently large, illustrate corresponding default license plate structure Model is that the possibility of target license plate structural model is very big, therefore this directly can be preset license plate structure model and be determined as target carriage Otherwise board structural model illustrates that a possibility that corresponding default license plate structure model is target license plate structural model is smaller, then says It is not suitable as the model of target license plate structural model in default license plate structure model before improving eyesight, needs to update default license plate Structural model re-starts matching.
For example, above-mentioned default license plate structure model includes: the license plate of single layer 2-5 type structure, single layer 2-3-2 type structure The license plate of license plate, bilayer 2-3-2 type structure is obtained by the calculating to total value is matched with the third of each default license plate structure model Third matching total value to the license plate of the double-deck 2-3-2 type structure is maximum, and is greater than third predetermined threshold value, it is determined that the double-deck 2-3-2 The license plate of type structure is target license plate structural model.If the third matching total value of the license plate of bilayer 2-3-2 type structure is maximum, but Be the license plate of the double-deck 2-3-2 type structure third matching total value be less than third predetermined threshold value, then need to update default license plate structure Model re-starts matching.
Wherein, the setting of the first preset threshold, the second preset threshold and third predetermined threshold value can be identical, can also not Together, also, the first preset threshold, the second preset threshold and third predetermined threshold value can be sent out based on the events such as character, multiword symbol are lacked Raw probability, and to the tolerance correlation that different event occurs.
Certainly, if the maximum value only gived in the first matching total value above is unsatisfactory for being greater than the first preset threshold, First carry out the process of the calculating of the second matching total value, the calculating for carrying out third matching total value again；In practical application example, may be used also To be if the maximum value in the first matching total value is unsatisfactory for being greater than the first preset threshold, the meter of third matching total value is first carried out Calculation, the calculating for carrying out the second matching total value again.Here it is not specifically limited.
If target license plate structural model is the corresponding default license plate structure model of the first matching total value, step 1116 is executed； If target license plate structural model is that the second matching total value or third match the corresponding default license plate structure model of total value, identification is determined As a result the number of characters of number of characters and target license plate structural model in is inconsistent, if target license plate structural model is that the second matching is total It is worth corresponding default license plate structure model, thens follow the steps 1119；If target license plate structural model is third, matching total value is corresponding Default license plate structure model, then follow the steps 1120.
Step 1116, by comparing the word in the quantity and recognition result of preset rectangle frame in target license plate structural model The quantity of symbol determines whether the number of characters in recognition result is consistent with the number of characters of target license plate structural model；If target license plate The quantity of preset rectangle frame is greater than the quantity of the character in recognition result in structural model, thens follow the steps 1117；If target The quantity of preset rectangle frame is less than the quantity of the character in recognition result in license plate structure model, thens follow the steps 1118.
Step 1117, it determines and lacks character in recognition result.
Step 1118, determine that there are redundant characters in recognition result.
Step 1119, determine that the first character in recognition result is redundant character.
Step 1120, character is lacked before determining the first character in recognition result.
Other steps are same or similar with embodiment illustrated in fig. 5, no longer repeat one by one here.
In above-mentioned embodiment illustrated in fig. 11, it can also be that presetting license plate model for some carries out step 1101 to step 1120 process, such as the license plate of single layer 2-5 type structure, the first matching total value is first calculated, if it is total to meet the first matching Value is greater than the condition of the first preset threshold, it is determined that the license plate of single layer 2-5 type structure is target license plate structural model；If discontented Foot first matches the condition that total value is greater than the first preset threshold, then is directed to the license plate of single layer 2-5 type structure, and it is total to calculate the second matching Value or third match total value, further judge whether the license plate of single layer 2-5 type structure can be used as target license plate structural model, If the license plate of single layer 2-5 type structure cannot be used as target license plate structural model, then be directed to the vehicle of single layer 2-3-2 type structure The default license plate structure model such as license plate of board, bilayer 2-3-2 type structure, carries out above-mentioned Matching and modification process, until determining mesh Mark license plate structure model.
Due in mutually isostructural license plate, there is also a large amount of license plate model, these models although structure having the same, Such as be the license plate of single layer 2-5 type structure, but character width in each model is different and/or the first center away from it is different and/ Or second center away from difference, then constitute the identical different license plate models of structure, the Matching and modification process of these models is also the same Suitable for embodiment shown in above-mentioned Fig. 5 and Figure 11.I will not elaborate.
The embodiment of the invention also provides a kind of license plate recognition devices, are the vehicle of the embodiment of the present invention with reference to Figure 12, Figure 12 The structural schematic diagram of board identification device, as shown in figure 12, license plate recognition device includes:
Identification module 1201 obtains the recognition result of license board information for carrying out Car license recognition to images to be recognized；
Matching module 1202, for will in the location information of character each in recognition result, with default license plate structure model it is each The location information of preset rectangle frame is matched, and determines the corresponding target license plate structural model of recognition result；
First judgment module 1203 judges to know for the matching result according to recognition result and target license plate structural model Whether the number of characters in other result is consistent with the number of characters of target license plate structural model；
Determining module 1204, if the judging result for first judgment module 1203 is number of characters and mesh in recognition result The number of characters for marking license plate structure model is inconsistent, it is determined that lacks character in recognition result or there are redundant characters.
Optionally, matching module 1202 specifically can be used for:
By each preset rectangle frame in the location information of character each in recognition result, with the first default license plate structure model Location information is matched, and the first default license plate structure model is to match priority most in default license plate structure model to be matched High default license plate structure model；
If successful match, it is determined that the first default license plate structure model is the corresponding target license plate structure mould of recognition result Type；
If matching is unsuccessful, the default vehicle of matching highest priority is rejected from default license plate structure model to be matched Board structural model, and returning to execution will be each in the location information of character each in recognition result, with the first default license plate structure model The location information of preset rectangle frame is matched.
Optionally, license plate recognition device can also include:
Screening module, for being distributed line number, screening character distribution line number and recognition result according to the character in recognition result In the identical default license plate structure model of character distribution line number.
Optionally, the width that license plate structure model includes preset rectangle frame is preset；
Matching module 1202, specifically can be also used for:
For each default license plate structure model, every a line internal moment shape frame is preset in license plate structure model from head to tail according to this Put in order, calculate the overlapping widths between the width of rectangle frame and the width of recognition result middle finger location character；
According to the default confidence level of overlapping widths and specific bit character, the corresponding matching value of specific bit character is calculated；
The corresponding matching value of characters all in recognition result is added, recognition result and the default license plate structure model are calculated Matching total value；
It determines and matches the maximum default license plate structure model of total value in each default license plate structure model as recognition result matching Target license plate structural model.
Optionally, specific bit character includes: and is in and the same row of rectangle frame in default license plate structure model in recognition result In first character of column position, recognition result in rectangle frame in default license plate structure model with moving to right one at arrangement position The second character and recognition result in rectangle frame in default license plate structure model with moving to left one at arrangement position Third character；
Matching module 1202, specifically can be also used for:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates rectangle frame The second of width and the width of the first overlapping widths of the width of the first character and rectangle frame and the width of the second character is Chong Die Width；
Row of the rectangle frame from head to tail in license plate structure model in every a line in addition to the first rectangle frame is preset according to this Column sequence, calculates the third overlapping widths of the width of rectangle frame and the width of third character；
According to the default confidence level of the first overlapping widths and the first character, corresponding first matching of the first character is calculated Value；
According to the default confidence level of the second overlapping widths and the second character, corresponding second matching of the second character is calculated Value；
According to the default confidence level of third overlapping widths and third character, the corresponding third matching of third character is calculated Value；
Corresponding first matching value of characters all in recognition result is added, recognition result and the default license plate structure are calculated First matching total value of model；
Corresponding second matching value of characters all in recognition result is added, recognition result and the default license plate structure are calculated Second matching total value of model；
The corresponding third matching value of characters all in recognition result is added, recognition result and the default license plate structure are calculated The third of model matches total value；
It determines in the first matching total value, the second matching total value and third matching total value of each default license plate structure model, The maximum value of all values；
By the corresponding default license plate structure model of maximum value, it is determined as the matched target license plate structural model of recognition result.
Optionally, specific bit character includes: and is in and the same row of rectangle frame in default license plate structure model in recognition result In first character of column position, recognition result in rectangle frame in default license plate structure model with moving to right one at arrangement position The second character and recognition result in rectangle frame in default license plate structure model with moving to left one at arrangement position Third character；
Matching module 1202, specifically can be also used for:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates rectangle frame First overlapping widths of the width of width and the first character；
According to the default confidence level of the first overlapping widths and the first character, corresponding first matching of the first character is calculated Value；
Corresponding first matching value of characters all in recognition result is added, recognition result and the default license plate structure are calculated First matching total value of model；
Judge whether maximum first matching total value is greater than the first preset threshold in each default license plate structure model；
If so, determining that the first matching maximum default license plate structure model of total value is to know in each default license plate structure model The other matched target license plate structural model of result；
If it is not, then executing:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates rectangle frame Second overlapping widths of the width of width and the second character；
According to the default confidence level of the second overlapping widths and the second character, corresponding second matching of the second character is calculated Value；
Corresponding second matching value of characters all in recognition result is added, recognition result and the default license plate structure are calculated Second matching total value of model；
Judge whether maximum second matching total value is greater than the second preset threshold in each default license plate structure model；
If so, determining that the second matching maximum default license plate structure model of total value is to know in each default license plate structure model The other matched target license plate structural model of result；
If it is not, then executing:
Row of the rectangle frame from head to tail in license plate structure model in every a line in addition to the first rectangle frame is preset according to this Column sequence, calculates the third overlapping widths of the width of rectangle frame and the width of third character；
According to the default confidence level of third overlapping widths and third character, the corresponding third matching of third character is calculated Value；
The corresponding third matching value of characters all in recognition result is added, recognition result and the default license plate structure are calculated The third of model matches total value；
Judge whether maximum third matching total value is greater than third predetermined threshold value in each default license plate structure model；
If so, determining that the third matching maximum default license plate structure model of total value is to know in each default license plate structure model The other matched target license plate structural model of result.
Optionally, first judgment module 1203 specifically can be used for:
If target license plate structural model is the corresponding default license plate structure model of the first matching total value, by comparing target The quantity of character in license plate structure model in the quantity and recognition result of preset rectangle frame determines the character in recognition result Whether number is consistent with the number of characters of target license plate structural model；
If target license plate structural model is that the second matching total value or third match the corresponding default license plate structure model of total value, Then determine that the number of characters of the number of characters and target license plate structural model in recognition result is inconsistent.
Optionally, determining module 1204 specifically can be used for:
If target license plate structural model is the corresponding default license plate structure model of the first matching total value and target license plate structure The quantity of preset rectangle frame is greater than the quantity of the character in recognition result in model, it is determined that lacks character in recognition result；
If target license plate structural model is the corresponding default license plate structure model of the first matching total value and target license plate structure The quantity of preset rectangle frame is less than the quantity of the character in recognition result in model, it is determined that there are extra words in recognition result Symbol；
If target license plate structural model is the corresponding default license plate structure model of the second matching total value, it is determined that recognition result In the first character be redundant character；
If target license plate structural model is that third matches the corresponding default license plate structure model of total value, it is determined that recognition result In the first character before lack character.
Optionally, license plate recognition device can also include:
Adding module, if for lacking character, character lacking in addition in recognition result；
Module is removed, if removing redundant character for there are redundant characters in recognition result.
Optionally, adding module specifically can be used for:
If lacking character after the tail position character of recognition result, identified from the position after the character of recognition result tail position Character, and the character that will identify that is added in recognition result；
If lacking character before the first character in recognition result, from the first character before position from identify character, And the character that will identify that is added in recognition result；
Module is removed, specifically can be used for:
If the tail position character of recognition result is redundant character, tail position character is removed；
If the first character in recognition result is redundant character, the first character is removed.
Optionally, license plate recognition device can also include:
Second judgment module is sentenced for the alphabetical distribution rule and digital distribution rule according to target license plate structural model Whether disconnected recognition result is wrong；
Modified module, if the judging result for the second judgment module be it is yes, according to preset alteration ruler, to identification As a result it modifies.
Optionally, the second judgment module specifically can be used for:
According to the arrangement position of rectangle frame where presetting letter in target license plate structural model, judge to be in recognition result Whether the character with the same arrangement position of rectangle frame where default letter is number；
If it has, then will be in recognition result and rectangle frame where default letter according to preset character similarity table The character change of same arrangement position is the corresponding capitalization of number；
According to the arrangement position of rectangle frame where preset number in target license plate structural model, judge to be in recognition result Whether the character with the same arrangement position of rectangle frame where preset number is capitalization；
If it has, then will be in recognition result and rectangle frame where preset number according to preset character similarity table The character change of same arrangement position is the corresponding number of capitalization.
As it can be seen that license plate recognition device provided in an embodiment of the present invention, firstly, according to the matching pair of the recognition result of license board information The target license plate structural model answered, so by comparing the number of characters in recognition result whether the word with target license plate structural model Symbol number unanimously judges with the presence or absence of character or multicharacter problem is lost in recognition result, if number of characters and mesh in recognition result The number of characters for marking license plate structure model is inconsistent, then illustrates to exist in recognition result and lose character or multicharacter problem, in this way, just Further recognition result can be adjusted, keep recognition result more accurate and reliable.
The embodiment of the invention also provides a kind of electronic equipment, and with reference to Figure 13, Figure 13 is that the electronics of the embodiment of the present invention is set Standby structural schematic diagram.As shown in figure 13, including processor 1301 and memory 1302, wherein
Memory 1302, for storing computer program；
Processor 1301 when for executing the program stored on memory 1302, realizes that the embodiment of the present invention is provided Licence plate recognition method all steps.
The communication bus that above-mentioned electronic equipment is mentioned can be Peripheral Component Interconnect standard (Peripheral Component Interconnect, PCI) bus or expanding the industrial standard structure (Extended Industry Standard Architecture, EISA) bus etc..The communication bus can be divided into address bus, data/address bus, control bus etc..For just It is only indicated with a thick line in expression, figure, it is not intended that an only bus or a type of bus.
Communication interface is for the communication between above-mentioned electronic equipment and other equipment.
Memory may include random access memory (Random Access Memory, RAM), also may include non-easy The property lost memory (Non-Volatile Memory, NVM), for example, at least a magnetic disk storage.Optionally, memory may be used also To be storage device that at least one is located remotely from aforementioned processor.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit, CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal Processor, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete Door or transistor logic, discrete hardware components.
In addition, the embodiment of the invention provides a kind of calculating corresponding to licence plate recognition method provided by above-described embodiment Machine readable storage medium storing program for executing, is stored with machine-executable instruction, and when being called and being executed by processor, machine-executable instruction promotes Processor realizes all steps of licence plate recognition method provided in an embodiment of the present invention.
For electronic equipment and computer readable storage medium embodiment, since the method content that it is related to is basic It is similar to embodiment of the method above-mentioned, so being described relatively simple, related place is referring to the part explanation of embodiment of the method It can.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that There is also other identical elements in process, method, article or equipment including the element.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device, For electronic equipment and computer readable storage medium embodiment, since it is substantially similar to the method embodiment, so description Fairly simple, the relevent part can refer to the partial explaination of embodiments of method.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention It is interior.

Claims (25)

1. a kind of licence plate recognition method, which is characterized in that the described method includes:
Car license recognition is carried out to images to be recognized, obtains the recognition result of license board information；
By the position of preset rectangle frame each in the location information of character each in the recognition result, with default license plate structure model Information is matched, and determines the corresponding target license plate structural model of the recognition result；
According to the matching result of the recognition result and the target license plate structural model, the character in the recognition result is judged Whether number is consistent with the number of characters of the target license plate structural model；
If NO, it is determined that lack character in the recognition result or there are redundant characters.
2. the method according to claim 1, wherein the position by character each in the recognition result is believed Breath, is matched with the location information of each preset rectangle frame in default license plate structure model, determines that the recognition result is corresponding Target license plate structural model, comprising:
By each preset rectangle frame in the location information of character each in the recognition result, with the first default license plate structure model Location information is matched, and the first default license plate structure model is that matching is preferential in default license plate structure model to be matched The highest default license plate structure model of grade；
If successful match, it is determined that the first default license plate structure model is the corresponding target license plate structure of the recognition result Model；
If matching is unsuccessful, the pre- of the matching highest priority is rejected from the default license plate structure model to be matched If license plate structure model, and return and execute the location information by character each in the recognition result, with the first default license plate The location information of each preset rectangle frame is matched in structural model.
3. method according to claim 1 or 2, which is characterized in that in the position by character each in the recognition result Confidence breath, is matched with the location information of each preset rectangle frame in default license plate structure model, determines the recognition result Before corresponding target license plate structural model, the method also includes:
It is distributed line number according to the character in the recognition result, the character point in screening character distribution line number and the recognition result The identical default license plate structure model of cloth line number.
4. according to the method described in claim 3, it is characterized in that, the default license plate structure model includes preset rectangle frame Width；
The location information by character each in the recognition result, with each preset rectangle frame in default license plate structure model Location information is matched, comprising:
For each default license plate structure model, row of every a line internal moment shape frame from head to tail in license plate structure model is preset according to this Column sequence, calculates the overlapping widths between the width of rectangle frame and the width of the recognition result middle finger location character；
According to the default confidence level of the overlapping widths and the specific bit character, corresponding of the specific bit character is calculated With value；
The corresponding matching value of characters all in the recognition result is added, the recognition result and the default license plate structure are calculated The matching total value of model；
The corresponding target license plate structural model of the determination recognition result, comprising:
It determines and matches the maximum default license plate structure model of total value in each default license plate structure model as recognition result matching Target license plate structural model.
5. according to the method described in claim 4, it is characterized in that, the specific bit character includes: in the recognition result In be in the first character of the same arrangement position of rectangle frame in default license plate structure model, the recognition result and default vehicle Rectangle frame is in and default with moving to right at arrangement position in one the second character and the recognition result in board structural model Rectangle frame is the same as the third character for moving to left one at arrangement position in license plate structure model；
It is described to preset every a line internal moment shape frame putting in order from head to tail in license plate structure model according to this, calculate rectangle frame Overlapping widths between width and the width of the recognition result middle finger location character, comprising:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates the width of rectangle frame With the second of the width of the width and second character of the first overlapping widths and rectangle frame of the width of first character Overlapping widths；
According to this, to preset arrangement of the rectangle frame from head to tail in license plate structure model in every a line in addition to the first rectangle frame suitable Sequence calculates the third overlapping widths of the width of rectangle frame and the width of the third character；
It is corresponding to calculate the specific bit character for the default confidence level according to the overlapping widths and the specific bit character Matching value, comprising:
According to the default confidence level of first overlapping widths and first character, first character corresponding the is calculated One matching value；
According to the default confidence level of second overlapping widths and second character, second character corresponding the is calculated Two matching values；
According to the default confidence level of the third overlapping widths and the third character, the third character corresponding is calculated Three matching values；
It is described to be added the corresponding matching value of characters all in the recognition result, calculate the recognition result and the default license plate The matching total value of structural model, comprising:
Corresponding first matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated First matching total value of structural model；
Corresponding second matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated Second matching total value of structural model；
The corresponding third matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated The third of structural model matches total value；
It is the recognition result that the maximum default license plate structure model of total value is matched in each default license plate structure model of determination Matched target license plate structural model, comprising:
It determines in the first matching total value, the second matching total value and third matching total value of each default license plate structure model, owns The maximum value of value；
By the corresponding default license plate structure model of the maximum value, it is determined as the matched target license plate structure mould of the recognition result Type.
6. according to the method described in claim 4, it is characterized in that, the specific bit character includes: in the recognition result In be in the first character of the same arrangement position of rectangle frame in default license plate structure model, the recognition result and default vehicle Rectangle frame is in and default with moving to right at arrangement position in one the second character and the recognition result in board structural model Rectangle frame is the same as the third character for moving to left one at arrangement position in license plate structure model；
The location information by character each in the recognition result, with each preset rectangle frame in default license plate structure model Location information is matched, and determines the corresponding target license plate structural model of the recognition result, comprising:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates the width of rectangle frame With the first overlapping widths of the width of first character；
According to the default confidence level of first overlapping widths and first character, first character corresponding the is calculated One matching value；
Corresponding first matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated First matching total value of structural model；
Judge whether maximum first matching total value is greater than the first preset threshold in each default license plate structure model；
If so, determining that the first matching maximum default license plate structure model of total value is the knowledge in each default license plate structure model The other matched target license plate structural model of result；
If it is not, then executing:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates the width of rectangle frame With the second overlapping widths of the width of second character；
According to the default confidence level of second overlapping widths and second character, second character corresponding the is calculated Two matching values；
Corresponding second matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated Second matching total value of structural model；
Judge whether maximum second matching total value is greater than the second preset threshold in each default license plate structure model；
If so, determining that the second matching maximum default license plate structure model of total value is the knowledge in each default license plate structure model The other matched target license plate structural model of result；
If it is not, then executing:
According to this, to preset arrangement of the rectangle frame from head to tail in license plate structure model in every a line in addition to the first rectangle frame suitable Sequence calculates the third overlapping widths of the width of rectangle frame and the width of the third character；
According to the default confidence level of the third overlapping widths and the third character, the third character corresponding is calculated Three matching values；
The corresponding third matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated The third of structural model matches total value；
Judge whether maximum third matching total value is greater than third predetermined threshold value in each default license plate structure model；
If so, determining that the third matching maximum default license plate structure model of total value is the knowledge in each default license plate structure model The other matched target license plate structural model of result.
7. method according to claim 5 or 6, which is characterized in that described according to the recognition result and the target carriage The matching result of board structural model, judge number of characters in the recognition result whether the word with the target license plate structural model It is consistent to accord with number, comprising:
If the target license plate structural model is the corresponding default license plate structure model of the first matching total value, by comparing described The quantity of the quantity of preset rectangle frame and the character in the recognition result, determines the identification in target license plate structural model Whether the number of characters in as a result is consistent with the number of characters of the target license plate structural model；
If the target license plate structural model is that the second matching total value or third match the corresponding default license plate structure model of total value, Then determine that the number of characters of the number of characters and the target license plate structural model in the recognition result is inconsistent.
8. method according to claim 5 or 6, which is characterized in that lack in the determination recognition result character or There are redundant characters, comprising:
If the target license plate structural model is the first matching corresponding default license plate structure model of total value and the target license plate The quantity of preset rectangle frame is greater than the quantity of the character in the recognition result in structural model, it is determined that the recognition result In lack character；
If the target license plate structural model is the first matching corresponding default license plate structure model of total value and the target license plate The quantity of preset rectangle frame is less than the quantity of the character in the recognition result in structural model, it is determined that the recognition result In there are redundant characters；
If the target license plate structural model is the corresponding default license plate structure model of the second matching total value, it is determined that the identification As a result the first character in is redundant character；
If the target license plate structural model is that third matches the corresponding default license plate structure model of total value, it is determined that the identification As a result lack character before the first character in.
9. the method according to claim 1, wherein lacking character in the determination recognition result or depositing After redundant character, the method also includes:
If lacking character in the recognition result, character lacking in addition；
If there are redundant characters in the recognition result, the redundant character is removed.
10. according to the method described in claim 9, it is characterized in that, character lacking in the addition, comprising:
If lacking character after the tail position character of the recognition result, from the position after the character of recognition result tail position Identify character, and the character that will identify that is added in the recognition result；
If lacking character before the first character in the recognition result, from the first character before position from identify word Symbol, and the character that will identify that is added in the recognition result；
The removal redundant character, comprising:
If the tail position character of the recognition result is redundant character, tail position character is removed；
If the first character in the recognition result is redundant character, the first character is removed.
11. the method according to claim 1, wherein in the corresponding target carriage of the determination recognition result After board structural model, the method also includes:
According to the alphabetical distribution rule of the target license plate structural model and digital distribution rule, whether the recognition result is judged It is wrong；
If it has, then modifying according to preset alteration ruler to the recognition result.
12. according to the method for claim 11, which is characterized in that described to be distributed according to the letter of target license plate structural model Rule and digital distribution rule, judge whether the recognition result is wrong, comprising:
According to the arrangement position of rectangle frame where presetting letter in the target license plate structural model, judge in the recognition result It whether is number in the character with the same arrangement position of rectangle frame where the default letter；
If it has, then will be in the recognition result and square where the default letter according to preset character similarity table The character change of the same arrangement position of shape frame is the corresponding capitalization of the number；
According to the arrangement position of rectangle frame where preset number in the target license plate structural model, judge in the recognition result It whether is capitalization in the character with the same arrangement position of rectangle frame where the preset number；
If it has, then will be in the recognition result and square where the preset number according to preset character similarity table The character change of the same arrangement position of shape frame is the corresponding number of the capitalization.
13. a kind of license plate recognition device, which is characterized in that described device includes:
Identification module obtains the recognition result of license board information for carrying out Car license recognition to images to be recognized；
Matching module, for that each in the location information of character each in the recognition result, with default license plate structure model will preset The location information of rectangle frame matched, determine the corresponding target license plate structural model of the recognition result；
First judgment module judges institute for the matching result according to the recognition result and the target license plate structural model Whether the number of characters stated in recognition result is consistent with the number of characters of the target license plate structural model；
Determining module, if the judging result for the first judgment module is number of characters and the mesh in the recognition result The number of characters for marking license plate structure model is inconsistent, it is determined that lacks character in the recognition result or there are redundant characters.
14. device according to claim 13, which is characterized in that the matching module is specifically used for:
By each preset rectangle frame in the location information of character each in the recognition result, with the first default license plate structure model Location information is matched, and the first default license plate structure model is that matching is preferential in default license plate structure model to be matched The highest default license plate structure model of grade；
If successful match, it is determined that the first default license plate structure model is the corresponding target license plate structure of the recognition result Model；
If matching is unsuccessful, the pre- of the matching highest priority is rejected from the default license plate structure model to be matched If license plate structure model, and return and execute the location information by character each in the recognition result, with the first default license plate The location information of each preset rectangle frame is matched in structural model.
15. device described in 3 or 14 according to claim 1, which is characterized in that described device further include:
Screening module, for being distributed line number, screening character distribution line number and the identification according to the character in the recognition result As a result the identical default license plate structure model of character distribution line number in.
16. device according to claim 15, which is characterized in that the default license plate structure model includes preset rectangle The width of frame；
The matching module, is specifically also used to:
For each default license plate structure model, row of every a line internal moment shape frame from head to tail in license plate structure model is preset according to this Column sequence, calculates the overlapping widths between the width of rectangle frame and the width of the recognition result middle finger location character；
According to the default confidence level of the overlapping widths and the specific bit character, corresponding of the specific bit character is calculated With value；
The corresponding matching value of characters all in the recognition result is added, the recognition result and the default license plate structure are calculated The matching total value of model；
It determines and matches the maximum default license plate structure model of total value in each default license plate structure model as recognition result matching Target license plate structural model.
17. device according to claim 16, which is characterized in that the specific bit character includes: in the recognition result In be in the first character of the same arrangement position of rectangle frame in default license plate structure model, the recognition result and default In license plate structure model rectangle frame with moved to right at arrangement position in one the second character and the recognition result in it is pre- If rectangle frame is the same as the third character for moving to left one at arrangement position in license plate structure model；
The matching module, is specifically also used to:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates the width of rectangle frame With the second of the width of the width and second character of the first overlapping widths and rectangle frame of the width of first character Overlapping widths；
According to this, to preset arrangement of the rectangle frame from head to tail in license plate structure model in every a line in addition to the first rectangle frame suitable Sequence calculates the third overlapping widths of the width of rectangle frame and the width of the third character；
According to the default confidence level of first overlapping widths and first character, first character corresponding the is calculated One matching value；
According to the default confidence level of second overlapping widths and second character, second character corresponding the is calculated Two matching values；
According to the default confidence level of the third overlapping widths and the third character, the third character corresponding is calculated Three matching values；
Corresponding first matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated First matching total value of structural model；
Corresponding second matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated Second matching total value of structural model；
The corresponding third matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated The third of structural model matches total value；
It determines in the first matching total value, the second matching total value and third matching total value of each default license plate structure model, owns The maximum value of value；
By the corresponding default license plate structure model of the maximum value, it is determined as the matched target license plate structure mould of the recognition result Type.
18. device according to claim 16, which is characterized in that the specific bit character includes: in the recognition result In be in the first character of the same arrangement position of rectangle frame in default license plate structure model, the recognition result and default In license plate structure model rectangle frame with moved to right at arrangement position in one the second character and the recognition result in it is pre- If rectangle frame is the same as the third character for moving to left one at arrangement position in license plate structure model；
The matching module, is specifically also used to:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates the width of rectangle frame With the first overlapping widths of the width of first character；
According to the default confidence level of first overlapping widths and first character, first character corresponding the is calculated One matching value；
Corresponding first matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated First matching total value of structural model；
Judge whether maximum first matching total value is greater than the first preset threshold in each default license plate structure model；
If so, determining that the first matching maximum default license plate structure model of total value is the knowledge in each default license plate structure model The other matched target license plate structural model of result；
If it is not, then executing:
Every a line internal moment shape frame putting in order from head to tail in license plate structure model is preset according to this, calculates the width of rectangle frame With the second overlapping widths of the width of second character；
According to the default confidence level of second overlapping widths and second character, second character corresponding the is calculated Two matching values；
Corresponding second matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated Second matching total value of structural model；
Judge whether maximum second matching total value is greater than the second preset threshold in each default license plate structure model；
If so, determining that the second matching maximum default license plate structure model of total value is the knowledge in each default license plate structure model The other matched target license plate structural model of result；
If it is not, then executing:
According to this, to preset arrangement of the rectangle frame from head to tail in license plate structure model in every a line in addition to the first rectangle frame suitable Sequence calculates the third overlapping widths of the width of rectangle frame and the width of the third character；
According to the default confidence level of the third overlapping widths and the third character, the third character corresponding is calculated Three matching values；
The corresponding third matching value of characters all in the recognition result is added, the recognition result and the default license plate are calculated The third of structural model matches total value；
Judge whether maximum third matching total value is greater than third predetermined threshold value in each default license plate structure model；
If so, determining that the third matching maximum default license plate structure model of total value is the knowledge in each default license plate structure model The other matched target license plate structural model of result.
19. device described in 7 or 18 according to claim 1, which is characterized in that the first judgment module is specifically used for:
If the target license plate structural model is the corresponding default license plate structure model of the first matching total value, by comparing described The quantity of the quantity of preset rectangle frame and the character in the recognition result, determines the identification in target license plate structural model Whether the number of characters in as a result is consistent with the number of characters of the target license plate structural model；
If the target license plate structural model is that the second matching total value or third match the corresponding default license plate structure model of total value, Then determine that the number of characters of the number of characters and the target license plate structural model in the recognition result is inconsistent.
20. any device described in 7 to 18 according to claim 1, which is characterized in that the determining module is specifically used for:
If the target license plate structural model is the first matching corresponding default license plate structure model of total value and the target license plate The quantity of preset rectangle frame is greater than the quantity of the character in the recognition result in structural model, it is determined that the recognition result In lack character；
If the target license plate structural model is the first matching corresponding default license plate structure model of total value and the target license plate The quantity of preset rectangle frame is less than the quantity of the character in the recognition result in structural model, it is determined that the recognition result In there are redundant characters；
If the target license plate structural model is the corresponding default license plate structure model of the second matching total value, it is determined that the identification As a result the first character in is redundant character；
If the target license plate structural model is that third matches the corresponding default license plate structure model of total value, it is determined that the identification As a result lack character before the first character in.
21. device according to claim 13, which is characterized in that described device further include:
Adding module, if for lacking character, character lacking in addition in the recognition result；
Module is removed, if removing the redundant character for there are redundant characters in the recognition result.
22. device according to claim 21, which is characterized in that the adding module is specifically used for:
If lacking character after the tail position character of the recognition result, from the position after the character of recognition result tail position Identify character, and the character that will identify that is added in the recognition result；
If lacking character before the first character in the recognition result, from the first character before position from identify word Symbol, and the character that will identify that is added in the recognition result；
The removal module, is specifically used for:
If the tail position character of the recognition result is redundant character, tail position character is removed；
If the first character in the recognition result is redundant character, the first character is removed.
23. device according to claim 13, which is characterized in that described device further include:
Second judgment module is sentenced for the alphabetical distribution rule and digital distribution rule according to the target license plate structural model Whether the recognition result that breaks is wrong；
Modified module, if the judging result for second judgment module be it is yes, according to preset alteration ruler, to described Recognition result is modified.
24. device according to claim 23, which is characterized in that second judgment module is specifically used for:
According to the arrangement position of rectangle frame where presetting letter in the target license plate structural model, judge in the recognition result It whether is number in the character with the same arrangement position of rectangle frame where the default letter；
If it has, then will be in the recognition result and square where the default letter according to preset character similarity table The character change of the same arrangement position of shape frame is the corresponding capitalization of the number；
According to the arrangement position of rectangle frame where preset number in the target license plate structural model, judge in the recognition result It whether is capitalization in the character with the same arrangement position of rectangle frame where the preset number；
If it has, then will be in the recognition result and square where the preset number according to preset character similarity table The character change of the same arrangement position of shape frame is the corresponding number of the capitalization.
25. a kind of electronic equipment, which is characterized in that including processor and memory, wherein
The memory, for storing computer program；
The processor when for executing the program stored on the memory, realizes that claim 1-12 is any described Method and step.
CN201810133583.3A 2018-02-09 2018-02-09 License plate recognition method and device Active CN110135416B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810133583.3A CN110135416B (en) 2018-02-09 2018-02-09 License plate recognition method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810133583.3A CN110135416B (en) 2018-02-09 2018-02-09 License plate recognition method and device

Publications (2)

Publication Number Publication Date
CN110135416A true CN110135416A (en) 2019-08-16
CN110135416B CN110135416B (en) 2021-06-04

Family

ID=67567689

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810133583.3A Active CN110135416B (en) 2018-02-09 2018-02-09 License plate recognition method and device

Country Status (1)

Country Link
CN (1) CN110135416B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111695563A (en) * 2020-06-10 2020-09-22 北京筑梦园科技有限公司 Single-layer and double-layer license plate recognition method, server and parking charging system
CN112329758A (en) * 2020-11-04 2021-02-05 深圳市极致科技股份有限公司 License plate-based fuzzy matching method and device, electronic equipment and storage medium
CN114419636A (en) * 2022-01-10 2022-04-29 北京百度网讯科技有限公司 Text recognition method, device, equipment and storage medium

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102693431A (en) * 2012-05-31 2012-09-26 信帧电子技术（北京）有限公司 Method and device for identifying type of white number plate
CN102722733A (en) * 2012-05-31 2012-10-10 信帧电子技术（北京）有限公司 Identification method and device of license plate types
CN104636748A (en) * 2013-11-14 2015-05-20 张伟伟 License plate recognition method and device
CN106683073A (en) * 2015-11-11 2017-05-17 杭州海康威视数字技术股份有限公司 License plate detection method, camera and server
CN106845487A (en) * 2016-12-30 2017-06-13 佳都新太科技股份有限公司 A kind of licence plate recognition method end to end

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102693431A (en) * 2012-05-31 2012-09-26 信帧电子技术（北京）有限公司 Method and device for identifying type of white number plate
CN102722733A (en) * 2012-05-31 2012-10-10 信帧电子技术（北京）有限公司 Identification method and device of license plate types
CN104636748A (en) * 2013-11-14 2015-05-20 张伟伟 License plate recognition method and device
CN106683073A (en) * 2015-11-11 2017-05-17 杭州海康威视数字技术股份有限公司 License plate detection method, camera and server
CN106845487A (en) * 2016-12-30 2017-06-13 佳都新太科技股份有限公司 A kind of licence plate recognition method end to end

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
ANA RIZA F. QUIROS等: "A Genetic Algorithm and Artificial Neural Network based Approach for the Machine Vision of Plate Segmentation and Character Recognition", 《IEEE》 *
郭燚平: "基于形状上下文的复杂车牌识别***", 《中国优秀硕士学位论文全文数据库（电子期刊）信息科技辑》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111695563A (en) * 2020-06-10 2020-09-22 北京筑梦园科技有限公司 Single-layer and double-layer license plate recognition method, server and parking charging system
CN111695563B (en) * 2020-06-10 2022-07-05 北京筑梦园科技有限公司 Single-layer and double-layer license plate recognition method, server and parking charging system
CN112329758A (en) * 2020-11-04 2021-02-05 深圳市极致科技股份有限公司 License plate-based fuzzy matching method and device, electronic equipment and storage medium
CN114419636A (en) * 2022-01-10 2022-04-29 北京百度网讯科技有限公司 Text recognition method, device, equipment and storage medium

Also Published As

Publication number Publication date
CN110135416B (en) 2021-06-04

Similar Documents

Publication Publication Date Title
CN110135416A (en) A kind of licence plate recognition method and device
CN106960195A (en) A kind of people counting method and device based on deep learning
CN105844205B (en) Character information recognition methods based on image procossing
CN108681692A (en) Increase Building recognition method in a kind of remote sensing images based on deep learning newly
CN106485199A (en) A kind of method and device of body color identification
CN108596338A (en) A kind of acquisition methods and its system of neural metwork training collection
CN109657664A (en) A kind of recognition methods, device and the electronic equipment of license plate type
CN104778238B (en) The analysis method and device of a kind of saliency
CN110245545A (en) A kind of character recognition method and device
CN109543753B (en) License plate recognition method based on self-adaptive fuzzy repair mechanism
CN105023025B (en) A kind of opener mark image sorting technique and system
CN108073926A (en) A kind of licence plate recognition method and device
CN110532880A (en) Screening sample and expression recognition method, neural network, equipment and storage medium
CN109839619A (en) Based on radar signal rough segmentation choosing method, system and the storage medium for adaptively dividing bucket
CN109829428A (en) Based on the video image pedestrian detection method and system for improving YOLOv2
CN103493067B (en) The method and apparatus for identifying the character of video
CN110472550A (en) A kind of text image shooting integrity degree judgment method and system
CN107578438A (en) Circle recognition methods, device and electronic equipment
JP2001052116A (en) Device and method for matching pattern stream, device and method for matching character string
CN108446702B (en) Image character segmentation method, device, equipment and storage medium
CN109034158A (en) A kind of licence plate recognition method, device and computer equipment
CN111191531A (en) Rapid pedestrian detection method and system
CN108595576A (en) It is a kind of based on the vehicle of database to scheme to search drawing method
CN106709489A (en) Processing method and device of character identification
CN116543189B (en) Target detection method, device, equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant