Specific implementation mode
It is with reference to the accompanying drawings and embodiments, right in order to make the object, technical solution and advantage of the application be more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
The inscape content of trademark image may be various, and approximate factor is also various, traditional quotient
Mark does not retrieve the determination method to inputting trade mark and search key:It is input picture file or be typing word.
The combination for the picture file and typing word content that the setting of search key and algorithm limitation pass thereon, this retrieval are crucial
The setting of word and algorithm can not reflect the indirect associated text information of input trade mark and the arbitrary local message of trademark image, these
The arbitrary local message of associated text information and trademark image connect may constitute similar mark production to input trade mark and sample trade mark
Raw to influence, the missing inspection to the partial information must easily cause the missing inspection mass defect of trade mark retrieval result.
In addition, traditional technology can be converted into machine by optical character identification (OCR) method to the text image of specification
The digital form of editable text, but its there is also following limitation or defects:When the text image to non-standard is identified,
The accuracy rate that it is identified is not high;Whether the pronunciation of the text image not shown directly in None- identified text image, spelling words intellectual
Information with meaning, graphical element coding and other reflection image form-pronunciation-meaning features;The word identified from image, as
When being used for identical or similar mark image retrieval for keyword, though certain retrieval effectiveness can be played, since it is short of it
His picture material description, thus cause identical or similar mark image missing inspection unavoidable.
It can define, the retrieval of traditional trade mark also rests on the mode of manual typing mostly, and working efficiency is low and works
Huge energy consumption is obvious.
And the embodiment of the present invention establishes sample image data by the trade mark and knowledge data information of the existing magnanimity of system
Library, sample image database include sample trademark database, trade mark inscape sample image database, word dictionary database
And word dictionary database, sample image data is converted, is divided, combined treatment, the image for obtaining sample image is special
Descriptor, associated text information, assembled unit data and the arbitrary local message of trademark image are levied, input trade mark is converted,
Segmentation, combined treatment obtain characteristics of image descriptor, associated text information, assembled unit data and the trademark image of input trade mark
As arbitrary local message, it is based on the assembled unit data and the arbitrary local message sample retrieval trademark database of trademark image,
The text and form-pronunciation-meaning for obtaining the image associated by matched preliminary search sample trade mark and the sample trade mark, having been recorded in trade mark
Characteristic information, the matched minimum unit of institute and assembled unit data, calculate between preliminary search sample trade mark and input trade mark
Shape, sound, justice, the individual event matching rate of search key, probability of mismatch and comprehensive approximate rate meet default individual event by comprehensive approximate rate
The sample retrieval trade mark that matching rate, probability of mismatch and comprehensive approximate rate and/or sequence ranking meet default ranking is ranked up, from
And obtain input trade mark retrieval result.Various embodiments of the present invention can improve in brand recognition retrieval identical or approximate image
With effect, to improve identical or similar mark recall rate, precision rate.
Specifically, brand recognition search method provided by the present application, can be applied to application environment as shown in Figure 1
In.Wherein, terminal 102 can be communicated with server 104 by network by network, convenient for obtaining input trade mark, sample quotient
The related data that mark, other sample trade marks and sample trademark database are related to, it should be noted that terminal 102 also can not be with clothes
Business device 104 is communicated, and related data can be stored in terminal 102 in advance, then is handled;Wherein, terminal 102 can
With but be not limited to various personal computers, laptop, smart mobile phone, tablet computer and portable wearable device, take
Business device 104 can be realized with the server cluster of the either multiple server compositions of independent server.
In one embodiment, as shown in Fig. 2, providing a kind of brand recognition search method, it is applied to Fig. 1 in this way
In terminal for illustrate, include the following steps:
Step S210 converts the image data of input trade mark by sample retrieval image data base, obtains input trade mark
Characteristics of image descriptor and associated text information;Sample image database pre-establishes, and includes the image of sample image
The database of feature descriptor, associated text information, minimum unit and assembled unit data;Assembled unit data are phenogram
As the data of arbitrary local message;
Wherein, input trade mark includes with the input trade mark of graphic form typing and with the input trade mark of written form typing;
Sample image includes with the sample image of graphic form typing and with the sample image of written form typing;That is the embodiment of the present invention
The trade mark of processing can be graphic form, or written form.
Sample image includes sample trademark pattern, appearance design pattern, the work of fine arts pattern of copyright registration, each Chinese written language
Pattern, each non-Chinese written language pattern and custom images;Sample image database includes sample trademark database, trade mark composition
Element sample image database, word dictionary database and word dictionary database.
Further, in order to realize effective trade mark retrieval, the inscape of trademark image should be considered from many aspects
Content, it is also various to influence trade mark approximation factor, to obtain preferable trade mark retrieval recall ratio, it is necessary to scientifically and rationally really
Determine search key and its algorithm, sample image database of the present invention pre-establishes, and includes the figure of sample image
As feature descriptor, the database of associated text information, minimum unit and assembled unit data;Wherein, assembled unit data
The arbitrary local message of trademark image can be characterized;And to sample image and input trade mark processing procedure, may include conversion,
The flows such as segmentation and combination.The above processing allows the invention to obtain more complete complete trade mark search key.
In various embodiments of the present invention, the associated text information of sample image includes the brand logo of the sample image recorded
Things title and sample image described in constituent encoder, sample image can recognize the text and form-pronunciation-meaning feature of the word of reading;Shape
Pronunciation and meaning feature includes the graphics shape form of expression or style, pronunciation, meaning and nearly word form, sound of sample image
Nearly word and the nearly word of justice.;
And characteristics of image descriptor is using the identical or similar character string of height, to having in input trade mark or sample image
There are same perception content or feature to be recorded, and uses different character strings to there is difference in input trade mark or sample image
The image feature representation form that perceived content or feature are recorded;Image feature representation form is description input trade mark or sample
The set of one or more groups of character strings of the characteristics of image of image.
In a specific embodiment, the image data that input trade mark is converted by sample retrieval image data base, obtains
Include to the step of characteristics of image descriptor and associated text information for inputting trade mark:
Extraction is with the characteristics of image descriptor of the input trade mark of graphic form typing;Sample is retrieved based on characteristics of image descriptor
The corresponding sample image of matched characteristics of image descriptor is considered as identical or high as the input image of trade mark by this image data base
It spends approximate image, and characteristics of image descriptor that sample image has recorded, associated text information is confirmed as with graphic form
The characteristics of image descriptor of the input trade mark of typing, associated text information;And
Based on written form typing input trade mark character search sample image database, by matched sample word
Characteristics of image descriptor that corresponding sample image has recorded, associated text information are confirmed as the input quotient with written form typing
Target characteristics of image descriptor, associated text information.
Specifically, above-mentioned realization process may include:To with graphic form indicate input trade mark picture file and with
The word of the input trade mark of written form typing is converted into the processing of characteristics of image descriptor and associated text information respectively.
Wherein, is converted by characteristics of image descriptor and is associated with for the picture file of the input trade mark indicated with graphic form
The method of the processing of text message, including:First, existing skill is used to the picture file of the input trade mark indicated with graphic form
Art method extracts its characteristics of image descriptor;Second, using the mass data information of the existing sample image recorded, based on figure
As feature descriptor sample retrieval image data base, obtains matched characteristics of image descriptor and the characteristics of image descriptor is right
The sample image and associated text information answered are believed using the information as the characteristics of image descriptor of input trade mark with associated text
Breath, wherein associated text information includes:Brand logo element in the figurative mark that has recorded of matched sample image compile
Yard, the things title described in figurative mark, trade mark style, pronunciation, meaning and its nearly word form, sound are close
Texts and the form-pronunciation-meaning features such as the nearly word of word, justice.
Characteristics of image descriptor and associated text information are converted into the word of the input trade mark recorded with written form
Processing method, including:First, based on using the word of the input trade mark of written form typing as keyword retrieval sample image
Database, the sample word to be matched;Second, find out the sample image and association text corresponding to the matched sample word
This information, wherein associated text information includes:Brand logo element in the figurative mark that the matched sample image has recorded is compiled
Yard, the things title described in figurative mark, to the characteristics of image descriptor that image is indicated, the writing shape of trade mark word
Texts and the form-pronunciation-meaning feature such as formula, pronunciation, meaning and the nearly word of its nearly word form, sound, the nearly word of justice.Wherein, the word includes
Chinese text, non-Chinese text, number, symbol.
Step S220, the characteristics of image descriptor and associated text information of segmentation input trade mark, obtain input trade mark respectively
Each characteristics of image descriptor minimum unit, each associated text information minimum unit;Characteristics of image descriptor minimum unit is pair
Answer one or more character strings of any image characteristic point represented by characteristics of image descriptor;Associated text information minimum unit
A word for any text message characteristic point represented by corresponding associated text information or multiple significant spelling words intellectuals;
Specifically, it is exactly to be carried out to the minimum unit of characteristics of image descriptor to do dividing processing to characteristics of image descriptor
Identification, each minimum unit of characteristics of image descriptor is split, and it is exactly to closing to do dividing processing to associated text information
The minimum unit of connection text message is identified, and each minimum unit of associated text information is split.
Wherein, characteristics of image descriptor minimum unit refers to:The character string of characteristics of image descriptor is generally used to refer to
The characteristic point of image, one or more character strings corresponding to each characteristic point are known as characteristics of image descriptor minimum unit.?
In one specific embodiment, characteristics of image descriptor is for indicating image contour line or image framework line any pixel point
The feature descriptor of position data and the correspondence of the conventional coordinates coordinates regional of any specification;Characteristics of image descriptor is most
Junior unit be the conventional coordinates of any specification any coordinates regional corresponding image contour line or one of image framework line
Or the position data of multiple pixels;
In addition, the character string of characteristics of image descriptor is generally used to refer to the characteristic point of image, corresponding to each characteristic point
One or more character strings be known as characteristics of image descriptor minimum unit.Image characteristic point described in characteristics of image descriptor
It is usually multiple, thus characteristics of image descriptor minimum unit can also be multiple.To inputting the characteristics of image description of trade mark
Symbol is split the process of processing:Each image characteristic point represented by characteristics of image descriptor is split,
Each or multiple character strings corresponding to each image characteristic point by characteristics of image descriptor are considered as characteristics of image descriptor
Minimum unit.
And associated text information minimum unit refers to:The word of associated text information is generally used to refer to text information
Characteristic point, it is minimum single that the significant spelling words intellectuals of one or more corresponding to each characteristic point are known as associated text information
Member.In a specific embodiment, associated text information minimum unit is the association represented by any word or spelling words intellectual
Text message corresponds to the data for having significant word or vocabulary.
Specifically, associated text information word is generally used to refer to the characteristic point of associated text information, each characteristic point institute
A corresponding word or multiple significant spelling words intellectuals are known as associated text information minimum unit.The spy of associated text information
Sign point is usually multiple, thus the minimum unit of associated text information can also be multiple.
And the process that processing is split to the associated text information for inputting trade mark may include:By associated text information institute
What each character features point indicated was split, each text corresponding to each character features point by associated text information
Word or multiple significant spelling words intellectuals are considered as associated text information minimum unit.Wherein, the word packet in associated text information
Include Chinese text, non-Chinese text (foreign languages of i.e. each languages), number and symbol.
Step S230 is respectively combined each characteristics of image descriptor of input trade mark according to default minimum unit rule of combination
Minimum unit, each associated text information minimum unit obtain characteristics of image descriptor combinations cell data, the association of input trade mark
Text message assembled unit data;
Specifically, it is exactly to be advised according to preset combination to minimum unit data to be combined processing to minimum unit data
It is then combined, obtains assembled unit data (and then embody trademark image arbitrary local message), wherein assembled unit data are
Multiple character strings of any local feature represented by corresponding described image feature descriptor or associated text information.
Isolated characteristics of image descriptor minimum unit or associated text information minimum unit may not have practical application to anticipate
Justice, but it is based on the embodiment of the present invention, each minimum unit is combined according to default minimum unit rule of combination, obtains combination
Cell data makes the combination of the characteristics of image descriptor minimum unit after combination or the combination of associated text information minimum unit have spy
Fixed meaning.
In a specific embodiment, it may include that characteristics of image descriptor is minimum single to preset minimum unit rule of combination
First rule of combination and associated text information minimum unit rule of combination;Characteristics of image descriptor combinations cell data includes being used for table
Show connected domain assembled unit data, for indicating line segment assembled unit data and for the string data of storage;Association text
This information assembled unit data include spelling words intellectual cell data, text pronunciation assembled unit data, word meaning assembled unit
Data and brand logo constituent encoder assembled unit data;
Specifically, can establish preset characteristics of image descriptor minimum unit according to the needs of application and be associated with text
This information minimum unit rule of combination;Further, by preset characteristics of image descriptor minimum unit and associated text information
Minimum unit rule of combination is combined, and obtains characteristics of image descriptor combinations cell data and associated text information assembled unit
Data.
It should be noted that the characteristics of image descriptor minimum unit data splitting acquired in the embodiment of the present invention can be used
In indicating a connected domain assembled unit data, it may also indicate that a line segment assembled unit data, also may indicate that for carrying out
Store the string data of processing.Associated text information minimum unit data splitting acquired in the embodiment of the present invention can be used for
The assembled unit data for indicating a vocabulary, may also indicate that the assembled unit data of a word, may also indicate that relatively independent part
A vocabulary or one group of word assembled unit data.
Further, in a specific embodiment, pre-set image feature descriptor minimum unit rule of combination can be with
The characteristics of image descriptor of characteristics of image descriptor minimum unit rule of combination and image framework line including image contour line is most
Junior unit rule of combination;
The characteristics of image descriptor minimum unit rule of combination of image contour line includes:It will be complete on any image contour line
Portion's line segment is confirmed as an image entire combination unit;It is a connected domain group by the closed loop line justification on any image contour line
Close unit;Line segment on the image contour line of any first default fixed length is confirmed as a line segment assembled unit;Wherein, first
The value range of default fixed length is more than or equal to 20% of the line segment overall length on image contour line;
The characteristics of image descriptor minimum unit rule of combination of image framework line includes:It will be complete on any image skeleton line
Portion's line segment is confirmed as an image entire combination unit;Continual line on any image skeleton line is confirmed as a connection
Domain assembled unit;Line segment on the image framework line of any second default fixed length is confirmed as a line segment assembled unit;Wherein,
The value range of second default fixed length is more than or equal to 20% of the line segment overall length on image framework line.
Assembled unit data obtained by combined processing and the arbitrary local message of trademark image are corresponding characteristics of image
Multiple character strings of any local feature represented by descriptor or associated text information.
It further, in a specific embodiment, can be with for the associated text information minimum unit of input trade mark
Following steps are taken to be combined:The word for splitting input trade mark one by one, obtains associated text information minimum unit;According to association
Text message minimum unit rule of combination is combined each associated text information minimum unit, obtains each spelling words intellectual unit number
According to;Associated text information minimum unit rule of combination includes:By size, color, languages are identical and the closely coupled word connect is true
Think a connected cypher unit;The connected cypher unit of each default word number fixed length is confirmed as part and combines list
Member;Wherein, the value range for presetting word number fixed length is in total alpha-numeric 20% or more the value of connected cypher unit;
The text pronunciation with word minimum unit assembled unit data match is obtained from word dictionary database, according to
Text pronunciation marks the text pronunciation in each word minimum unit assembled unit data, and it is single to obtain the combination of text pronunciation minimum unit
Metadata;
The word combination with each word minimum unit assembled unit data match is obtained from word dictionary database, is obtained
To word meaning assembled unit data;
Each brand logo constituent encoder of trade mark will be inputted, is confirmed as brand logo constituent encoder assembled unit number
According to.Wherein, brand logo constituent encoder refers to foundation《VIENNA AGREEMENT FOR ESTABLISHING AND INTERNATIONAL CLASSIFICATION OF THE FIGURATIVE ELEMENTS OF MARKS》It is generated
A kind of brand logo element dividing tool is made of brand logo element by the list of major class, group and group classification, wherein wrapping
Include brand logo element numerals and brand logo element name.
Step S240, based on characteristics of image descriptor combinations cell data, associated text information assembled unit data retrieval
Sample trademark database in sample image database obtains matched each preliminary search sample trade mark and preliminary search sample
Each characteristics of image descriptor minimum unit, each associated text information minimum unit of trade mark;
Wherein, matching refers to the assembled unit data that are obtained through aforementioned processing of input trade mark (i.e. trademark image is arbitrary
Local message) the assembled unit data (i.e. the arbitrary local message of trademark image) that are recorded with sample trademark database are identical
, and then the corresponding sample trade mark of assembled unit data that the record can be obtained.
Specifically, the input trade mark feature descriptor assembled unit data that preceding method is obtained, associated text information
Assembled unit data retrieve the sample trademark database in sample image database as search key, acquisition
The image associated by preliminary search sample trade mark and the sample trade mark matched, the text recorded in trade mark and form-pronunciation-meaning feature letter
Breath, minimum unit and assembled unit data.
Step S250 believes according to each characteristics of image descriptor minimum unit of preliminary search sample trade mark, each associated text
It ceases minimum unit and inputs each characteristics of image descriptor minimum unit, each associated text information minimum unit of trade mark, obtain
Individual event approximation rate;Individual event matching rate is handled, preliminary search sample trade mark is obtained and inputs the comprehensive approximate rate of trade mark;
It should be noted that minimum unit matching rate refers between sample trade mark and input trade mark in shape, sound, justice, inspection
The minimum unit accounting to match respectively in terms of rope keyword;Minimum unit probability of mismatch refers to sample trade mark and input trade mark
Unmatched minimum unit accounting is distinguished in terms of shape, sound, justice, search key.Wherein, input trade mark is in shape, sound, justice, inspection
The minimum unit of rope keyword can be indicated with the characteristics of image descriptor minimum unit of its correspondence image.
In a specific embodiment, during associated text information minimum unit may include Chinese minimum unit and be non-
Literary minimum unit;Individual event approximation rate may include Chinese individual event approximation rate, non-Chinese individual event approximation rate and characteristics of image individual event
Approximate rate;
Believed according to each characteristics of image descriptor minimum unit of preliminary search sample trade mark, each associated text in step S250
It ceases minimum unit and inputs each characteristics of image descriptor minimum unit, each associated text information minimum unit of trade mark, obtain
The step of individual event approximation rate includes:
Obtain the sum and characteristics of image descriptor of total, the non-Chinese minimum unit of the Chinese minimum unit of input trade mark
The sum of minimum unit, Chinese minimum unit total number, the non-Chinese of preliminary search sample trade mark matching input trade mark are minimum single
First total number and characteristics of image descriptor minimum unit total number, preliminary search sample trade mark mismatch the Chinese of input trade mark most
Junior unit total number, non-Chinese minimum unit total number and characteristics of image descriptor minimum unit total number;
Chinese minimum unit matching rate is obtained based on following formula:
Ma1=(Ua1÷U01) × 100%
Wherein, Ma1Indicate Chinese minimum unit matching rate, U01Indicate the sum of the Chinese minimum unit of input trade mark, Ua1
Indicate the Chinese minimum unit total number of preliminary search sample trade mark matching input trade mark;
Non- Chinese minimum unit matching rate is obtained based on following formula:
Ma2=(Ua2÷U02) × 100%
Wherein, Ma2Indicate non-Chinese minimum unit matching rate, U02Indicate the total of the non-Chinese minimum unit of input trade mark
Number, Ua2Indicate the non-Chinese minimum unit total number of preliminary search sample trade mark matching input trade mark;
Characteristics of image descriptor minimum unit matching rate is obtained based on following formula:
Ma0=(Ua0÷U00) × 100%
Wherein, Ma0Indicate characteristics of image descriptor minimum unit matching rate, U00Indicate the characteristics of image description of input trade mark
Accord with the sum of minimum unit, Ua0Indicate that the characteristics of image descriptor minimum unit of preliminary search sample trade mark matching input trade mark closes
It counts;
Chinese minimum unit probability of mismatch is obtained based on following formula:
Mi1=(Uc1÷U01) × 100%+ (n1-1)×ω1
Wherein, Mi1Indicate Chinese minimum unit probability of mismatch, U01Indicate the sum of the Chinese minimum unit of input trade mark,
Uc1Indicate that preliminary search sample trade mark mismatches the Chinese minimum unit total number of input trade mark, n1Indicate preliminary search sample quotient
Mark combines the place's number not matched that on line, ω in Chinese minimum unit with input trade mark1Number n at expression1Flexible strategy;Wherein,
ω1Value range be less than or equal to 80%;
Non- Chinese minimum unit probability of mismatch is obtained based on following formula:
Mi2=(Uc2÷U02) × 100%+ (n2-1)×ω2
Wherein, Mi2Indicate non-Chinese minimum unit probability of mismatch, U02Indicate the total of the non-Chinese minimum unit of input trade mark
Number, Uc2Indicate that preliminary search sample trade mark mismatches the non-Chinese minimum unit total number of input trade mark, n2Indicate preliminary search
Sample trade mark combines the place's number not matched that on line, ω with input trade mark in non-Chinese minimum unit2Number n at expression2Power
Number;Wherein, ω2Value range be less than or equal to 80%;
Characteristics of image descriptor minimum unit probability of mismatch is obtained based on following formula:
Mi0=(Uc0÷U00) × 100%+ (n0-1)×ω0
Wherein, Mi0Indicate characteristics of image descriptor minimum unit probability of mismatch, U00Indicate that the characteristics of image of input trade mark is retouched
State the sum of symbol minimum unit, Uc0Indicate that preliminary search sample trade mark mismatches the minimum list of characteristics of image descriptor of input trade mark
First total number, n0Indicate that preliminary search sample trade mark combines institute on line with input trade mark in characteristics of image descriptor minimum unit
The place's number not matched that, ω0Number n at expression0Flexible strategy;Wherein, ω0Value range be less than or equal to 80%;
Based on following formula, Chinese individual event approximation rate is obtained:
M1=Ma1-Mi1×β1
Wherein, M1Indicate that Chinese individual event approximation rate, β 1 indicate Mi1Flexible strategy;Wherein, the value range of β 1 is to be less than or wait
In 80%;
Based on following formula, non-Chinese individual event approximation rate is obtained
M2=Ma2-Mi2×β2
Wherein, M2Indicate that non-Chinese individual event approximation rate, β 2 indicate Mi2Flexible strategy;Wherein, the value range of β 2 be less than or
Equal to 80%;
Based on following formula, characteristics of image individual event approximation rate is obtained:
M0=Ma0-Mi0×β0
Wherein, M0Indicate characteristics of image individual event approximation rate, β0Indicate Mi0Flexible strategy;Wherein, β0Value range be less than or
Equal to 80%.
Further, it is based on following formula, obtains comprehensive approximate rate:
M=(M1+M2+M0)÷μ
Wherein, μ indicates M1、M2、M0It is not 0 item number.
In a specific embodiment, non-Chinese minimum unit is English minimum unit;Non- Chinese minimum unit matching
Rate is English minimum unit matching rate;Non- Chinese minimum unit probability of mismatch is English minimum unit probability of mismatch;Non- Chinese is single
The approximate rate of item is English individual event approximation rate;
Wherein, characteristics of image descriptor minimum unit combination line is characteristics of image line;Chinese minimum unit combines line
For the minimum unit of the corresponding form-pronunciation-meaning feature composition of Chinese trade mark word, by the path line to be formed that puts in order;Non- Chinese is most
Junior unit combination line is the minimum unit of the corresponding form-pronunciation-meaning feature composition of non-Chinese trade mark word, to be formed by putting in order
Path line.
Such as:Assuming that the word content for recognizing reading in trade mark is " blue global village ", shape feature information is minimum single
Member is that each word of reading can be recognized in the trade mark, from left to right " blue --- color --- ground --- ball --- village ", or from the right side to
It is exactly minimum unit combination line that the recognizing of left " village --- ball --- ground --- color --- is blue ", which reads track sequence line,.
Based on the sample trademark image that the above-mentioned retrieval of the present invention matches, be assembled unit data to input trade mark and
The matching result that the arbitrary local message of trademark image is generated as search key embodies Chinese, English and characteristics of image
The common point of the assembled unit of descriptor minimum unit is the concentrated expression of trademark image feature.
In various embodiments of the present invention, calculates preliminary search sample trade mark and inputs the comprehensive approximate rate between trade mark,
The prior art can be referred to realize, for example, using 201710553009.9 patent of invention of application number《A kind of trade mark inquiry result is close
Like degree evaluation and sort method, device》The method of middle discussion obtains.
Step S260, the preliminary search sample trade mark for meeting preset requirement to comprehensive approximate rate are ranked up, are retrieved
As a result.
In a specific embodiment, the preliminary search sample quotient that comprehensive approximate rate is greater than or equal to 30% is filtered out
Mark, and the preliminary search sample trade mark filtered out is ranked up, and sequence ranking is taken to be less than or equal to the preliminary inspection within 500
Rope sample trade mark is as retrieval result.
In practical applications, default minimum unit matching rate, probability of mismatch, comprehensive approximate rate and pre- can be needed according to application
If sequence ranking, usually, preset minimum unit matching rate more than 30% value, preset minimum unit probability of mismatch less than
70% value, preset comprehensive approximation rate is more than 30% value, and predetermined order ranking is less than 500 values.
Predetermined order refers to being ranked up with the comprehensive approximate rate that matched sample trade mark is obtained.Predetermined order will be met
The matched sample trade mark of ranking is considered as or height similar trade mark identical as input trade mark;
Aforementioned obtained same sample retrieval trade mark, may match with multiple search keys, that is, cause multiple inspections
Hitch fruit has the record of repetition trade mark registration number, these duplicate messages trade mark retrieval work be meaningless, cope with its into
The processing of row deduplication.The specific method of deduplication processing is, by the trade mark record of the same number of registration of same merchandise classification by aforementioned
Calculated comprehensive approximate rate is ranked up, and only takes highest 1 record of comprehensive approximate rate, it is same to delete remaining same merchandise classification
The trade mark of one number of registration records.
And after abovementioned steps calculate, it can will meet the matched sample trade mark of predetermined order ranking as sample retrieval
Trade mark, and reported its ranking results as trade mark retrieval result.
Above-mentioned brand recognition search method, can be by the trade mark and knowledge data information of the existing magnanimity of system, from shape sound
The presumption that the right way of conduct carries out text identification and its form-pronunciation-meaning characteristic information in face of input trade mark obtains, can be by with graphic form table
The word point of the picture file of the input trade mark or sample trade mark that show and the input trade mark or sample trade mark that are recorded with written form
Be not converted into the processing of characteristics of image descriptor and associated text information, to the handling result be split, combined treatment,
The arbitrary local message of assembled unit data and trademark image of input trade mark or sample trade mark is obtained, is believed with the association of big data
Breath estimates containing for the pronunciation, spelling words intellectual for identifying the image not directly displayed out in the image for inputting trade mark or sample trade mark
Justice, brand logo constituent encoder and other reflection image form-pronunciation-meaning features, with acquired assembled unit data and trademark image
Arbitrary local message effectively overcomes manual entry to be in the past not easy to the word in input trademark image as search key
The part combination etc. for carrying out nearly word form, the nearly word of sound, the nearly word of justice, figure constitution carry out it is exhaustive, easily cause search key disunity,
The defect that extraction information is easily omitted;The text image of specification and the text image of non-standard can effectively be identified, can be overcome
Traditional technology method has easily caused the defect of trade mark search key omission, can effectively solve the problem that search key in trade mark retrieval
The comprehensive sex chromosome mosaicism of automation and intelligentification and accuracy of acquisition is realized from previous craft and is entered into intelligent automatic identification typing
Across, the accuracy rate of its identification is improved, identical or similar mark matching effect in brand recognition retrieval is improved, it is identical to improve
Or the recall rate, precision rate of similar mark, trade mark retrieval working efficiency can be effectively improved.
For the technical solution that the present invention is further explained, spy is with practical application brand recognition search method of the present invention
Example as shown in figure 3, providing a kind of brand recognition search method, is applied in Fig. 1 in this way in one embodiment
It illustrates, includes the following steps for terminal:
Step S310, establishes sample image database, carries out feature extraction, segmentation, combined treatment to sample image, obtains
The assembled unit data of sample image are simultaneously stored in sample image database;
In a specific embodiment, the step of establishing sample image database can specifically include:
Each sample image is collected, the characteristics of image descriptor of each sample image is extracted and store;
The associated text information of typing sample image;
Characteristics of image descriptor is split and according to characteristics of image descriptor minimum unit rule of combination combination
Processing, obtains each characteristics of image descriptor minimum unit and each characteristics of image descriptor combinations cell data;
The word in the associated text information of sample image is split one by one, obtains associated text information minimum unit;According to
Associated text information minimum unit rule of combination is combined each associated text information minimum unit, obtains each spelling words intellectual list
Metadata;Associated text information minimum unit rule of combination includes:By size, color, languages are identical and the closely coupled text connect
Word is confirmed as a connected cypher unit;The connected cypher unit of each default word number fixed length is confirmed as local group
Close unit;Wherein, the value range for presetting word number fixed length is to be taken in connected cypher unit total alpha-numeric 20% or more
Value;
The text pronunciation with word minimum unit assembled unit data match is obtained from word dictionary database, according to
Text pronunciation marks the text pronunciation in each word minimum unit assembled unit data, and it is single to obtain the combination of text pronunciation minimum unit
Metadata;
The word combination with each word minimum unit assembled unit data match is obtained from word dictionary database, is obtained
To word meaning assembled unit data;
By each brand logo constituent encoder of sample trade mark, it is confirmed as brand logo constituent encoder assembled unit number
According to.
In a specific embodiment, the feature extraction to sample image with reference to described in following step S320 is processed
Journey, with sample image be process object, sample image is split, combined treatment, obtain the minimum unit of sample image with
And assembled unit data.
Step S320 converts input trade mark, is divided, combined treatment, and the characteristics of image description of input trade mark is obtained
Symbol, associated text information and assembled unit data.
Specifically, as shown in Figure 4, Figure 5, providing several input trade marks at random, the first exemplary diagram seems Huawei Technologies
The sample trademark pattern of Co., Ltd, the second exemplary diagram seem the figurative mark that " Great Wall " is made of Li Shu Ti word, these figures
Sample can become the embodiment of the present invention and input trade mark.
In an embodiment of the present invention, it is further illustrated in conjunction with Fig. 4, Fig. 5 and sample image (or input trade mark) is done respectively
The extraction of characteristics of image descriptor is divided, the specific implementation process of combined treatment:
One, conversion processing is carried out to input trade mark, including:
Characteristics of image descriptor and associated text are converted into the picture file of the input trade mark indicated with graphic form
The method of the processing of information:
First, the sample image indicated with graphic form or the picture file for inputting trade mark are carried using art methods
Take its characteristics of image descriptor;
By taking Fig. 4 as an example, the patent of invention application No. is 201710553007X can be used《A kind of image contour line descriptor
Acquisition methods, device》The characteristics of image descriptor or image contour line descriptor of extraction, wherein the contour line be based on 10 ×
The characteristics of image descriptor of the conventional coordinates of 10 specifications is:
3,4,5,15,25,35,45,55,65,55,45,44,34,24,23,13;
6,7,8,18,28,27,37,47,56,66,56,46,36,26,16;
12,23,33,34,44,54,55,65,64,54,53,43,42,32,31,21,22;
19,29,30,40,50,49,48,58,57,67,66,56,57,47,37,38,28,29;
41,42,52,53,54,64,65,64,63,62,61,51;
49,50,60,70,69,68,67,57,58,59;
62,63,64,65,74,73,83,82,72;
67,68,69,70,80,79,89,88,78,77;
81,82,92,91;
82,83,93,94,84,94,93,92;
84,85,95,96,95,94;
85,95;
86,96,97,87,97,98,88,98,97,87,97,96;
88,89,90,89,90,100,99,100,99,98;
90,100.
The characteristics of image descriptor of conventional coordinates that the contour line is based on 20 × 20 specifications is:
7,8,9,30,50,70,90,110,130,150,170,190,210,230,250,230,229,209,189,
188,168,148,147,127,107,106,86,66,46,26,27;
12,13,14,34,35,55,75,95,115,114,134,154,174,173,193,212,232,231,251,
231,211,191,171,151,131,111,91,71,51,52,32;
44,64,85,105,106,126,127,147,167,168,188,208,209,229,249,248,228,227,
206,205,185,184,164,163,143,142,122,102,82,83,63;
58,78,98,99,119,139,159,179,178,198,197,196,216,215,235,234,233,253,
252,232,233,213,193,194,174,154,155,135,115,116,96,97,77;
161,162,182,183,184,204,205,225,226,227,247,248,269,268,267,266,265,
264,263,243,242,222,221,201,181;
179,180,200,220,240,260,259,258,278,277,276,275,274,273,253,254,234,
235,236,216,217,197,198,199;
263,264,265,266,267,268,269,288,287,307,306,325,324,304,303,283;
273,274,275,276,277,278,279,299,298,318,317,337,336,315,314,294,293;
321,341,342,343,323,324,344,364,384,383,363,362,361,381,361,341;
324,325,345,365,385,386,367,347,327,347,367,387,386,385,384,364,344;
329,330,350,370,371,391,390,370,369,388,368,348,349;349,350,370,369;
331,332,352,372,373,353,333,334,354,374,375,355,335,336,356,376,375,
395,394,374,354,353,373,393,392,372,371,351;
337,338,339,359,358,357,358,359,379,378,377,398,399,398,397,377,376,
356,357;
340,360,380,400,380,360.
Fig. 6 is the conventional coordinates coordinates regional of the position data of pixel and 10 × 10 specifications on Fig. 4 image contour lines
Correspondence image.
Fig. 7 is the conventional coordinates coordinates regional of the position data of pixel and 20 × 20 specifications on Fig. 4 image contour lines
Correspondence image.
By taking Fig. 5 as an example, the patent of invention application No. is 201710553007X can be used《A kind of image contour line descriptor
Acquisition methods, device》The characteristics of image descriptor or image contour line descriptor of extraction, wherein the contour line be based on 10 ×
The characteristics of image descriptor of the conventional coordinates of 10 specifications is:
6,7,17,27,37,27,28,18,8,9,19,29,30,40,39,49,39,40,50,60,59,69,70,80,
90,100,99,89,79,89,88,98,88,78,88,87,97,96,86,87,77,67,77,76,75,65,66,56,46,
36,26,16;
38,48;
47,57;
58,68;
58,59,69,79,78,68;
2,12,22,23,13,14,4,14,24,23,33,32,42,43,44,34,35,45,55,54,53,63,64,
74,75,85,95,94,84,74,73,83,93,92,82,72,62,52,51,41,31,41,42,32,22,12;
52,53,52,53,63,73,72,62;
9,10,20,19,29,19.
The characteristics of image descriptor of conventional coordinates that the contour line is based on 20 × 20 specifications is:
16,17,37,57,77,97,98,118,119,120,140,160,159,158,157,177,197,198,178,
158,159,179,199,219,239,238,258,278,279,299,319,320,340,360,380,400,399,398,
378,358,338,337,317,337,357,356,376,356,355,335,315,316,315,295,315,335,334,
354,374,373,372,352,332,333,313,293,294,274,273,293,292,312,311,291,290,270,
251,252,232,212,192,191,171,151,131,132,112,92,72,52,32,33,53,73,93,113,133,
134,114,94,95,115,116,96,76,56,36;
155,156,176,175;
173,174,194,214,234,233,213,193;
215,216,236,256,276,275,255,235;
216,217,237,257,277,297,296,276,256,236;
3,4,24,44,64,84,85,65,66,46,47,27,28,48,68,88,87,107,106,126,125,124,
144,164,165,166,167,168,148,149,169,189,209,208,207,206,205,225,226,246,247,
267,268,288,289,309,310,330,350,370,390,389,388,368,367,347,327,307,306,326,
345,365,364,384,383,363,343,323,303,283,263,243,223,203,202,222,221,201,181,
161,141,142,162,163,143,123,103,83,63,43,23;
204,205,204,224,225,245,265,266,286,285,305,304,284,264,244,224;
18,19,39,59,79,78,98,97,77,78,58,38.
Fig. 8 is the conventional coordinates coordinates regional of the position data of pixel and 10 × 10 specifications on Fig. 5 image contour lines
Correspondence image.
Fig. 9 is the conventional coordinates coordinates regional of the position data of pixel and 20 × 20 specifications on Fig. 5 image contour lines
Correspondence image.
Second, using the mass data information of the existing sample trade mark recorded, sample is retrieved based on characteristics of image descriptor
This image data base obtains the sample image corresponding to matched characteristics of image descriptor and the characteristics of image descriptor and association
Text message, using the information as the characteristics of image descriptor and associated text information of input trade mark, wherein associated text information
Including:Brand logo constituent encoder in the figurative mark that has recorded of matched sample image, described in figurative mark
Things title, texts and the form-pronunciation-meaning such as trade mark style, pronunciation, meaning and the nearly word of its nearly word form, sound, the nearly word of justice
Feature.
Characteristics of image descriptor and associated text information are converted into the word with the input trade mark of written form typing
The method of processing, including:
First, based on using the word of the input trade mark of written form typing as keyword retrieval sample image database, obtain
To the sample word to match;By taking Fig. 5 as an example, the word of the input trade mark of typing is " Great Wall ", is examined for keyword with " Great Wall "
Rope sample image database, the record for the sample word " Great Wall " that can be matched.
Second, find out the sample image and associated text information corresponding to the matched sample word, wherein associated text
Information includes:Brand logo constituent encoder in the figurative mark that the matched sample image has recorded, described in figurative mark
Things title, to the characteristics of image descriptor that image is indicated, trade mark style, pronunciation, meaning and its
Texts and the form-pronunciation-meaning features such as the nearly word of the nearly word of nearly word form, sound, justice.Wherein, the word includes Chinese text, non-Chinese text
Word, number, symbol.
In upper example, through retrieve can get " Great Wall " word corresponding to sample trade mark and associated text information, Fig. 5 can be
One of corresponding sample trade mark, associated text information includes:The characteristics of image of the various written form institute composition pictures of " Great Wall " word
Descriptor, texts and the form-pronunciation-meaning feature such as " Great Wall " text pronunciation, meaning and the nearly word of its nearly word form, sound, the nearly word of justice.
Two, dividing processing is done respectively to characteristics of image descriptor and associated text information;
It is exactly that the minimum unit of characteristics of image descriptor is identified to do dividing processing to characteristics of image descriptor, will be schemed
As each minimum unit of feature descriptor is split, it is exactly to associated text information to do dividing processing to associated text information
Minimum unit be identified, each minimum unit of associated text information is split.
In example as above, the characteristics of image descriptor of image be used to indicate the position data of a certain pixel of image contour line with
The characteristic point of this image of the correspondence of conventional coordinates coordinates regional of a certain specification therefore can be by each specification
The position data of one coordinates regional corresponding image contour line one or more pixel of conventional coordinates is considered as image spy
Levy descriptor minimum unit.
For Fig. 7, in the characteristics of image descriptor of conventional coordinates of the contour line based on 10 × 10 specifications " 3,4,
5,15,25,35,45,55,65,55,45,44,34,24,23,13 " characteristics of image descriptor minimum unit is respectively descriptor
In each number, i.e.,:"3","4","5","15","25","35","45","55","65","55","45","44",
“34”、“24”、“23”、“13”。
For another example for Fig. 7, in the characteristics of image descriptor of conventional coordinates of the contour line based on 20 × 20 specifications " 7,
8,9,30,50,70,90,110,130,150,170,190,210,230,250,230,229,209,189,188,168,148,
147,127,107,106,86,66,46,26,27 " characteristics of image descriptor minimum unit is respectively each in descriptor
Number, i.e.,:"7","8","9","30","50","70","90","110","130","150","170","190","210",
“230”、“250”、“230”、“229”、“209”、“189”、“188”、“168”、“148”、“147”、“127”、“107”、
“106”、“86”、“66”、“46”、“26”、“27”。
For using input " Great Wall " word as input trade mark, in vpg connection, associated text information is " Great Wall ", " long
Each word is the minimum units that the minimum unit of the input trade mark, i.e. " length " and " city " are respectively the input trade mark in city ".
Three, processing is combined to minimum unit data;
It is exactly to carry out group according to preset rule of combination to minimum unit data to be combined processing to minimum unit data
It closes, obtains assembled unit data and the arbitrary local message of trademark image, wherein the assembled unit data are corresponding described image
Multiple character strings of any local feature represented by feature descriptor or associated text information.
It obtains assembled unit data and the arbitrary local message specific method of trademark image includes:
First, according to the needs of application, establish preset characteristics of image descriptor minimum unit rule of combination, wherein pre-
If characteristics of image descriptor minimum unit rule of combination specifically include:
The characteristics of image descriptor minimum unit rule of combination of image contour line includes:1) on each image contour line
Whole line segments are considered as an image entire combination unit;2) the closed loop line on each image contour line is considered as a connected domain group
Close unit;3) line segment on the image contour line of every 1 first default fixed length is considered as a line segment assembled unit, wherein first is pre-
Setting length can be in 20% and the above value of its line segment overall length.
The characteristics of image descriptor minimum unit rule of combination of image framework line includes:1) on each image framework line
Whole line segments are considered as an image entire combination unit;2) continual line is considered as a connection on each image framework line
Domain assembled unit;3) line segment on the image framework line of every 1 second default fixed length is considered as a line segment assembled unit, wherein the
Two default fixed length can be in 20% and the above value of its line segment overall length.
Second, it is by aforementioned preset characteristics of image descriptor minimum unit rule of combination, characteristics of image descriptor is minimum
Unit is combined, and obtains the assembled unit data and the arbitrary local message of trademark image of characteristics of image descriptor respectively.
In some embodiments of the invention, acquired image feature descriptor assembled unit data can be used to indicate that one
A connected domain assembled unit data may also indicate that a line segment assembled unit data.Wherein, each connected domain assembled unit data
It is exactly the arbitrary local message of image represented by the characteristics of image descriptor.
For the contour line is based on the characteristics of image descriptor of the conventional coordinates of 10 × 10 specifications in Fig. 5, Mei Gelian
Logical domain assembled unit or the arbitrary local message of image include as follows:
" 3,4,5,15,25,35,45,55,65,55,45,44,34,24,23,13 ",
" 6,7,8,18,28,27,37,47,56,66,56,46,36,26,16 ",
" 12,23,33,34,44,54,55,65,64,54,53,43,42,32,31,21,22 ",
" 19,29,30,40,50,49,48,58,57,67,66,56,57,47,37,38,28,29 ",
" 41,42,52,53,54,64,65,64,63,62,61,51 ",
" 49,50,60,70,69,68,67,57,58,59 ",
" 62,63,64,65,74,73,83,82,72 ",
" 67,68,69,70,80,79,89,88,78,77 ",
" 81,82,92,91 ",
" 82,83,93,94,84,94,93,92 ",
" 84,85,95,96,95,94 ",
" 85,95 ",
" 86,96,97,87,97,98,88,98,97,87,97,96 ",
" 88,89,90,89,90,100,99,100,99,98 ",
" 90,100 ".
For different information objects, illustrate the identification segmentation and combination of its associated text information minimum unit respectively as follows
Handle specific method:
1, the word content of reading, i.e., the word contained in trade mark can be recognized.
The word contained in trade mark includes:Chinese written language, domestic minority language, foreign language, domestic ethnic group
Word and foreign language can also several different languages words in the side of subdivision by different languages.The word contained in trade mark is divided
It is exactly that word is split one by one by word that trade mark contains to cut processing, keeps the word that each word becomes the trade mark minimum single
Member, word number are exactly its minimum unit number;To the word contained in trade mark do trade mark word minimum unit combined treatment be exactly by
It is combined according to following rule of combination, by spelling words intellectual as a result, obtaining word minimum unit assembled unit data respectively:
Trade mark word minimum unit rule of combination includes:1) each size, color, languages are identical and closely coupled connect
Word is considered as a connected cypher unit;2) local group on the connected cypher unit of each default word number fixed length
Close unit, wherein default word number fixed length can be in total alpha-numeric 20% or more the value of connected cypher unit.
2, the text pronunciation of reading can be recognized;
To the word minimum unit assembled unit data of aforementioned acquisition, the text to match can be obtained from word dictionary database
Character pronunciation, to its text pronunciation of its label character, the text pronunciation minimum unit assembled unit data of acquisition.
3, the spelling words intellectual meaning of reading can be recognized;
It, can be from word dictionary number to the word minimum unit assembled unit data and trade mark entirety spelling words intellectual of aforementioned acquisition
The word combination to match is obtained according to library, which is considered as the significant spelling words intellectual of trade mark, it is impossible to which trade mark is complete
Portion's word progress word combination is considered as spelling words intellectual of the trade mark without meaning, wherein can trade mark whole word be carried out word group
The each significant spelling words intellectual closed is considered as a spelling words intellectual meaning minimum unit assembled unit data.
4, brand logo constituent encoder;
When the brand logo constituent encoder of each trade mark has multiple, using each brand logo constituent encoder as
Brand logo constituent encoder minimum unit assembled unit data.
In a specific embodiment, sample trademark database is established, it, will be to sample branding data by abovementioned steps
It converted, divided, the assembled unit data of sample trade mark obtained by combined treatment and the arbitrary local message of trademark image and will
It is stored in sample trademark database.
Step S330 is based on assembled unit data retrieval sample trademark database, obtains matched preliminary search sample quotient
Text and form-pronunciation-meaning characteristic information, the matched minimum list of institute recorded in image, trade mark associated by mark and the sample trade mark
Member and assembled unit data.
Step S340 is calculated between preliminary search sample trade mark and input trade mark in shape, the list of sound, justice, search key
The approximate rate of item and comprehensive approximate rate.
Step S350 meets default individual event approximation rate and comprehensive approximate rate by comprehensive approximate rate and/or sequence ranking meets
The sample retrieval trade mark of default ranking is ranked up, and reports retrieval result.
In a specific embodiment, the specific implementation of step S330 to step S350 can refer to input trade mark
Processing procedure described in step S240 to step S260, is process object to input trade mark, and retrieval matching is carried out to input trade mark,
And obtain final retrieval result.
Brand recognition search method of the present invention, can be by the trade mark and knowledge data information of the existing magnanimity of system, from shape
The presumption for carrying out text identification and its form-pronunciation-meaning characteristic information in terms of the pronunciation and meaning to input trademark image obtains, can be by with picture
The picture file for the input trade mark or sample trade mark that form indicates and input trade mark or sample trade mark with written form record
Word is converted into the processing of characteristics of image descriptor and associated text information respectively, to the handling result be split, group
Conjunction is handled, and the arbitrary local message of assembled unit data and trademark image of input trade mark or sample trade mark is obtained, with big data
Related information presumption identify the pronunciation of image not directly displayed out in the image of input trade mark or sample trade mark, group of text
Meaning, brand logo constituent encoder and other reflection image form-pronunciation-meaning features of conjunction, with acquired assembled unit data and quotient
The arbitrary local message of logo image effectively overcomes manual entry to be in the past not easy in input trademark image as search key
Word carry out the part combination etc. of nearly word form, the nearly word of sound, the nearly word of justice, figure constitution and carry out exhaustive, easily cause search key
The defect that disunity, extraction information are easily omitted;The text image of specification and the text image of non-standard can effectively be identified,
Traditional technology method can be overcome to have easily caused the defect of trade mark search key omission, can effectively solve the problem that and retrieved in trade mark retrieval
The automation and intelligentification and the comprehensive sex chromosome mosaicism of accuracy that keyword obtains, realize from previous craft and are entered into intelligent automatic identification
The leap of typing improves the accuracy rate of its identification, improves identical or similar mark matching effect in brand recognition retrieval, to carry
High identical or similar mark recall rate, precision rate can effectively improve trade mark retrieval working efficiency.
It should be understood that although each step in the flow chart of Fig. 2, Fig. 3 is shown successively according to the instruction of arrow,
Be these steps it is not that the inevitable sequence indicated according to arrow executes successively.Unless expressly stating otherwise herein, these steps
There is no stringent sequences to limit for rapid execution, these steps can execute in other order.And in Fig. 2, Fig. 3 at least
A part of step may include that either these sub-steps of multiple stages or stage are not necessarily in same a period of time to multiple sub-steps
Quarter executes completion, but can execute at different times, the execution in these sub-steps or stage be sequentially also not necessarily according to
Secondary progress, but can either the sub-step of other steps or at least part in stage in turn or replace with other steps
Ground executes.
In one embodiment, as shown in Figure 10, a kind of brand recognition retrieval device is provided, including:
Conversion module 110 is inputted for the image data by sample retrieval image data base conversion input trade mark
The characteristics of image descriptor and associated text information of trade mark;Sample image database pre-establishes, including sample image
Characteristics of image descriptor, associated text information, minimum unit and assembled unit data database;Assembled unit data are
Characterize the data of the arbitrary local message of image;
Divide module 120, for the characteristics of image descriptor and associated text information of segmentation input trade mark respectively, obtains defeated
Enter each characteristics of image descriptor minimum unit, each associated text information minimum unit of trade mark;Characteristics of image descriptor is minimum single
Member is one or more character strings of any image characteristic point represented by correspondence image feature descriptor;Associated text information is most
Junior unit is a word of any text message characteristic point represented by corresponding associated text information or multiple significant texts
Word combines;
Composite module 130, for according to default minimum unit rule of combination, being respectively combined each characteristics of image of input trade mark
Descriptor minimum unit, each associated text information minimum unit obtain the characteristics of image descriptor combinations unit number of input trade mark
According to, associated text information assembled unit data;
Module 140 is retrieved, for based on characteristics of image descriptor combinations cell data, associated text information assembled unit number
According to the sample trademark database in sample retrieval image data base, matched each preliminary search sample trade mark and preliminary inspection are obtained
Each characteristics of image descriptor minimum unit, each associated text information minimum unit of rope sample trade mark;
Obtain approximate rate module 150, for according to each characteristics of image descriptor minimum unit of preliminary search sample trade mark,
Each characteristics of image descriptor minimum unit of each associated text information minimum unit and input trade mark, each associated text information are most
Junior unit obtains individual event approximation rate;Individual event matching rate is handled, it is approximate with the synthesis of trade mark is inputted to obtain preliminary search sample trade mark
Rate;
Sorting module 160, the preliminary search sample trade mark for meeting preset requirement to comprehensive approximate rate are ranked up, obtain
To retrieval result.
In a specific embodiment, input trade mark includes with the input trade mark of graphic form typing and with written form
The input trade mark of typing;Sample image includes with the sample image of graphic form typing and with the sample graph of written form typing
Picture;
Sample image includes sample trademark pattern, appearance design pattern, the work of fine arts pattern of copyright registration, each Chinese written language
Pattern, each non-Chinese written language pattern and custom images;Sample image database further includes trade mark inscape sample image
Database, word dictionary database and word dictionary database;
The associated text information of sample image includes brand logo constituent encoder, the sample image of the sample image recorded
Described things title and sample image can recognize the text and form-pronunciation-meaning feature of the word of reading;Form-pronunciation-meaning feature includes sample graph
The graphics shape form of expression or style of picture, pronunciation, meaning and the nearly word of nearly word form, sound and the nearly word of justice;
Characteristics of image descriptor is using the identical or similar character string of height, to having in input trade mark or sample image
Same perception content or feature are recorded, and use different character strings to having the not same feeling in input trade mark or sample image
The image feature representation form that perhaps feature is recorded in knowing;Image feature representation form is description input trade mark or sample graph
The set of one or more groups of character strings of the characteristics of image of picture;
Default minimum unit rule of combination includes characteristics of image descriptor minimum unit rule of combination and associated text information
Minimum unit rule of combination;Characteristics of image descriptor combinations cell data includes for indicating connected domain assembled unit data, using
In expression line segment assembled unit data and for the string data of storage;Associated text information assembled unit data include text
Word assembled unit data, text pronunciation assembled unit data, word meaning assembled unit data and brand logo constituent encoder
Assembled unit data;
Further include establishing database module, for establishing sample image database.
In a specific embodiment, the word in associated text information includes the outer national literature of Chinese text, each languages
Word, number and symbol;
Database module is established, for collecting each sample image, extracts and store the characteristics of image description of each sample image
Symbol;The associated text information of typing sample image;Characteristics of image descriptor is split and according to characteristics of image descriptor
The combined treatment of minimum unit rule of combination obtains each characteristics of image descriptor minimum unit and each characteristics of image set of descriptors
Close cell data;The word in the associated text information of sample image is split one by one, obtains associated text information minimum unit;It presses
Each associated text information minimum unit is combined according to associated text information minimum unit rule of combination, obtains each spelling words intellectual
Cell data;Associated text information minimum unit rule of combination includes:By size, color, languages are identical and closely coupled connect
Word is confirmed as a connected cypher unit;The connected cypher unit of each default word number fixed length is confirmed as part
Assembled unit;Wherein, the value range for presetting word number fixed length is in connected cypher unit total alpha-numeric 20% or more
Value;The text pronunciation with word minimum unit assembled unit data match is obtained from word dictionary database, according to text
Character pronunciation marks the text pronunciation in each word minimum unit assembled unit data, obtains text pronunciation minimum unit assembled unit
Data;The word combination with each word minimum unit assembled unit data match is obtained from word dictionary database, is obtained
Word meaning assembled unit data;By each brand logo constituent encoder of sample trade mark, it is confirmed as brand logo element volume
Code character closes cell data.
In a specific embodiment, conversion module, for extracting with the image of the input trade mark of graphic form typing
Feature descriptor;It is based on characteristics of image descriptor sample retrieval image data base, matched characteristics of image descriptor is corresponding
Sample image is considered as or height approximate image identical as the input image of trade mark, and the characteristics of image that sample image has been recorded
Descriptor, associated text information are confirmed as with the characteristics of image descriptor of the input trade mark of graphic form typing, associated text letter
Breath;And
Based on written form typing input trade mark character search sample image database, by matched sample word
Characteristics of image descriptor that corresponding sample image has recorded, associated text information are confirmed as the input quotient with written form typing
Target characteristics of image descriptor, associated text information.
In a specific embodiment, pre-set image feature descriptor minimum unit rule of combination may include image wheel
The characteristics of image descriptor minimum unit combination of the characteristics of image descriptor minimum unit rule of combination and image framework line of profile
Rule;
The characteristics of image descriptor minimum unit rule of combination of image contour line includes:It will be complete on any image contour line
Portion's line segment is confirmed as an image entire combination unit;It is a connected domain group by the closed loop line justification on any image contour line
Close unit;Line segment on the image contour line of any first default fixed length is confirmed as a line segment assembled unit;Wherein, first
The value range of default fixed length is more than or equal to 20% of the line segment overall length on image contour line;
The characteristics of image descriptor minimum unit rule of combination of image framework line includes:It will be complete on any image skeleton line
Portion's line segment is confirmed as an image entire combination unit;Continual line on any image skeleton line is confirmed as a connection
Domain assembled unit;Line segment on the image framework line of any second default fixed length is confirmed as a line segment assembled unit;Wherein,
The value range of second default fixed length is more than or equal to 20% of the line segment overall length on image framework line.
In a specific embodiment, characteristics of image descriptor is for indicating that image contour line or image framework line are appointed
The feature descriptor of the position data of one pixel and the correspondence of the conventional coordinates coordinates regional of any specification;
Characteristics of image descriptor minimum unit is any coordinates regional corresponding image of the conventional coordinates of any specification
The position data of one or more pixels of contour line or image framework line;
Associated text information minimum unit has for the associated text information correspondence represented by any word or spelling words intellectual
The word of meaning or the data of vocabulary.
In a specific embodiment, associated text information minimum unit include Chinese minimum unit and non-Chinese most
Junior unit;Individual event approximation rate includes Chinese individual event approximation rate, non-Chinese individual event approximation rate and characteristics of image individual event approximation rate;
Approximate rate module is obtained, is used for:
Obtain the sum and characteristics of image descriptor of total, the non-Chinese minimum unit of the Chinese minimum unit of input trade mark
The sum of minimum unit, Chinese minimum unit total number, the non-Chinese of preliminary search sample trade mark matching input trade mark are minimum single
First total number and characteristics of image descriptor minimum unit total number, preliminary search sample trade mark mismatch the Chinese of input trade mark most
Junior unit total number, non-Chinese minimum unit total number and characteristics of image descriptor minimum unit total number;
Chinese minimum unit matching rate is obtained based on following formula:
Ma1=(Ua1÷U01) × 100%
Wherein, Ma1Indicate Chinese minimum unit matching rate, U01Indicate the sum of the Chinese minimum unit of input trade mark, Ua1
Indicate the Chinese minimum unit total number of preliminary search sample trade mark matching input trade mark;
Non- Chinese minimum unit matching rate is obtained based on following formula:
Ma2=(Ua2÷U02) × 100%
Wherein, Ma2Indicate non-Chinese minimum unit matching rate, U02Indicate the total of the non-Chinese minimum unit of input trade mark
Number, Ua2Indicate the non-Chinese minimum unit total number of preliminary search sample trade mark matching input trade mark;
Characteristics of image descriptor minimum unit matching rate is obtained based on following formula:
Ma0=(Ua0÷U00) × 100%
Wherein, Ma0Indicate characteristics of image descriptor minimum unit matching rate, U00Indicate the characteristics of image description of input trade mark
Accord with the sum of minimum unit, Ua0Indicate that the characteristics of image descriptor minimum unit of preliminary search sample trade mark matching input trade mark closes
It counts;
Chinese minimum unit probability of mismatch is obtained based on following formula:
Mi1=(Uc1÷U01) × 100%+ (n1-1)×ω1
Wherein, Mi1Indicate Chinese minimum unit probability of mismatch, U01Indicate the sum of the Chinese minimum unit of input trade mark,
Uc1Indicate that preliminary search sample trade mark mismatches the Chinese minimum unit total number of input trade mark, n1Indicate preliminary search sample quotient
Mark combines the place's number not matched that on line, ω in Chinese minimum unit with input trade mark1Number n at expression1Flexible strategy;Wherein,
ω1Value range be less than or equal to 80%;
Non- Chinese minimum unit probability of mismatch is obtained based on following formula:
Mi2=(Uc2÷U02) × 100%+ (n2-1)×ω2
Wherein, Mi2Indicate non-Chinese minimum unit probability of mismatch, U02Indicate the total of the non-Chinese minimum unit of input trade mark
Number, Uc2Indicate that preliminary search sample trade mark mismatches the non-Chinese minimum unit total number of input trade mark, n2Indicate preliminary search
Sample trade mark combines the place's number not matched that on line, ω with input trade mark in non-Chinese minimum unit2Number n at expression2Power
Number;Wherein, ω2Value range be less than or equal to 80%;
Characteristics of image descriptor minimum unit probability of mismatch is obtained based on following formula:
Mi0=(Uc0÷U00) × 100%+ (n0-1)×ω0
Wherein, Mi0Indicate characteristics of image descriptor minimum unit probability of mismatch, U00Indicate that the characteristics of image of input trade mark is retouched
State the sum of symbol minimum unit, Uc0Indicate that preliminary search sample trade mark mismatches the minimum list of characteristics of image descriptor of input trade mark
First total number, n0Indicate that preliminary search sample trade mark combines institute on line with input trade mark in characteristics of image descriptor minimum unit
The place's number not matched that, ω0Number n at expression0Flexible strategy;Wherein, ω0Value range be less than or equal to 80%;
Based on following formula, Chinese individual event approximation rate is obtained:
M1=Ma1-Mi1×β1
Wherein, M1Indicate that Chinese individual event approximation rate, β 1 indicate Mi1Flexible strategy;Wherein, the value range of β 1 is to be less than or wait
In 80%;
Based on following formula, non-Chinese individual event approximation rate is obtained
M2=Ma2-Mi2×β2
Wherein, M2Indicate that non-Chinese individual event approximation rate, β 2 indicate Mi2Flexible strategy;Wherein, the value range of β 2 be less than or
Equal to 80%;
Based on following formula, characteristics of image individual event approximation rate is obtained:
M0=Ma0-Mi0×β0
Wherein, M0Indicate characteristics of image individual event approximation rate, β0Indicate Mi0Flexible strategy;Wherein, β0Value range be less than or
Equal to 80%.
In a specific embodiment, approximate rate module is obtained to be additionally operable to:
Based on following formula, comprehensive approximate rate is obtained:
M=(M1+M2+M0)÷μ
Wherein, μ indicates M1、M2、M0It is not 0 item number.
In a specific embodiment, non-Chinese minimum unit is English minimum unit;Non- Chinese minimum unit matching
Rate is English minimum unit matching rate;Non- Chinese minimum unit probability of mismatch is English minimum unit probability of mismatch;Non- Chinese is single
The approximate rate of item is English individual event approximation rate;
It is characteristics of image line that characteristics of image descriptor minimum unit, which combines line,;Chinese minimum unit combination line is Chinese
Minimum unit that the corresponding form-pronunciation-meaning feature of trade mark word is constituted, by the path line to be formed that puts in order;Non- Chinese minimum unit
Combination line is the minimum unit of the corresponding form-pronunciation-meaning feature composition of non-Chinese trade mark word, by the track to be formed that puts in order
Line.
In a specific embodiment, sorting module, it is first more than or equal to 30% for filtering out comprehensive approximate rate
Sample retrieval trade mark is walked, and the preliminary search sample trade mark filtered out is ranked up, and sequence ranking is taken to be less than or equal to 500
Within preliminary search sample trade mark as retrieval result.
Above each module can be embedded in or in the form of hardware independently of in the processor in computer equipment, can also be with soft
Part form is stored in the memory in computer equipment, and the corresponding behaviour of the above modules is executed in order to which processor calls
Make.
The specific of device, which is retrieved, about brand recognition limits the limit that may refer to above for brand recognition search method
Fixed, details are not described herein.Modules in above-mentioned brand recognition retrieval device can fully or partially through software, hardware and its
It combines to realize.Above-mentioned each module can be embedded in or in the form of hardware independently of in the processor in computer equipment, can also
It is stored in a software form in the memory in computer equipment, in order to which processor calls the above modules of execution corresponding
Operation.
In one embodiment, a kind of computer equipment is provided, which can be terminal, internal structure
Figure can be as shown in figure 11.The computer equipment includes processor, memory, network interface, the number connected by system bus
According to library, display screen and input unit.Wherein, the processor of the computer equipment is for providing calculating and control ability.The calculating
The memory of machine equipment includes non-volatile memory medium, built-in storage.The non-volatile memory medium be stored with operating system,
Computer program and database.The built-in storage is the operation of the operating system and computer program in non-volatile memory medium
Environment is provided.The database of the computer equipment is for storing trade mark sample image and database data.The computer is set
Standby network interface is used to communicate by network connection with external terminal.To realize when the computer program is executed by processor
A kind of image text and form-pronunciation-meaning characteristic recognition method.The display screen of the computer equipment can be liquid crystal display or electronics
The input unit of ink display screen, the computer equipment can be the touch layer covered on display screen, can also be that computer is set
Button, trace ball or the Trackpad being arranged on standby shell, can also be external keyboard, Trackpad or mouse etc..
It will be understood by those skilled in the art that structure shown in Figure 11, only with the relevant part of application scheme
The block diagram of structure, does not constitute the restriction for the computer equipment being applied thereon to application scheme, and specific computer is set
Standby may include either combining certain components than more or fewer components as shown in the figure or being arranged with different components.
In one embodiment, a kind of computer equipment, including memory and processor are provided, is stored in memory
Computer program, the processor realize each step in above-mentioned brand recognition search method when executing computer program.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program realizes each step in above-mentioned brand recognition search method when being executed by processor.
One of ordinary skill in the art will appreciate that realizing all or part of flow in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, computer program can be stored in a non-volatile computer and can be read
In storage medium, the computer program is when being executed, it may include such as the flow of the embodiment of above-mentioned each method.Wherein, the application
Any reference to memory, storage, database or other media used in each embodiment provided, may each comprise non-
Volatibility and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM (PROM),
Electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include arbitrary access
Memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms, such as static
RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhanced SDRAM
(ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) directly RAM (RDRAM), straight
Connect memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above example can be combined arbitrarily, to keep description succinct, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield is all considered to be the range of this specification record.
Only several embodiments of the present invention are expressed for above example, the description thereof is more specific and detailed, but can not
Therefore it is construed as limiting the scope of the patent.It should be pointed out that for those of ordinary skill in the art,
Under the premise of not departing from present inventive concept, various modifications and improvements can be made, these are all within the scope of protection of the present invention.
Therefore, the protection domain of patent of the present invention should be determined by the appended claims.