CN106777050A - It is a kind of based on bag of words and to take into account the footwear stamp line expression and system of semantic dependency - Google Patents
It is a kind of based on bag of words and to take into account the footwear stamp line expression and system of semantic dependency Download PDFInfo
- Publication number
- CN106777050A CN106777050A CN201611130372.1A CN201611130372A CN106777050A CN 106777050 A CN106777050 A CN 106777050A CN 201611130372 A CN201611130372 A CN 201611130372A CN 106777050 A CN106777050 A CN 106777050A
- Authority
- CN
- China
- Prior art keywords
- footwear
- semantic
- stamp
- print image
- primitive
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
- G06F16/5838—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content using colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/50—Extraction of image or video features by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Library & Information Science (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Processing Or Creating Images (AREA)
- Image Analysis (AREA)
Abstract
Based on bag of words and the footwear stamp line expression and system of semantic dependency are taken into account the invention discloses a kind of.Specific steps of the present invention include:S1. footwear stamp print image semantic vocabulary relation table is built;S2. footwear stamp print image is gathered;S3. footwear stamp print image primitive is extracted;S4. the textural characteristics of footwear stamp print image primitive are extracted;S5. footwear stamp print image primitive semantic classes is judged;S6. footwear stamp print image semanteme frequency histogram is counted;S7. obtaining can express the footwear stamp print image semantic meaning representation histogram of image, semantic correlation.The present invention effectively avoids interference of the semantic dependency to semantic meaning representation by building semantic dependency model, improves the accuracy of semantic meaning representation.
Description
Technical field
It is specifically a kind of based on bag of words and to take into account semantic dependency the present invention relates to technical field of image processing
Footwear stamp line expression and system.
Background technology
The method commonly used in image expression field includes:
(1) the image, semantic expression of view-based access control model bag of words.Its general principle is to regard image as unordered feature list
Set of words, by counting the number of times that each feature word occurs in the picture, obtains feature word frequencies histogram vectors, with this
As the expression of image.
(2) the image, semantic expression of view-based access control model language model.Image is viewed as having certain arrangement by visual language model
The document of order, can complete the semantic meaning representation of entire image by Local Co-occurrence frequency in image and spatial relationship.
(3) semantic meaning representation based on study encoding model.Its main representative is sparse coding.Sparse coding is to utilize image
The combination of feature bases describes picture material.
At present, algorithm above still suffers from some problems for image expression:
(1) algorithm does not consider that semanteme has certain contact (i.e. in the presence of semantic phase in specific feature aspect at present
Closing property), and the consequence such as cause quantization error serious, so as to semantic tagger, semantic meaning representation produces extreme influence.
(2) some study encoding models only have the coding layer of individual layer, and the visual dictionary for learning lacks the selection to feature
Property, reduce the semantic resolving power of picture material.
(3) semantic results of the algorithm generally to marking are no at present further feeds back and adjusts, more or dependence
In the selection of low-level image feature.
The content of the invention
In view of the deficiency that prior art is present, the invention aims to providing one kind based on bag of words and taking into account semanteme
The footwear stamp line expression of correlation.
To achieve these goals, technical solution of the present invention is as follows:
It is a kind of based on bag of words and to take into account the footwear stamp line expression of semantic dependency, it is characterised in that specific step
Suddenly include:
S1, in advance structure footwear stamp print image semantic vocabulary relation table;
S2, collection footwear stamp print image;
S3, the footwear stamp print image to being gathered carry out the extraction of footwear stamp print image primitive, to obtain corresponding footwear stamp
Print image primitive;
S4, the Wavelet Fourier plum forests feature for extracting footwear stamp print image primitive;
S5, successively according to the Wavelet Fourier plum forests feature and footwear stamp line of each footwear stamp print image primitive for being extracted
Image, semantic lexical relation table carries out characteristic matching, and is determined corresponding to the footwear stamp print image primitive based on obtained matching degree
Semantic classes;
S6, the number of times that each semanteme occurs in the footwear stamp print image is counted, to obtain footwear stamp print image language
Adopted frequency histogram;
S7, based on the footwear stamp print image semanteme frequency histogram and footwear stamp print image semantic vocabulary relation table for being obtained
Acquisition can express the footwear stamp print image semantic meaning representation histogram of image, semantic correlation;
A kind of based on bag of words and the footwear stamp line table of semantic dependency is taken into account another object of the present invention is to provide
Up to system, the system includes:
Relation table builds module, based on the geometry that footwear are patterned, builds footwear stamp print image semantic vocabulary and closes
It is table;
Extraction module, the footwear stamp print image to gathering carries out footwear stamp print image primitive and extracts, onestep extraction footwear of going forward side by side
The textural characteristics of stamp print image primitive;
Primitive semantic classes determination module, according to footwear stamp print image semantic relation table and footwear stamp print image primitive line
The comparing result of feature is managed, the semantic classes corresponding to footwear stamp print image primitive is judged;
Image expression module, statistics footwear stamp print image semanteme frequency histogram, and further treatment acquisition can take into account
The footwear stamp print image semantic meaning representation histogram of image, semantic correlation.
Compared with prior art, beneficial effects of the present invention:
(1) present invention determines footwear stamp line semantic word in specific aspect by footwear stamp line semantic vocabulary relation table
Semantic dependency, so as to effectively prevent interference of the semantic dependency to semantic meaning representation.
(2) the semantic dependency processing mode that the present invention is used effectively forms a kind of feedback mechanism, can be to semantic tagger
Result exercises supervision and adjusts, and improves the accuracy of semantic meaning representation.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are this hairs
Some bright embodiments, for those of ordinary skill in the art, without having to pay creative labor, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is that footwear stamp line semantic vocabulary relation table of the present invention builds flow chart;
Fig. 2 is that footwear stamp line of the present invention expresses flow chart;
Fig. 3 is footwear stamp line expression system structure chart of the present invention;
Fig. 4 is that footwear stamp line expression system relation table of the present invention builds modular structure;
Fig. 5 is footwear stamp line expression system extraction module structure chart of the present invention;
Fig. 6 is footwear stamp line semantic relation schematic diagram of the present invention.
Specific embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
A part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
Based on bag of words and the footwear stamp line expression and system of semantic dependency are taken into account the invention provides a kind of,
Based on bag of words and take into account semantic dependency footwear stamp line expression flow as shown in Fig. 2 its step includes:
S1, in advance structure footwear stamp print image semantic vocabulary relation table, construction step include:
S1.1, structure semantic vocabulary table
S1.1.1, the geometry in advance corresponding to foundation footwear stamp print image are classified and sorted each shape
Set respectively and be designated as R respectively with its one-to-one semantic label, institute's semantic tags1,R2,…,RN, it represents footwear stamp line
True semanteme, the geometry can such as setback type, lattice block type, class it is round;
Footwear stamp print image primitive in S1.1.2, delimitation footwear stamp print image, and to the footwear stamp under same semantic label
Line is clustered to obtain each corresponding footwear stamp line set of letters, in footwear watermark image, independent connected region is claimed
Be footwear stamp line primitive, it is the minimum independent individuality that decorative pattern distribution is constituted in footwear stamp print image, due to footwear print into
Trace body, the difference for wearing time, the walking habits of people and imaging modality, same footwear stamp line have different on image
Comprising multiple footwear stamp line words under the form of expression, therefore each semantic label, the footwear stamp line under same semantic label is entered
Row cluster, footwear stamp line word is referred to as by all kinds of representative primitives.It is designated as Rl={ α1,α2,...αn, wherein RlRepresent semantic mark
Sign RlUnder footwear stamp line set of letters, αiRepresent footwear stamp line word;
S1.1.3, footwear stamp line semanteme word is constituted based on the corresponding footwear stamp line set of letters of each semantic label for being obtained
Remittance table, it is designated as Sv={ R1,R2,...,RN, wherein R1,R2,...,RNSemantic label R is represented respectively1,R2,…,RNUnder it is each
Footwear stamp line set of letters;
S1.2, differentiation semantic dependency
It is residual due to the abrasion that the more and different user of decorative pattern classification in footwear stamp print image causes to decorative pattern image
Lack the difference of equivalent damage, cause to exist between semantic label to connect each other (i.e. semantic dependency).In order to be able to build exactly
Go out semantic vocabulary relation table, it is necessary to carry out the judgement of semantic dependency first.Comprise the following steps that:
S1.2.1, the footwear stamp line primitive of marked semantic label is taken as the trainer shoe stamp line for differentiating semantic dependency
Primitive image;
S1.2.2, to trainer shoe stamp line primitive image zooming-out Wavelet Fourier plum forests feature;
S1.2.3, the trainer shoe stamp print image textural characteristics that will be obtained and footwear stamp line semantic vocabulary table SvMiddle footwear stamp
The footwear stamp print image textural characteristics that line word is extracted carry out similarity comparison, and mark trainer shoe stamp line primitive figure again
The semantic label of picture;
S1.2.4, the semantic label according to new mark and the comparing result of former semantic label, differentiate semantic relation, institute's predicate
Adopted relation includes:
Synonym:Assuming that certain class primitive set Rl={ α1,α2,...αnIt corresponds to semantic label for Rl, primitive set Rk=
{β1,β2,...βnIt corresponds to semantic label for Rk.Two class primitives are extracted into feature and semantic classes is marked again, if RkIn it is each
Individual primitive can also be labeled as Rl, while RlIn each primitive can also be labeled as Rk, then it is considered that Rl、RkTwo kinds of semantemes exist
This visual signature aspect is synonym.The two class images as shown in a groups in Fig. 6, because its similarity higher causes two
Classification is extremely easily obscured during differentiation, according to being defined as synonym.
Upper hyponym:Assuming that certain class primitive set Rl={ α1,α2,...αnIt corresponds to semantic label for Rl, primitive set Rk
={ β1,β2,...βnIt corresponds to semantic label for Rk.Two class primitives extraction feature is marked into semantic classes again, if RlIn it is each
Individual primitive can also be labeled as Rk, but RkIn each primitive cannot be labeled as Rl, then it is considered that RkIt is RlSuperordinate term, Rl
It is RkHyponym.The two class images as shown in the b groups in Fig. 6, First Kind Graph picture can be due to factors such as incomplete or abrasions in b
It is judged into Equations of The Second Kind.And Equations of The Second Kind is difficult to be judged to the first kind due to its particularity.Upper hyponym is then called according to definition.
Fallibility word:It refer to the semantic word for being easy to mutually obscure with other semantemes.Such as 2 class primitive image institutes in c groups in Fig. 6
Show, First Kind Graph picture was not only similar to setback shape but also similar to curved;Equations of The Second Kind image is not only similar but also similar to petal to circle.
Primitive image of type is easily mutually obscured with other semantic classes in this.It is embodied in following two aspects:
Certain class primitive Rl={ α1,α2,...αnIts corresponding semantic label be Rl, after extraction feature is marked again, if
RlIn each primitive easily mark is difficult labeled as R for the primitive in label, but other semantic labelsl, then R is claimedlFor can
Letter fallibility word.Credible fallibility word exist itself semantic purity it is high the characteristics of (mark RlThe primitive one of semantic label is set to RlIt is semantic
Classification figure).
For semantic label A, the primitive in the corresponding primitive set of other semantic labels extracts feature respectively to be carried out again
After mark semantic classes, easily mark isl, then R is claimedlReferred to as insincere fallibility word.Insincere fallibility word is vulnerable to other classifications
Influence, classification purity is relatively low.
S1.3, structure semantic vocabulary relation table
S1.3.1, by footwear stamp line semantic vocabulary table Sv={ R1、R2、...RNIn synonym merge, obtain new
Footwear stamp line semantic vocabulary table Sns={ w1,w2,…,wi,…,wK, wherein wiRepresent footwear stamp line word, i=1,2 ... K, K
Represent footwear stamp line number of words, and the new footwear stamp line semantic vocabulary table SnsIn it is mutually different between each footwear stamp line word
Justice;
S1.3.2, structure footwear stamp line semantic vocabulary relation table Sr={ P1,P2,…,Pi,…,PK, wherein PiIt is individual five yuan
Group, it includes the corresponding word of semantic label, semantic label, part of speech, association vocabulary and weight;The part of speech includes up/down justice
Word, trusted/untrusted fallibility word;Semantic vocabulary relation table SrIt is the basis of shoe sole print semantic meaning representation.The weight initial value
It is 1, and part of speech in footwear stamp line semantic vocabulary relation table is updated, it by part of speech is credible fallibility that its update mode is
The weight of the semantic classes that the weights of the semantic classes of word are improved, part of speech is insincere fallibility word keeps constant.According to language
Adopted lexical relation table Sr, with shoe sole print image in, the frequency that each word occurs is printed representing footwear.It is as shown in Figure 1 footwear stamp
Line semantic vocabulary relation table builds flow
S2, collection footwear stamp print image;
S3, the footwear stamp print image to being gathered carry out the extraction of footwear stamp print image primitive, to obtain corresponding footwear stamp
Print image primitive;Be divided into primitive in respective affiliated area by the connection characteristic according to footwear stamp print image primitive, then chooses
Area takes its external square as target area, the footwear stamp print image primitive for as extracting more than the connected domain of certain value.Here
Choosing area can be prevented effectively from the interference of partial noise more than the footwear stamp print image primitive of certain value.
S4, to footwear stamp line primitive extract Wavelet Fourier plum forests feature.
S5, textural characteristics successively according to each footwear stamp print image primitive for being extracted and footwear stamp print image semanteme word
The relation table that converges carries out characteristic matching, and determines the semantic category corresponding to the footwear stamp print image primitive based on obtained matching degree
Not;
Import footwear stamp print image semantic relation table, obtain table in semantic label, the corresponding word of semantic label, part of speech,
Association five information of vocabulary and weight;
The footwear stamp print image primitive textural characteristics that will be extracted and footwear stamp line word institute in footwear stamp line semantic vocabulary table
The footwear stamp print image primitive textural characteristics of extraction are matched, by matching degree score or be similarity score to semantic label
It is ranked up, when similarity score is more than certain threshold value, then this footwear stamp print image primitive is demarcated as the semantic label, it is no
Then it is judged to refusal identification.Threshold value mentioned herein determines according to part of speech, when part of speech is insincere fallibility word, threshold value compared with
Height, if part of speech is non-insincere fallibility word, threshold value is relatively low.Specifically, when part of speech is non-insincere fallibility word, and it is somebody's turn to do
Divide and this footwear stamp print image primitive is then demarcated as the semantic label more than certain threshold value, be otherwise then judged to refusal identification;Work as word
Property for insincere fallibility word when, then need to improve threshold value and be mixed into the semanteme limiting non-similar semantic footwear stamp print image primitive
In classification, that is, when judging whether the score is more than part of speech for non-insincere fallibility word more than another threshold value, and another threshold value
Set threshold value;It is that this footwear stamp print image primitive is demarcated as the semantic label, is otherwise judged to refusal identification.
S6, the number of times that each semanteme occurs in the footwear stamp print image is counted, to obtain footwear stamp print image language
Adopted frequency histogram.
S7, under normal circumstances, frequency histogram can be used as the semantic meaning representation of the footwear stamp print image.But to take into account semantic phase
Guan Xing, it will be further processed to histogram, specific method is as follows:
It is the semantic label of hyponym that part of speech is searched in semantic vocabulary relation table, and it is right in histogram to search the label
The frequency answered, when its frequency is more than certain value, is assigned to association vocabulary (i.e. its superordinate term), by the semantic label by its frequency
The frequency is set to 0.Conversely, when its frequency is less than certain value, it is believed that the hyponym is smaller on image, semantic expression influence, does not do
Treatment.(for processing not by the way of directly merging because a superordinate term may be correspondingly multiple lower adopted for upper hyponym
Word, directly merging can influence the expression of other hyponyms.)
The semantic frequency histogram for obtaining is multiplied with the weighted value in semantic vocabulary relation table, final footwear stamp is obtained
Print image semantic meaning representation histogram.
The footwear stamp print image semanteme frequency histogram obtained after treatment is straight based on former footwear stamp print image semanteme frequency
What side's figure and footwear stamp print image semantic vocabulary relation table were obtained, semantic dependency has been taken into account, therefore this patent prints as footwear
The semantic meaning representation of decorative pattern image.
Be illustrated in figure 3 the present invention offer it is a kind of based on bag of words and take into account semantic dependency footwear stamp line expression
System, including:Relation table builds module, is used to build footwear stamp print image semantic vocabulary relation table;Extraction module, to by adopting
The footwear stamp print image that collects extracts footwear stamp print image primitive and goes forward side by side the textural characteristics of onestep extraction footwear stamp print image primitive;
Primitive semantic classes determination module, the successively textural characteristics according to each footwear stamp print image primitive for being extracted and footwear stamp line
Image, semantic lexical relation table carries out characteristic matching, and is determined corresponding to the footwear stamp print image primitive based on obtained matching degree
Semantic classes;Image expression module, is used to count footwear stamp print image semantic meaning representation histogram, and further obtain histogram
Obtain the footwear stamp print image footwear stamp print image semantic meaning representation histogram that can take into account semantic dependency.
The structure of relation table structure module in Fig. 3 is illustrated in figure 4, it includes:Semantic vocabulary table builds module, to structure
Build footwear stamp print image semantic vocabulary table;Semantic dependency discrimination module, is used to judge the semantic pass of footwear stamp print image primitive
System;Semantic vocabulary relation table builds module, is used to build semantic vocabulary relation table.
Extraction module structure in Fig. 3 is illustrated in figure 5, it includes:Primitive extraction module, for extracting footwear stamp print image
Primitive;First characteristic extracting module, for extracting footwear stamp print image primitive textural characteristics.
In sum, based on bag of words and the footwear stamp line expression side of semantic dependency is taken into account the invention provides a kind of
Method and system, its purposes are not limited only to the extraction of footwear stamp line, in fields such as image separation, image retrieval, image procossings, can answer
With.It is the image expression method for being designed and being taken into account image, semantic correlation based on bag of words, by footwear stamp print image
The histogrammic further treatment of semantic meaning representation, not only effectively prevent interference of the semantic dependency to semantic meaning representation, and effectively
A kind of feedback mechanism is formd, semantic tagger result can be exercised supervision and be adjusted, while semantic meaning representation accuracy is improved
Also leading position of the footwear stamp print image textural characteristics during semantic tagger is largely controlled.
Finally it should be noted that:Various embodiments above is merely illustrative of the technical solution of the present invention, rather than its limitations;To the greatest extent
Pipe has been described in detail with reference to foregoing embodiments to the present invention, it will be understood by those within the art that:Its according to
The technical scheme described in foregoing embodiments can so be modified, or which part or all technical characteristic are entered
Row equivalent;And these modifications or replacement, the essence of appropriate technical solution is departed from various embodiments of the present invention technology
The scope of scheme.
Claims (8)
1. it is a kind of based on bag of words and to take into account the footwear stamp line expression of semantic dependency, it is characterised in that specific steps
Including:
S1, in advance structure footwear stamp print image semantic vocabulary relation table;
S2, collection footwear stamp print image;
S3, the footwear stamp print image to being gathered carry out the extraction of footwear stamp print image primitive, to obtain corresponding footwear stamp line figure
As primitive;
S4, the Wavelet Fourier plum forests feature for extracting footwear stamp print image primitive;
S5, successively according to the Wavelet Fourier plum forests feature and footwear stamp print image of each footwear stamp print image primitive for being extracted
Semantic vocabulary relation table carries out characteristic matching, and determines the language corresponding to the footwear stamp print image primitive based on obtained matching degree
Adopted classification;
S6, the number of times that each semanteme occurs in the footwear stamp print image is counted, to obtain the semantic frequency of footwear stamp print image
Rate histogram;
S7, based on obtained footwear stamp print image semanteme frequency histogram and footwear stamp print image semantic vocabulary relation table obtain
The footwear stamp print image semantic meaning representation histogram of image, semantic correlation can be expressed.
2. it is according to claim 1 it is a kind of based on bag of words and take into account the footwear stamp line expression of semantic dependency,
Characterized in that, the S1 includes the step of building footwear stamp print image semantic vocabulary relation table:
S1.1, structure semantic vocabulary table
S1.1.1, the geometry in advance corresponding to foundation footwear stamp print image are classified and to sorted each shape point
R She Ding not be respectively designated as with its one-to-one semantic label, institute's semantic tags1,R2,…,RN, wherein N is positive integer;
Footwear stamp print image primitive in S1.1.2, delimitation footwear stamp print image, and the footwear stamp line under same semantic label is entered
Row clusters to obtain each corresponding footwear stamp line set of letters;Wherein, comprising multiple footwear stamp lines under each semantic label
Word, is designated as Rl={ α1,α2,..,αn, wherein RlRepresent semantic label RlUnder footwear stamp line set of letters, αiRepresent semantic
Label R1Under whole footwear stamp line word i=1,2, n, l=1,2, N;
S1.1.3, footwear stamp line semantic vocabulary is constituted based on the corresponding footwear stamp line set of letters of each semantic label for being obtained
Table, it is designated as Sv={ R1,R2,...,RN, wherein R1,R2,...,RNSemantic label R is represented respectively1,R2,…,RNUnder each footwear
Stamp line set of letters;
S1.2, differentiation semantic dependency
S1.2.1, the footwear stamp line primitive of marked semantic label is taken as the trainer shoe stamp line primitive for differentiating semantic dependency
Image;
S1.2.2, the Wavelet Fourier plum forests feature for extracting trainer shoe stamp line primitive image;
S1.2.3, the trainer shoe stamp print image Wavelet Fourier plum forests feature that will be obtained and footwear stamp line semantic vocabulary table SvIn
The Wavelet Fourier plum forests feature of the footwear stamp print image that footwear stamp line word is extracted, and trainer shoe stamp line base is marked again
The semantic label of first image;
S1.2.4, the semantic label according to new mark and the comparing result of former semantic label, differentiate semantic relation;
The semantic relation includes synonym, upper hyponym, fallibility word;
S1.3, structure semantic vocabulary relation table
S1.3.1, by footwear stamp line semantic vocabulary table Sv={ R1、R2、...RNIn synonym merge, obtain new footwear print
Decorative pattern semantic vocabulary table Sns={ w1,w2,…,wi,…,wK, wherein wiRepresent footwear stamp line word, i=1,2K, K table
Show footwear stamp line number of words, and the new footwear stamp line semantic vocabulary table SnsIn it is mutually different between each footwear stamp line word
Justice;
S1.3.2, structure footwear stamp line semantic vocabulary relation table Sr={ P1,P2,…,Pi,…,PK, wherein PiIt is a five-tuple,
It includes the corresponding word of semantic label, semantic label, part of speech, association vocabulary and weight;The part of speech include up/down justice word,
Trusted/untrusted fallibility word;The weight initial value is 1, and part of speech in footwear stamp line semantic vocabulary relation table is carried out
Update, its update mode be the weights of the semantic classes that part of speech is credible fallibility word are improved, part of speech be insincere fallibility
The weight of the semantic classes of word keeps constant.
3. it is according to claim 1 it is a kind of based on bag of words and take into account the footwear stamp line expression of semantic dependency,
Characterized in that, the S3 includes the step of extracting footwear stamp line primitive image:
Footwear stamp print image is divided to respective affiliated area by S3.1 according to the connection characteristic of footwear stamp print image;
S3.2 chooses connected region of the area more than certain value, takes its external square type as target area, the footwear print as extracted
Decorative pattern picture element.
4. it is according to claim 1 it is a kind of based on bag of words and take into account the footwear stamp line expression of semantic dependency,
Characterized in that, S5 includes the step of judging footwear stamp print image primitive semantic classes:
S5.1 obtains footwear stamp print image semantic vocabulary relation table by step S1, obtains semantic label, semantic label correspondence in table
Word, part of speech, association five information of vocabulary and weight;
The footwear stamp print image primitive textural characteristics that S5.2 will be extracted and footwear stamp line word institute in footwear stamp line semantic vocabulary table
The footwear stamp print image primitive textural characteristics of extraction are matched, by matching degree score or be similarity score to semantic label
It is ranked up, when similarity score is more than certain threshold value, then this footwear stamp print image primitive is demarcated as the semantic label, it is no
Then it is judged to refusal identification.
5. it is according to claim 2 a kind of based on bag of words and to take into account the footwear stamp print image expression side of semantic dependency
Method, it is characterised in that the S7 includes:
S7.1, to search part of speech in footwear stamp line semantic vocabulary relation table be the semantic label of hyponym, and search the label and exist
The corresponding frequency in footwear stamp print image semanteme frequency histogram, when its frequency is not less than a certain setting value, its frequency is assigned
Association vocabulary is given, while the frequency of the semantic label is set into 0, when its frequency is less than the setting value, is not processed;
S7.2, by treatment after semantic frequency histogram be multiplied with the weighted value in semantic vocabulary relation table, obtain final language
Justice expression histogram.
6. it is a kind of based on bag of words and to take into account the footwear stamp line expression system of semantic dependency, it is characterised in that the system
Including:
Relation table builds module, is used to build footwear stamp print image semantic vocabulary relation table;
Extraction module, is used to extract the textural characteristics of footwear stamp print image primitive;
Primitive semantic classes determination module, is used to judge the semantic classes corresponding to footwear stamp print image primitive;
Image expression module, is used to obtain the footwear stamp print image semantic meaning representation histogram that can express image, semantic correlation.
7. according to claim 6 based on bag of words and to take into account the footwear stamp line expression system of semantic dependency, it is special
Levy and be, relation table builds module to be included:
Semantic vocabulary table builds module, is used to build footwear stamp print image semantic vocabulary table;
Semantic dependency discrimination module, is used to judge the semantic relation of footwear stamp print image primitive;
Semantic vocabulary relation table builds module, is used to build semantic vocabulary relation table.
8. according to claim 6 based on bag of words and to take into account the footwear stamp line expression system of semantic dependency, it is special
Levy and be, the extraction module includes:
Primitive extraction module, for extracting footwear stamp print image primitive;
Primitive feature extraction module, for extracting footwear stamp print image primitive textural characteristics.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611130372.1A CN106777050B (en) | 2016-12-09 | 2016-12-09 | It is a kind of based on bag of words and to take into account the shoes stamp line expression and system of semantic dependency |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611130372.1A CN106777050B (en) | 2016-12-09 | 2016-12-09 | It is a kind of based on bag of words and to take into account the shoes stamp line expression and system of semantic dependency |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106777050A true CN106777050A (en) | 2017-05-31 |
CN106777050B CN106777050B (en) | 2019-09-06 |
Family
ID=58874893
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611130372.1A Active CN106777050B (en) | 2016-12-09 | 2016-12-09 | It is a kind of based on bag of words and to take into account the shoes stamp line expression and system of semantic dependency |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106777050B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107301426A (en) * | 2017-06-14 | 2017-10-27 | 大连海事大学 | A kind of multi-tag clustering method of shoe sole print image |
CN109598000A (en) * | 2018-12-28 | 2019-04-09 | 百度在线网络技术(北京)有限公司 | Semantic relation recognition methods, device, computer equipment and storage medium |
CN110992397A (en) * | 2019-10-21 | 2020-04-10 | 浙江大华技术股份有限公司 | Personnel entrance and exit trajectory tracking method and system, computer equipment and storage medium |
CN113537391A (en) * | 2021-08-06 | 2021-10-22 | 大连海事大学 | Shoe print image clustering method guided by interactive text semantic attributes |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105760507A (en) * | 2016-02-23 | 2016-07-13 | 复旦大学 | Cross-modal subject correlation modeling method based on deep learning |
CN105844299A (en) * | 2016-03-23 | 2016-08-10 | 浙江理工大学 | Image classification method based on bag of words |
CN105930789A (en) * | 2016-04-18 | 2016-09-07 | 电子科技大学 | Human body behavior recognition based on logarithmic Euclidean space BOW (bag of words) model |
-
2016
- 2016-12-09 CN CN201611130372.1A patent/CN106777050B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105760507A (en) * | 2016-02-23 | 2016-07-13 | 复旦大学 | Cross-modal subject correlation modeling method based on deep learning |
CN105844299A (en) * | 2016-03-23 | 2016-08-10 | 浙江理工大学 | Image classification method based on bag of words |
CN105930789A (en) * | 2016-04-18 | 2016-09-07 | 电子科技大学 | Human body behavior recognition based on logarithmic Euclidean space BOW (bag of words) model |
Non-Patent Citations (2)
Title |
---|
MIN-QUAN JING ET.AL.: ""A NOVEL METHOD FOR SHOEPRINTS RECOGNITION AND"", 《IEEE》 * |
朱道广: ""基于视觉词袋模型的图像分类研究"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107301426A (en) * | 2017-06-14 | 2017-10-27 | 大连海事大学 | A kind of multi-tag clustering method of shoe sole print image |
CN107301426B (en) * | 2017-06-14 | 2020-06-30 | 大连海事大学 | Multi-label clustering method for sole pattern images |
CN109598000A (en) * | 2018-12-28 | 2019-04-09 | 百度在线网络技术(北京)有限公司 | Semantic relation recognition methods, device, computer equipment and storage medium |
CN109598000B (en) * | 2018-12-28 | 2023-06-16 | 百度在线网络技术(北京)有限公司 | Semantic relation recognition method, semantic relation recognition device, computer equipment and storage medium |
CN110992397A (en) * | 2019-10-21 | 2020-04-10 | 浙江大华技术股份有限公司 | Personnel entrance and exit trajectory tracking method and system, computer equipment and storage medium |
CN110992397B (en) * | 2019-10-21 | 2023-07-21 | 浙江大华技术股份有限公司 | Personnel access track tracking method, system, computer equipment and storage medium |
CN113537391A (en) * | 2021-08-06 | 2021-10-22 | 大连海事大学 | Shoe print image clustering method guided by interactive text semantic attributes |
CN113537391B (en) * | 2021-08-06 | 2023-09-05 | 大连海事大学 | Shoe print image clustering method guided by interactive text semantic attribute |
Also Published As
Publication number | Publication date |
---|---|
CN106777050B (en) | 2019-09-06 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Xian et al. | Latent embeddings for zero-shot classification | |
Wu et al. | Self-critical reasoning for robust visual question answering | |
Cornia et al. | Show, control and tell: A framework for generating controllable and grounded captions | |
Lu et al. | Neural baby talk | |
CN107391703B (en) | The method for building up and system of image library, image library and image classification method | |
Plummer et al. | Phrase localization and visual relationship detection with comprehensive image-language cues | |
Shi et al. | Weakly supervised object localization using things and stuff transfer | |
Olszewski | Generalized feature extraction for structural pattern recognition in time-series data | |
Farhadi et al. | Transfer learning in sign language | |
CN109815801A (en) | Face identification method and device based on deep learning | |
CN104992142A (en) | Pedestrian recognition method based on combination of depth learning and property learning | |
Zhang et al. | Discriminative bimodal networks for visual localization and detection with natural language queries | |
CN106777050B (en) | It is a kind of based on bag of words and to take into account the shoes stamp line expression and system of semantic dependency | |
CN106815604A (en) | Method for viewing points detecting based on fusion of multi-layer information | |
CN105389326B (en) | Image labeling method based on weak matching probability typical relevancy models | |
JP6928206B2 (en) | Data identification method based on associative clustering deep learning neural network | |
CN107145514B (en) | Chinese sentence pattern classification method based on decision tree and SVM mixed model | |
Chen et al. | What's in a name? First names as facial attributes | |
CN109886161A (en) | A kind of road traffic index identification method based on possibility cluster and convolutional neural networks | |
CN111985581A (en) | Sample-level attention network-based few-sample learning method | |
CN109213853A (en) | A kind of Chinese community's question and answer cross-module state search method based on CCA algorithm | |
CN106897669A (en) | A kind of pedestrian based on consistent iteration various visual angles transfer learning discrimination method again | |
Li et al. | Dating ancient paintings of Mogao Grottoes using deeply learnt visual codes | |
CN107220598A (en) | Iris Texture Classification based on deep learning feature and Fisher Vector encoding models | |
Pezzelle et al. | Is the red square big? MALeViC: Modeling adjectives leveraging visual contexts |
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 |