CN110210582A - A kind of Chinese handwriting identifying method based on part cooperation presentation class - Google Patents
A kind of Chinese handwriting identifying method based on part cooperation presentation class Download PDFInfo
- Publication number
- CN110210582A CN110210582A CN201910521515.9A CN201910521515A CN110210582A CN 110210582 A CN110210582 A CN 110210582A CN 201910521515 A CN201910521515 A CN 201910521515A CN 110210582 A CN110210582 A CN 110210582A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- vector
- chinese character
- dictionary
- handwritten chinese
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/24—Character recognition characterised by the processing or recognition method
- G06V30/242—Division of the character sequences into groups prior to recognition; Selection of dictionaries
- G06V30/244—Division of the character sequences into groups prior to recognition; Selection of dictionaries using graphical properties, e.g. alphabet type or font
- G06V30/245—Font recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/28—Character recognition specially adapted to the type of the alphabet, e.g. Latin alphabet
- G06V30/287—Character recognition specially adapted to the type of the alphabet, e.g. Latin alphabet of Kanji, Hiragana or Katakana characters
Abstract
The present invention provides a kind of Chinese handwriting identifying method based on part cooperation presentation class, comprising the steps of: S1, chooses handwritten Chinese character library, extracts the feature vector of all handwritten Chinese characters and classification in the handwritten Chinese character library, establish set of eigenvectors A；Extract the feature vector y of handwritten Chinese character to be identified；S2, a feature vector most like with y is found out in every category feature vector, establish dictionary D1；S3, y is solved in D1In cooperation rarefaction representation vectorPass throughY is sought to D1The first reconstructed residual；Dictionary D is established according to the N category feature vector in the first reconstructed residual selected characteristic vector set A2；S4, y is solved in D2In cooperation rarefaction representation vectorPass throughY is sought to D2The second reconstructed residual, according to second reconstructed residual judge y correspond to A in feature vector classification, realize Handwritten Chinese Character Recognition.The present invention reduces the scale of dictionary used when Handwritten Chinese Character Recognition, also improves accuracy of identification while reducing Algorithms T-cbmplexity.
Description
Technical field
The present invention relates to optical character recognition technology, in particular to a kind of automatic identifying method of handwritten Chinese character.
Background technique
OCR (optical character identification Optical Character Recognition) refers to (such as to be swept using electronic equipment
Retouch instrument or digital camera) character on paper is obtained, its shape is determined by detecting dark, bright mode, then with character recognition side
Shape is translated into the process of computword by method；That is, it is directed to printed character, it will be in paper document using optical mode
Text conversion becomes the image file of black and white lattice, and passes through identification software for the text conversion in image into text formatting, supplies
The technology that word processor is further edited and processed.OCR technique is widely used in typing and processing bank money, text money
Material, archives folder, official documents and correspondence etc. can replace the manual typing of people, save a large amount of manpowers.Usually with final discrimination, recognition speed
Important evidence as evaluation and test OCR technique.It is OCR most important how except mistake or using auxiliary information raising recognition correct rate
Project.
Handwritten Kanji recognition technology belongs to the field OCR.Obvious Handwritten Chinese Character Recognition is hand-written English difficult many identifying.
First, the classification of English character is few, in total 62 characters (capital and small letter of 26 English alphabets adds ten Arabic numerals again), and
Chinese in total 50, more than 000 Chinese character commonly just has more than 3000；Second, there are many font, the books of different fonts for identical Chinese character
WriteMode has larger difference；Everyone writing style of third is also different, and actual writing effect is unfavorable for machine recognition.
Therefore people use accuracy and stability that thousand and one way goes improvement to identify in the past few decades.
In recent years, compressed perception theory influences, and rarefaction representation is introduced into pattern recognition problem, proposes sparse by people
Presentation class algorithm (Sparse Representation-based Classification, SRC), nowadays the algorithm is extensive
Classification and Identification applied to figure.During handwritten Kanji recognition, some scholars carry out classifier while learning dictionary
Then training classifies to image sparse coding with obtained classifier.Some scholars are obtaining multiple specified category dictionaries
Under the premise of, classified according to reconstructed residual of the feature vector under different category dictionaries, achieves relatively good classifying quality.But
It is in the prior art, when by rarefaction representation classifier Handwritten Chinese Character Recognition, since dictionary scale is excessive and rarefaction representation point
It is solved in class algorithm and minimizes L1The calculation amount of norm is excessive, and causes time complexity high, to affect recognition efficiency.
Summary of the invention
The object of the present invention is to provide a kind of Chinese handwriting identifying methods based on part cooperation presentation class, in OCR mistake
Cheng Zhong, by concentrating extraction unit that feature vector is divided to construct dictionary from the feature vector in handwritten Chinese character library, by handwritten Chinese character to be measured
Feature vector corresponds in certain category feature vector of constructed dictionary, realizes Handwritten Chinese Character Recognition.Handwritten Chinese character of the invention is known
Other method can reduce computation complexity, realize quick Handwritten Chinese Character Recognition, and improve the accuracy rate of identification.
In order to achieve the above object, the present invention provides a kind of handwritten Kanji recognition side based on part cooperation presentation class
Method, comprising the steps of:
S1, handwritten Chinese character library is chosen, extracts the feature vector of all handwritten Chinese characters and classification in the handwritten Chinese character library, builds
Vertical set of eigenvectors A=[Ai]i∈[1,k]；AiFor the i-th category feature vector in A,M is AiIn each feature vector
Dimension, niFor AiThe number of middle feature vector, k are the feature vector classification sum extracted from handwritten Chinese character library；It extracts wait know
The feature vector y of other handwritten Chinese character, wherein y ∈ Rm；
S2, a feature vector most like with y is found out in every category feature vector, construct dictionary D1；
S3, it is based on L2Norm solves y in D1In cooperation rarefaction representation vector Pass throughSeek y pairs
D1The first reconstructed residual；Dictionary D is constructed according to the N category feature vector in the first reconstructed residual selected characteristic vector set A2,
D2It also is N category feature vector most like with y in A；
S4, it is based on L2Norm solves y in D2In cooperation rarefaction representation vector K ' is that the N class is special
The feature vector sum for including in sign vector；Pass throughY is sought to D2The second reconstructed residual；It is residual according to second reconstruct
Difference judges that y corresponds to feature vector classification in A, realizes Handwritten Chinese Character Recognition.
The step S2 includes:
S21, Gaussian kernel K (y, A are calculatedij)=exp (- | | y-Aij||2/ σ), i ∈ [1, k], j ∈ [1, ni], Aij∈Ai, one
A feature vector in a Gaussian kernel corresponding A；Wherein K () indicates gaussian kernel function；δ is the Europe of all feature vectors in y and A
The mean value of formula distance；
S22, from Gaussian kernel collectionOne the smallest Gaussian kernel of middle selection enables
Feature vector diFeature vector corresponding equal to the smallest Gaussian kernel；di∈Ai, i ∈ [1, k]；
S23, construction dictionary D1={ d1,d2,...,dk}。
δ in step S21 is specifically referred to:
Wherein ApqA feature vector being characterized in vector set A；Indicate y and ApqEuclidean distance；N is in A
The number of all feature vectors,
In the step S3, specifically include:
S31, cooperation rarefaction representation vector is calculated In
I-th of element correspond to di；Wherein λ1It is regularization parameter, I is unit matrix, i ∈ [1, k]；
S32, the first reconstructed residual is calculatedri(y) corresponding di、Ai, i ∈ [1, k]；Its
InBeing willIn not with diResulting k dimensional vector after corresponding element is set as 0；||·||2Indicate L2Norm；
S33, from the first reconstructed residual set R1={ r1(y),...,rk(y) } N number of the smallest first reconstructed residual is selected；
S34, N category feature vector corresponding with N number of the smallest first reconstructed residual in A is denoted as A ' respectively1、...、
A′N；Construct dictionary D2={ A '1..., A 'N}。
In the step S4, specifically include:
S41, cooperation rarefaction representation vector is calculatedWherein λ2It is regularization ginseng
Number, I is unit matrix；Dimension be equal to k ', k ' be dictionary D2The number of included feature vector；In a member
The corresponding D of element2In a feature vector；
S42, the second reconstructed error is calculatedrt' (y) corresponds to dictionary D2In one kind it is special
Levy vector A 't, whereinBeing willIn not with A 'tResulting k ' dimensional vector after corresponding element is set as 0, t ∈ [1,
N]；
If S43, r 'p(y)=min { r1′(y),...,r′N(y) }, [1, N] p ∈ then judges that y belongs to dictionary D2In one
Category feature vector A 'p, realize Handwritten Chinese Character Recognition.
The N ∈ [10,15].
The handwritten Chinese character library is the Off-line Handwritten Chinese sample database CASIA-HWDB1.0 that Institute of Automation, CAS provides.
Extracted in the step S1 by LBP image detection algorithm the feature of handwritten Chinese character library and handwritten Chinese character to be identified to
Amount.
Compared with prior art, the Chinese handwriting identifying method of the invention based on part cooperation presentation class, mentions first
The feature vector of all handwritten Chinese characters and classification in the handwritten Chinese character library are taken, set of eigenvectors A is obtained；Extract the hand-written Chinese to be measured
The feature vector y of word.Then by gaussian kernel function found out from every category feature vector of A a feature most like with y to
Amount is to construct dictionary D1.Then in dictionary D1In find out the N number of feature vector increasingly similar with y, in A with N number of feature to
Measure corresponding N category feature vector construction dictionary D2.Finally by judging dictionary D2In a kind of feature vector most like with y, realize
Handwritten Chinese Character Recognition.Chinese handwriting identifying method of the invention, not only reduces computation complexity, improves Handwritten Chinese Character Recognition
Speed, further improve the precision of Handwritten Chinese Character Recognition.
Detailed description of the invention
In order to illustrate more clearly of technical solution of the present invention, attached drawing needed in description will be made simply below
It introduces, it should be apparent that, the accompanying drawings in the following description is one embodiment of the present of invention, and those of ordinary skill in the art are come
It says, without creative efforts, is also possible to obtain other drawings based on these drawings:
Fig. 1 is the Chinese handwriting identifying method step schematic diagram of the invention based on part cooperation presentation class.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The present invention provides a kind of Chinese handwriting identifying method based on part cooperation presentation class, as shown in Figure 1, including step
It is rapid:
S1, handwritten Chinese character library is chosen, extracts the feature vector of all handwritten Chinese characters and classification in the handwritten Chinese character library, builds
Vertical set of eigenvectors A=[Ai]i∈[1,k]；AiFor the i-th category feature vector in A,M is AiIn each feature vector
Dimension, niFor AiThe number of middle feature vector, k are the feature vector classification sum extracted from handwritten Chinese character library；Pass through LBP
Image detection algorithm extracts the feature vector y of handwritten Chinese character to be identified, wherein y ∈ Rm.The handwritten Chinese character library be the Chinese Academy of Sciences from
The Off-line Handwritten Chinese sample database CASIA-HWDB1.0 provided by dynamicization.
The a certain category feature vector A in set of eigenvectors A can be used in feature vector y in perfect conditioniLinearly Representation, and y
With the feature vector linear independence of other classifications.Both y can be expressed as
Wherein Vi,jFor AiIn a feature vector, αi,j∈R,j∈[1,ni]。
Y can also be indicated by the feature vector of all k types:
Y=Ax0∈Rm (2)
x0The rarefaction representation vector for being y at A, in formula (2), x0In each element correspond to A in a feature to
Amount.Under ideal conditions:
N is the number of all feature vectors in A,When y can pass through AiWhen identification, in x0The inside, removing pair
It should be in AiElement other than, other elements are all 0.That is, obtaining rarefaction representation of the y at A by solution formula (2)
Vector may know that y corresponds to the classification of all feature vectors in A, realize Handwritten Chinese Character Recognition.
Since Chinese character amount is huge in handwritten Chinese character library, the set of eigenvectors A established according to handwritten Chinese character library is very huge, if
The corresponding relationship of certain category feature vector in y and A is solved, computation complexity is high, and speed is very slow.Therefore need to reduce in A with y
The quantity of the feature vector of comparison.But in order to abide by the basic assumption of the rarefaction representation (sample size of class where feature vector
When sufficient, this feature vector can linearly be expressed by the feature vector of place class), so can only be in A according to feature vector
Classification deletes A.Such as the i-th category feature vector A can be deleted completelyi, but A cannot be deletediIn Partial Feature vector.
S2, a feature vector most like with y is found out in every category feature vector, construct dictionary D1。
The step S2 specifically includes:
S21, Gaussian kernel K (y, A are calculatedij)=exp (- | | y-Aij||2/ σ), i ∈ [1, k], j ∈ [1, ni], Aij∈Ai, one
A feature vector in a Gaussian kernel corresponding A；Wherein K () indicates gaussian kernel function；δ is the Europe of all feature vectors in y and A
The mean value of formula distance；δ is specifically referred to:
Wherein ApqA feature vector being characterized in vector set A；Indicate y and ApqEuclidean distance.
Gaussian kernel function can better the distance between measuring vector, can more accurately examine similar between outgoing vector
Degree.If y and AijVery close to then K (y, Aij) can be closer in 1, whereas if if y and AijSimilarity is not high, then K
(y,Aij) it can be closer to 0.Especially in the case where lacking priori knowledge, gaussian kernel function can have preferably than other kernel functions
Test effect.
S22, from Gaussian kernel collectionOne the smallest Gaussian kernel of middle selection enables special
Levy vector diFeature vector corresponding equal to the smallest Gaussian kernel；di∈Ai, i ∈ [1, k]；
S23, construction dictionary D1={ d1,d2,...,dk}。
S3, it is based on L2Norm solves y in D1In cooperation rarefaction representation vector Pass throughSeek y
To D1The first reconstructed residual；Dictionary is constructed according to the N category feature vector in the first reconstructed residual selected characteristic vector set A
D2, D2It also is N category feature vector most like with y in A.
In the step S3, specifically include:
S31, cooperation rarefaction representation vector is calculated In
I-th of element corresponds to di；Wherein λ1It is regularization parameter, I is unit matrix, i ∈ [1, k]；
In traditional rarefaction representation classification, pass through L1Norm solves y=Ax0Solution For rarefaction representation of the y at A
Vector；Specifically,For the solution for meeting following formula:
Wherein | | | |1L is sought in expression1Norm, | | | |2L is sought in expression2Norm；ε is the tolerance of error, and the error is
Since the image of optical identification is easy by caused by influence of noise.
Zhang et al. starts to produce bosom to cognitive question traditional in rarefaction representation by the mechanism of analysis SRC algorithm
It doubts, they think the rarefaction representation that the cooperation between the also sample to work in algorithm indicates, rather than only thought in the past.
That is when solving rarefaction representation vector, the collaborative between sample be also it is equally very important, to propose
Cooperation rarefaction representation sorting algorithm (Collaborative Sparse Representation based
Classification, CSRC).Therefore, y=Ax0SolutionIt can change into and pass through L2Norm solves,It is y at A
Cooperate rarefaction representation vector；Specifically,For the solution for meeting following formula:
Pass through L2Norm solves rarefaction representation vector of the y at A, will not reduce the discrimination of y, but solution well
Determined L1Excessive problem is consumed when norm algorithm, to get the favour of people.
In order to further reduce the computation complexity of algorithm, recognition efficiency is promoted, we carry out formula (5) excellent
Change, formula (5) is converted into the form of regularization least square method, i.e.,
Wherein, λ is a regularization parameter, then carries out derivation transformation to this formula (6) again, obtains
Enable P=(ATA+λI)-1AT, because P only has relationship with dictionary A and regularization parameter λ, and it is not related with y.
Then P can be precomputed and using it as projection matrix, and can also willIt is described as to obtain after being placed on y on P
The projection arrived.So the also just no longer very large L of computation amount1Norm goes to solve rarefaction representation vector of the y at A, directly
Connect using formula (7) the cooperation rarefaction representation vector for calculating y at A, thus also substantially reduce over due to when consume it is excessively high
Brought negative effect.
According to formula (7), y is in dictionary D1Under rarefaction representation vectorIt calculates in the following way:
S32, the first reconstructed residual is calculatedri(y) corresponding di、Ai, i ∈ [1, k]；Its
InBeing willIn not with diResulting k dimensional vector after corresponding element is set as 0；ri(y) smaller, show y and di
It is more similar.
In this application embodiment,WhereinIn first
A element corresponds to d1, then
S33, from the first reconstructed residual set R1={ r1(y),...,rk(y) } N number of the smallest first reconstructed residual is selected；
S34, N category feature vector corresponding with N number of the smallest first reconstructed residual in A is denoted as A ' respectively1、...、
A′N；Construct dictionary D2={ A '1..., A 'N}.Number is it was demonstrated that when working as [10,15] N ∈ through a large number of experiments, the dictionary enough made
D2The demand for reducing the resolution of computation complexity and guarantee y can be met simultaneously.It is method of the invention below in HWDB1.0 hand
In writing of Chinese characters library when every class sample number measures 7, influence of the different N values to discrimination
N=10, r in this application embodiment1(y), r3(y), r5(y), r7(y), r9(y), r10(y), r11(y), r12(y),
r13(y), r14(y) it is the smallest 10 the first reconstructed residuals, then constructs dictionary D2={ A1, A3, A5, A7, A9, A10, A11, A12,
A13, A14}。
S4, it is based on L2Norm solves y in D2In cooperation rarefaction representation vector K ' is that the N class is special
The feature vector sum for including in sign vector；Pass throughY is sought to D2The second reconstructed residual；It is residual according to second reconstruct
Difference judges that y corresponds to feature vector classification in A, realizes Handwritten Chinese Character Recognition.
In the step S4, specifically include:
S41, cooperation rarefaction representation vector is calculatedWherein λ2It is regularization ginseng
Number, I is unit matrix；Dimension be equal to k ', k ' be dictionary D2The number of included feature vector；In a member
The corresponding D of element2In a feature vector；
S42, the second reconstructed error is calculatedrt' (y) corresponds to dictionary D2In one kind it is special
Levy vector A 't, whereinBeing willIn not with A 'tResulting k ' dimensional vector after corresponding element is set as 0, t ∈
[1,N]；
If S43, r 'p(y)=min { r1′(y),...,r′N(y) }, [1, N] p ∈ then judges that y belongs to dictionary D2In one
Category feature vector A 'p, realize Handwritten Chinese Character Recognition.
Compared with prior art, the Chinese handwriting identifying method of the invention based on part cooperation presentation class, mentions first
The feature vector of all handwritten Chinese characters and classification in the handwritten Chinese character library are taken, set of eigenvectors A is obtained；Extract the hand-written Chinese to be measured
The feature vector y of word.Then by gaussian kernel function found out from every category feature vector of A a feature most like with y to
Amount is to construct dictionary D1.Then in dictionary D1In find out the N number of feature vector increasingly similar with y, in A with N number of feature to
Measure corresponding N category feature vector construction dictionary D2.Finally by judging dictionary D2In a kind of feature vector most like with y, realize
Handwritten Chinese Character Recognition.Chinese handwriting identifying method of the invention, not only reduces computation complexity, improves Handwritten Chinese Character Recognition
Speed, further improve the precision of Handwritten Chinese Character Recognition.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can readily occur in various equivalent modifications or replace
It changes, these modifications or substitutions should be covered by the protection scope of the present invention.Therefore, protection scope of the present invention should be with right
It is required that protection scope subject to.
Claims (8)
1. a kind of Chinese handwriting identifying method based on part cooperation presentation class, which is characterized in that include step:
S1, handwritten Chinese character library is chosen, extracts the feature vector of all handwritten Chinese characters and classification in the handwritten Chinese character library, established special
Levy vector set A=[Ai]i∈[1,k]；AiFor the i-th category feature vector in A,M is AiIn each feature vector dimension
Number, niFor AiThe number of middle feature vector, k are the feature vector classification sum extracted from handwritten Chinese character library；Extract hand to be identified
The feature vector y of writing of Chinese characters, wherein y ∈ Rm；
S2, a feature vector most like with y is found out in every category feature vector, construct dictionary D1；
S3, it is based on L2Norm solves y in D1In cooperation rarefaction representation vectorPass throughY is sought to D1
The first reconstructed residual；Dictionary D is constructed according to the N category feature vector in the first reconstructed residual selected characteristic vector set A2, D2
It also is N category feature vector most like with y in A；
S4, it is based on L2Norm solves y in D2In cooperation rarefaction representation vectorK ' be the N category feature to
The feature vector sum for including in amount；Pass throughY is sought to D2The second reconstructed residual；Sentenced according to second reconstructed residual
Disconnected y corresponds to feature vector classification in A, realizes Handwritten Chinese Character Recognition.
2. the Chinese handwriting identifying method as described in claim 1 based on part cooperation presentation class, which is characterized in that described
Step S2 includes:
S21, Gaussian kernel K (y, A are calculatedij)=exp (- | | y-Aij||2/ σ), i ∈ [1, k], j ∈ [1, ni], Aij∈Ai, a height
A feature vector in this core corresponding A；Wherein K () indicates gaussian kernel function；δ be all feature vectors in y and A it is European away from
From mean value；
S22, from Gaussian kernel collectionOne the smallest Gaussian kernel of middle selection, enable feature to
Measure diFeature vector corresponding equal to the smallest Gaussian kernel；di∈Ai, i ∈ [1, k]；
S23, construction dictionary D1={ d1,d2,...,dk}。
3. the Chinese handwriting identifying method as claimed in claim 2 based on part cooperation presentation class, which is characterized in that step
δ in S21 is specifically referred to:
Wherein ApqA feature vector being characterized in vector set A；Indicate y and ApqEuclidean distance；N is to own in A
The number of feature vector,
4. the Chinese handwriting identifying method as described in claim 1 based on part cooperation presentation class, which is characterized in that described
In step S3, specifically include:
S31, cooperation rarefaction representation vector is calculated In i-th
A element corresponds to di；Wherein λ1It is regularization parameter, I is unit matrix, i ∈ [1, k]；
S32, the first reconstructed residual is calculatedri(y) corresponding di、Ai, i ∈ [1, k]；WhereinBeing willIn not with diResulting k dimensional vector after corresponding element is set as 0；||·||2Indicate L2Norm；
S33, from the first reconstructed residual set R1={ r1(y),...,rk(y) } N number of the smallest first reconstructed residual is selected；
S34, N category feature vector corresponding with N number of the smallest first reconstructed residual in A is denoted as A ' respectively1、...、A′N；
Construct dictionary D2={ A '1..., A 'N}。
5. the Chinese handwriting identifying method as claimed in claim 4 based on part cooperation presentation class, which is characterized in that described
In step S4, specifically include:
S41, cooperation rarefaction representation vector is calculatedWherein λ2It is regularization parameter, I is
Unit matrix；Dimension be equal to k ', k ' be dictionary D2The number of included feature vector；In an element it is corresponding
D2In a feature vector；
S42, the second reconstructed error is calculatedrt' (y) corresponds to dictionary D2In a category feature to
Measure A 't, whereinBeing willIn not with A 'tResulting k ' dimensional vector after corresponding element is set as 0, t ∈ [1, N]；
If S43, r 'p(y)=min { r1′(y),...,r′N(y) }, [1, N] p ∈ then judges that y belongs to dictionary D2In a category feature
Vector A 'p, realize Handwritten Chinese Character Recognition.
6. the Chinese handwriting identifying method as described in claim 1 based on part cooperation presentation class, which is characterized in that described
N∈[10,15]。
7. the Chinese handwriting identifying method as described in claim 1 based on part cooperation presentation class, which is characterized in that described
Handwritten Chinese character library is the Off-line Handwritten Chinese sample database CASIA-HWDB1.0 that Institute of Automation, CAS provides.
8. the Chinese handwriting identifying method as described in claim 1 based on part cooperation presentation class, which is characterized in that described
The feature vector in handwritten Chinese character library and handwritten Chinese character to be identified is extracted in step S1 by LBP image detection algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910521515.9A CN110210582A (en) | 2019-06-17 | 2019-06-17 | A kind of Chinese handwriting identifying method based on part cooperation presentation class |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910521515.9A CN110210582A (en) | 2019-06-17 | 2019-06-17 | A kind of Chinese handwriting identifying method based on part cooperation presentation class |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110210582A true CN110210582A (en) | 2019-09-06 |
Family
ID=67792923
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910521515.9A Pending CN110210582A (en) | 2019-06-17 | 2019-06-17 | A kind of Chinese handwriting identifying method based on part cooperation presentation class |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110210582A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102262723A (en) * | 2010-05-24 | 2011-11-30 | 汉王科技股份有限公司 | Face recognition method and device |
| CN107729886A (en) * | 2017-11-24 | 2018-02-23 | 北京小米移动软件有限公司 | The processing method and processing device of facial image |
| CN108182712A (en) * | 2017-12-07 | 2018-06-19 | 西安万像电子科技有限公司 | Image processing method, apparatus and system |
| US20180173415A1 (en) * | 2013-06-09 | 2018-06-21 | Apple Inc. | Managing real-time handwriting recognition |
| CN108875696A (en) * | 2018-07-05 | 2018-11-23 | 五邑大学 | The Off-line Handwritten Chinese Recognition method of convolutional neural networks is separated based on depth |
| US10296817B1 (en) * | 2018-03-30 | 2019-05-21 | Gyrfalcon Technology Inc. | Apparatus for recognition of handwritten Chinese characters |
| CN109784266A (en) * | 2019-01-09 | 2019-05-21 | 江西理工大学应用科学学院 | A kind of handwritten Kanji recognition algorithm of multi-model hypergraph |
-
2019
- 2019-06-17 CN CN201910521515.9A patent/CN110210582A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102262723A (en) * | 2010-05-24 | 2011-11-30 | 汉王科技股份有限公司 | Face recognition method and device |
| US20180173415A1 (en) * | 2013-06-09 | 2018-06-21 | Apple Inc. | Managing real-time handwriting recognition |
| CN107729886A (en) * | 2017-11-24 | 2018-02-23 | 北京小米移动软件有限公司 | The processing method and processing device of facial image |
| CN108182712A (en) * | 2017-12-07 | 2018-06-19 | 西安万像电子科技有限公司 | Image processing method, apparatus and system |
| US10296817B1 (en) * | 2018-03-30 | 2019-05-21 | Gyrfalcon Technology Inc. | Apparatus for recognition of handwritten Chinese characters |
| CN108875696A (en) * | 2018-07-05 | 2018-11-23 | 五邑大学 | The Off-line Handwritten Chinese Recognition method of convolutional neural networks is separated based on depth |
| CN109784266A (en) * | 2019-01-09 | 2019-05-21 | 江西理工大学应用科学学院 | A kind of handwritten Kanji recognition algorithm of multi-model hypergraph |
Non-Patent Citations (3)
| Title |
|---|
| ARICA 等: "an overview of character recognition focused on off-line handwriting", 《IEEE TRANS ON SYSTEM MAN AND CYBERNETICS》 * |
| 张军等: "基于压缩感知的手写汉字识别研究", 《西安航空学院学报》 * |
| 施志刚等: "基于改进局部协作表示的快速人脸识别算法", 《计算机工程与设计》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Roy et al. | HMM-based Indic handwritten word recognition using zone segmentation | |
| Liwicki et al. | IAM-OnDB-an on-line English sentence database acquired from handwritten text on a whiteboard | |
| Javed et al. | Segmentation free nastalique urdu ocr | |
| Jayadevan et al. | Offline recognition of Devanagari script: A survey | |
| Singh | Optical character recognition techniques: a survey | |
| Singh et al. | Offline script identification from multilingual indic-script documents: a state-of-the-art | |
| Fischer | Handwriting recognition in historical documents | |
| Kumar et al. | Classification of characters and grading writers in offline handwritten Gurmukhi script | |
| Hansen | A Matlab Project in Optical Character Recognition (OCR) | |
| Malakar et al. | A holistic approach for handwritten Hindi word recognition | |
| Assabie et al. | Offline handwritten Amharic word recognition | |
| Bukhari et al. | High performance layout analysis of Arabic and Urdu document images | |
| Elgammal et al. | A graph-based segmentation and feature extraction framework for Arabic text recognition | |
| Inunganbi et al. | Recognition of handwritten Meitei Mayek script based on texture feature | |
| US8340428B2 (en) | Unsupervised writer style adaptation for handwritten word spotting | |
| Panda et al. | Odia offline typewritten character recognition using template matching with unicode mapping | |
| Jena et al. | Implementation of linear discriminant analysis for Odia numeral recognition | |
| Antony et al. | Haar features based handwritten character recognition system for Tulu script | |
| Aravinda et al. | Template matching method for Kannada handwritten recognition based on correlation analysis | |
| Nadeem et al. | Character recognition using template matching | |
| Mane et al. | Handwritten character recognition using elastic matching and PCA | |
| CN110210582A (en) | A kind of Chinese handwriting identifying method based on part cooperation presentation class | |
| Bharathi et al. | Improvement of Telugu OCR by segmentation of Touching Characters | |
| Revathi et al. | A survey on OCR for Telugu language | |
| Awan et al. | Handwritten Sindhi character recognition using neural networks |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190906 |
|
| RJ01 | Rejection of invention patent application after publication |