CN107886090A - A kind of single sample face recognition method, system, equipment and readable storage medium storing program for executing - Google Patents

A kind of single sample face recognition method, system, equipment and readable storage medium storing program for executing Download PDF

Info

Publication number
CN107886090A
CN107886090A CN201711352665.9A CN201711352665A CN107886090A CN 107886090 A CN107886090 A CN 107886090A CN 201711352665 A CN201711352665 A CN 201711352665A CN 107886090 A CN107886090 A CN 107886090A
Authority
CN
China
Prior art keywords
reference data
multiple dimensioned
single sample
data
face
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201711352665.9A
Other languages
Chinese (zh)
Other versions
CN107886090B (en
Inventor
张莉
徐晓祥
王邦军
凌兴宏
姚望舒
张召
李凡长
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Suzhou University
Original Assignee
Suzhou University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Suzhou University filed Critical Suzhou University
Priority to CN201711352665.9A priority Critical patent/CN107886090B/en
Publication of CN107886090A publication Critical patent/CN107886090A/en
Application granted granted Critical
Publication of CN107886090B publication Critical patent/CN107886090B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Computer Interaction (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)

Abstract

This application discloses single sample face recognition method, system, equipment and readable storage medium storing program for executing, wherein, this method includes：The face image data marked is obtained, obtains reference data；By carrying out multiple dimensioned supporting vector conversion to the reference data, the reference data multiple dimensioned higher-dimension projection accordingly is obtained；The feature of the multiple dimensioned higher-dimension projection is extracted, obtains the reference data multiple dimensioned high dimensional feature sequence accordingly；Using the multiple dimensioned high dimensional feature sequence, back projection is carried out to the reference data and data to be tested respectively, obtains the reference data and the corresponding virtual image of data to be tested；According to virtual image corresponding to the reference data, the classification accuracy of the calculating corresponding virtual image of data to be tested, to complete the recognition of face to single sample.Organically combined it can be seen that single sample face recognition method provided by the invention expands feature extraction and sample, so as to improve the accuracy rate of single sample recognition of face.

Description

A kind of single sample face recognition method, system, equipment and readable storage medium storing program for executing
Technical field
The present invention relates to field of face identification, more particularly to a kind of single sample face recognition method, system, equipment and readable Storage medium.
Background technology
Face recognition technology refers to that computer to the facial image detected, is carried out with the face database stored in computer After matching, respective identity identification is carried out.Single sample recognition of face refers to there was only 1 width facial image in training set as training sample This.There was only 1 width or a small amount of people mostly in practical application, such as facial image database of public security, airport and department of customs Face image.Research shows that the performance of how many pairs of recognitions of face of number of training purpose has a significant impact.Many outstanding faces are known Other algorithm performance when handling single sample recognition of face problem drastically declines or even failed.Single sample recognition of face problem is ground Study carefully the difficult point and hot issue in always recognition of face research.
In the prior art, solution to the problems described above is generally divided into two classes.One kind is gone out from the angle of feature selecting Hair, find and select the feature for comparing sample size robust to carry out recognition of face, another kind of gone out from the angle of exptended sample Hair, generates multiple virtual samples, so as to reduce the influence of sample size.But prior art is general only from the aspect of single, Cause the accuracy rate to single sample recognition of face relatively low.
Therefore, how to improve the accuracy rate of single sample recognition of face is those skilled in the art's urgent problem to be solved.
The content of the invention
In view of this, it is an object of the invention to provide a kind of single sample face recognition method, system, equipment and readable deposit Storage media, improve the accuracy rate of single sample recognition of face.Its concrete scheme is as follows：
A kind of single sample face recognition method, including：
The face image data marked is obtained, obtains reference data；
By carrying out multiple dimensioned supporting vector conversion to the reference data, it is multiple dimensioned accordingly to obtain the reference data Higher-dimension projects；
The feature of the multiple dimensioned higher-dimension projection is extracted, obtains the reference data multiple dimensioned high dimensional feature sequence accordingly Row；
Using the multiple dimensioned high dimensional feature sequence, back projection is carried out to the reference data and data to be tested respectively, Obtain the reference data and the corresponding virtual image of data to be tested；
According to virtual image corresponding to the reference data, the classification of the corresponding virtual image of data to be tested is calculated Accuracy rate, to complete the recognition of face to single sample.
Optionally, described to obtain the face image data marked, obtaining the process of reference data includes：
The face image data marked and corresponding classification information are obtained, obtains reference data.
Optionally, it is described by carrying out multiple dimensioned supporting vector conversion to the reference data, obtain the reference data The process of corresponding multiple dimensioned higher-dimension projection includes：
Multi-scale filtering is carried out to the reference data, obtains the corresponding Graham matrix of the reference data；
According to the Graham matrix, the multiple dimensioned source for calculating the Graham matrix becomes matrix, obtains the reference number Become matrix according to corresponding multiple dimensioned source；
The corresponding supporting vector of matrix computations is become according to the multiple dimensioned source and crosses filter, it is corresponding to obtain the reference data Supporting vector crosses filter；
Using vector filter, corresponding multiple dimensioned higher-dimension projection matrix is calculated, it is corresponding to obtain the reference data Multiple dimensioned higher-dimension projection matrix.
Optionally, it is described that multi-scale filtering is carried out to the reference data, obtain the corresponding Graham of the reference data The process of matrix includes：
Multi-scale filtering is carried out to the reference data using gaussian kernel function, obtains the corresponding Gray of the reference data Nurse matrix.
Optionally, the feature of the extraction multiple dimensioned higher-dimension projection, it is multiple dimensioned accordingly to obtain the reference data The process of high dimensional feature sequence includes：
Extract the central row of the multiple dimensioned higher-dimension projection matrix, obtain the reference data it is corresponding it is multiple dimensioned support to Change of variable characteristic sequence.
Optionally, the virtual image according to corresponding to the reference data, it is empty accordingly that the data to be tested are calculated Intend the classification accuracy of image, included with completing the process of the recognition of face to single sample：
According to virtual image corresponding to the reference data, it is corresponding to calculate the data to be tested using arest neighbors method The classification accuracy of virtual image, to complete the recognition of face to single sample.
Accordingly, invention additionally discloses a kind of single sample face identification system, including：
Reference data acquisition module, for obtaining the face image data marked, obtain reference data；
Higher-dimension projects acquisition module, for by carrying out multiple dimensioned supporting vector conversion to the reference data, obtaining institute State reference data multiple dimensioned higher-dimension projection accordingly；
High dimensional feature retrieval module, for extracting the feature of the multiple dimensioned higher-dimension projection, obtain the reference number According to corresponding multiple dimensioned high dimensional feature sequence；
Virtual image acquisition module, for utilizing the multiple dimensioned high dimensional feature sequence, respectively to the reference data and Data to be tested carry out back projection, obtain the reference data and the corresponding virtual image of data to be tested；
Classification accuracy computing module, for the virtual image according to corresponding to the reference data, calculate described to be tested The classification accuracy of the corresponding virtual image of data, to complete the recognition of face to single sample.
Accordingly, present invention also offers a kind of single sample face recognition device, single sample face recognition device bag Include memory, processor and the single sample recognition of face journey that is stored on the memory and can run on the processor Sequence, the step of single sample recognition of face program is arranged for carrying out above-mentioned single sample face recognition method.
Accordingly, present invention also offers a kind of computer-readable recording medium, on the computer-readable recording medium Single sample recognition of face program is stored with, single sample recognition of face program realizes above-mentioned single sample when being executed by processor The step of face identification method.
Single sample face recognition method provided by the invention, including：The face image data marked is obtained, is referred to Data；By carrying out multiple dimensioned supporting vector conversion to reference data, reference data multiple dimensioned higher-dimension projection accordingly is obtained；Carry The feature of multiple dimensioned higher-dimension projection is taken, obtains reference data multiple dimensioned high dimensional feature sequence accordingly；It is special using multiple dimensioned higher-dimension Sequence is levied, back projection is carried out to reference data and data to be tested respectively, reference data is obtained and data to be tested is empty accordingly Intend image；The virtual image according to corresponding to reference data, the classification accuracy of the corresponding virtual image of data to be tested is calculated, with Complete the recognition of face to single sample.It can be seen that single sample face recognition method provided by the invention passes through the reference to having marked Data carry out multiple dimensioned supporting vector conversion, feature extraction, obtain corresponding multiple dimensioned high dimensional feature sequence, then, using more Yardstick high dimensional feature sequence, back projection is carried out to reference data and data to be tested respectively, obtains reference data and number to be tested According to corresponding virtual image；The virtual image according to corresponding to reference data, calculate point of the corresponding virtual image of data to be tested Class accuracy rate, i.e., feature extraction of the prior art and sample are expanded and organically combined, so as to improve single sample face The accuracy rate of identification.
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 There is the required accompanying drawing used in technology description to be briefly described, it should be apparent that, drawings in the following description are only this The embodiment of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis The accompanying drawing of offer obtains other accompanying drawings.
Fig. 1 is a kind of flow chart of single sample face recognition method provided in an embodiment of the present invention；
Fig. 2 is a kind of specific implementation flow chart of single sample face recognition method provided in an embodiment of the present invention；
Fig. 3 is the virtual image figure that a kind of single sample face recognition method provided in an embodiment of the present invention obtains；
Fig. 4 is a kind of structural representation of single sample face identification system provided in an embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made Embodiment, belong to the scope of protection of the invention.
The embodiment of the invention discloses a kind of single sample face recognition method, as shown in figure 1, including：
Step S11：The face image data marked is obtained, obtains reference data.
Need to illustrate, obtain the face image data marked, obtained reference data includes but is not limited to Need to identify the data under the expression of face, illumination, rotation status.Obtaining needs the face image data that identifies and corresponding Classification information, obtain reference data.
Step S12：By carrying out multiple dimensioned supporting vector conversion to reference data, it is multiple dimensioned accordingly to obtain reference data Higher-dimension projects.
Need to illustrate, by carrying out multiple dimensioned supporting vector conversion to reference data, obtain reference data phase The process for the multiple dimensioned higher-dimension projection answered includes：
Multi-scale filtering is carried out to reference data, obtains the corresponding Graham matrix of reference data；
According to Graham matrix, the multiple dimensioned source for calculating Graham matrix becomes matrix, obtains reference data more chis accordingly Degree source becomes matrix；
The corresponding supporting vector of matrix computations is become according to multiple dimensioned source and crosses filter, obtains the corresponding supporting vector of reference data Cross filter；
Using vector filter, corresponding multiple dimensioned higher-dimension projection matrix is calculated, it is multiple dimensioned accordingly to obtain reference data Higher-dimension projection matrix.
It should be further stated that above-mentioned steps with algorithm be prior art, will not be repeated here.
It should be further stated that above-mentioned carry out multi-scale filtering to reference data, the corresponding lattice of reference data are obtained The process of rem matrix can be：
Multi-scale filtering is carried out to reference data using gaussian kernel function, obtains the corresponding Graham matrix of reference data.
Step S13：The feature of multiple dimensioned higher-dimension projection is extracted, obtains reference data multiple dimensioned high dimensional feature sequence accordingly Row.
Need to illustrate, extract the feature of multiple dimensioned higher-dimension projection, obtain reference data multiple dimensioned height accordingly Dimensional feature sequence process includes：
The central row of multiple dimensioned higher-dimension projection matrix is extracted, obtaining reference data, multiple dimensioned supporting vector conversion is special accordingly Levy sequence.
It should be further stated that above-mentioned steps with algorithm be prior art, will not be repeated here.
Step S14：Using multiple dimensioned high dimensional feature sequence, back projection is carried out to reference data and data to be tested respectively, Obtain reference data and the corresponding virtual image of data to be tested.
Step S15：The virtual image according to corresponding to reference data, calculate the classification of the corresponding virtual image of data to be tested Accuracy rate, to complete the recognition of face to single sample.
Need to illustrate, the above-mentioned virtual image according to corresponding to reference data, it is corresponding to calculate data to be tested The classification accuracy of virtual image, can be to complete the process of the recognition of face to single sample：
The virtual image according to corresponding to reference data, the corresponding virtual image of data to be tested is calculated using arest neighbors method Classification accuracy, to complete recognition of face to single sample.
It should be further stated that face corresponding to face corresponding to the reference data marked and data to be tested is Same face.
It can be seen that single sample face recognition method provided in an embodiment of the present invention is more by being carried out to the reference data marked The conversion of yardstick supporting vector, feature extraction, corresponding multiple dimensioned high dimensional feature sequence is obtained, it is then, special using multiple dimensioned higher-dimension Sequence is levied, back projection is carried out to reference data and data to be tested respectively, reference data is obtained and data to be tested is empty accordingly Intend image；The virtual image according to corresponding to reference data, the classification accuracy of the corresponding virtual image of data to be tested is calculated, i.e., Feature extraction of the prior art and sample are expanded and organically combined, so as to improve the accurate of single sample recognition of face Rate.
The embodiment of the invention also discloses a kind of specific embodiment of single sample face recognition method, as shown in Figure 2.
The specific embodiment of the invention is on the premise of with single sample face recognition method provided by the invention, in U.S. AR people Tested on face data set.Need to illustrate, AR human face datas are concentrated and contain 2600 facial images, specifically Ground, facial image one share 100 classes, have 26 images per class.
The specific embodiment of the invention chooses preceding 5 class in AR data sets, and the reference number marked is used as using first per class According to other 25 are used as test data, and exponent number takes 4.Need to illustrate, this 26 pictures contains people and do not shared the same light According to pictures corresponding to various situations such as, different shooting angles, different expressions.This experiment randomly selects an image from every class As training sample, other images are as test sample.
The training process and test process that the specific embodiment of the invention includes；Wherein, training process includes：
(11) the known reference data markedWherein X is view data, and y is classification information, Ntr It is training samples number, as classification number.In the present embodiment, Ntr=5.Wherein, it is some there was only a pictures per class image The face image data of people, corresponding test image are these people in expression, illumination, the facial image number under rotation status According to.Need to illustrate, the reference data marked is training data.
(12) Graham matrix is produced by carrying out multi-scale filtering to the above-mentioned reference data marked.Specifically, orderr For the yardstick exponent number of decomposition.To jth rank, l=(2 is generatedj+1)×(2j+ 1) individual rectangular block
{-2j,-2j+1,...,2j}×{-2j,-2j+1,...,2j, j=1,2 ..., r.
And the coordinate points in rectangular block are designated as:
Wherein, mxi,myi∈{-2j,-2j+1,...,2j}。
Then, multiple dimensioned gaussian kernel function is utilized：
Produce a series of multiple dimensioned Graham matrix { K1,K2,...,Kr, wherein jth rank Graham matrix KjI-th Row pth column element is kj(xi,xp).In the specific embodiment of the invention, r=4, σ=0.5.
(13) multi-scale filtering in (12) is utilized to produce the multiple dimensioned source change matrix of Graham matrix computations
{Ω1,Ω2,...,Ωr},Ωj=Kj+CjI,Cj>=0, j=1,2 ..., r.
Wherein, I is unit matrix, CjIt is j rank regularization factors.In the specific embodiment of the invention, Cj=21-j。
(14) multiple dimensioned source change matrix computations supporting vector in (13) is utilized to cross filter：
{A1,A2,...,Ar},Aj=Ωj -1, j=1,2 ..., r
Wherein, e is complete 1 vector.
(15) supporting vector in (14) is utilized to cross the multiple dimensioned higher-dimension projection matrix of filter calculating：
{Q1,Q2,...,Qr},Qj=Aj(I-eBj T), j=1,2 ..., r.
(16) multiple dimensioned higher-dimension projection matrix Q in (15) is extractedjCentral row, and be reconstructed into (2j+1)×(2j+ 1) conduct The multiple dimensioned sub- SV of supporting vector transform characteristicsj, composition sequence { SV1,SV2,...,SVr}。
(17) the reference data X to having markedi, according to multiple dimensioned supporting vector transform characteristics sequence { SV in (16)1, SV2,...,SVrProduce multiple dimensioned supporting vector conversion virtual image S corresponding to the reference data markedir；Wherein, i=1, 2 ..., Ntr, Sij=LPij*SVj, LPi1=Xi, LPi(j+1)=LPij-Sij, j=l, 2 ..., r.
Test process includes：
(21) input test data X'.According to multiple dimensioned supporting vector transform characteristics subsequence { SV in step (16)1, SV2,...,SVrProduce the multiple dimensioned supporting vector conversion virtual image S' of test datar, as shown in Figure 3；Wherein, S'j= LPj*SVj, LP1=X', LPj+1=LPj-Sj, j=1,2 ..., r.
(22) according to the multiple dimensioned supporting vector conversion virtual graph image set { S obtained in the reference data markedir, i= 1,2 ..., Ntr, the classification for obtaining treating the multiple dimensioned supporting vector conversion virtual image of test data using arest neighbors method are accurate True rate.
The beneficial effect of the embodiment of the present invention can pass through following experimental verification：
By single sample face recognition method proposed by the present invention, the training of single recognition of face is carried out with AR human face datas collection And test.Experiment shows that the present invention can effectively complete feature selecting to every class face, finds out key feature, produces virtual sample This, as shown in Figure 3.By taking 3 faces in a kind of data in AR data sets as an example, it is 4 to take exponent number, depicts the preceding more chis of 3 rank respectively Spend supporting vector conversion virtual image sequence.Corresponding experiment effect is found：List proposed by the present invention based on supporting vector conversion The accuracy rate of sample face recognition method is 75.5%, more than the accuracy rate and LU decomposition methods of singular value decomposition method 63.5% 65.25% accuracy rate, thus there is higher nicety of grading.
Accordingly, a kind of single sample face identification system is also disclosed in the embodiment of the present invention, as shown in figure 4, including：
Reference data acquisition module 11, for obtaining the face image data marked, obtain reference data；
Higher-dimension projects acquisition module 12, for by carrying out multiple dimensioned supporting vector conversion to reference data, being referred to Data multiple dimensioned higher-dimension projection accordingly；
High dimensional feature retrieval module 13, for extracting the feature of multiple dimensioned higher-dimension projection, it is corresponding to obtain reference data Multiple dimensioned high dimensional feature sequence；
Virtual image acquisition module 14, for utilizing multiple dimensioned high dimensional feature sequence, respectively to reference data and to be tested Data carry out back projection, obtain reference data and the corresponding virtual image of data to be tested；
Classification accuracy computing module 15, for the virtual image according to corresponding to reference data, calculate data phase to be tested The classification accuracy for the virtual image answered, to complete the recognition of face to single sample.
Accordingly, present invention also offers a kind of single sample face recognition device, single sample face recognition device includes depositing Reservoir, processor and storage on a memory and the single sample recognition of face program that can run on a processor, single sample face Recognizer is arranged for carrying out the step of above-mentioned single sample face recognition method.
Accordingly, present invention also offers a kind of computer-readable recording medium, stored on computer-readable recording medium There is single sample recognition of face program, single sample recognition of face program realizes above-mentioned single sample recognition of face when being executed by processor The step of method.
Finally, it is to be noted that, herein, such as first and second or the like relational terms be used merely to by One entity or operation make a distinction with another entity or operation, and not necessarily require or imply these entities or operation Between any this actual relation or order be present.Moreover, term " comprising ", "comprising" or its any other variant meaning Covering including for nonexcludability, so that process, method, article or equipment including a series of elements not only include that A little key elements, but also the other element including being not expressly set out, or also include for this process, method, article or The intrinsic key element of equipment.In the absence of more restrictions, the key element limited by sentence "including a ...", is not arranged Except other identical element in the process including the key element, method, article or equipment being also present.
A kind of single sample face recognition method, system, equipment and readable storage medium storing program for executing provided by the present invention are entered above Go and be discussed in detail, specific case used herein is set forth to the principle and embodiment of the present invention, and the above is implemented The explanation of example is only intended to help the method and its core concept for understanding the present invention；Meanwhile for the general technology people of this area Member, according to the thought of the present invention, there will be changes in specific embodiments and applications, in summary, this explanation Book content should not be construed as limiting the invention.

Claims (9)

  1. A kind of 1. single sample face recognition method, it is characterised in that including：
    The face image data marked is obtained, obtains reference data；
    By carrying out multiple dimensioned supporting vector conversion to the reference data, the reference data multiple dimensioned higher-dimension accordingly is obtained Projection；
    The feature of the multiple dimensioned higher-dimension projection is extracted, obtains the reference data multiple dimensioned high dimensional feature sequence accordingly；
    Using the multiple dimensioned high dimensional feature sequence, back projection is carried out to the reference data and data to be tested respectively, obtained The reference data and the corresponding virtual image of data to be tested；
    According to virtual image corresponding to the reference data, the classification for calculating the corresponding virtual image of data to be tested is accurate Rate, to complete the recognition of face to single sample.
  2. 2. single sample face recognition method according to claim 1, it is characterised in that described to obtain the face figure marked As data, obtaining the process of reference data includes：
    The face image data marked and corresponding classification information are obtained, obtains reference data.
  3. 3. single sample face recognition method according to claim 1, it is characterised in that described by the reference data Multiple dimensioned supporting vector conversion is carried out, obtaining the process of the reference data multiple dimensioned higher-dimension projection accordingly includes：
    Multi-scale filtering is carried out to the reference data, obtains the corresponding Graham matrix of the reference data；
    According to the Graham matrix, the multiple dimensioned source for calculating the Graham matrix becomes matrix, obtains the reference data phase The multiple dimensioned source answered becomes matrix；
    The corresponding supporting vector of matrix computations is become according to the multiple dimensioned source and crosses filter, the reference data is obtained and supports accordingly Vector filter；
    Using vector filter, corresponding multiple dimensioned higher-dimension projection matrix is calculated, it is more accordingly to obtain the reference data Yardstick higher-dimension projection matrix.
  4. 4. single sample face recognition method according to claim 3, it is characterised in that described to be carried out to the reference data Multi-scale filtering, obtaining the process of the corresponding Graham matrix of the reference data includes：
    Multi-scale filtering is carried out to the reference data using gaussian kernel function, obtains the corresponding Graham square of the reference data Battle array.
  5. 5. single sample face recognition method according to claim 1, it is characterised in that the extraction multiple dimensioned higher-dimension The feature of projection, obtaining the process of the reference data multiple dimensioned high dimensional feature sequence accordingly includes：
    The central row of the multiple dimensioned higher-dimension projection matrix is extracted, obtaining the reference data, multiple dimensioned supporting vector becomes accordingly Change characteristic sequence.
  6. 6. single sample face recognition method according to any one of claim 1 to 5, it is characterised in that described in the basis Virtual image corresponding to reference data, the classification accuracy of the corresponding virtual image of data to be tested is calculated, with completion pair The process of the recognition of face of single sample includes：
    According to virtual image corresponding to the reference data, it is virtual accordingly to calculate the data to be tested using arest neighbors method The classification accuracy of image, to complete the recognition of face to single sample.
  7. A kind of 7. single sample face identification system, it is characterised in that including：
    Reference data acquisition module, for obtaining the face image data marked, obtain reference data；
    Higher-dimension projects acquisition module, for by carrying out multiple dimensioned supporting vector conversion to the reference data, obtaining the ginseng Examine data multiple dimensioned higher-dimension projection accordingly；
    High dimensional feature retrieval module, for extracting the feature of the multiple dimensioned higher-dimension projection, obtain the reference data phase The multiple dimensioned high dimensional feature sequence answered；
    Virtual image acquisition module, for utilizing the multiple dimensioned high dimensional feature sequence, respectively to the reference data and to be measured Try data and carry out back projection, obtain the reference data and the corresponding virtual image of data to be tested；
    Classification accuracy computing module, for the virtual image according to corresponding to the reference data, calculate the data to be tested The classification accuracy of corresponding virtual image, to complete the recognition of face to single sample.
  8. 8. a kind of single sample face recognition device, it is characterised in that single sample face recognition device includes memory, processing Device and the single sample recognition of face program that is stored on the memory and can run on the processor, single sample people Face recognizer is arranged for carrying out the step of single sample face recognition method as any one of claim 1 to 6.
  9. 9. a kind of computer-readable recording medium, it is characterised in that single sample is stored with the computer-readable recording medium Recognition of face program, realized when single sample recognition of face program is executed by processor such as any one of claim 1 to 6 institute The step of single sample face recognition method stated.
CN201711352665.9A 2017-12-15 2017-12-15 Single-sample face recognition method, system, equipment and readable storage medium Active CN107886090B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711352665.9A CN107886090B (en) 2017-12-15 2017-12-15 Single-sample face recognition method, system, equipment and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711352665.9A CN107886090B (en) 2017-12-15 2017-12-15 Single-sample face recognition method, system, equipment and readable storage medium

Publications (2)

Publication Number Publication Date
CN107886090A true CN107886090A (en) 2018-04-06
CN107886090B CN107886090B (en) 2021-07-30

Family

ID=61771585

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711352665.9A Active CN107886090B (en) 2017-12-15 2017-12-15 Single-sample face recognition method, system, equipment and readable storage medium

Country Status (1)

Country Link
CN (1) CN107886090B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103413119A (en) * 2013-07-24 2013-11-27 中山大学 Single sample face recognition method based on face sparse descriptors
CN103839051A (en) * 2014-03-05 2014-06-04 华侨大学 Single-sample vein recognition method based on 2DPCA and partition LBP
CN106778522A (en) * 2016-11-25 2017-05-31 江南大学 A kind of single sample face recognition method extracted based on Gabor characteristic with spatial alternation
CN106897700A (en) * 2017-02-27 2017-06-27 苏州大学 A kind of single sample face recognition method and system
CN106951819A (en) * 2016-08-19 2017-07-14 南京理工大学 The single sample face recognition method screened based on sparse probability distribution and multistage classification
CN107239741A (en) * 2017-05-10 2017-10-10 杭州电子科技大学 A kind of single sample face recognition method based on sparse reconstruct

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103413119A (en) * 2013-07-24 2013-11-27 中山大学 Single sample face recognition method based on face sparse descriptors
CN103839051A (en) * 2014-03-05 2014-06-04 华侨大学 Single-sample vein recognition method based on 2DPCA and partition LBP
CN106951819A (en) * 2016-08-19 2017-07-14 南京理工大学 The single sample face recognition method screened based on sparse probability distribution and multistage classification
CN106778522A (en) * 2016-11-25 2017-05-31 江南大学 A kind of single sample face recognition method extracted based on Gabor characteristic with spatial alternation
CN106897700A (en) * 2017-02-27 2017-06-27 苏州大学 A kind of single sample face recognition method and system
CN107239741A (en) * 2017-05-10 2017-10-10 杭州电子科技大学 A kind of single sample face recognition method based on sparse reconstruct

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
XU YAN: "《Single sample face recognition based on sample augments and MSD fusion》", 《IEEE XPLORE》 *
ZHANG YAN ET AL.: "《Virtual multi-pose faces produced method with single sample on homography transformation》", 《IEEE XPLORE》 *

Also Published As

Publication number Publication date
CN107886090B (en) 2021-07-30

Similar Documents

Publication Publication Date Title
Chang et al. Unsupervised transfer learning via multi-scale convolutional sparse coding for biomedical applications
Van Der Maaten Barnes-hut-sne
Ganin et al. -fields: Neural network nearest neighbor fields for image transforms
CN109214273A (en) Facial image comparison method, device, computer equipment and storage medium
CN111680678B (en) Target area identification method, device, equipment and readable storage medium
Li et al. Local log-euclidean covariance matrix (l 2 ecm) for image representation and its applications
JP2013012190A (en) Method of approximating gabor filter as block-gabor filter, and memory to store data structure for access by application program running on processor
CN104966081B (en) Spine image-recognizing method
CN108875767A (en) Method, apparatus, system and the computer storage medium of image recognition
WO2019246397A1 (en) A fully fourier space spherical convolutional neural network based on clebsch-gordan transforms
CN107967461A (en) The training of SVM difference models and face verification method, apparatus, terminal and storage medium
CN103839066A (en) Feature extraction method from biological vision
CN110084268A (en) Image comparison method, face identification method and device, computer storage medium
Wang et al. Building correlations between filters in convolutional neural networks
Lu et al. Similar handwritten chinese character recognition based on CNN-SVM
CN101599077B (en) Method for retrieving three-dimensional object
CN102708589A (en) Three-dimensional target multi-viewpoint view modeling method on basis of feature clustering
CN107967331A (en) It is matched with the data preparation method of high-performance computer structure
Kekre et al. Query based Image Retrieval using Kekre’s, DCT and Hybrid wavelet Transform over 1st and 2nd Moment
CN112149747A (en) Hyperspectral image classification method based on improved Ghost3D module and covariance pooling
Nurzynska et al. Evaluation of Keypoint Descriptors for Flight Simulator Cockpit Elements: WrightBroS Database
CN107886090A (en) A kind of single sample face recognition method, system, equipment and readable storage medium storing program for executing
CN115661472A (en) Image duplicate checking method and device, computer equipment and storage medium
CN115588220A (en) Two-stage multi-scale self-adaptive low-resolution face recognition method and application
CN106709502B (en) Multi-feature fusion recognition method based on voting method

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