CN103839051B - The method of single sample hand vein recognition based on 2DPCA and subregion LBP - Google Patents
The method of single sample hand vein recognition based on 2DPCA and subregion LBP Download PDFInfo
- Publication number
- CN103839051B CN103839051B CN201410077823.4A CN201410077823A CN103839051B CN 103839051 B CN103839051 B CN 103839051B CN 201410077823 A CN201410077823 A CN 201410077823A CN 103839051 B CN103839051 B CN 103839051B
- Authority
- CN
- China
- Prior art keywords
- image
- lbp
- single sample
- subregion
- 2dpca
- 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.)
- Expired - Fee Related
Links
Landscapes
- Collating Specific Patterns (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The present invention is a kind of vein identification method based on 2DPCA features and subregion LBP features, overcome the problem of single specimen discerning performance is relatively low, just with palm vein image sample, avoid the trouble for gathering a variety of biometric samples, test result indicates that, the present invention has been largely overcoming the influence of single sample situation in palm vein identification, improves the discrimination in the case of single sample, so as to ensure that the practicality of single sample palm vein identification.
Description
Technical field
The invention belongs to palm vein identification technology field, it is related to a kind of single sample vein based on 2DPCA and subregion LBP
Know method for distinguishing.
Background technology
With the development and progress of modern society, requirement of the people to social informatization and networking security is increasingly
It is high.Vein identification is as a kind of current newest biometrics identification technology, also correspondingly by domestic and international research institution
More and more pay attention to company.So-called hand vein recognition, refers to carry out body using the vein blood vessel under skin as identity characteristic
The technology of part identification.Vein is as one kind of blood vessel, and it is than artery close to skin, it is easier to is detected and obtained by near infrared ray.
The curve of the vein pattern complexity suitable with branch, everyone difference is fully aware of, according to statistics, and it is similar that palm vein is distributed
Rate is only less than 0.00008%, and compared to the identification of fingerprint, iris etc., it has higher accuracy.Vein blood vessel is located at body
Inside table, its changes in microstructure is increased less with the age, and is difficult to forge or operation change, once epidermis can be avoided
Defect undermined that fingerprint, personal recognition carries out；Compared to DNA, iris recognition, its gatherer process is very friendly.One
Individual typical vein recognition system is broadly divided into two parts：One is registration part, including image preprocessing, feature extraction and spy
Levy the generation of database；Two be verification portion, including image preprocessing, feature extraction and is carried out with the feature in property data base
Matching judgment.
But, in current hand vein recognition research, most is the algorithm research based on multisample, the identification of algorithm
Rate can be improved with the increase of training sample number, but when only single sample, recognition performance can be then remarkably decreased.For
Single specimen discerning performance deficiency is solved, concern is primarily with multi-modal biological characteristic fusion method for current research.But,
In practical application, the situation for lacking a variety of biometric samples is frequently encountered.
The content of the invention
The purpose of the present invention in order to avoid gather the troubles of a variety of biometric samples there is provided one kind based on 2DPCA and
Subregion LBP single sample palm vein recognition methods, effectively improves single sample palm vein discrimination.
A kind of single sample palm vein recognition methods based on 2DPCA and subregion LBP of the present invention, comprises the following steps：
Step 1, generation 2DPCA feature databases：
Step 11, single sample palm vein image original to every in training image storehouse are adopted by image resampling method
Sample 4 generations, 4 virtual subgraphs, then recycle the virtual image generating algorithm of singular value disturbance, to original single sample
The singular value of palm vein image carries out 4 other 4 virtual subgraphs of generation of disturbance；
Step 12, using 2DPCA algorithms vein pattern is extracted from 8 virtual subgraphs of generation, obtain original single sample
The projection properties image formation 2DPCA feature databases of this palm vein image；
Step 2, generation subregion LBP feature databases：
Step 21, by the ROI region of the original single sample palm vein image of every in training image storehouse by the decile of row 2 and
The pattern of the decile of row 2, is divided into 4 equal-sized subgraphs；
Step 22, circle LBP operator extraction textural characteristics are utilized to each subgraph, its step is：To in subgraph
Each pixel, in window of the radius for 8 circular LBP operators, the gray value using window center point is threshold value in window
16 sampled point pixels make binary conversion treatment, obtain 16 bits, and calculation formula is：
WhereinZp、ZcRepresent the pixel value and the picture of LBP operator central points of sampled point in LBP operators
Element value；
Step 23, the LBP textural characteristics to obtaining, further extract its equivalent formulations, i.e., when LBP textural characteristics institute is right
The circulation binary number answered from 0 to 1 or from 1 to 0 be up to saltus step twice when, the corresponding binary number of LBP textural characteristics is just
Referred to as equivalent formulations；
Step 24, the Uniform LBP histograms of the corresponding each subgraph of original single sample palm vein image are connected
Pick up to form the vein pattern expression figure of entire image, form subregion LBP feature databases；
Step 3, double-deck screening：
Step 31, the 2DPCA features of extraction test image are matched with 2DPCA feature databases by nearest neighbor method, select phase
Like before degree ranking 10% sample, then using this 10% sample class, choose corresponding subregion LBP feature databases and be used as step
32 search sample space；
The subregion LBP feature databases reduced in step 32, the subregion LBP features of extraction test image and step 31 are by arest neighbors
Method selects best match pattern class, i.e., the sample minimum with subregion LBP characteristic distances is selected from the subregion LBP feature databases of diminution
Classification, matching process terminates.
Described image resampling method is new to obtain by being sampled to original single sample palm vein image
Virtual subgraph, detailed process is that the original single sample palm vein image of every in training image storehouse is divided into P × Q sizes
Sub-block, the pixel of then being sampled from the sub-block of each P × Q sizes constitutes the one of original single sample palm vein image
Virtual subgraph is opened, described P and Q are the length and width of sampling interval respectively, P=Q=2 are made, if original single sample palm vein figure
The size of picture is M × N, and M, N are the length and width of image respectively, then the virtual subgraph Iij generated by image resampling
(m, n) is:
Iij(m,n)=I((m-1)×P+i,(n-1)×Q+j) （1）
Formula（1）In, 1≤m≤M/P, 1≤n≤N/Q, 1≤i≤P, 1≤j≤Q.
The described virtual image generation method disturbed based on singular value is by original single sample palm vein image
Singular value disturbed to generate new virtual subgraph, it is assumed that I (x, y) represent a M × N vein image gray value
Distribution, wherein x ∈ [1, M], y ∈ [1, N] pass through formula（2）Obtain a new virtual subgraph P：
P=UΣnVT（2）
Formula（2）In, n represents coefficient of disturbance, and U, V are expressed as an orthogonal matrix, the transposition of T representing matrixs, Σ tables
Show the singular value that value on a diagonal matrix, its diagonal is image array I (x, y), U, Σ, V is by image array I (x, y)
Singular value decomposition formula（3）Determine：I=UΣVT（3）.
Described 2DPCA algorithms are directly to obtain covariance matrix using virtual subgraph, try to achieve the preceding d of covariance matrix
Characteristic vector composition projection matrix corresponding to individual larger characteristic value, then all projects to above-mentioned 8 virtual subgraphs
State in characteristic vector, obtain the projection properties image of original single sample palm vein image.
The present invention is a kind of vein identification method based on 2DPCA features and subregion LBP features, overcomes single specimen discerning
The problem of performance is relatively low, just with palm vein image sample, it is to avoid the troubles of a variety of biometric samples of collection, experiment
As a result show, the present invention has been largely overcoming the influence of single sample situation in palm vein identification, improves single sample
In the case of discrimination, so as to ensure that the practicality of single sample palm vein identification.
Brief description of the drawings
Fig. 1 is the palm vein identification process figure of the present invention；
Fig. 2 is the virtual image of image resampling generation of the present invention（P=Q=2）；
Fig. 3 is the basic LBP operators in the present invention；
Fig. 4 is Simulation identification result diagram in the present invention.
Below in conjunction with the drawings and specific embodiments, the invention will be further described.
Embodiment
As shown in figure 1, a kind of single sample palm vein recognition methods based on 2DPCA and subregion LBP of the present invention, specific bag
Include following steps：
Step 1, generation 2DPCA feature databases：
Step 11, single sample palm vein image original to every in training image storehouse are adopted by image resampling method
Sample 4 generations, 4 virtual subgraphs, then recycle the virtual image generating algorithm of singular value disturbance, to original single sample
The singular value of palm vein image carries out disturbance 4 times, to generate other 4 virtual subgraphs；
Wherein image resampling method is to obtain new void by being sampled to original single sample palm vein image
Intend subgraph, detailed process is as follows, the original single sample palm vein image of every in training image storehouse is divided into P × Q big
Small sub-block a, pixel of then being sampled from the sub-block of each P × Q sizes constitutes original single sample palm vein image
One virtual subgraph, described P and Q are the length and width of sampling interval respectively, and the present invention uniformly makes P=Q=2, image resampling
As a result as shown in Fig. 2 the size for setting original single sample palm vein image is M × N, M, N are the length and width of image respectively, then pass through
Cross a virtual subgraph I of image resampling generationij(m, n) is:
Iij(m,n)=I((m-1)×P+i,(n-1)×Q+j) （1）
Formula（1）In, 1≤m≤M/P, 1≤n≤N/Q, 1≤i≤P, 1≤j≤Q；
The virtual image generation method wherein disturbed based on singular value is by original single sample palm vein image
Singular value is disturbed to generate new virtual subgraph, and detailed process is as follows：Assuming that I (x, y) represents M × N vein
The grey value profile of image, wherein x ∈ [1, M], y ∈ [1, N] pass through formula（2）Obtain a new virtual subgraph P：
P=UΣnVT（2）
Formula（2）In, n represents coefficient of disturbance, and U, V are expressed as an orthogonal matrix, the transposition of T representing matrixs, Σ tables
Show the singular value that value on a diagonal matrix, its diagonal is image array I (x, y), U, Σ, V is by image array I (x, y)
Singular value decomposition formula（3）Determine：I=UΣVT（3）；
Step 12, using 2DPCA algorithms vein pattern is extracted from 8 virtual subgraphs of generation, obtain original single sample
The projection properties image of this palm vein image forms 2DPCA feature databases；
Described 2DPCA algorithms are directly to obtain covariance matrix using virtual subgraph, try to achieve the preceding d of covariance matrix
Characteristic vector composition projection matrix corresponding to individual larger characteristic value, then all projects to above-mentioned 8 virtual subgraphs
State in characteristic vector, obtain the projection properties image of original single sample palm vein image；
Step 2, generation subregion LBP feature databases：
Step 21, by the ROI region of the original single sample palm vein image of every in training image storehouse by the decile of row 2 and
The pattern of the decile of row 2, is divided into 4 equal-sized subgraphs；
Step 22, circle LBP operator extraction textural characteristics are utilized to each subgraph, its step is：To in subgraph
Each pixel, in window of the radius for 8 circular LBP operators, the gray value using window center point is threshold value in window
16 sampled point pixels make binary conversion treatment, obtain 16 bits；Method And Principle such as Fig. 3 of one simple LBP operator
Shown, calculation formula is as follows：
WhereinZp、ZcRepresent the pixel value and the picture of LBP operator central points of sampled point in LBP operators
Element value；
Step 23, the LBP textural characteristics to obtaining, further extract its equivalent formulations (Uniform Pattern), i.e.,
When the circulation binary number corresponding to LBP textural characteristics is from 0 to 1 or from 1 to 0 be up to saltus step twice, the LBP textural characteristics
Corresponding binary number is known as equivalent formulations；
Step 24, the Uniform LBP histograms of the corresponding each subgraph of original single sample palm vein image are connected
Pick up to form the vein pattern expression figure of entire image, form subregion LBP feature databases；
Step 3, double-deck screening：
Step 31, the 2DPCA features of extraction test image are matched with 2DPCA feature databases by nearest neighbor method, select phase
Like before degree ranking 10% sample, then using this 10% sample class, choose corresponding subregion LBP feature databases and be used as step
32 search sample space；
The subregion LBP feature databases reduced in step 32, the subregion LBP features of extraction test image and step 31 are by arest neighbors
Method selects best match pattern class, i.e., the sample minimum with subregion LBP characteristic distances is selected from the subregion LBP feature databases of diminution
Classification, matching process terminates.
Fig. 4 is the Simulation identification effect diagram of the present invention, it can be seen that the present invention using double-deck screening technique compared to
Only with 2DPCA or LBP vein identification method, the raising highly significant of discrimination, average recognition rate can reach
93.116%, with good practical value.
It is described above, only it is present pre-ferred embodiments, is not intended to limit the scope of the present invention, therefore
Any subtle modifications, equivalent variations and modifications that every technical spirit according to the present invention is made to above example, still belong to
In the range of technical solution of the present invention.
Claims (4)
1. a kind of single sample palm vein recognition methods based on 2DPCA and subregion LBP, it is characterised in that comprise the following steps：
Step 1, generation 2DPCA feature databases：
Step 11, single sample palm vein image original to every in training image storehouse pass through image resampling method sampling 4
4 virtual subgraphs of secondary generation, then recycle the virtual image generating algorithm of singular value disturbance, to original single sample palm
The singular value of vein image carries out 4 other 4 virtual subgraphs of generation of disturbance；
Step 12, using 2DPCA algorithms vein pattern is extracted from 8 virtual subgraphs of generation, obtain original single sample hand
The projection properties image formation 2DPCA feature databases of vena metacarpea image；
Step 2, generation subregion LBP feature databases：
Step 21, by the ROI region of the original single sample palm vein image of every in training image storehouse press the decile of row 2 and row 2
The pattern of decile, is divided into 4 equal-sized subgraphs；
Step 22, circle LBP operator extraction textural characteristics are utilized to each subgraph, its step is：To each in subgraph
Pixel, in window of the radius for 8 circular LBP operators, by threshold value of the gray value of window center point to 16 in window
Sampled point pixel makees binary conversion treatment, obtains 16 bits, and calculation formula is：
WhereinZp、ZcRepresent the pixel value and the pixel of LBP operator central points of sampled point in LBP operators
Value, P represents the number of pixel in the windows of circular LBP operators, and R represents the radius of the window of circular LBP operators, and I represents son
Image；
Step 23, the LBP textural characteristics to obtaining, further extract its equivalent formulations, i.e., when corresponding to LBP textural characteristics
Circulate binary number from 0 to 1 or from 1 to 0 be up to saltus step twice when, the corresponding binary number of LBP textural characteristics is known as
Equivalent formulations；
Step 24, the Uniform LBP histograms of the corresponding each subgraph of original single sample palm vein image are connected
To form the vein pattern expression figure of entire image, subregion LBP feature databases are formed；
Step 3, double-deck screening：
Step 31, the 2DPCA features of extraction test image are matched with 2DPCA feature databases by nearest neighbor method, select similarity
10% sample before ranking, then using this 10% sample class, chooses corresponding subregion LBP feature databases and is used as step 32
Search sample space；
The subregion LBP feature databases reduced in step 32, the subregion LBP features of extraction test image and step 31 are selected by nearest neighbor method
Go out best match pattern class, i.e., the sample class minimum with subregion LBP characteristic distances is selected from the subregion LBP feature databases of diminution
Not, matching process terminates.
2. a kind of single sample palm vein recognition methods based on 2DPCA and subregion LBP according to claim 1, it is special
Levy and be：Described image resampling method is to obtain new void by being sampled to original single sample palm vein image
Intend subgraph, detailed process is that the original single sample palm vein image of every in training image storehouse is divided into P × Q sizes
Sub-block a, pixel of then being sampled from the sub-block of each P × Q sizes constitutes one of original single sample palm vein image
Virtual subgraph, described P and Q are the length and width of sampling interval respectively, P=Q=2 are made, if original single sample palm vein figure
The size of picture is M × N, and M, N are the length and width of image respectively, then the virtual subgraph I generated by image resamplingij
(m, n) is:
Iij(m, n)=I ((m-1) × P+i, (n-1) × Q+j) (1)
In formula (1), 1≤m≤M/P, 1≤n≤N/Q, 1≤i≤P, 1≤j≤Q.
3. a kind of single sample palm vein recognition methods based on 2DPCA and subregion LBP according to claim 1, it is special
Levy and be：The described virtual image generation method disturbed based on singular value is by original single sample palm vein image
Singular value is disturbed to generate new virtual subgraph, it is assumed that I (x, y) represents the gray value point of M × N vein image
Cloth, wherein x ∈ [1, M], y ∈ [1, N], M and N represent the length and width of vein image respectively, and a new void is obtained by formula (2)
Intend subgraph P：
P=U ΣnVT (2)
In formula (2), n represents coefficient of disturbance, and U, V are expressed as an orthogonal matrix, and the transposition of T representing matrixs, Σ represents one
The singular value that value on individual diagonal matrix, its diagonal is image array I (x, y), U, Σ, V is strange by image array I's (x, y)
Different value decomposition formula (3) determines：I=U Σ VT (3)
I in formula (3) is vein image, it be in a two-dimentional image array, the image array each element I (x,
Y) gray value that vein image is put at (x, y) is represented.
4. a kind of single sample palm vein recognition methods based on 2DPCA and subregion LBP according to claim 1, it is special
Levy and be：Described 2DPCA algorithms are directly to obtain covariance matrix using virtual subgraph, try to achieve the preceding d of covariance matrix
Characteristic vector composition projection matrix corresponding to individual larger characteristic value, then all projects to above-mentioned 8 virtual subgraphs
State in characteristic vector, obtain the projection properties image of original single sample palm vein image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410077823.4A CN103839051B (en) | 2014-03-05 | 2014-03-05 | The method of single sample hand vein recognition based on 2DPCA and subregion LBP |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410077823.4A CN103839051B (en) | 2014-03-05 | 2014-03-05 | The method of single sample hand vein recognition based on 2DPCA and subregion LBP |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103839051A CN103839051A (en) | 2014-06-04 |
| CN103839051B true CN103839051B (en) | 2017-07-21 |
Family
ID=50802530
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410077823.4A Expired - Fee Related CN103839051B (en) | 2014-03-05 | 2014-03-05 | The method of single sample hand vein recognition based on 2DPCA and subregion LBP |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103839051B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105512656A (en) * | 2014-09-22 | 2016-04-20 | 郭进锋 | Palm vein image collection method |
| CN107886090B (en) * | 2017-12-15 | 2021-07-30 | 苏州大学 | Single-sample face recognition method, system, equipment and readable storage medium |
| CN114022914B (en) * | 2021-11-11 | 2023-06-20 | 江苏理工学院 | Palmprint recognition method based on fusion depth network |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254165A (en) * | 2011-08-12 | 2011-11-23 | 北方工业大学 | Hand back vein identification method based on fusion of structural coding features and texture coding features |
-
2014
- 2014-03-05 CN CN201410077823.4A patent/CN103839051B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254165A (en) * | 2011-08-12 | 2011-11-23 | 北方工业大学 | Hand back vein identification method based on fusion of structural coding features and texture coding features |
Non-Patent Citations (5)
| Title |
|---|
| ".多模态手部生物特征识别技术的研究.多模态手部生物特征识别技术的研究";张睿;《中国优秀硕士学位论文全文数据库 信息科技辑》;20080815(第8期);73-77 * |
| "Sampled Two-dimensional LDA for Face Recognition with One Training";Hongtao Yin 等;《Innovative Computing, Information and Control, 2006. ICICIC "06. First International Conference on》;20060901;113-116 * |
| "基于三层虚拟图像生成的单样本人脸识别";张建明 等;《计算机工程》;20100531;第36卷(第9期);187-189 * |
| "基于虚拟图像的单样本人脸识别方法";许孝勇;《计算机工程》;20120131;第38卷(第1期);143-145 * |
| "自适应加权LBP的单样本人脸识别方法";赵汝哲 等;《计算机工程与应用》;20120425;第48卷(第31期);146-149 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103839051A (en) | 2014-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107392082B (en) | Small-area fingerprint comparison method based on deep learning | |
| CN103902978B (en) | Face datection and recognition methods | |
| CN110288018A (en) | A kind of WiFi personal identification method merging deep learning model | |
| Kaucha et al. | Early detection of lung cancer using SVM classifier in biomedical image processing | |
| CN107729820A (en) | A kind of finger vein identification method based on multiple dimensioned HOG | |
| Fronitasari et al. | Palm vein recognition by using modified of local binary pattern (LBP) for extraction feature | |
| CN107844736A (en) | iris locating method and device | |
| CN108764359A (en) | The k means clustering methods of printenv | |
| CN106228118A (en) | The finger vein identification method that a kind of characteristic point and bianry image combine | |
| CN106203497A (en) | A kind of finger vena area-of-interest method for screening images based on image quality evaluation | |
| CN110163182A (en) | A kind of hand back vein identification method based on KAZE feature | |
| CN103839051B (en) | The method of single sample hand vein recognition based on 2DPCA and subregion LBP | |
| Bruno et al. | Palmprint principal lines extraction | |
| CN113112498A (en) | Grape leaf scab identification method based on fine-grained countermeasure generation network | |
| Meng et al. | Multi-classification of breast cancer histology images by using gravitation loss | |
| CN103903017B (en) | A kind of face identification method based on adaptive soft histogram local binary patterns | |
| Raghavendra et al. | An efficient finger vein indexing scheme based on unsupervised clustering | |
| CN105373781A (en) | Binary image processing method for identity authentication | |
| CN110263726A (en) | A kind of finger vein identification method and device based on depth correlation feature learning | |
| Surbiryala et al. | Finger vein indexing based on binary features | |
| CN110136112B (en) | Computer-aided detection algorithm based on mammary gland X-ray photography calcification | |
| CN105512682B (en) | A kind of security level identification recognition methods based on Krawtchouk square and KNN-SMO classifier | |
| Li et al. | Online personal verification by palmvein image through palmprint-like and palmvein information | |
| CN111209850A (en) | Method for generating applicable multi-device identification finger vein image based on improved cGAN network | |
| CN111127407A (en) | Fourier transform-based style migration counterfeit image detection device and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170721 Termination date: 20200305 |