CN110516594A - A kind of guard method and its protective device for referring to vein and can cancelling feature templates - Google Patents
A kind of guard method and its protective device for referring to vein and can cancelling feature templates Download PDFInfo
- Publication number
- CN110516594A CN110516594A CN201910796837.4A CN201910796837A CN110516594A CN 110516594 A CN110516594 A CN 110516594A CN 201910796837 A CN201910796837 A CN 201910796837A CN 110516594 A CN110516594 A CN 110516594A
- Authority
- CN
- China
- Prior art keywords
- vein
- template
- cancel
- feature
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
-
- 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
- G06V40/40—Spoof detection, e.g. liveness detection
- G06V40/45—Detection of the body part being alive
-
- 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
- G06V40/14—Vascular patterns
Abstract
The invention discloses a kind of finger veins can cancel the guard method and its protective device of feature templates, method includes the following steps: will refer to that each neighborhood difference maps in vein image, obtains difference excitation figure；Judgement refers to vein direction, and calculates curvature, obtains geometrical characteristic figure；Difference excitation figure is combined with geometrical characteristic figure, is obtained two dimensional character figure, is obtained the eigenmatrix of both direction；To eigenmatrix dimensionality reduction, feature vector is obtained；Random number vector is generated, random matrix is obtained；Feature vector is unified into eigenmatrix, then by feature vector and random matrix inner product operation, acquisition can cancel template；Template can be cancelled to project to same sub-spaces, establish objective function, generation refers to that vein can cancel template；It will refer to that vein can cancel template matching, and calculate the similarity of test sample and training sample, classify to test sample.The present invention improves recognition performance and revocation/reusability, increases irreversibility, diversity and safety.
Description
Technical field
A kind of guard method more particularly to a kind of finger vein the present invention relates to biometrics identification technology field can be cancelled
The guard method of feature templates, the finger vein for further relating to the guard method can cancel the protective device of feature templates.
Background technique
Living things feature recognition at present is widely used in every field, and adjoint identity information safety problem is more and more prominent
Out, so that biological information protection becomes one of the main direction of studying of current information security.Biological characteristic has uniqueness,
Stability, i.e. everyone biological characteristic are different from, and will not be changed within some time.Uniqueness answers it extensively
For user identity identification；Stability, so that the characteristic information of user is once issued, it, can not even if user characteristics are stolen
Regenerate new biological characteristic.And can cancel template is raw biometric to be mapped to new feature space, then carry out spy
Template can also be recalled and be issued again even if characteristic information is revealed by sign matching.Therefore, biometric templates can be cancelled
It protects most important.
In recent years, template method can be cancelled and be successfully applied to palmmprint, fingerprint, the features such as face.Template protection side can be cancelled
Method is broadly divided into three categories: 1) based on the method for Random Maps, this method will extract the primitive character template of biology, pass through difference
Mapping scheme, primitive character template is projected to new feature space, will not influence the distance between raw biometric, can
Property and discrimination to ensure safety, but the disadvantage is that the transformation data after directly saving Random Maps are deposited as biometric templates
Restore raw information attacking using cross-matched or inverse transformation.2) based on the method for ECG, this method and external bio-identification
Technology is compared, and the inherence of electrocardiogram and dynamic characteristic and its intrinsic life indicate so that their easily stolen or forgeries, this increasing
The difficulty of Replay Attack is added.3) based on the method for biohash, this method generates RP matrix pair using the specific token of user
The raw biometric information of extraction carries out mapping transformation, and the binary data after quantization saves as irreversible biological characteristic mould
Plate ensure that the irreversibility of template to a certain extent.Especially under hidden conditional, can get close to zero etc. mistakes
Rate.But if token loses, recognition performance is substantially reduced.In conclusion these existing recognition performances can preferably cancel life
Object feature templates method cannot be guaranteed safety, and guarantee that the method recognition performance of biometric templates safety is poor.
Summary of the invention
It cannot be guaranteed to identify safety to solve the high method of existing biometric templates method recognition performance, and guarantee
The poor technical problem of the method recognition performance of biometric templates safety, the present invention, which provides a kind of finger vein, can cancel character modules
The guard method and its protective device of plate.
The present invention is implemented with the following technical solutions: a kind of guard method for referring to vein and can cancelling feature templates comprising
Following steps:
(1) the finger vein image progress feature extraction to vein sample is referred to；Wherein, the finger vein sample is test sample
Or training sample；Feature extracting method the following steps are included:
(1.1) each neighborhood difference in the finger vein image is mapped according to deflection, obtains difference excitation figure；
(1.2) according to the vein sample, judgement refers to vein direction, and calculates the curvature for referring to vein direction, with
Obtain two geometrical characteristic figures of double variable curvatures；
(1.3) difference excitation figure is constructed into Joint Distribution two dimensional character with two geometrical characteristic figures respectively, and respectively
Two Joint Distribution two dimensional character figures are obtained, both direction is obtained and obtains eigenmatrix；
(2) according to the eigenmatrix of both direction, generation refers to that vein can cancel template；Wherein, the finger vein can be cancelled
Template the following steps are included:
(2.1) PCA dimensionality reduction is carried out to two eigenmatrixes, obtains feature vector；
(2.2) one group of random number vector is generated by User Token, obtains random matrix；
(2.3) described eigenvector in each direction is first unified into an eigenmatrix, then by described eigenvector
Inner product operation is carried out with the column vector of the random matrix, generates the template on corresponding direction, is obtained desirable in both direction
Disappear template；
(2.4) first the template of cancelling on different directions is projected to same sub-spaces and is merged, then in not Tongfang
Objective function is established in upward cancelling between template, and is generated the finger vein and can be cancelled template；
(3) first the test sample, the training sample are formed by and refer to that vein can cancel template and match, and counted
The similarity for calculating the test sample Yu the training sample determines range further according to the similarity and a similarity, right
The test sample is classified.
The present invention while considering to refer to the directional information and curvature of vein by using the mode of double variable curvatures, can
Preferably matching refers to the line feature in vein image, and the feature vector after dimensionality reduction is combined guarantor with the pseudo random matrix of generation
Card can cancel the irreversible of template, and by merging the template of cancelling of both direction to keep correlation in class
Irreversible simultaneously, solving the high method of existing biometric templates method recognition performance cannot be guaranteed safety, and guarantee
The poor technical problem of the method recognition performance of biometric templates safety, improves the error rates such as discrimination and reduction, and stabilization has
Effect property it is higher, recognition performance is more preferable, and cannot in any manner the token of real user or biological information not
In the case where existing simultaneously, restores raw biometric data, irreversibility can be improved, can be improved the safety of template, together
When also improve and have revocation/reusability and multifarious technical effect.
As a further improvement of the foregoing solution, in step (1.2), the finger vein is judged in Gabor filter
Direction, and the curvature calculation method the following steps are included:
(1.2.1) defines two-dimensional coordinate system；
(1.2.2) calculates the warp factor one in the two-dimensional coordinate system middle finger vein direction:
In formula, (x, y) is the point coordinate on the finger vein direction, and σ is standard deviation；
(1.2.3) calculates the warp factor two in the two-dimensional coordinate system middle finger vein direction:
In formula, u is the frequency of sine wave, and f is the curvature of Gabor filter, i2=-1；θ is the finger vein direction
Angle, and calculation formula are as follows:
In formula, nθThe direction number defined on [0,2 π] for Gabor filter；
(1.2.4) calculates the convolution of the warp factor one with the warp factor two, and described in convolution results are used as
Curvature.
As a further improvement of the foregoing solution, in step (1.2), also Gabor filter is standardized,
And standardization formula are as follows:
In formula, m and n respectively indicate the number for referring to vein direction and curvature,For the bending
The real part of degree.
As a further improvement of the foregoing solution, also quiet to the finger negated by Gabor filter in step (1.2)
Arteries and veins image is filtered, while extracting the finger vein direction and curvature feature；Wherein, Filtering Formula are as follows:
In formula,Indicate convolution as a result,Indicate it is described refer to vein image negate as a result,Indicate volume
Product operation, j indicate the number of Gabor filter.
As a further improvement of the foregoing solution, in step (1.2), also using maximum and time big response as being located at
The characteristic value of point between two Gabor filters, and representation formula are as follows:
In formula, O1θ is worked as in expressionmAnd fnWhen reaching best simultaneously and most matched with the finger vein image, corresponding Gabor filter
The quantized value of wave device；O2θ is worked as in expressionmAnd fnWhen one of them is reached preferably, the quantized value of corresponding Gabor filter.
As a further improvement of the foregoing solution, in step (1.1), grey scale difference between neighborhood territory pixel and center pixel
(xi-xc) be respectively with the component in vertical direction in the horizontal directionWithAnd the calculation formula of component is respectively as follows:
It defines difference and motivates operator are as follows:
Wherein, p is neighborhood territory pixel number, ξ (xc)∈[-π,π]。
As a further improvement of the foregoing solution, in step (2.2), the acquisition methods of the random matrix include following
Step:
(2.2.1) defines following situations:
Good situation:Each user has its rtAnd biological characteristic；rtTo register
Stage and Qualify Phase, the same seed generate a token rt；
The situation of difference:All users use identical rtAnd biological characteristic；
(2.2.2) uses rtGenerate one group of pseudo-random variableWherein, m > n, column
The dimension of vector is identical as the row vector dimension of the eigenmatrix；
(2.2.3) generates m random column variable, and forms the matrix R of random m*n；Wherein, each element normalization point
Cloth is in [- 1,1] section；
(2.2.4) converts orthonormal basis for matrix R by Orthogonal Method, to obtain the random matrix.
As a further improvement of the foregoing solution, the objective function are as follows:
s.t w1 TCXXw1=w2 TCYYw2=1
Wherein, CXX,CYYIt is the auto-covariance matrix of the feature templates X, Y of two different directions respectively, c indicates original sample
Collect classification number, ni indicates every class sample number；w1,w2ForThe feature templates institute of two different directions to be solved is right
The projection matrix answered.
As a further improvement of the foregoing solution, in step (3), pass through normalizated correlation coefficient Size Match score meter
Calculate the similarity；Wherein, the calculation formula of the similarity are as follows:
In formula, A, B are two feature vectors, A=(a1,a2,a3......an), B=(b1,b2,b3......bn)；μA
(μB) be feature vector A (B) mean value, σA(σB) be A (B) standard deviation, l is the length value of A or B, and the value of NCC is in -1 and 1
Between.
The present invention also provides the protective devices that a kind of finger vein can cancel feature templates, using above-mentioned any finger
Vein can cancel the guard method of feature templates comprising:
Characteristic extracting module is used for the finger vein image progress feature extraction to vein sample is referred to；Wherein, the finger is quiet
Arteries and veins sample is test sample or training sample；The characteristic extracting module includes difference exciting unit, double variable curvature filters
Unit extracts feature unit；The difference exciting unit is used for each neighborhood difference in the finger vein image according to deflection
It is mapped, obtains difference excitation figure；Double variable curvature filter cells are used for according to the vein sample, and judgement refers to quiet
Arteries and veins direction, and the curvature for referring to vein direction is calculated, to obtain two geometrical characteristic figures of double variable curvatures；The extraction
Feature unit is used to difference excitation figure constructing Joint Distribution two dimensional character with two geometrical characteristic figures respectively, and obtains respectively
Two Joint Distribution two dimensional character figures are taken, both direction is obtained and obtains eigenmatrix；
Template generation module can be cancelled, be used for the eigenmatrix according to both direction, generation refers to that vein can cancel template；
Wherein, the template generation module of cancelling includes dimensionality reduction unit, random number generation unit, template generation unit, fusion generation
Unit；The dimensionality reduction unit is used to carry out PCA dimensionality reduction to two eigenmatrixes, obtains feature vector；The generating random number list
Member obtains random matrix for generating one group of random number vector by User Token；The template generation unit is used for first will be every
Described eigenvector on a direction is unified into an eigenmatrix, then by the column of described eigenvector and the random matrix to
Amount carries out inner product operation, generates the template on corresponding direction, obtains and cancels template in maximum and time general orientation；The fusion
Generation unit is for first projecting the template of cancelling on different directions to same sub-spaces and merge, then in different directions
On cancel and establish objective function between template, and generate the finger vein and can cancel template；And match cognization unit,
Refer to that vein can cancel template and match for being first formed by the test sample, the training sample, and described in calculating
The similarity of test sample and the training sample determines range further according to the similarity and a similarity, to the survey
Sample is originally classified.
Cancel template protection method compared to existing, finger vein of the invention can cancel the guard method of feature templates
And its protective device has the advantages that
1, this refers to that vein can cancel the guard methods of feature templates and can be improved the recognition performance of template.It is identified by verifying
Performance, the present invention are verified from good situation and bad situation.It is in the discrimination of good situation, the guard method
100%, etc. error rates reach zero, this is identical as the guard method of the prior art.In the situation of difference, the discrimination of the guard method
With etc. error rates be significantly improved compared to existing guard method, this is because guard method of the invention simultaneously consider refer to
The directional information and curvature of vein can preferably match the line feature referred in vein image, therefore have the mistakes such as lower
Accidentally rate, and performance is more stable effectively.
2, this refer to vein can cancel feature templates guard method can increase generation template irreversibility.The present invention
Guard method, by PCA dimensionality reduction, by feature vector dimension drop to can the maximum dimension for keeping similar degree in the class, such line
Property conversion process be equivalent to primitive character matrix guaranteed by projecting to new subspace it is preliminary irreversible.Such
Under the conditions of, even if attacker obtains the token of real user and can cancel template, illegal user is also difficult extensive from low-dimensional matrix
Original eigenmatrix is arrived again.The process that eigenmatrix after dimensionality reduction is combined with random matrix is equivalent to image and added by the present invention
Salt, intrinsic dimensionality is unknown number after the dimensionality reduction of feature vector, and random matrix has multiple vectors, i.e., is only only less than unknown number
The equation of number, find out in this way come solution of equations have it is infinite multiple, therefore, it is difficult to directly by pseudo random number restore protected
The template of shield.The present invention use be based on improved canonical correlation analysis, generate both direction cancel template after, by melting
The feature templates for closing the two directions make fused feature templates further increase irreversibility.
3, this refers to that vein can cancel the guard methods of feature templates and can be improved revocation/reusability.Assuming that real user
Token and template can be cancelled obtained by attacker, system has received the report of template loss, it is easy to cancel original template
And use family token instead again.It is all not due to issuing template again using the random matrix that the User Token after replacement generates every time
With, therefore obtain that template can be cancelled and be also different.And the present invention issues a new template again, every time it is newly generated can
Cancelling template can avoid falling into the region where user's primary template as far as possible, to guarantee the safety for the template newly issued
Property.
4, this refers to that vein can cancel the guard methods of feature templates and can increase the diversity that can cancel template.The present invention makes
Random matrix is generated with User Token, different application programs and different users register real user letter in different times
The token used when breath is all different, so that user is also different by the random matrix that token generates every time, therefore is sent out
The cancel template and finger vein pattern information of cloth are associated, but have certain difference.Therefore, when attacker is really used
Template and token are cancelled in family, can not be used for other applications even if the template of an application program can be obtained, from
And both ensure that the safety of biological template, also increase the diversity that can cancel template.
Detailed description of the invention
Fig. 1 is the flow chart that the finger vein of the embodiment of the present invention 1 can cancel the guard method of feature templates；
Fig. 2 is the frame diagram that the finger vein of the embodiment of the present invention 1 can cancel the guard method of feature templates；
Fig. 3 is the flow chart that the guard method that finger vein shown in FIG. 1 can cancel feature templates carries out feature extraction；
Fig. 4 is that the finger vein of the embodiment of the present invention 2 can cancel very matched distribution situation in the guard method of feature templates
Schematic diagram；
Fig. 5 is that the finger vein of the embodiment of the present invention 2 can cancel false matched distribution situation in the guard method of feature templates
Schematic diagram；
Fig. 6 is that the finger vein of the embodiment of the present invention 2 can cancel a kind of distribution of FAR-FRR in the guard method of feature templates
Situation schematic diagram；
Fig. 7 is that the finger vein of the embodiment of the present invention 2 can cancel the another kind point of FAR-FRR in the guard method of feature templates
Cloth situation schematic diagram.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, not
For limiting the present invention.
Embodiment 1
Fig. 1 and Fig. 2 is please referred to, a kind of guard method for referring to vein and can cancelling feature templates is present embodiments provided,
Template, matching and identification can be cancelled by feature extraction, generation to realize, can be improved the performance and peace for referring to vein pattern template
Quan Xing.Wherein, it includes following three steps (step (1)-that the finger vein of the present embodiment, which can cancel the guard method of feature templates,
(3))。
(1) the finger vein image progress feature extraction to vein sample is referred to.Wherein, refer to that vein sample is test sample or instruction
Practice sample.In the present embodiment, it is cut from original image and refers to that the size of the ROI of vein is 96*150.Wherein, feature extracting method
Include the following steps (step (1.1)-(1.3)).
(1.1) it will refer to that each neighborhood difference is mapped according to deflection in vein image, and obtain difference excitation figure.Original difference
Substantially isotropic Laplacian operator is encouraged in shunt excitation, and grey scale change information is caused not make full use of.And the present embodiment
It is different relative to the position of central pixel point from each neighborhood territory pixel point in order to effectively reflect the grey scale change of image.In
In practical calculating, each neighborhood difference is mapped according to direction angle information, i.e., grey scale difference between neighborhood territory pixel and center pixel
(xi-xc) and horizontal square to angle.Define grey scale difference (x between neighborhood territory pixel and center pixeli-xc) in the horizontal direction
Above and the component in vertical direction is respectivelyWithAnd the calculation formula of component is respectively as follows:
New difference excitation operator is defined as a result, are as follows:
Wherein, p is neighborhood territory pixel number, ξ (xc)∈[-π,π]。
After improving difference excitation operator, refer to that the noise of vein image is relatively fewer, refers to that vein main line is prominent, it is some tiny
Streakline is filtered out.It can not only reasonably reflect the situation of change of regional area pixel grey scale, and expand different finger veins
Between difference, reduce the influence of noise.
(1.2) according to vein sample, judgement refers to vein direction, and calculates the curvature for referring to vein direction, can with acquisition pair
Two geometrical characteristic figures of variable curvature.In the present embodiment, judgement refers to vein direction in Gabor filter, due to traditional
Gabor filter can only extract the direction character of vein, cannot reflect bending degree in a certain direction, therefore the present embodiment
Building variable curvature Gabor filter calculates its bending degree while judgement refers to vein direction.Wherein, the meter of curvature
Calculation method the following steps are included:
(1.2.1) defines two-dimensional coordinate system；
The warp factor one in (1.2.2) calculating two-dimensional coordinate system middle finger vein direction:
In formula, (x, y) is the point coordinate referred on vein direction, and σ is standard deviation；
The warp factor two in (1.2.3) calculating two-dimensional coordinate system middle finger vein direction:
In formula, u is the frequency of sine wave, and f is the curvature of Gabor filter, i2=-1；θ is the angle for referring to vein direction,
And calculation formula are as follows:
In formula, nθThe direction number defined on [0,2 π] for Gabor filter；
(1.2.4) calculates the convolution of warp factor one and warp factor two, and using convolution results as curvature, it may be assumed that
Due to referring to that the gray value of vein image is positive number, and the gray value of vein curve is smaller than the pixel value of surrounding, quiet
There are larger differences for arteries and veins curve and the gray value of background, therefore are also standardized to Gabor filter, and at standardization
Manage formula are as follows:
In formula, m and n respectively indicate the number for referring to vein direction and curvature,For the reality of curvature
Portion.
In variable curvature Gabor filter after standardization other than effective coverage all be 0, filter cannot directly and
Image directly carries out convolution operation.Therefore, the finger vein image negated is filtered by Gabor filter, is extracted simultaneously
Refer to vein direction and curvature feature；Wherein, Filtering Formula are as follows:
In formula,Indicate convolution as a result,Indicate refer to vein image negate as a result,Indicate convolution behaviour
Make, j indicates the number of Gabor filter.
Due to filter limited amount, structure of certain point may be between two filters, can will be maximum and time big
Characteristic value of the response as the point being located between two Gabor filters, and representation formula are as follows:
In formula, O1θ is worked as in expressionmAnd fnWhen reaching best simultaneously and most being matched with finger vein image, corresponding Gabor filter
Quantized value；O2θ is worked as in expressionmAnd fnWhen one of them is reached preferably, the quantized value of corresponding Gabor filter.
(1.3) difference excitation figure is constructed into Joint Distribution two dimensional character with two geometrical characteristic figures respectively, and obtained respectively
Two Joint Distribution two dimensional character figures obtain both direction and obtain eigenmatrix；According to Joint Distribution feature extraction, this joint point
Cloth feature extraction obtains difference excitation respectively with the Joint Distribution of both direction after referring to feature extraction, both direction can be obtained
Eigenmatrix.Each direction character matrix is combined by each sample by the feature row vector that feature extraction obtains.It please join
Fig. 3 is read, for each finger vein sample, obtains the geometrical characteristic figure O of improved difference excitation figure ξ and variable curvature Gabor1、
O2.ξ is constructed into Joint Distribution two dimensional character with each geometrical characteristic figure respectively, as shown in Figure 3.It is mentioned according to Joint Distribution feature
It takes, the eigenmatrix of both direction can be obtained.Each direction character matrix is passed through the feature that feature extraction obtains by each sample
Row vector is combined.
(2) according to the eigenmatrix of both direction, generation refers to that vein can cancel template.Wherein, refer to that vein can cancel template
Include the following steps (step (2.1)-(2.4)).
(2.1) PCA dimensionality reduction is carried out to two eigenmatrixes, obtains feature vector.For each image zooming-out feature to
Amount carries out PCA dimensionality reduction by the way of controlling variance, obtains feature vector F, the feature vector after dimensionality reduction makes the feature to be formed
Matrix is irreversible, tentatively is slightly difficult to restore original characteristic information in this way.
(2.2) one group of random number vector is generated by User Token, obtains random matrix.In the present embodiment, token is used
Generate one group of random number vector, and random matrix acquisition methods the following steps are included:
(2.2.1) defines following situations:
Good situation:Each user has its rtAnd biological characteristic；rtTo register
Stage and Qualify Phase, the same seed generate a token rt；
The situation of difference:All users use identical rtAnd biological characteristic；
(2.2.2) uses rtGenerate one group of pseudo-random variableWherein, m > n, column
The dimension of vector is identical as the row vector dimension of eigenmatrix；
(2.2.3) generates m random column variable, and forms the matrix R of random m*n；Wherein, each element normalization point
Cloth is in [- 1,1] section；
(2.2.4) converts orthonormal basis for matrix R by Orthogonal Method, to obtain random matrix.
In registration phase and Qualify Phase, the same seed generates one by Blum_Blum_Shub bit generator
A token rt,rtIt is stored in USB or E-card.The situation of good situation and difference uses r respectivelytGenerate random matrix R.
(2.3) feature vector in each direction is first unified into an eigenmatrix, then by feature vector and random square
The column vector of battle array carries out inner product operation, generates the template on corresponding direction, obtains and cancels template in both direction.Wherein,
(<f,r1>,<f,r2>,...,<f,rn>),where m>n。
(2.4) first the template of cancelling on different directions is projected to same sub-spaces and is merged, then in not Tongfang
Objective function is established in upward cancelling between template, and is generated and referred to that vein can cancel template.Traditional canonical correlation analysis
It (CCA) is to maximize two different modalities sample set overall relevancies.The present embodiment cancels template for what different directions obtained
Same sub-spaces are projected to, so that new template of cancelling further has irreversibility, are obtained in different directions desirable
Objective function is established between the template that disappears:
s.t w1 TCXXw1=w2 TCYYw2=1
Wherein, CXX,CYYIt is the auto-covariance matrix of the feature templates X, Y of two different directions respectively, c indicates original sample
Collect classification number, ni indicates every class sample number；w1,w2ForThe feature templates institute of two different directions to be solved is right
The projection matrix answered can be calculated by lagrange multiplier approach, finally beIt is middle maximum special
The corresponding feature vector solution procedure of value indicative.
The feature templates of different directions are projected to same sub-spaces by addition classification information by the above method, so that
Feature templates and primary template have irreversibility after fusion, that is, generate and refer to that vein can cancel template code.
(3) first test sample, training sample are formed by and refer to that vein can cancel template and match, and calculate test specimens
The similarity of this and training sample, determines range further according to similarity and a similarity, classifies to test sample.At this
In embodiment, the similarity between vein pattern is calculated by normalizated correlation coefficient Size Match score.Wherein, similarity
Calculation formula are as follows:
A, B are two feature vectors, A=(a1,a2,a3......an), B=(b1,b2,b3......bn)。μA(μB) it is special
Levy the mean value of vector A (B), σA(σB) be A (B) standard deviation, l is the length value of A or B, and the value of NCC is between -1 and 1.In
In the present embodiment, if NCC is close to 1, it means that two finger vein images may be identical；Otherwise, the two images have
It may be different.
In conclusion the guard method that the finger vein of the present embodiment can cancel feature templates has the advantage that
1, this refers to that vein can cancel the guard methods of feature templates and can be improved the recognition performance of template.It is identified by verifying
Performance, the present embodiment are verified from good situation and bad situation.It is in the discrimination of good situation, the guard method
100%, etc. error rates reach zero, this is identical as the guard method of the prior art.In the situation of difference, the discrimination of the guard method
With etc. error rates be significantly improved compared to existing guard method, this is because the guard method of the present embodiment considers simultaneously
The directional information and curvature for referring to vein can preferably match the line feature referred in vein image, therefore have lower etc.
Error rate, and performance is more stable effectively.
2, this refer to vein can cancel feature templates guard method can increase generation template irreversibility.This implementation
Feature vector dimension is dropped to the maximum dimension for keeping similar degree in the class of energy by PCA dimensionality reduction by the guard method of example, such
Linear transform process be equivalent to primitive character matrix guaranteed by projecting to new subspace it is preliminary irreversible.In this way
Under conditions of, even if attacker obtains the token of real user and can cancel template, illegal user is also difficult from low-dimensional matrix
It is restored to original eigenmatrix.The process that eigenmatrix after dimensionality reduction is combined with random matrix is equivalent to figure by the present embodiment
As adding salt, intrinsic dimensionality is unknown number after the dimensionality reduction of feature vector, and random matrix has multiple vectors, i.e., only only less than not
Know the equation of several numbers, find out in this way come solution of equations have it is infinite multiple, therefore, it is difficult to directly by pseudo random number recovery
Shielded template.The present embodiment use be based on improved canonical correlation analysis, generate both direction cancel template after,
Feature templates by merging the two directions make fused feature templates further increase irreversibility.
3, this refers to that vein can cancel the guard methods of feature templates and can be improved revocation/reusability.Assuming that real user
Token and template can be cancelled obtained by attacker, system has received the report of template loss, it is easy to cancel original template
And use family token instead again.It is all not due to issuing template again using the random matrix that the User Token after replacement generates every time
With, therefore obtain that template can be cancelled and be also different.And the present embodiment issues a new template again, it is newly generated every time
Template can be cancelled can avoid falling into the region where user's primary template as far as possible, to guarantee the safety for the template newly issued
Property.
4, this refers to that vein can cancel the guard methods of feature templates and can increase the diversity that can cancel template.The present embodiment
User Token has been used to generate random matrix, different application programs and different users register real user in different times
The token used when information is all different, so that user is also different by the random matrix that token generates every time, therefore
The cancel template and finger vein pattern information of publication are associated, but have certain difference.Therefore, when attacker obtains really
User's cancels template and token, can not be used for other applications even if the template of an application program can be obtained,
To both ensure that the safety of biological template, the diversity that can cancel template is also increased.
Embodiment 2
A kind of guard method for referring to vein and can cancelling feature templates is present embodiments provided, on the basis of embodiment 1
Carry out experimental evaluation.Recognition performance in order to verify the guard method is more effective and can cancel tightened up irreversible, the sheet of template
Embodiment passes through in four recognition performance, diversity, revocation/reusability, irreversibility aspect explanations.
(1) recognition performance
Refer to that vein can cancel the safety of template, cannot using sacrifice discrimination and etc. error rates as cost.Verify performance point
The situation of preferably the case where and difference.Good situation refers to that same user is passed through in registration phase and Qualify Phase by the same seed
Random bit generator generates token.There is different token in different user, different application.Difference situation refer to, all users
There is identical token.In the case where good, discrimination 100%, etc. error rates reach zero, all improvement relevant with biohash
The discrimination of method with wait error rates all consistent with the present embodiment, therefore good situation is not analyzed.In the case of difference, this
Embodiment propose method discrimination and etc. error rates it is more preferable compared to other methods.Experimental result is as shown in table 1, table 2.
Table 1 compares EER (%) in the experimental data table of two databases respectively
Table 2 compares EER (%) in the experimental data table of two databases respectively
| Method | PolyU | SDUMLA-FV |
| PG-Gabor | 0.45 | 1.35 |
| PG-ASAVE | 0.43 | 1.1 |
| RD | -- | 1.10 |
| S2DPHC | 16.52 | 10.31 |
| Proposed method | 0.1042 | 0.3747 |
Analytical table 1, table 2.Table 1 is respectively modified to verify the validity for the feature extracting method that the present embodiment is proposed
Feature extracting method obtains recognition result.Because LDN, HOG, LDP method mainly extract the marginal information and gradient letter of image
Breath, it is so prominent for referring to the marginal information in vein image not, therefore performance is restricted.WLBP is more sensitive to illumination, influences
Recognition effect.LWLD method has carried out linear filtering to finger vein image, and constructs line weber feature histogram on its basis,
Effectively improve recognition performance.The error rates such as LTP method are relatively high, since this method only considered the gray value of pixel,
Illumination effect can not effectively be overcome.The method of the present embodiment considers the direction for referring to vein and curvature, can be better
With the line feature referred in vein image, good effect is reached.Table 2, first three groups data source is in can cancel Fusion Features side
Cancelling for method refers to vein result.It is main relatively after characteristic processing and irreversible conversion, method performance comparison.Wherein,
S2DPHC cannot portray the direction for referring to vein because extracting the fusion of four direction feature merely through Gabor filter well
Feature, and template is converted by random matrix, irreversibility is weaker, thus EER higher.The method of the present embodiment has lower etc.
Error rate is more stable compared to other methods effective.
Similar to refer to that Intervenous characteristic matching is known as true matching, difference refers to that Intervenous characteristic matching is known as false matching.It is logical
Cross the distribution situation of true and false matching score, can qualitative balancing method performance quality.Fig. 4 and Fig. 5 is please referred to, which is situated between
The Genuine-impostor match distribution situation of template can be cancelled by having continued.Because there is translation in the finger vein in database,
Deformation, the problems such as rotating and is fuzzy, although very matching score and the false distribution for matching score is independent.But there are lesser intersections
Region, overall performance are preferable.
Fig. 6 and Fig. 7 is please referred to, between FAR expression class in test, the number of mistake receiving.FRR is indicated in intra-class testing
False Rejects number.Wherein, as FAR=FRR for etc. error rates (EER), EER is smaller, and representation method performance is more stable.From this
As can be seen that the guard method of the present embodiment has reached the error rates such as lower and had good stability in two figures.
(2) irreversibility
Cancel template by what the guard method generated, cannot in any manner in the token of real user or
In the presence of when biological information difference, restore raw biometric data.
For the eigenmatrix that the mm dimensional feature vector of feature extraction is formed, 1) pass through PCA dimensionality reduction, by feature vector dimension
The m dimension that maximum can keep similar degree in the class is dropped to, characteristic dimension mm > > m, such linear transform process is equivalent to will be original
Eigenmatrix is guaranteed preliminary irreversible by projecting to new subspace.In such a situa-tion, even if attacker obtains
The token of real user and template can be cancelled, illegal user is also difficult to be restored to original eigenmatrix from low-dimensional matrix.2)
The process that eigenmatrix after dimensionality reduction is combined with random matrix is equivalent to image and adds salt, feature dimensions after the dimensionality reduction of feature vector
Several is m, i.e. m number is unknown, and random matrix has n vector to only have n equation, and n < m, is found out in this way
Solution of equations has infinite multiple.Therefore, it is difficult to directly restore shielded template by PRN.3) using based on improved allusion quotation
Type correlation analysis, generate both direction cancel template after, the feature templates by merging the two directions to merge
Feature templates afterwards increase irreversibility.
(3) revocation/reusability
Assuming that the token of real user and can cancel template and obtained by attacker, system has received the report of template loss,
It is easy to cancel original template and uses family token instead again.Since publication template uses the User Token after replacement again every time
The random matrix of generation is all different, therefore is obtained that template can be cancelled and be also different.System is sent out again through this embodiment
One new template of cloth, each newly generated template of cancelling can avoid falling into the area where user's primary template as far as possible
Domain, to guarantee the safety for the template newly issued.
(4) diversity
The present embodiment has used User Token token to generate random matrix, and the different user of different application programs is not
With time register real user information when the token that uses all be different so that user every time by token generate with
Machine matrix is also different, therefore the cancel template and finger vein pattern information issued are associated, but has certain difference.
Therefore when what attacker obtained real user cancels template and token, even if the template that can obtain an application program
It cannot be used for other applications, thus both ensure that the safety of biological template also illustrated the diversity that can cancel template.
In addition, feature extraction uses double variable curvature gabor filters, which considers directional information and bending degree very simultaneously
Good reflects texture structure, to rotation and translation more robust.Refer to the experiment on intravenous data library in Poly_U and SDUMLA-FV
The result shows that the present embodiment has better performance compared to other methods in two databases.
Embodiment 3
The protective devices of feature templates can be cancelled by present embodiments providing a kind of finger vein, the device using embodiment 1 or
Finger vein in embodiment 2 can cancel the guard method of feature templates.Wherein, the finger vein in the present embodiment can cancel character modules
The protective device of plate includes characteristic extracting module, can cancel template generation module and match cognization unit.
Characteristic extracting module is used for the finger vein image progress feature extraction to vein sample is referred to.Wherein, refer to vein sample
For test sample or training sample, characteristic extracting module includes difference exciting unit, double variable curvature filter cells, extracts spy
Levy unit.Difference exciting unit obtains difference and swashs for that will refer to that each neighborhood difference is mapped according to deflection in vein image
Encourage figure.Double variable curvature filter cells are used for according to vein sample, and judgement refers to vein direction, and calculates and refer to the curved of vein direction
Curvature, to obtain two geometrical characteristic figures of double variable curvatures.Extract feature unit be used for by difference excitation figure respectively with two
Geometrical characteristic figure constructs Joint Distribution two dimensional character, and obtains two Joint Distribution two dimensional character figures respectively, obtains both direction
Obtain eigenmatrix.
Template generation module can be cancelled for the eigenmatrix according to both direction, generation refers to that vein can cancel template.Its
In, can cancel template generation module includes dimensionality reduction unit, random number generation unit, template generation unit, fusion generation unit.Drop
It ties up unit to be used to carry out PCA dimensionality reduction to two eigenmatrixes, obtains feature vector.Random number generation unit is used to enable by user
Board generates one group of random number vector, obtains random matrix.Template generation unit is for first joining feature vector in each direction
An eigenmatrix is synthesized, then feature vector and the column vector of random matrix are subjected to inner product operation, is generated on corresponding direction
Template obtains and cancels template in maximum and time general orientation.Generation unit is merged to be used for first by cancelling on different directions
Template is projected to same sub-spaces and is merged, then cancelling in different directions establishes objective function between template, and
Generation refers to that vein can cancel template.
Match cognization unit refers to that vein can cancel template progress for being first formed by test sample, training sample
Match, and calculate the similarity of test sample and training sample, range is determined further according to similarity and a similarity, to test specimens
This is classified.
Embodiment 4
The present embodiment provides a kind of terminals comprising memory, processor and storage are on a memory and can
The computer program run on a processor.Realize that the finger vein of embodiment 1 can cancel feature templates when processor executes program
Guard method the step of.
The method of embodiment 1 is such as designed to independently operated program in use, can be applied in the form of software,
On computer terminals, terminal can be computer, smart phone, control system and other internet of things equipment for installation
Deng.The method of embodiment 1 can also be designed to the program of embedded operation, and installation on computer terminals, is such as mounted on monolithic
On machine.
Embodiment 5
The present embodiment provides a kind of computer readable storage mediums, are stored thereon with computer program.Program is by processor
When execution, finger vein the step of can cancelling the guard method of feature templates of embodiment 1 is realized.
The method of embodiment 1 is such as designed to computer-readable storage medium in use, can be applied in the form of software
Matter can independently operated program, computer readable storage medium can be USB flash disk, is designed to U-shield, be designed to by USB flash disk by outer
Start the program of entire method in triggering.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (10)
1. a kind of refer to that vein can cancel the guard methods of feature templates, which is characterized in that itself the following steps are included:
(1) the finger vein image progress feature extraction to vein sample is referred to；Wherein, the finger vein sample is test sample or instruction
Practice sample；Feature extracting method the following steps are included:
(1.1) each neighborhood difference in the finger vein image is mapped according to deflection, obtains difference excitation figure；
(1.2) according to the vein sample, judgement refers to vein direction, and calculates the curvature for referring to vein direction, to obtain
Two geometrical characteristic figures of double variable curvatures；
(1.3) difference excitation figure is constructed into Joint Distribution two dimensional character with two geometrical characteristic figures respectively, and obtained respectively
Two Joint Distribution two dimensional character figures obtain both direction and obtain eigenmatrix；
(2) according to the eigenmatrix of both direction, generation refers to that vein can cancel template；Wherein, the finger vein can cancel template
The following steps are included:
(2.1) PCA dimensionality reduction is carried out to two eigenmatrixes, obtains feature vector；
(2.2) one group of random number vector is generated by User Token, obtains random matrix；
(2.3) described eigenvector in each direction is first unified into an eigenmatrix, then by described eigenvector and institute
The column vector for stating random matrix carries out inner product operation, generates the template on corresponding direction, obtains and cancels mould in both direction
Plate；
(2.4) first the template of cancelling on different directions is projected to same sub-spaces and is merged, then in different directions
Cancel and establish objective function between template, and generate the finger vein and can cancel template；
(3) first the test sample, the training sample are formed by and refer to that vein can cancel template and match, and calculate institute
The similarity for stating test sample Yu the training sample determines range further according to the similarity and a similarity, to described
Test sample is classified.
2. referring to that vein can cancel the guard method of feature templates as described in claim 1, which is characterized in that in step (1.2)
In, judge the finger vein direction in Gabor filter, and the curvature calculation method the following steps are included:
(1.2.1) defines two-dimensional coordinate system；
(1.2.2) calculates the warp factor one in the two-dimensional coordinate system middle finger vein direction:
In formula, (x, y) is the point coordinate on the finger vein direction, and σ is standard deviation；
(1.2.3) calculates the warp factor two in the two-dimensional coordinate system middle finger vein direction:
In formula, u is the frequency of sine wave, and f is the curvature of Gabor filter, i2=-1；θ is the angle for referring to vein direction,
And calculation formula are as follows:
In formula, nθThe direction number defined on [0,2 π] for Gabor filter；
(1.2.4) calculates the convolution of the warp factor one and the warp factor two, and using convolution results as the bending
Degree.
3. referring to that vein can cancel the guard method of feature templates as claimed in claim 2, which is characterized in that in step (1.2)
In, also Gabor filter is standardized, and standardization formula are as follows:
In formula, m and n respectively indicate the number for referring to vein direction and curvature,For the curvature
Real part.
4. referring to that vein can cancel the guard method of feature templates as claimed in claim 3, which is characterized in that in step (1.2)
In, also the finger vein image negated is filtered by Gabor filter, while extracting the finger vein direction and curvature
Feature；Wherein, Filtering Formula are as follows:
In formula,Indicate convolution as a result,Indicate it is described refer to vein image negate as a result,Indicate convolution behaviour
Make, j indicates the number of Gabor filter.
5. referring to that vein can cancel the guard method of feature templates as claimed in claim 2, which is characterized in that in step (1.2)
In, maximum and secondary big response is also used as to the characteristic value of the point between two Gabor filters, and representation formula
Are as follows:
In formula, O1θ is worked as in expressionmAnd fnWhen reaching best simultaneously and most matched with the finger vein image, corresponding Gabor filter
Quantized value；O2θ is worked as in expressionmAnd fnWhen one of them is reached preferably, the quantized value of corresponding Gabor filter.
6. referring to that vein can cancel the guard method of feature templates as described in claim 1, which is characterized in that in step (1.1)
In, grey scale difference (x between neighborhood territory pixel and center pixeli-xc) be respectively with the component in vertical direction in the horizontal direction
WithAnd the calculation formula of component is respectively as follows:
It defines difference and motivates operator are as follows:
Wherein, p is neighborhood territory pixel number, ξ (xc)∈[-π,π]。
7. referring to that vein can cancel the guard method of feature templates as described in claim 1, which is characterized in that in step (2.2)
In, the acquisition methods of the random matrix the following steps are included:
(2.2.1) defines following situations:
Good situation:Each user has its rtAnd biological characteristic；rtFor in registration phase and
Qualify Phase, the same seed generate a token rt；
The situation of difference:All users use identical rtAnd biological characteristic；
(2.2.2) uses rtGenerate one group of pseudo-random variableWherein, m > n, column vector
Dimension is identical as the row vector dimension of the eigenmatrix；
(2.2.3) generates m random column variable, and forms the matrix R of random m*n；Wherein, each element normalization is distributed in
[- 1,1] section；
(2.2.4) converts orthonormal basis for matrix R by Orthogonal Method, to obtain the random matrix.
8. referring to that vein can cancel the guard method of feature templates as described in claim 1, which is characterized in that the objective function
Are as follows:
s.t w1 TCXXw1=w2 TCYYw2=1
Wherein, CXX,CYYIt is the auto-covariance matrix of the feature templates X, Y of two different directions respectively, c indicates original sample collection classification
Number, ni indicate every class sample number；w1,w2ForThrowing corresponding to the feature templates of two different directions to be solved
Shadow matrix.
9. referring to that vein can cancel the guard method of feature templates as described in claim 1, which is characterized in that in step (3),
The similarity is calculated by normalizated correlation coefficient Size Match score；Wherein, the calculation formula of the similarity are as follows:
In formula, A, B are two feature vectors, A=(a1,a2,a3......an), B=(b1,b2,b3......bn)；μA(μB) it is special
Levy the mean value of vector A (B), σA(σB) be A (B) standard deviation, l is the length value of A or B, and the value of NCC is between -1 and 1.
10. a kind of protective device for referring to vein and can cancelling feature templates, application is as described in any one of claim 1-9
Refer to that vein can cancel the guard method of feature templates, characterized in that it comprises:
Characteristic extracting module is used for the finger vein image progress feature extraction to vein sample is referred to；Wherein, the finger vein sample
This is test sample or training sample；The characteristic extracting module include difference exciting unit, double variable curvature filter cells,
Extract feature unit；The difference exciting unit is for reflecting each neighborhood difference in the finger vein image according to deflection
It penetrates, obtains difference excitation figure；Double variable curvature filter cells are used for according to the vein sample, and judgement refers to vein side
To, and the curvature for referring to vein direction is calculated, to obtain two geometrical characteristic figures of double variable curvatures；The extraction feature
Unit is used to difference excitation figure constructing Joint Distribution two dimensional character with two geometrical characteristic figures respectively, and obtains two respectively
A Joint Distribution two dimensional character figure obtains both direction and obtains eigenmatrix；
Template generation module can be cancelled, be used for the eigenmatrix according to both direction, generation refers to that vein can cancel template；Its
In, the template generation module of cancelling includes dimensionality reduction unit, random number generation unit, template generation unit, fusion generation list
Member；The dimensionality reduction unit is used to carry out PCA dimensionality reduction to two eigenmatrixes, obtains feature vector；The random number generation unit
For generating one group of random number vector by User Token, random matrix is obtained；The template generation unit is used for first will be each
Described eigenvector on direction is unified into an eigenmatrix, then by the column vector of described eigenvector and the random matrix
Inner product operation is carried out, the template on corresponding direction is generated, obtains and cancels template in maximum and time general orientation；The fusion life
At unit for first the template of cancelling on different directions to be projected to same sub-spaces and merged, then in different directions
Cancel and establish objective function between template, and generate the finger vein and can cancel template；And
Match cognization unit, template can be cancelled by being used to that the test sample, the training sample to be first formed by finger vein
It is matched, and calculates the similarity of the test sample Yu the training sample, it is similar further according to the similarity and one
It spends and determines range, classify to the test sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910796837.4A CN110516594B (en) | 2019-08-27 | 2019-08-27 | Protection method and protection device for finger vein feature template capable of being cancelled |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910796837.4A CN110516594B (en) | 2019-08-27 | 2019-08-27 | Protection method and protection device for finger vein feature template capable of being cancelled |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110516594A true CN110516594A (en) | 2019-11-29 |
| CN110516594B CN110516594B (en) | 2022-03-18 |
Family
ID=68627217
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910796837.4A Active CN110516594B (en) | 2019-08-27 | 2019-08-27 | Protection method and protection device for finger vein feature template capable of being cancelled |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110516594B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112651304A (en) * | 2020-12-11 | 2021-04-13 | 西安电子科技大学 | Revocable palm print template generation method, device, equipment and storage medium based on feature fusion |
| CN113033700A (en) * | 2021-04-19 | 2021-06-25 | 安徽大学 | Finger vein feature template protection method and device |
| CN113269020A (en) * | 2021-03-08 | 2021-08-17 | 重庆邮电大学 | Fingertip vein image feature identification positioning method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080178002A1 (en) * | 2006-10-13 | 2008-07-24 | Shinji Hirata | System, Server, Terminal and Tamper Resistant Device for Authenticating a User |
| US20130088327A1 (en) * | 2011-10-06 | 2013-04-11 | Hitachi, Ltd. | Template delivery type cancelable biometric authentication system and method therefor |
| CN104268533A (en) * | 2014-10-10 | 2015-01-07 | 南昌航空大学 | Non-contact revocable palm print bimodal authentication method |
| CN106446867A (en) * | 2016-10-13 | 2017-02-22 | 济南大学 | Double-factor palmprint identification method based on random projection encryption |
| CN108197577A (en) * | 2018-01-08 | 2018-06-22 | 安徽大学 | The finger venous image characteristic extraction method of joint Sobel and MFRAT |
| CN108429614A (en) * | 2018-01-05 | 2018-08-21 | 杭州电子科技大学 | A kind of fuzzy vault implementation method merged with face characteristic grade based on fingerprint |
| CN109886202A (en) * | 2019-02-22 | 2019-06-14 | 济南大学 | Voidable palmmprint contention code characteristic recognition method based on IoM |
-
2019
- 2019-08-27 CN CN201910796837.4A patent/CN110516594B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080178002A1 (en) * | 2006-10-13 | 2008-07-24 | Shinji Hirata | System, Server, Terminal and Tamper Resistant Device for Authenticating a User |
| US20130088327A1 (en) * | 2011-10-06 | 2013-04-11 | Hitachi, Ltd. | Template delivery type cancelable biometric authentication system and method therefor |
| CN104268533A (en) * | 2014-10-10 | 2015-01-07 | 南昌航空大学 | Non-contact revocable palm print bimodal authentication method |
| CN106446867A (en) * | 2016-10-13 | 2017-02-22 | 济南大学 | Double-factor palmprint identification method based on random projection encryption |
| CN108429614A (en) * | 2018-01-05 | 2018-08-21 | 杭州电子科技大学 | A kind of fuzzy vault implementation method merged with face characteristic grade based on fingerprint |
| CN108197577A (en) * | 2018-01-08 | 2018-06-22 | 安徽大学 | The finger venous image characteristic extraction method of joint Sobel and MFRAT |
| CN109886202A (en) * | 2019-02-22 | 2019-06-14 | 济南大学 | Voidable palmmprint contention code characteristic recognition method based on IoM |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112651304A (en) * | 2020-12-11 | 2021-04-13 | 西安电子科技大学 | Revocable palm print template generation method, device, equipment and storage medium based on feature fusion |
| CN113269020A (en) * | 2021-03-08 | 2021-08-17 | 重庆邮电大学 | Fingertip vein image feature identification positioning method and system |
| CN113033700A (en) * | 2021-04-19 | 2021-06-25 | 安徽大学 | Finger vein feature template protection method and device |
| CN113033700B (en) * | 2021-04-19 | 2022-11-08 | 安徽大学 | Finger vein feature template protection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110516594B (en) | 2022-03-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Discriminative and robust competitive code for palmprint recognition | |
| Wu et al. | A secure palm vein recognition system | |
| EP2883190B1 (en) | Texture features for biometric authentication | |
| Tome et al. | The 1st competition on counter measures to finger vein spoofing attacks | |
| Barpanda et al. | Iris recognition with tunable filter bank based feature | |
| Jaswal et al. | DeepKnuckle: revealing the human identity | |
| CN105138972A (en) | Face authentication method and device | |
| CN110516594A (en) | A kind of guard method and its protective device for referring to vein and can cancelling feature templates | |
| Alheeti | Biometric iris recognition based on hybrid technique | |
| Zhang et al. | 3D Biometrics | |
| Labati et al. | Touchless fingerprint biometrics | |
| Laadjel et al. | Combining Fisher locality preserving projections and passband DCT for efficient palmprint recognition | |
| Srivastava | Personal identification using iris recognition system, A Review | |
| CN105678150A (en) | User authority managing method | |
| CN105701462A (en) | Identity identification method | |
| CN205318544U (en) | Device and system are prevented cheaing by ATM based on three dimensional face identification | |
| Chen et al. | Iris recognition using 3D co-occurrence matrix | |
| Rahulkar et al. | Fast discrete curvelet transform based anisotropic iris coding and recognition using k-out-of-n: A fused post-classifier | |
| Kumar et al. | Palmprint Recognition in Eigen-space | |
| Hegde et al. | Human authentication using finger knuckle print | |
| Ahmad et al. | Palmprint recognition using local and global features | |
| CN105701411A (en) | Information secure transmission method | |
| Beukes | Hand vein-based biometric authentication with limited training samples | |
| Czajka et al. | Iris recognition system based on Zak-Gabor wavelet packets | |
| Sushma et al. | Multi Biometric Template Protection using Random Projection and Adaptive Bloom Filter |
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 |