CN106971157A - Fingerprint and face coupled identification method based on multiple linear regression associative memory model - Google Patents

Abstract

The invention discloses a kind of fingerprint based on multiple linear regression associative memory model and face coupled identification method, comprise the following steps：S1：Gather fingerprint picture and face picture；S2：Respectively obtain fingerprint picture and the associative memory input matrix and output matrix of face picture；S3：Build multiple linear regression fingerprint picture identification model and multiple linear regression face picture identification model with regression parameter；S4：Regression parameter is calculated, multiple linear regression fingerprint picture identification model, multiple linear regression face picture identification model is obtained；S5：Fingerprint picture and face picture are identified.Beneficial effect：Identity information realizes Multiple recognition, and reliability is high, and associative memory and multiple linear regression model are combined, and picture is changed into parameter, and safety coefficient is high, and recognition effect is good, good to identity information protecting effect, and crypticity is high.

Description

Fingerprint and face coupled identification method based on multiple linear regression associative memory model

Technical field

It is specifically a kind of to be based on multiple linear regression associative memory the present invention relates to view data Techniques of preserving field The fingerprint and face coupled identification method of model.

Background technology

With the development in big data epoch, people generally preserve daily photo even identity document photo to database In, personal information is easily so stolen by hacker, to be peddled or carried out illegal activity, the private letter of such people Breath is compromised, and daily life is easily disturbed or even is involved among crime dramas, easily causes inconvenience.

Need to carry out the mechanism that identity information is verified at some, it will usually which the identity information for gathering people carries out identity Checking.The face information or finger print information of such as collection people.Information content is few, and identification checkup confidence level is not high.And right It is typically directly to preserve when the identity information of people is preserved, not to carrying out processing preservation to photo, safety coefficient is low, body Part information is easily stolen takes.When identity information database once meeting with the attack of criminal, identity information is extremely easily stolen Carry out illegal activity is fetched, gives people to cause unnecessary trouble.

The content of the invention

In view of the above-mentioned problems, the invention provides a kind of fingerprint and face based on multiple linear regression associative memory model Coupled identification method, combining association memory and multiple linear regression model, a series of ginsengs are converted into by face picture, fingerprint picture Number is preserved, and realizes that dual identity is verified, safe, good reliability,.

To reach above-mentioned purpose, the concrete technical scheme that the present invention is used is as follows：

A kind of fingerprint and face coupled identification method based on multiple linear regression associative memory model, its key are to wrap Include following steps：

S1：The fingerprint picture and face picture of collection crowd, is grouped to the fingerprint picture and face picture collected Numbering；

S2：By setting binary map luminance threshold, the obtained all fingerprint pictures of step S1 and face picture are processed as Two-value picture, obtains the associative memory input matrix and output matrix of fingerprint picture and the associative memory input matrix of face picture And output matrix；

S3：Using cell neural network structure, two multiple linear regression picture recognition moulds for carrying regression parameter are built Type, respectively multiple linear regression fingerprint picture identification model and multiple linear regression face picture identification model；

S4：According to the obtained fingerprint pictures of step S2 and associative memory input matrix, output matrix and the step of face picture Two multiple linear regression picture recognition models that rapid S3 is obtained, the regression parameter in calculation procedure S3, final true respectively Determine multiple linear regression fingerprint picture identification model, multiple linear regression face picture identification model；

S5：Based on autoassociative memories criterion, fingerprint picture and face picture are identified respectively.

Further, when carrying out authentication, it is necessary to which the fingerprint picture and face picture that obtain the user simultaneously are carried out Dual verification.

When having, one of picture verification is problematic, then verification failure.Dual verification is realized, it is with a high credibility, improve body The reliability that part information is preserved.

Further, the binary map luminance threshold K=(0,1,2,3...255)；

Include m group pictures in step sl, corresponding numbering is 1,2,3...m, and all fingerprint pictures and face picture are equal Individual pixel is arranged by N rows M to constitute, pixel number is n=N × M；

If the binary map matrix of fingerprint is input matrix Γ '=(X ' of associative memory¹,X′²,…,X′ⁱ,…,X′^m), i ∈ { 1,2 ..., m }, represents the input vector of pixel composition all in the binary map of the i-th width fingerprint, wherein, Represent the two of the i-th width fingerprint It is worth the input value of j-th of pixel in figure；

If the binary map matrix of fingerprint is the output matrix of associative memory The output vector of pixel composition all in the binary map of the i-th width fingerprint is represented, wherein, Represent the two-value in the i-th width fingerprint The output valve of j-th of pixel in figure；

If the binary map matrix of face is the input matrix Γ "=(X " of associative memory¹,X″²,…,X″ⁱ,…,X″^m), i ∈ { 1,2 ..., m }, represents the input vector of pixel composition all in the binary map of the i-th width face, wherein, Represent in the i-th width face The input value of j-th of pixel in binary map；

If the binary map matrix of face is the output matrix of associative memory The output vector of pixel composition all in the binary map of the i-th width face is represented, wherein, Represent the two of the i-th width face It is worth the output valve of j-th of pixel in figure.

Further describe, step S3 particular content is：：

The multiple linear regression fingerprint picture identification model built based on cell neural network structure, be specially：

Wherein, i=1,2 ..., N；J=1,2 ..., M；k₁(i, r)=max { 1-i ,-r }；k₂(i, r)=min { N-i, r }； l₁(j, r)=max { 1-j ,-r }；l₂(i, r)=min { M-i, r }；R is to represent template radius,For input parameter,It is defeated Go out,It is input,It is offset,It is input template；

Input templateExpression formula it is as follows：

It is 1 to make r, then in equation (2)

Formula (1) is rewritten as：

Y '=A ' X '+V ' (3)

Wherein, input vector：

Output vector

Offset

Dot-blur pattern A ' and offset V ' is the regression parameter, dot-blur pattern A '=(a '_ij)_n×nShape can be written as Formula：

Wherein,

The multiple linear regression face picture identification model built based on cell neural network structure, be specially：

Wherein, i=1,2 ..., N；J=1,2 ..., M；k₁(i, r)=max { 1-i ,-r }；k₂(i, r)=min { N-i, r }； l₁(j, r)=max { 1-j ,-r }；l₂(i, r)=min { M-i, r }；R is to represent template radius,For input parameter,It is defeated Go out,It is input,It is offset,It is input template；

Input templateExpression formula it is as follows：

It is 1 to make r, then in equation (7)

Formula (6) is rewritten as：

Y "=A " X "+V " (8)

Wherein, input vector：

Output vector

Offset

Dot-blur pattern A " and offset V " is the regression parameter, dot-blur pattern A "=(a "_ij)_n×nShape can be written as Formula：

Wherein,

Further describe, unknown regression parameter in step S4 in calculation procedure S3 is concretely comprised the following steps：S41：Make to Amount

Make Y'=((Y "¹)^T,(Y″²)^T,…,(Y″ⁱ)^T,…,(Y″^m)^T)^T；Y "=((Y "¹)^T,(Y″²)^T,…,(Y″ⁱ )^T,…,(Y″^m)^T)^T；

Wherein Y 'ⁱWith Y "ⁱThe row of pixel composition all in the i-th width fingerprint binary map and face binary map are represented respectively Vector；

L ∈ { 1,2 ..., m } and q ∈ { 1,2 ..., N } are made,

It can then be obtained by formula (3)：

X ' L '=Y ' (11)

Formula (8) can be obtained：

X " L "=Y " (12)

Then

Wherein, L ' and L " is constant.

S42：In autoassociative memories criterion, exist：

By the fingerprint picture associative memory input matrix Γ '=(X ' obtained in step S2¹,X′²,…,X′ⁱ,…,X′^m) and Output matrixBring into formula (11), the fingerprint picture association obtained in step S2 is remembered Recall input matrix Γ "=(X "¹,X″²,…,X″ⁱ,…,X″^m) and output matrixBring into In formula (12), and Γ ' is converted into X ', willY' is converted into, Γ " is converted into X ", willY " is converted into, X ' L are obtained₁ =Y' and X " L₂=Y ", draws L '=L₁=pinv (X ') Y' and L "=L₂=pinv (X ") Y "；

S43：The skew v ' that step S42 is obtained_iAnd input templateBring formula (4), (5) into, obtain dot-blur pattern A ' and Offset V ', draws multiple linear regression fingerprint picture identification model；

The skew v " that step S42 is obtained_iAnd input templateBring formula (9), (10) into, respectively, obtain dot-blur pattern A " With offset V ", multiple linear regression face picture identification model is obtained.

Further describe, to concretely comprising the following steps that fingerprint picture and face picture are identified in step S5：

S51：Fingerprint picture and face picture are obtained, fingerprint picture and the input matrix of face picture is respectively obtained；

S52：Fingerprint picture input matrix is input in multiple linear regression fingerprint picture identification model and obtains fingerprint image The model output matrix of piece, face picture input matrix is input in multiple linear regression face picture identification model and obtains people The model output matrix of face picture；

S53：The model output matrix that the input matrix that step S51 is obtained is obtained with step S52 is respectively to fingerprint picture Matched with face picture；

S54：If fingerprint picture the match is successful rate is H₁, face picture the match is successful rate is H₂, judge authentication matching degree Whether H is more than match settings value h, wherein H=H₁×H₂, h=0~1；If so, for the match is successful, otherwise it fails to match.Wherein Meet H₁≥h₀, H₂≥h₀。

Beneficial effects of the present invention：Autoassociative memories and multiple linear regression model are combined, by fingerprint picture and people Face picture changes into series of parameters and preserved, and identity information includes fingerprint and face picture, and checking reliability is high and right The preserving type crypticity of picture is strong, and safety coefficient is high, effectively prevents that people's identity information is compromised；Using by picture through model The form of parameter is changed into, simple and convenient, practicality is good, and picture recognition effect is good, face picture and the protection of fingerprint picture are imitated It is really good.

Brief description of the drawings

Fig. 1 is the image identification method flow chart of the present invention；

Fig. 2 is multiple linear regression fingerprint picture identification model location parameter resolution principle figure；

Fig. 3 is multiple linear regression face picture identification model location parameter resolution principle figure.

Embodiment

The embodiment and operation principle to the present invention are described in further detail below in conjunction with the accompanying drawings.

It will be seen from figure 1 that a kind of fingerprint and face coupled identification side based on multiple linear regression associative memory model Method, comprises the following steps：

S1：The fingerprint picture and face picture of collection crowd, is grouped to the fingerprint picture and face picture collected Numbering；

When carrying out identity information preservation, it is necessary to which the fingerprint picture and face picture that obtain the user simultaneously carry out dual core It is right.

If one of picture the match is successful rate is less than H setting value h, by identity information recognition failures, identifying system can It is high by property, improve identity information and preserve reliability.

S2：By setting binary map luminance threshold, fingerprint picture and face picture are processed as two-value picture, fingerprint is obtained The associative memory input matrix and output matrix of picture and the associative memory input matrix and output matrix of face picture；

The binary map luminance threshold K=(0,1,2,3...255)；In the present embodiment, K=100 is set.

Include m group pictures in step sl, corresponding numbering is 1,2,3...m, and all fingerprint pictures and face picture are equal Individual pixel is arranged by N rows M to constitute, pixel number is n=N × M；

If the binary map matrix of fingerprint is input matrix Γ '=(X ' of associative memory¹,X′²,…,X′ⁱ,…,X′^m), i ∈ { 1,2 ..., m }, represents the input vector of pixel composition all in the binary map of the i-th width fingerprint, wherein, Represent the two of the i-th width fingerprint It is worth the input value of j-th of pixel in figure；

If the binary map matrix of fingerprint is the output matrix of associative memory The output vector of pixel composition all in the binary map of the i-th width fingerprint is represented, wherein, Represent the two-value in the i-th width fingerprint The output valve of j-th of pixel in figure；

If the binary map matrix of face is the input matrix Γ "=(X " of associative memory¹,X″²,…,X″ⁱ,…,X″^m), i ∈ { 1,2 ..., m }, represents the input vector of pixel composition all in the binary map of the i-th width face, wherein, Represent in the i-th width face The input value of j-th of pixel in binary map；

If the binary map matrix of face is the output matrix of associative memory The output vector of pixel composition all in the binary map of the i-th width face is represented, wherein, Represent the two of the i-th width face It is worth the output valve of j-th of pixel in figure.

In the present embodiment, black pixel point (0) is mapped as -1 in two-value picture, by white pixel point in two-value picture (255) it is mapped as 1.

S3：Using cell neural network structure, two multiple linear regression picture recognition moulds for carrying regression parameter are built Type, respectively multiple linear regression fingerprint picture identification model and multiple linear regression face picture identification model；

In the present embodiment, multiple linear regression fingerprint picture identification model is built and many based on cell neural network structure First linear face picture identification model, be specially：

The multiple linear regression fingerprint picture identification model built based on cell neural network structure, be specially：

Wherein, i=1,2 ..., N；J=1,2 ..., M；k₁(i, r)=max { 1-i ,-r }；k₂(i, r)=min { N-i, r }； l₁(j, r)=max { 1-j ,-r }；l₂(i, r)=min { M-i, r }；R is to represent template radius,For input parameter,It is defeated Go out,It is input,It is offset,It is input template；

Input templateExpression formula it is as follows：

It is 1, input template to make rExpression formula in

By formulaIt is rewritten as：

Y '=A ' X '+V '

Wherein, input vector：

Output vector

Offset

Dot-blur pattern A ' and offset V ' is the regression parameter, dot-blur pattern A '=(a '_ij)_n×nShape can be written as Formula：

Wherein,

The multiple linear regression face picture identification model built based on cell neural network structure, be specially：

Wherein, i=1,2 ..., N；J=1,2 ..., M；k₁(i, r)=max { 1-i ,-r }；k₂(i, r)=min { N-i, r }； l₁(j, r)=max { 1-j ,-r }；l₂(i, r)=min { M-i, r }；R is to represent template radius,For input parameter,It is defeated Go out,It is input,It is offset,It is input template；

Input templateExpression formula it is as follows：

It is 1 to make r, then input templateExpression formula in

FormulaIt is rewritten as：

Y "=A " X "+V "

Wherein, input vector：

Output vector

Offset

Dot-blur pattern A " and offset V " is the regression parameter, dot-blur pattern A "=(a "_ij)_n×nShape can be written as Formula：

Wherein,

S4：According to the obtained fingerprint pictures of step S2 and associative memory input matrix, output matrix and the step of face picture Two multiple linear regression picture recognition models that rapid S3 is obtained, the regression parameter in calculation procedure S3, final true respectively Determine multiple linear regression fingerprint picture identification model, multiple linear regression face picture identification model；

Regression parameter in step S4 in calculation procedure S3 is concretely comprised the following steps：

S41：Order vector

Make Y '=((Y '¹)^T, (Y '²)^T..., (Y 'ⁱ)^T..., (Y '^m)^T)^T, Y "=((Y "¹)^T, (Y "²)^T..., (Y "ⁱ )^T..., (Y "^m)^T)^T；

Wherein, i ∈ { 1,2 ..., m }, Y 'ⁱWith Y "ⁱRepresent all in the i-th width fingerprint binary map and face binary map respectively The column vector of pixel composition；

L ∈ { 1,2 ..., m } and q ∈ { 1,2 ..., N } are made,

It can then be obtained by formula Y '=A ' X '+V '：

X ' L '=Y'

It can be obtained by formula Y "=A " X "+V "：

X " L "=Y "

Therefore,

Wherein, L ' and L " is constant；

S42：In autoassociative memories criterion, exist：

The fingerprint picture associative memory input matrix Γ ' obtained in step S2 is converted into X ', output matrixIt is converted into Y', brings formula X ' L '=Y' into, obtains L '；By the face picture associative memory input matrix Γ " obtained in step S2 conversions For X ", output matrixY " is converted into, formula X " L "=Y " is brought into, obtains L ".V ' and input template must be offset according to L ' According to L " v must be offset " and input template

S43：The skew v ' that step S42 is obtained_iAnd input templateBring formula into

Dot-blur pattern A ' and offset V ' are obtained, is drawn Multiple linear regression fingerprint picture identification model；

The skew v " that step S42 is obtained_iAnd input templateBring formula into

Respectively, dot-blur pattern A " and offset V " is obtained, Obtain multiple linear regression face picture identification model.

S5：Based on autoassociative memories criterion, fingerprint picture and face picture are identified respectively.Specially：

S51：Fingerprint picture and face picture are obtained, fingerprint picture and the input matrix of face picture is respectively obtained；

S52：Fingerprint picture input matrix is input in multiple linear regression fingerprint picture identification model and obtains fingerprint image The model output matrix of piece, face picture input matrix is input in multiple linear regression face picture identification model and obtains people The model output matrix of face picture；

S53：The model output matrix that the input matrix that step S51 is obtained is obtained with step S52 is respectively to fingerprint picture Matched with face picture；

S54：If fingerprint picture the match is successful rate is H₁, face picture the match is successful rate is H₂, judge authentication matching degree Whether H is more than match settings value h, wherein H=H₁×H₂, h=0~1；If so, for the match is successful, otherwise it fails to match, wherein also Need to meet H₁≥h₀, H₂≥h₀, wherein h₀=0.92

In the present embodiment, h=0.9, h₀=0.92

It should be pointed out that described above is not limitation of the present invention, the present invention is also not limited to the example above, What those skilled in the art were made in the essential scope of the present invention changes, is modified, adds or replaces, and also should Belong to protection scope of the present invention.

Claims (6)

1. a kind of fingerprint and face coupled identification method based on multiple linear regression associative memory model, it is characterised in that including Following steps：

S1：The fingerprint picture and face picture of collection crowd, packet numbering is carried out to the fingerprint picture and face picture collected；

S2：By setting binary map luminance threshold, the obtained all fingerprint pictures of step S1 and face picture are processed as two-value Picture, obtains the associative memory input matrix and output matrix of fingerprint picture and the associative memory input matrix of face picture and defeated Go out matrix；

S3：Using cell neural network structure, two multiple linear regression picture recognition models for carrying regression parameter are built, point Wei not multiple linear regression fingerprint picture identification model and multiple linear regression face picture identification model；

S4：According to the obtained fingerprint pictures of step S2 and associative memory input matrix, output matrix and the step S3 of face picture Two obtained multiple linear regression picture recognition models, the regression parameter in calculation procedure S3, is finally determined many respectively First linear regression fingerprint picture identification model, multiple linear regression face picture identification model；

S5：Based on autoassociative memories criterion, fingerprint picture and face picture are identified respectively.

2. fingerprint and face coupled identification side according to claim 1 based on multiple linear regression associative memory model Method, it is characterised in that：Carry out authentication when, it is necessary to obtain simultaneously the user fingerprint picture and face picture carry out it is dual Verification.

3. fingerprint and face coupled identification side according to claim 1 based on multiple linear regression associative memory model Method, it is characterised in that：The binary map luminance threshold K=(0,1,2,3 ..., 255)；

Include m group pictures in step sl, corresponding numbering is 1,2,3...m, and all fingerprint pictures and face picture are by N rows The pixel composition of M row, pixel number is n=N × M；

If the binary map matrix of fingerprint is input matrix Γ '=(X ' of associative memory¹,X′²,…,X′ⁱ,…,X′^m), i ∈ 1, 2 ..., m }, the input vector of pixel composition all in the binary map of the i-th width fingerprint is represented, wherein, Represent the two of the i-th width fingerprint It is worth the input value of j-th of pixel in figure；

If the binary map matrix of fingerprint is the output matrix of associative memory The output vector of pixel composition all in the binary map of the i-th width fingerprint is represented, wherein, Represent the two-value in the i-th width fingerprint The output valve of j-th of pixel in figure；

If the binary map matrix of face is the input matrix Γ "=(X " of associative memory¹,X″²,…,X″ⁱ,…,X″^m), i ∈ 1, 2 ..., m }, the input vector of pixel composition all in the binary map of the i-th width face is represented, wherein, Represent in the i-th width face The input value of j-th of pixel in binary map；

If the binary map matrix of face is the output matrix of associative memory The output vector of pixel composition all in the binary map of the i-th width face is represented, wherein, Represent the two of the i-th width face It is worth the output valve of j-th of pixel in figure.

4. fingerprint and face coupled identification side according to claim 3 based on multiple linear regression associative memory model Method, it is characterised in that step S3 particular content is：

The multiple linear regression fingerprint picture identification model built based on cell neural network structure, be specially：

$\stackrel{\‾}{y_{ij}^{\′}}=\underset{k=k_1(i,r)}{\overset{k_2(i,r)}{\Σ}}\underset{l=l_1(j,r)}{\overset{l_2(j,r)}{\Σ}}{\stackrel{\‾}{a^{\′}}}_{kl}{\stackrel{\‾}{x^{\′}}}_{(k+i)(l+j)}+{\stackrel{\‾}{v^{\′}}}_{ij}---\left(1\right)$

Wherein, i=1,2 ..., N；J=1,2 ..., M；k₁(i, r)=max { 1-i ,-r }；k₂(i, r)=min { N-i, r }；l₁ (j, r)=max { 1-j ,-r }；l₂(i, r)=min { M-i, r }；R is to represent template radius,For input parameter,It is output,It is input,It is offset,It is input template；

Input templateExpression formula it is as follows：

It is 1 to make r, then in equation (2)

$\stackrel{\‾}{a^{\′}}={\left[\begin{array}{ccc}{\stackrel{\‾}{a^{\′}}}_{-1,-1}& {\stackrel{\‾}{a^{\′}}}_{-1,0}& {\stackrel{\‾}{a^{\′}}}_{-1,1}\\ {\stackrel{\‾}{a^{\′}}}_{0,-1}& {\stackrel{\‾}{a^{\′}}}_{0,0}& {\stackrel{\‾}{a^{\′}}}_{0,1}\\ {\stackrel{\‾}{a^{\′}}}_{1,-1}& {\stackrel{\‾}{a^{\′}}}_{1,0}& {\stackrel{\‾}{a^{\′}}}_{1,1}\end{array}\right]}_{3\×3}.$

Formula (1) is rewritten as：

Y '=A ' X '+V ' (3)

Wherein, input vector：

$X^{\′}={(x_1^{\′},x_2^{\′},...,x_i^{\′},...,x_n^{\′})}^T={({\stackrel{\‾}{x^{\′}}}_{11},{\stackrel{\‾}{x^{\′}}}_{12},...,{\stackrel{\‾}{x^{\′}}}_{1M},{\stackrel{\‾}{x^{\′}}}_{21},...,{\stackrel{\‾}{x^{\′}}}_{NM})}^T,i\∈\{1,2,...,n\}$

Output vector

$Y^{\′}={(y_1^{\′},y_2^{\′},...,y_i^{\′},...,y_n^{\′})}^T={({\stackrel{\‾}{y^{\′}}}_{11},{\stackrel{\‾}{y^{\′}}}_{12},...,{\stackrel{\‾}{y^{\′}}}_{1M},{\stackrel{\‾}{y^{\′}}}_{21},...,{\stackrel{\‾}{y^{\′}}}_{NM})}^T,i\∈\{1,2,...,n\}$

Offset

$V^{\′}={(v_1^{\′},v_2^{\′},...,v_i^{\′},...,v_n^{\′})}^T={({\stackrel{\‾}{v^{\′}}}_{11},{\stackrel{\‾}{v^{\′}}}_{12},...,{\stackrel{\‾}{v^{\′}}}_{1M},{\stackrel{\‾}{v^{\′}}}_{21},...,{\stackrel{\‾}{v^{\′}}}_{NM})}^T,i\∈\{1,2,...,n\}---\left(4\right)$

Dot-blur pattern A ' and offset V ' is the regression parameter, dot-blur pattern A '=(a_i′_j)_n×nForm can be written as：

Wherein,

The multiple linear regression face picture identification model built based on cell neural network structure, be specially：

$\stackrel{\‾}{y_{ij}^{\′\′}}=\underset{k=k_1(i,r)}{\overset{k_2(i,r)}{\Σ}}\underset{l=l_1(j,r)}{\overset{l_2(j,r)}{\Σ}}{\stackrel{\‾}{a^{\′\′}}}_{kl}{\stackrel{\‾}{x^{\′\′}}}_{(k+i)(l+j)}+{\stackrel{\‾}{v^{\′\′}}}_{ij}---\left(6\right)$

Wherein, i=1,2 ..., N；J=1,2 ..., M；k₁(i, r)=max { 1-i ,-r }；k₂(i, r)=min { N-i, r }；l₁ (j, r)=max { 1-j ,-r }；l₂(i, r)=min { M-i, r }；R is to represent template radius,For input parameter,It is defeated Go out,It is input,It is offset,It is input template；

Input templateExpression formula it is as follows：

It is 1 to make r, then in equation (7)

$\stackrel{\‾}{a^{\′\′}}={\left[\begin{array}{ccc}{\stackrel{\‾}{a^{\′\′}}}_{-1,-1}& {\stackrel{\‾}{a^{\′\′}}}_{-1,0}& {\stackrel{\‾}{a^{\′\′}}}_{-1,1}\\ {\stackrel{\‾}{a^{\′\′}}}_{0,-1}& {\stackrel{\‾}{a^{\′\′}}}_{0,0}& {\stackrel{\‾}{a^{\′\′}}}_{0,1}\\ {\stackrel{\‾}{a^{\′\′}}}_{1,-1}& {\stackrel{\‾}{a^{\′\′}}}_{1,0}& {\stackrel{\‾}{a^{\′\′}}}_{1,1}\end{array}\right]}_{3\×3}.$

Formula (6) is rewritten as：

Y "=A " X "+V " (8)

Wherein, input vector：

$X^{\′\′}={(x_1^{\′\′},x_2^{\′\′},...,x_i^{\′\′},...,x_n^{\′\′})}^T={({\stackrel{\‾}{x^{\′\′}}}_{11},{\stackrel{\‾}{x^{\′\′}}}_{12},...,{\stackrel{\‾}{x^{\′\′}}}_{1M},{\stackrel{\‾}{x^{\′\′}}}_{21},...,{\stackrel{\‾}{x^{\′\′}}}_{NM})}^T,i\∈\{1,2,...,n\}$

Output vector

$Y^{\′\′}={(y_1^{\′\′},y_2^{\′\′},...,y_i^{\′\′},...,y_n^{\′\′})}^T={({\stackrel{\‾}{y^{\′\′}}}_{11},{\stackrel{\‾}{y^{\′\′}}}_{12},...,{\stackrel{\‾}{y^{\′\′}}}_{1M},{\stackrel{\‾}{y^{\′\′}}}_{21},...,{\stackrel{\‾}{y^{\′\′}}}_{NM})}^T,i\∈\{1,2,...,n\}$

Offset

$V^{\′\′}={(v_1^{\′\′},v_2^{\′\′},...,v_i^{\′\′},...,v_n^{\′\′})}^T={({\stackrel{\‾}{v^{\′\′}}}_{11},{\stackrel{\‾}{v^{\′\′}}}_{12},...,{\stackrel{\‾}{v^{\′\′}}}_{1M},{\stackrel{\‾}{v^{\′\′}}}_{21},...,{\stackrel{\‾}{v^{\′\′}}}_{NM})}^T,i\∈\{1,2,...,n\}---\left(9\right)$

Dot-blur pattern A " and offset V " is the regression parameter, dot-blur pattern A "=(a "_ij)_n×nForm can be written as：

Wherein,

5. fingerprint and face coupled identification side according to claim 4 based on multiple linear regression associative memory model Method, it is characterised in that regression parameter in step S4 in calculation procedure S3 is concretely comprised the following steps：

S41：Order vector

$L^{\′\′}={(v^{\′\′},{\stackrel{\‾}{a^{\′\′}}}_{-1,-1},{\stackrel{\‾}{a^{\′\′}}}_{-1,0},{\stackrel{\‾}{a^{\′\′}}}_{-1,1},{\stackrel{\‾}{a^{\′\′}}}_{0,-1},{\stackrel{\‾}{a^{\′\′}}}_{0,0},{\stackrel{\‾}{a^{\′\′}}}_{0,1},{\stackrel{\‾}{a^{\′\′}}}_{1,-1},{\stackrel{\‾}{a^{\′\′}}}_{1,0},{\stackrel{\‾}{a^{\′\′}}}_{1,1})}_{10\×1}^T;$

Make Y'=((Y '¹)^T,(Y′²)^T,…,(Y′ⁱ)^T,…,(Y′^m)^T)^T, Y "=((Y "¹)^T,(Y″²)^T,…,(Y″ⁱ)^T,…,(Y ″^m)^T)^T；

Wherein, i ∈ { 1,2 ..., m }, Y 'ⁱWith Y "ⁱPixel all in the i-th width fingerprint binary map and face binary map is represented respectively The column vector of point composition；

L ∈ { 1,2 ..., m } and q ∈ { 1,2 ..., N } are made,

$X_q^{\′l}={\left(\begin{array}{ccc}0& {\mathrm{\α}}_{(q-1)M+1}^{\′l}& {\mathrm{\α}}_{(q-1)M+2}^{\′l}\\ {\mathrm{\α}}_{(q-1)M+1}^{\′l}& {\mathrm{\α}}_{(q-1)M+2}^{\′l}& {\mathrm{\α}}_{(q-1)M+3}^{\′l}\\ {\mathrm{\α}}_{(q-1)M+2}^{\′l}& {\mathrm{\α}}_{(q-1)M+3}^{\′l}& {\mathrm{\α}}_{(q-1)M+4}^{\′l}\\ .& .& .\\ .& .& .\\ .& .& .\\ {\mathrm{\α}}_{qM-2}^{\′l}& {\mathrm{\α}}_{qM-1}^{\′l}& {\mathrm{\α}}_{qM}^{\′l}\\ {\mathrm{\α}}_{qM-1}^{\′l}& {\mathrm{\α}}_{qM}^{\′l}& 0\end{array}\right)}_{M\×3},$

$X^{\′l}={\left(\begin{array}{ccc}0& X_1^{\′l}& X_2^{\′l}\\ X_1^{\′l}& X_2^{\′l}& X_3^{\′l}\\ X_2^{\′l}& X_3^{\′l}& X_4^{\′l}\\ .& .& .\\ .& .& .\\ .& .& .\\ X_q^{\′l}& X_{q+1}^{\′l}& X_{q+2}^{\′l}\\ .& .& .\\ .& .& .\\ .& .& .\\ X_{N-2}^{\′l}& X_{N-1}^{\′l}& X_N^{\′l}\\ X_{N-1}^{\′l}& X_N^{\′l}& 0\end{array}\right)}_{n\×9},$

$X^{\′}={\left[\begin{array}{cc}1& X^{\′1}\\ 1& X^{\′2}\\ .& .\\ .& .\\ .& .\\ 1& X^{\′l}\\ .& .\\ .& .\\ .& .\\ 1& X^{\′m}\end{array}\right]}_{(n\×m)\×10};$

$X_q^{\′\′l}={\left(\begin{array}{ccc}0& {\mathrm{\α}}_{(q-1)M+1}^{\′\′l}& {\mathrm{\α}}_{(q-1)M+2}^{\′\′l}\\ {\mathrm{\α}}_{(q-1)M+1}^{\′\′l}& {\mathrm{\α}}_{(q-1)M+2}^{\′\′l}& {\mathrm{\α}}_{(q-1)M+3}^{\′\′l}\\ {\mathrm{\α}}_{(q-1)M+2}^{\′\′l}& {\mathrm{\α}}_{(q-1)M+3}^{\′\′l}& {\mathrm{\α}}_{(q-1)M+4}^{\′\′l}\\ .& .& .\\ .& .& .\\ .& .& .\\ {\mathrm{\α}}_{qM-2}^{\′\′l}& {\mathrm{\α}}_{qM-1}^{\′\′l}& {\mathrm{\α}}_{qM}^{\′\′l}\\ {\mathrm{\α}}_{qM-1}^{\′\′l}& {\mathrm{\α}}_{qM}^{\′\′l}& 0\end{array}\right)}_{M\×3},$

$X^{\′\′l}={\left(\begin{array}{ccc}0& X_1^{\′\′l}& X_2^{\′\′l}\\ X_1^{\′\′l}& X_2^{\′\′l}& X_3^{\′\′l}\\ X_2^{\′\′l}& X_3^{\′\′l}& X_4^{\′\′l}\\ .& .& .\\ .& .& .\\ .& .& .\\ X_q^{\′\′l}& X_{q+1}^{\′\′l}& X_{q+2}^{\′\′l}\\ .& .& .\\ .& .& .\\ .& .& .\\ X_{N-2}^{\′\′l}& X_{N-1}^{\′\′l}& X_N^{\′\′l}\\ X_{N-1}^{\′\′l}& X_N^{\′\′l}& 0\end{array}\right)}_{n\×9},$

$X^{\′\′}={\left[\begin{array}{cc}1& X^{\′\′1}\\ 1& X^{\′\′2}\\ .& .\\ .& .\\ .& .\\ 1& X^{\′\′l}\\ .& .\\ .& .\\ .& .\\ 1& X^{\′\′m}\end{array}\right]}_{(n\×m)\×10}.$

It can then be obtained by formula (3)：

X ' L '=Y'(11)

Formula (8) can be obtained：

X " L "=Y " (12)

Therefore,

$L^{\′}=pinv\left(X^{\′}\right)\·Y^{\′}={(v^{\′},{\stackrel{\‾}{a^{\′}}}_{-1,-1},{\stackrel{\‾}{a^{\′}}}_{-1,0},{\stackrel{\‾}{a^{\′}}}_{-1,1},{\stackrel{\‾}{a^{\′}}}_{0,-1},{\stackrel{\‾}{a^{\′}}}_{0,0},{\stackrel{\‾}{a^{\′}}}_{0,1},{\stackrel{\‾}{a^{\′}}}_{1,-1},{\stackrel{\‾}{a^{\′}}}_{1,0},{\stackrel{\‾}{a^{\′}}}_{1,1})}_{10\×1}^T,---\left(13\right)$

$L^{\′\′}=pinv\left(X^{\′\′}\right)\·Y^{\′\′}={(v^{\′\′},{\stackrel{\‾}{a^{\′\′}}}_{-1,-1},{\stackrel{\‾}{a^{\′\′}}}_{-1,0},{\stackrel{\‾}{a^{\′\′}}}_{-1,1},{\stackrel{\‾}{a^{\′\′}}}_{0,-1},{\stackrel{\‾}{a^{\′\′}}}_{0,0},{\stackrel{\‾}{a^{\′\′}}}_{0,1},{\stackrel{\‾}{a^{\′\′}}}_{1,-1},{\stackrel{\‾}{a^{\′\′}}}_{1,0},{\stackrel{\‾}{a^{\′\′}}}_{1,1})}_{10\×1}^T,---\left(14\right)$

Wherein, L ' and L " is constant；

S42：In autoassociative memories criterion, exist：

${\mathrm{\Γ}}^{\′}={\hat{Y}}^{\′}---\left(15\right)$

${\mathrm{\Γ}}^{\′\′}={\hat{Y}}^{\′\′}---\left(16\right)$

The fingerprint picture associative memory input matrix Γ ' obtained in step S2 is converted into X ', output matrixIt is converted into Y', band Enter formula (11), obtain L '；By the face picture associative memory input matrix Γ " being converted into X " obtained in step S2, square is exported Battle arrayY " is converted into, formula (12) is brought into, obtains L ".V ' and input template must be offset according to L 'According to L " v must be offset " and Input template

S43：The skew v ' that step S42 is obtained_iAnd input templateBring formula (4), (5) into, obtain dot-blur pattern A ' and skew V ' is measured, multiple linear regression fingerprint picture identification model is drawn；

The skew v " that step S42 is obtained_iAnd input templateBring formula (9), (10) into, respectively, obtain dot-blur pattern A " and inclined Shifting amount V ", obtains multiple linear regression face picture identification model.

6. the fingerprint and face based on multiple linear regression associative memory model according to claim 1 or 2 or 3 or 4 or 5 Coupled identification method, it is characterised in that to concretely comprising the following steps that fingerprint picture and face picture are identified in step S5：

S51：Fingerprint picture and face picture are obtained, fingerprint picture and the input matrix of face picture is respectively obtained；

S52：Fingerprint picture input matrix is input in multiple linear regression fingerprint picture identification model and obtains fingerprint picture Model output matrix, face picture input matrix is input in multiple linear regression face picture identification model and obtains face figure The model output matrix of piece；

S53：The model output matrix that the input matrix that step S51 is obtained is obtained with step S52 is respectively to fingerprint picture and people Face picture is matched；

S54：If fingerprint picture the match is successful rate is H₁, face picture the match is successful rate is H₂, judge that authentication matching degree H is It is no to be more than match settings value h, wherein H=H₁×H₂, h=0~1；If so, for the match is successful, otherwise it fails to match.