CN106971157B - Identity coupling identification method based on multiple linear regression association memory model - Google Patents

Identity coupling identification method based on multiple linear regression association memory model Download PDF

Info

Publication number
CN106971157B
CN106971157B CN201710175033.3A CN201710175033A CN106971157B CN 106971157 B CN106971157 B CN 106971157B CN 201710175033 A CN201710175033 A CN 201710175033A CN 106971157 B CN106971157 B CN 106971157B
Authority
CN
China
Prior art keywords
picture
fingerprint
matrix
input
face
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN201710175033.3A
Other languages
Chinese (zh)
Other versions
CN106971157A (en
Inventor
韩琦
翁腾飞
刘晋
刘洋
吴政阳
谯自强
黄军建
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Science and Technology
Chongqing University of Education
Original Assignee
Chongqing University of Science and Technology
Chongqing University of Education
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Chongqing University of Science and Technology, Chongqing University of Education filed Critical Chongqing University of Science and Technology
Priority to CN201710175033.3A priority Critical patent/CN106971157B/en
Publication of CN106971157A publication Critical patent/CN106971157A/en
Application granted granted Critical
Publication of CN106971157B publication Critical patent/CN106971157B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1347—Preprocessing; Feature extraction
    • 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
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1365—Matching; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/168—Feature extraction; Face representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/16—Human faces, e.g. facial parts, sketches or expressions
    • G06V40/172—Classification, e.g. identification

Landscapes

  • Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Health & Medical Sciences (AREA)
  • Collating Specific Patterns (AREA)

Abstract

The invention discloses a fingerprint and face coupling identification method based on a multiple linear regression association memory model, which comprises the following steps: s1: collecting a fingerprint picture and a face picture; s2: respectively obtaining an association memory input matrix and an association memory output matrix of the fingerprint picture and the face picture; s3: constructing a multiple linear regression fingerprint picture identification model with regression parameters and a multiple linear regression face picture identification model; s4: calculating regression parameters to obtain a multiple linear regression fingerprint picture identification model and a multiple linear regression face picture identification model; s5: and identifying the fingerprint picture and the face picture. Has the advantages that: the identity information realizes multiple recognition, the reliability is high, the associative memory and the multiple linear regression model are combined, the picture is converted into parameters, the safety factor is high, the recognition effect is good, the identity information protection effect is good, and the secrecy is high.

Description

Identity coupling identification method based on multiple linear regression association memory model
Technical Field
The invention relates to the technical field of image data storage, in particular to an identity coupling identification method based on a multiple linear regression association memory model.
Background
With the development of the big data era, people usually store daily photos and even identity document photos in a database, so that private information is easy to steal by hackers to sell or conduct illegal criminal activities, the private information of people is leaked, daily life is easily disturbed and even involved in criminal events, and inconvenience is easily caused.
In some organizations that need to perform identity information verification, people's identity information is collected to perform identity verification. Such as collecting human face information or fingerprint information of a person. The information quantity is small, and the identity verification reliability is not high. When the identity information of people is stored, the identity information is usually directly stored, the photos are not processed and stored, the safety factor is low, and the identity information is easy to steal. Once the identity information database is attacked by lawbreakers, the identity information is very easy to be stolen for illegal activities, and unnecessary troubles are caused to people.
Disclosure of Invention
Aiming at the problems, the invention provides an identity coupling identification method based on a multiple linear regression association memory model, which combines the association memory and the multiple linear regression model to convert a human face picture and a fingerprint picture into a series of parameters for storage, realizes double identity verification and has high safety and good reliability.
In order to achieve the purpose, the invention adopts the following specific technical scheme:
an identity coupling identification method based on a multiple linear regression association memory model is characterized by comprising the following steps:
s1: collecting fingerprint pictures and face pictures of people, and numbering the collected fingerprint pictures and face pictures in groups;
s2: processing all the fingerprint pictures and the face pictures obtained in the step S1 into binary pictures by setting a binary picture brightness threshold value to obtain an association memory input matrix and an output matrix of the fingerprint pictures and an association memory input matrix and an output matrix of the face pictures;
s3: constructing two multiple linear regression image recognition models with regression parameters by using a cellular neural network structure, wherein the two multiple linear regression image recognition models are a multiple linear regression fingerprint image recognition model and a multiple linear regression face image recognition model respectively;
s4: respectively calculating the regression parameters in the step S3 according to the associative memory input matrix and the output matrix of the fingerprint picture and the face picture obtained in the step S2 and the two multiple linear regression picture identification models obtained in the step S3, and finally determining a multiple linear regression fingerprint picture identification model and a multiple linear regression face picture identification model;
s5: and respectively identifying the fingerprint picture and the face picture based on a self-association memory criterion.
Further, when performing identity authentication, a fingerprint image and a face image of a user need to be acquired at the same time for double check.
If one of the pictures has a problem in checking, the checking fails. Double checking is realized, the reliability is high, and the reliability of identity information storage is improved.
Further, the binary image brightness threshold K is (0, 1,2,3.. 255);
in step S1, M groups of pictures are included, the corresponding numbers are 1,2,3.. M, all the fingerprint pictures and the face pictures are composed of N rows and M columns of pixel points, and the number of the pixel points is N ═ nxm;
let the binary map matrix of the fingerprint be an input matrix ' ═ X ' of associative memory '1,X′2,…,X′i,…,X′m) I e {1,2, …, m }, represents an input vector consisting of all the pixels in the binary map of the ith fingerprint, wherein,
Figure GDA0002690266650000031
αj′irepresenting the input value of the jth pixel point in the binary image of the ith fingerprint;
setting binary image matrix of fingerprint as output matrix of associative memory
Figure GDA0002690266650000033
i ∈ {1,2, …, m }, and represents an output vector composed of all pixel points in the binary map of the ith fingerprint, wherein Y'i＝(y1′i，y2′i，…，yj′i，…，yn′i)T，i∈{1，2，…，m}，j{1，2，…，n}，yj′iThe output value of the jth pixel point in the binary image of the ith fingerprint is represented;
the binary image matrix of the face is set as an input matrix ═ (X ″) of associative memory1,X″2,…,X″i,…,X″m) I ∈ {1,2, …, m }, which represents an input vector composed of all the pixel points in the binary image of the ith human face, wherein X ″, wherei＝(α1″i，α2″i，…，αj″i，…，αn″i)T，i∈{1，2，…，m}，j∈{1，2，…，n}，αj″iThe input value of the jth pixel point in the ith human face binary image is represented;
setting binary image matrix of human face as output matrix of associative memory
Figure GDA0002690266650000038
i belongs to {1,2, …, m }, and represents an output vector consisting of all pixel points in the binary image of the ith human face, wherein Y ″)i＝(y1″i，y2″i，…，yj″i，…，yn″i)T，i∈{1，2，…，m}，j∈{1，2，…，n}，yj″iAnd the output value of the jth pixel point in the ith human face binary image is represented.
Further, the specific content of step S3 is as follows: :
the multivariate linear regression fingerprint picture identification model constructed based on the cellular neural network structure specifically comprises the following steps:
Figure GDA00026902666500000311
wherein iN＝1,2，…,N；jM＝1,2，…,M；k1(iN,r)＝max{1-iN,-r}；k2(iN,r)＝min{N-iN,r}；l1(jM,r)＝max{1-jM,-r}；l2(jM,r)＝min{M-jMR }; r is a value representing the radius of the template,
Figure GDA0002690266650000041
in order to input the parameters, the user can select the parameters,
Figure GDA0002690266650000042
is the output of the computer system,
Figure GDA0002690266650000043
is an input to the computer system that is,
Figure GDA0002690266650000044
is the amount of the offset that is,
Figure GDA0002690266650000045
is an input template;
input template
Figure GDA0002690266650000046
The expression of (a) is as follows:
Figure GDA0002690266650000047
let r be 1, then in equation (2)
Figure GDA0002690266650000048
Formula (1) is rewritten as:
Y′＝A′X′+V′ (3)
wherein, the input vector:
Figure GDA0002690266650000049
output vector
Figure GDA00026902666500000410
Offset amount
Figure GDA00026902666500000411
The memory matrix A 'and the offset V' are the regression parameters, and the memory matrix A 'is (a'ij)n×nThe following can be written:
Figure GDA0002690266650000051
wherein the content of the first and second substances,
Figure GDA0002690266650000052
Figure GDA0002690266650000053
Figure GDA0002690266650000054
the multivariate linear regression human face picture recognition model constructed based on the cellular neural network structure specifically comprises the following steps:
Figure GDA0002690266650000055
wherein iN＝1,2，…,N；jM＝1,2，…,M；k1(iN,r)＝max{1-iN,-r}；k2(iN,r)＝min{N-iN,r}；l1(jM,r)＝max{1-jM,-r}；l2(jM,r)＝min{M-jMR }; r is a value representing the radius of the template,
Figure GDA0002690266650000061
in order to input the parameters, the user can select the parameters,
Figure GDA0002690266650000062
is the output of the computer system,
Figure GDA0002690266650000063
is an input to the computer system that is,
Figure GDA0002690266650000064
is the amount of the offset that is,
Figure GDA0002690266650000065
is an input template;
input template
Figure GDA0002690266650000066
The expression of (a) is as follows:
Figure GDA0002690266650000067
let r be 1, then in equation (7)
Figure GDA0002690266650000068
Equation (6) is rewritten as:
Y″＝A″X″+V″ (8)
wherein, the input vector:
Figure GDA0002690266650000069
output vector
Figure GDA00026902666500000610
Offset amount
Figure GDA00026902666500000611
The memory matrix A ' and the offset V ' are the regression parameters, and the memory matrix A ' ═ aij)n×nThe following can be written:
Figure GDA0002690266650000071
wherein the content of the first and second substances,
Figure GDA0002690266650000072
Figure GDA0002690266650000073
Figure GDA0002690266650000074
further, the specific steps of calculating the unknown regression parameters in step S3 in step S4 are as follows: s41: order vector
Figure GDA0002690266650000075
Figure GDA0002690266650000076
Let 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'iAnd Y ″)iRespectively representing column vectors formed by all pixel points in the ith fingerprint binary image and the face binary image;
let l ∈ {1,2, …, m } and q ∈ {1,2, …, N },
Figure GDA0002690266650000081
Figure GDA0002690266650000082
Figure GDA0002690266650000083
Figure GDA0002690266650000091
Figure GDA0002690266650000092
Figure GDA0002690266650000093
then from equation (3) one can derive:
X′·L′＝Y' (11)
equation (8) can be found:
X″·L″＝Y″ (12)
then
Figure GDA0002690266650000094
Figure GDA0002690266650000095
Wherein L 'and L' are constants.
S42: among the self-associative memory criteria, there are:
Figure GDA0002690266650000101
Figure GDA0002690266650000102
the fingerprint picture obtained in step S2 is associated with the memory input matrix '((X))'1,X′2,…,X′i,…,X′m) And an output matrix
Figure GDA0002690266650000103
Substituting the formula (11), the fingerprint picture associative memory input matrix ″ (X ″) obtained in step S2 is stored in the storage unit1,X″2,…,X″i,…,X″m) And an output matrix
Figure GDA0002690266650000104
Substituted into equation (12) and converting 'to X', will
Figure GDA0002690266650000105
Conversion to Y', "to X", will
Figure GDA0002690266650000106
Conversion to Y 'gives X'. L1Y' and X ″. L2Y 'to obtain L' ═ L1Pinv (X '). Y' and L ″ -L2＝pinv(X″)·Y″；
S43: converting the offset v 'obtained in step S42'jAnd an input template
Figure GDA0002690266650000107
Substituting the formula (4) and the formula (5) to obtain a memory matrix A 'and an offset V', and obtaining a multiple linear regression fingerprint image identification model;
the offset v ″' obtained in step S42jAnd an input template
Figure GDA0002690266650000108
Substituting the formulas (9) and (10), respectively obtaining a memory matrix A 'and an offset V', and obtaining a multiple linear regression face picture recognition model.
Further, the specific steps of identifying the fingerprint picture and the face picture in step S5 are as follows:
s51: acquiring a fingerprint picture and a face picture, and respectively acquiring input matrixes of the fingerprint picture and the face picture;
s52: inputting the fingerprint picture input matrix into a multiple linear regression fingerprint picture identification model to obtain a model output matrix of the fingerprint picture, and inputting the face picture input matrix into the multiple linear regression face picture identification model to obtain a model output matrix of the face picture;
s53: matching the input matrix obtained in the step S51 with the model output matrix obtained in the step S52 to the fingerprint picture and the human face picture respectively;
s54: let the success rate of matching fingerprint pictures be H1The success rate of matching the face picture is H2Judging whether the identity verification matching degree H is larger than a matching set value H or not, wherein H is H1×H2H is 0-1; if so, the matching is successful, otherwise, the matching is failed. Wherein H is to be satisfied1≥h0，H2≥h0。
The invention has the beneficial effects that: the method combines the self-associative memory with the multivariate linear regression model, converts the fingerprint picture and the face picture into a series of parameters for storage, has high verification reliability because the identity information comprises the fingerprint picture and the face picture, has strong secrecy of the storage mode of the pictures and high safety factor, and effectively prevents the identity information of people from being leaked; the form of converting the picture into the parameters through the model is adopted, so that the method is simple and convenient, good in practicability, good in picture identification effect and good in face picture and fingerprint picture protection effect.
Drawings
FIG. 1 is a flow chart of a picture recognition method of the present invention;
FIG. 2 is a schematic diagram of a position parameter solution for a multiple linear regression fingerprint image recognition model;
FIG. 3 is a schematic diagram of a position parameter solution of a multiple linear regression face picture recognition model.
Detailed Description
The following provides a more detailed description of the embodiments and the operation of the present invention with reference to the accompanying drawings.
As shown in fig. 1, an identity coupling identification method based on a multiple linear regression association memory model includes the following steps:
s1: collecting fingerprint pictures and face pictures of people, and numbering the collected fingerprint pictures and face pictures in groups;
when identity information is stored, a fingerprint picture and a face picture of a user need to be acquired at the same time for double check.
If the matching success rate of one picture is lower than the H set value H, the identification of the identity information fails, the reliability of the identification system is high, and the storage reliability of the identity information is improved.
S2: processing the fingerprint picture and the face picture into binary pictures by setting a binary picture brightness threshold value to obtain an association memory input matrix and an output matrix of the fingerprint picture and an association memory input matrix and an output matrix of the face picture;
the binary image brightness threshold K is (0, 1,2,3.. 255); in the present embodiment, K is set to 100.
In step S1, M groups of pictures are included, the corresponding numbers are 1,2,3.. M, all the fingerprint pictures and the face pictures are composed of N rows and M columns of pixel points, and the number of the pixel points is N ═ nxm;
let the binary map matrix of the fingerprint be an input matrix ' ═ X ' of associative memory '1,X′2,…,X′i,…,X′m) I e {1,2, …, m }, represents an input vector consisting of all the pixels in the binary map of the ith fingerprint, wherein,
Figure GDA0002690266650000121
αj′irepresenting the input value of the jth pixel point in the binary image of the ith fingerprint;
setting binary image matrix of fingerprint as output matrix of associative memory
Figure GDA0002690266650000123
i is belonged to {1,2, …, m }, and represents that all pixel points in the binary image of the ith fingerprint consist ofOutput vector of, where Y'i＝(y1′i，y2′i，…，yj′i，…，yn′i)T，i∈{1，2，…，m}，j{1，2，…，n}，yj′iThe output value of the jth pixel point in the binary image of the ith fingerprint is represented;
the binary image matrix of the face is set as an input matrix ═ (X ″) of associative memory1,X″2,…,X″i,…,X″m) I ∈ {1,2, …, m }, which represents an input vector composed of all the pixel points in the binary image of the ith human face, wherein X ″, wherei＝(α1″i，α2″i，…，αj″i，…，αn″i)T，i∈{1，2，…，m}，j∈{1，2，…，n}，αj″iThe input value of the jth pixel point in the ith human face binary image is represented;
setting binary image matrix of human face as output matrix of associative memory
Figure GDA0002690266650000128
i belongs to {1,2, …, m }, and represents an output vector consisting of all pixel points in the binary image of the ith human face, wherein Y ″)i＝(y1″i，y2″i，…，yj″i，…，yn″i)T，i∈{1，2，…，m}，j∈{1，2，…，n}，yj″iAnd the output value of the jth pixel point in the ith human face binary image is represented.
In this embodiment, a black pixel (0) in the binary picture is mapped to-1, and a white pixel (255) in the binary picture is mapped to 1.
S3: constructing two multiple linear regression image recognition models with regression parameters by using a cellular neural network structure, wherein the two multiple linear regression image recognition models are a multiple linear regression fingerprint image recognition model and a multiple linear regression face image recognition model respectively;
in this embodiment, a multivariate linear regression fingerprint image recognition model and a multivariate linear face image recognition model are built based on a cellular neural network structure, specifically:
the multivariate linear regression fingerprint picture identification model constructed based on the cellular neural network structure specifically comprises the following steps:
Figure GDA0002690266650000131
wherein iN＝1,2，…,N；jM＝1,2，…,M；k1(iN,r)＝max{1-iN,-r}；k2(iN,r)＝min{N-iN,r}；l1(jM,r)＝max{1-jM,-r}；l2(jM,r)＝min{M-jMR }; r is a value representing the radius of the template,
Figure GDA0002690266650000132
in order to input the parameters, the user can select the parameters,
Figure GDA0002690266650000133
is the output of the computer system,
Figure GDA0002690266650000134
is an input to the computer system that is,
Figure GDA0002690266650000135
is the amount of the offset that is,
Figure GDA0002690266650000136
is an input template;
input template
Figure GDA0002690266650000137
The expression of (a) is as follows:
Figure GDA0002690266650000138
let r be 1, input the template
Figure GDA0002690266650000139
In the expression of
Figure GDA00026902666500001310
Will be a formula
Figure GDA0002690266650000141
The rewrite is:
Y′＝A′X′+V′
wherein, the input vector:
Figure GDA0002690266650000142
output vector
Figure GDA0002690266650000143
Offset amount
Figure GDA0002690266650000144
The memory matrix A 'and the offset V' are the regression parameters, and the memory matrix A 'is (a'ij)n×nThe following can be written:
Figure GDA0002690266650000145
wherein the content of the first and second substances,
Figure GDA0002690266650000146
Figure GDA0002690266650000151
Figure GDA0002690266650000152
the multivariate linear regression human face picture recognition model constructed based on the cellular neural network structure specifically comprises the following steps:
Figure GDA0002690266650000153
wherein iN＝1,2，…,N；jM＝1,2，…,M；k1(iN,r)＝max{1-iN,-r}；k2(iN,r)＝min{N-iN,r}；l1(jM,r)＝max{1-jM,-r}；l2(jM,r)＝min{M-jMR }; r is a value representing the radius of the template,
Figure GDA0002690266650000154
in order to input the parameters, the user can select the parameters,
Figure GDA0002690266650000155
is the output of the computer system,
Figure GDA0002690266650000156
is an input to the computer system that is,
Figure GDA0002690266650000157
is the amount of the offset that is,
Figure GDA0002690266650000158
is an input template;
input template
Figure GDA0002690266650000159
The expression of (a) is as follows:
Figure GDA00026902666500001510
let r be 1, then input the template
Figure GDA00026902666500001511
In the expression of
Figure GDA0002690266650000161
Formula (II)
Figure GDA0002690266650000162
The rewrite is:
Y″＝A″X″+V″
wherein, the input vector:
Figure GDA0002690266650000163
output vector
Figure GDA0002690266650000164
Offset amount
Figure GDA0002690266650000165
The memory matrix A ' and the offset V ' are the regression parameters, and the memory matrix A ' ═ aij)n×nThe following can be written:
Figure GDA0002690266650000166
wherein the content of the first and second substances,
Figure GDA0002690266650000171
Figure GDA0002690266650000172
Figure GDA0002690266650000173
s4: respectively calculating the regression parameters in the step S3 according to the associative memory input matrix and the output matrix of the fingerprint picture and the face picture obtained in the step S2 and the two multiple linear regression picture identification models obtained in the step S3, and finally determining a multiple linear regression fingerprint picture identification model and a multiple linear regression face picture identification model;
the specific steps of calculating the regression parameters in step S3 in step S4 are:
s41: order vector
Figure GDA0002690266650000174
Figure GDA0002690266650000175
Let Y ═ be ((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'iAnd Y ″)iRespectively representing column vectors formed by all pixel points in the ith fingerprint binary image and the face binary image;
let l ∈ {1,2, …, m } and q ∈ {1,2, …, N },
Figure GDA0002690266650000181
Figure GDA0002690266650000182
Figure GDA0002690266650000183
Figure GDA0002690266650000184
Figure GDA0002690266650000191
Figure GDA0002690266650000192
then it can be derived from the formula Y '═ a' X '+ V':
X′·L′＝Y'
available from the formula Y ═ a "X" + V ":
X″·L″＝Y″
therefore, the temperature of the molten metal is controlled,
Figure GDA0002690266650000193
Figure GDA0002690266650000194
wherein L 'and L' are constants;
s42: among the self-associative memory criteria, there are:
Figure GDA0002690266650000195
Figure GDA0002690266650000196
converting the fingerprint picture associative memory input matrix' obtained in the step S2 intoX', output matrix
Figure GDA0002690266650000197
Converting into Y ', and substituting into a formula X ', L ', or Y ', to obtain L '; converting the face picture associative memory input matrix 'obtained in the step S2 into X', and outputting the matrix
Figure GDA0002690266650000201
The conversion is Y ", and the formula X ″, L ═ Y ″, is substituted to obtain L". Deriving the offset v 'from L' and the input template
Figure GDA0002690266650000202
Deriving the offset v 'from L' and the input template
Figure GDA0002690266650000203
S43: converting the offset v 'obtained in step S42'jAnd an input template
Figure GDA0002690266650000204
Substituting into formula
Figure GDA0002690266650000205
Figure GDA0002690266650000206
Obtaining a memory matrix A 'and an offset V', and obtaining a multiple linear regression fingerprint image identification model;
the offset v ″' obtained in step S42jAnd an input template
Figure GDA0002690266650000207
Substituting into formula
Figure GDA0002690266650000208
Figure GDA0002690266650000209
Respectively obtaining a memory matrix A 'and an offset V', and obtaining a multiple linear regression face picture recognition model.
S5: and respectively identifying the fingerprint picture and the face picture based on a self-association memory criterion. The method specifically comprises the following steps:
s51: acquiring a fingerprint picture and a face picture, and respectively acquiring input matrixes of the fingerprint picture and the face picture;
s52: inputting the fingerprint picture input matrix into a multiple linear regression fingerprint picture identification model to obtain a model output matrix of the fingerprint picture, and inputting the face picture input matrix into the multiple linear regression face picture identification model to obtain a model output matrix of the face picture;
s53: matching the input matrix obtained in the step S51 with the model output matrix obtained in the step S52 to the fingerprint picture and the human face picture respectively;
s54: let the success rate of matching fingerprint pictures be H1The success rate of matching the face picture is H2Judging whether the identity verification matching degree H is larger than a matching set value H or not, wherein H is H1×H2H is 0-1; if so, matching is successful, otherwise, matching fails, wherein H is also required to be satisfied1≥h0，H2≥h0Wherein h is0＝0.92
In the present embodiment, h is 0.9, h0＝0.92
It should be noted that the above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and those skilled in the art may make variations, modifications, additions or substitutions within the spirit and scope of the present invention.

Claims (3)

1. An identity coupling identification method based on a multiple linear regression association memory model is characterized by comprising the following steps:
s1: collecting fingerprint pictures and face pictures of people, and numbering the collected fingerprint pictures and face pictures in groups;
s2: processing all the fingerprint pictures and the face pictures obtained in the step S1 into binary pictures by setting a binary picture brightness threshold value to obtain an association memory input matrix and an output matrix of the fingerprint pictures and an association memory input matrix and an output matrix of the face pictures;
the binary image brightness threshold K is (0, 1,2,3, ·, 255);
in step S1, M groups of pictures are included, the corresponding numbers are 1,2,3.. M, all the fingerprint pictures and the face pictures are composed of N rows and M columns of pixel points, and the number of the pixel points is N ═ nxm;
let the binary map matrix of the fingerprint be an input matrix ' ═ X ' of associative memory '1,X′2,…,X′i,…,X′m) I e {1,2, …, m }, represents an input vector consisting of all the pixels in the binary map of the ith fingerprint, wherein,
Figure FDA0002690266640000011
αj′irepresenting the input value of the jth pixel point in the binary image of the ith fingerprint;
setting binary image matrix of fingerprint as output matrix of associative memory
Figure FDA0002690266640000012
Figure FDA0002690266640000013
And an output vector composed of all pixel points in the binary image representing the ith fingerprint, wherein,
Figure FDA0002690266640000014
Figure FDA0002690266640000015
the output value of the jth pixel point in the binary image of the ith fingerprint is represented;
the binary image matrix of the face is set as an input matrix ═ (X ″) of associative memory1,X″2,…,X″i,…,X″m) I e {1,2, …, m }, represents an input vector composed of all the pixels in the binary image of the ith face, wherein,
Figure FDA0002690266640000021
αj″ithe input value of the jth pixel point in the ith human face binary image is represented;
setting binary image matrix of human face as output matrix of associative memory
Figure FDA0002690266640000023
Figure FDA0002690266640000024
And an output vector composed of all pixel points in the binary image representing the ith human face, wherein,
Figure FDA0002690266640000025
yj″ithe output value of the jth pixel point in the ith human face binary image is represented;
s3: constructing two multiple linear regression image recognition models with regression parameters by using a cellular neural network structure, wherein the two multiple linear regression image recognition models are a multiple linear regression fingerprint image recognition model and a multiple linear regression face image recognition model respectively;
the specific content of step S3 is:
the multivariate linear regression fingerprint picture identification model constructed based on the cellular neural network structure specifically comprises the following steps:
Figure FDA0002690266640000027
wherein iN＝1,2，…,N；jM＝1,2，…,M；k1(iN,r)＝max{1-iN,-r}；k2(iN,r)＝min{N-iN,r}；l1(jM,r)＝max{1-jM,-r}；l2(jM,r)＝min{M-jMR }; r is a value representing the radius of the template,
Figure FDA0002690266640000028
in order to input the parameters, the user can select the parameters,
Figure FDA0002690266640000029
is the output of the computer system,
Figure FDA00026902666400000210
is an input to the computer system that is,
Figure FDA00026902666400000211
is the amount of the offset that is,
Figure FDA00026902666400000212
is an input template;
input template
Figure FDA00026902666400000213
The expression of (a) is as follows:
Figure FDA00026902666400000214
let r be 1, then in equation (2)
Figure FDA0002690266640000031
Formula (1) is rewritten as:
Y′＝A′X′+V′ (3)
wherein, the input vector:
Figure FDA0002690266640000032
output vector
Figure FDA0002690266640000033
Offset amount
Figure FDA0002690266640000034
The memory matrix A 'and the offset V' are the regression parameters, and the memory matrix A 'is (a'ij)n×nThe following can be written:
Figure FDA0002690266640000035
wherein the content of the first and second substances,
Figure FDA0002690266640000036
Figure FDA0002690266640000041
Figure FDA0002690266640000042
the multivariate linear regression human face picture recognition model constructed based on the cellular neural network structure specifically comprises the following steps:
Figure FDA0002690266640000043
wherein iN＝1,2，…,N；jM＝1,2，…,M；k1(iN,r)＝max{1-iN,-r}；k2(iN,r)＝min{N-iN,r}；l1(jM,r)＝max{1-jM,-r}；l2(jM,r)＝min{M-jMR }; r is a value representing the radius of the template,
Figure FDA0002690266640000044
in order to input the parameters, the user can select the parameters,
Figure FDA0002690266640000045
is the output of the computer system,
Figure FDA0002690266640000046
is an input to the computer system that is,
Figure FDA0002690266640000047
is the amount of the offset that is,
Figure FDA0002690266640000048
is an input template;
input template
Figure FDA0002690266640000049
The expression of (a) is as follows:
Figure FDA00026902666400000410
let r be 1, then in equation (7)
Figure FDA0002690266640000051
Equation (6) is rewritten as:
Y″＝A″X″+V″ (8)
wherein, the input vector:
Figure FDA0002690266640000052
output vector
Figure FDA0002690266640000053
Offset amount
Figure FDA0002690266640000054
The memory matrix A 'and the offset V' are the regression parameters, the memory matrix
Figure FDA0002690266640000055
The following can be written:
Figure FDA0002690266640000056
wherein the content of the first and second substances,
Figure FDA0002690266640000057
Figure FDA0002690266640000061
Figure FDA0002690266640000062
s4: respectively calculating the regression parameters in the step S3 according to the associative memory input matrix and the output matrix of the fingerprint picture and the face picture obtained in the step S2 and the two multiple linear regression picture identification models obtained in the step S3, and finally determining a multiple linear regression fingerprint picture identification model and a multiple linear regression face picture identification model;
the specific steps of calculating the regression parameters in step S3 in step S4 are:
s41: order directionMeasurement of
Figure FDA0002690266640000063
Figure FDA0002690266640000064
Let Y ═ be ((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'iAnd Y ″)iRespectively representing column vectors formed by all pixel points in the ith fingerprint binary image and the face binary image;
let l ∈ {1,2, …, m } and q ∈ {1,2, …, N },
Figure FDA0002690266640000071
Figure FDA0002690266640000072
Figure FDA0002690266640000073
Figure FDA0002690266640000074
Figure FDA0002690266640000081
Figure FDA0002690266640000082
then from equation (3) one can derive:
X′·L′＝Y' (11)
equation (8) can be found:
X″·L″＝Y″ (12)
therefore, the temperature of the molten metal is controlled,
Figure FDA0002690266640000083
Figure FDA0002690266640000084
wherein L 'and L' are constants;
s42: among the self-associative memory criteria, there are:
Figure FDA0002690266640000085
Figure FDA0002690266640000086
converting the fingerprint picture associative memory input matrix 'obtained in the step S2 into X', and outputting the matrix
Figure FDA0002690266640000087
Converting into Y ', substituting into a formula (11) to obtain L'; converting the face picture associative memory input matrix 'obtained in the step S2 into X', and outputting the matrix
Figure FDA0002690266640000091
Converting into Y 'and substituting into a formula (12) to obtain L'; deriving the offset v 'from L' and the input template
Figure FDA0002690266640000092
Deriving the offset v 'from L' and the input template
Figure FDA0002690266640000093
S43: converting the offset v 'obtained in step S42'jAnd an input template
Figure FDA0002690266640000094
Substituting the formula (4) and the formula (5) to obtain a memory matrix A 'and an offset V', and obtaining a multiple linear regression fingerprint image identification model;
the offset v ″' obtained in step S42jAnd an input template
Figure FDA0002690266640000095
Substituting the formula (9) and the formula (10), and respectively obtaining a memory matrix A 'and an offset V', and obtaining a multiple linear regression face picture recognition model;
s5: respectively identifying the fingerprint picture and the face picture based on a self-association memory criterion;
the method combines the self-associative memory with the multivariate linear regression model, converts the fingerprint picture and the face picture into a series of parameters for storage, has high verification reliability because the identity information comprises the fingerprint picture and the face picture, has strong secrecy of the storage mode of the pictures and high safety factor, and effectively prevents the identity information of people from being leaked; the form of converting the picture into the parameters through the model is adopted, so that the method is simple and convenient, good in practicability, good in picture identification effect and good in face picture and fingerprint picture protection effect.
2. The identity coupling identification method based on the multiple linear regression association memory model as claimed in claim 1, wherein: when the identity authentication is performed, a fingerprint picture and a face picture of a user need to be acquired at the same time for double check.
3. The identity coupling identification method based on the multiple linear regression association memory model according to claim 1 or 2, wherein the specific steps of identifying the fingerprint picture and the face picture in the step S5 are as follows:
s51: acquiring a fingerprint picture and a face picture, and respectively acquiring input matrixes of the fingerprint picture and the face picture;
s52: inputting the fingerprint picture input matrix into a multiple linear regression fingerprint picture identification model to obtain a model output matrix of the fingerprint picture, and inputting the face picture input matrix into the multiple linear regression face picture identification model to obtain a model output matrix of the face picture;
s53: matching the input matrix obtained in the step S51 with the model output matrix obtained in the step S52 to the fingerprint picture and the human face picture respectively;
s54: let the success rate of matching fingerprint pictures be H1The success rate of matching the face picture is H2Judging whether the identity verification matching degree H is larger than a matching set value H or not, wherein H is H1×H2H is 0-1; if so, the matching is successful, otherwise, the matching is failed.
CN201710175033.3A 2017-03-22 2017-03-22 Identity coupling identification method based on multiple linear regression association memory model Expired - Fee Related CN106971157B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710175033.3A CN106971157B (en) 2017-03-22 2017-03-22 Identity coupling identification method based on multiple linear regression association memory model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710175033.3A CN106971157B (en) 2017-03-22 2017-03-22 Identity coupling identification method based on multiple linear regression association memory model

Publications (2)

Publication Number Publication Date
CN106971157A CN106971157A (en) 2017-07-21
CN106971157B true CN106971157B (en) 2020-12-04

Family

ID=59329538

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710175033.3A Expired - Fee Related CN106971157B (en) 2017-03-22 2017-03-22 Identity coupling identification method based on multiple linear regression association memory model

Country Status (1)

Country Link
CN (1) CN106971157B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108399375B (en) * 2018-02-07 2020-10-13 厦门瑞为信息技术有限公司 Identity recognition method based on associative memory
CN113223356B (en) * 2021-05-13 2022-12-13 深圳市技成科技有限公司 Skill training and checking system for PLC control technology

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1211770A (en) * 1998-03-27 1999-03-24 杨振宁 Human body physiological characteristics antifake verification method correlated with identification card mobile data base
US7996156B2 (en) * 2002-03-07 2011-08-09 The United States Of America As Represented By The Secretary, Department Of Health And Human Services Methods for predicting properties of molecules
US9042606B2 (en) * 2006-06-16 2015-05-26 Board Of Regents Of The Nevada System Of Higher Education Hand-based biometric analysis
US8660321B2 (en) * 2008-11-19 2014-02-25 Nec Corporation Authentication system, apparatus, authentication method, and storage medium with program stored therein
CN101576443B (en) * 2009-06-16 2011-01-05 北京航空航天大学 Life prediction method of accelerated life test based on grey RBF neural network
CN101777131B (en) * 2010-02-05 2012-05-09 西安电子科技大学 Method and device for identifying human face through double models
CN105787420B (en) * 2014-12-24 2020-07-14 北京三星通信技术研究有限公司 Method and device for biometric authentication and biometric authentication system
CN104809450B (en) * 2015-05-14 2018-01-26 郑州大学 Wrist vena identification system based on online extreme learning machine
CN106096560A (en) * 2016-06-15 2016-11-09 广州尚云在线科技有限公司 A kind of face alignment method

Also Published As

Publication number Publication date
CN106971157A (en) 2017-07-21

Similar Documents

Publication Publication Date Title
US10713532B2 (en) Image recognition method and apparatus
US20220029799A1 (en) System and method for creating one or more hashes for biometric authentication in real-time
US20180330179A1 (en) System and method for biometric identification
US8542886B2 (en) System for secure face identification (SCIFI) and methods useful in conjunction therewith
US20170262472A1 (en) Systems and methods for recognition of faces e.g. from mobile-device-generated images of faces
CN110751025A (en) Business handling method, device, equipment and medium based on face recognition
CN111860147A (en) Pedestrian re-identification model optimization processing method and device and computer equipment
WO2019047567A1 (en) Service provision method, device, storage medium and computing apparatus
CN111339897B (en) Living body identification method, living body identification device, computer device, and storage medium
CN110795714A (en) Identity authentication method and device, computer equipment and storage medium
WO2023071812A1 (en) Biometric extraction method and device for secure multi‑party computation system
WO2020172870A1 (en) Method and apparatus for determining motion trajectory of target object
CN112330331A (en) Identity verification method, device and equipment based on face recognition and storage medium
CN112001285B (en) Method, device, terminal and medium for processing beauty images
CN112231038B (en) Work order information display method, device, computer equipment and storage medium
CN106971157B (en) Identity coupling identification method based on multiple linear regression association memory model
US20090316960A1 (en) Mobile electronic device security protecting system and method
CN108875611B (en) Video motion recognition method and device
CN113420665A (en) Method, device and equipment for generating confrontation face image and training face recognition model
CN117275138A (en) Identity authentication method, device, equipment and storage medium based on automatic teller machine
WO2024045421A1 (en) Image protection method and related device
CN106980833B (en) Face recognition method based on multivariate linear regression association memory
CN116824676A (en) Digital identity information generation method, application method, device, system and equipment
CN117037244A (en) Face security detection method, device, computer equipment and storage medium
CN109450878B (en) Biological feature recognition method, device and system

Legal Events

Date Code Title Description
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
CF01 Termination of patent right due to non-payment of annual fee
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20201204