CN117542086A - Palm print palm vein multi-mode identity authentication method, device, storage medium and equipment - Google Patents
Palm print palm vein multi-mode identity authentication method, device, storage medium and equipment Download PDFInfo
- Publication number
- CN117542086A CN117542086A CN202210914591.8A CN202210914591A CN117542086A CN 117542086 A CN117542086 A CN 117542086A CN 202210914591 A CN202210914591 A CN 202210914591A CN 117542086 A CN117542086 A CN 117542086A
- Authority
- CN
- China
- Prior art keywords
- palm
- comparison
- vein
- palm print
- score
- 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.)
- Pending
Links
- 210000003462 vein Anatomy 0.000 title claims abstract description 440
- 238000000034 method Methods 0.000 title claims abstract description 82
- 238000003860 storage Methods 0.000 title claims abstract description 15
- 238000012549 training Methods 0.000 claims abstract description 128
- 238000009826 distribution Methods 0.000 claims abstract description 85
- 230000004927 fusion Effects 0.000 claims abstract description 48
- 239000011159 matrix material Substances 0.000 claims description 43
- 238000010606 normalization Methods 0.000 claims description 30
- 238000000513 principal component analysis Methods 0.000 claims description 26
- 238000004364 calculation method Methods 0.000 claims description 19
- 230000015654 memory Effects 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 8
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 230000000694 effects Effects 0.000 abstract description 22
- 238000005457 optimization Methods 0.000 abstract description 5
- 230000009467 reduction Effects 0.000 abstract description 5
- 230000008569 process Effects 0.000 description 9
- 238000000605 extraction Methods 0.000 description 7
- 230000006872 improvement Effects 0.000 description 7
- 230000007547 defect Effects 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 102000001554 Hemoglobins Human genes 0.000 description 2
- 108010054147 Hemoglobins Proteins 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012847 principal component analysis method Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 241001391944 Commicarpus scandens Species 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 208000007502 anemia Diseases 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000002203 pretreatment Methods 0.000 description 1
- 210000004243 sweat Anatomy 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
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
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
-
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/809—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of classification results, e.g. where the classifiers operate on the same input data
-
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/817—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level by voting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/95—Pattern authentication; Markers therefor; Forgery detection
-
- 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
- G06V40/14—Vascular patterns
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Vascular Medicine (AREA)
- Collating Specific Patterns (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The application discloses a palm print palm vein multi-mode identity authentication method, a palm print palm vein multi-mode identity authentication device, a storage medium and a device, and belongs to the field of biological feature recognition. According to the method, the palm print comparison result and the palm vein comparison result are fused through the set palm print comparison weight and the palm vein comparison weight, and identity authentication is carried out according to the decision fusion result obtained through fusion. The palm print comparison weight/palm vein comparison weight is determined according to the distribution of palm print/palm vein comparison scores of the palm print/palm vein training sample set and the rejection rate of the palm print/palm vein training sample set. The distribution of the palm print/palm vein comparison score can increase the distribution difference of the palm print/palm vein comparison score of the same person and the palm print/palm vein comparison score of different persons, thereby improving the authentication classification effect. The rejection rate may be used to guide the generation of weights such that optimization of weights proceeds toward a target direction that facilitates a reduction in rejection rate. The method and the device improve the accuracy of multi-mode recognition of palm print and palm veins.
Description
Technical Field
The application relates to the field of biological feature recognition, in particular to a palm print palm vein multi-mode identity authentication method, a device, a storage medium and equipment.
Background
Palm print recognition and palm vein recognition are important biological feature recognition methods, and have more applications, particularly in scenes such as higher user data privacy level or contactless recognition, and the like. The palm print identification and collection is to carry out coding identification on the line information of three main palm prints and the line information of some disordered palm prints on the palm. Palm vein recognition is performed by illuminating the palm portion with near infrared light, absorbing near infrared light by hemoglobin, and transmitting the light source through the avascular portion.
Palm print recognition uses lines on the surface of the palm, and the palm lines are exposed for a long time, so that the palm lines are easy to break and the like, and the collection of palm print images can be influenced. In addition, the palm sweat, wetness and other problems can also affect the palm print acquisition and recognition effect. Similar to fingerprint identification, living detection of palmprints is a great difficulty, and gray products and black products can forge palmprints easily to attack palmprint authentication systems.
The palm vein identification is to absorb near infrared light for imaging by hemoglobin, so that the palm vein has a good living body detection effect and cannot be damaged. Furthermore, palm perspiration does not affect imaging of the palmar veins. However, for some anemic people, the palm vein image is difficult to acquire, and the recognition effect of the palm vein image can be affected.
The biometric identification is currently mostly a single-mode scheme, the multi-mode biometric identification by utilizing different biometric features is an emerging development direction, and in many applications of the multi-mode biometric identification, the identification results are connected in series or in parallel, namely, the palm print and the palm vein are authenticated and then considered to be successful, or one authentication is carried out, namely, the palm print and the palm vein are authenticated and then considered to be successful.
The palm print and palm vein multi-mode identification method simply combines two identification results, cannot realize the complementation of the palm print and the palm vein, cannot better solve the defects of palm print identification and palm vein identification, and cannot improve the accuracy of palm print and palm vein identification.
Disclosure of Invention
In order to overcome the defects of the prior art, the application provides a palm print palm vein multi-mode identity authentication method, device, storage medium and equipment, and the accuracy of palm print palm vein multi-mode identification is improved.
The technical scheme provided by the application is as follows:
in a first aspect, the present application provides a palmprint palmvein multimodal identity authentication method, the method comprising:
acquiring two groups of biological characteristic data, wherein each group of biological characteristic data comprises palm print characteristic data and palm vein characteristic data;
comparing the two palm print characteristic data in the two groups of biological characteristic data to obtain a palm print comparison score; comparing the two palm vein feature data in the two groups of biological feature data to obtain palm vein comparison scores;
determining a palm print comparison result according to the palm print comparison score and a set palm print comparison threshold, determining a palm vein comparison result according to the palm vein comparison score and the set palm vein comparison threshold, and carrying out weighted voting on the palm print comparison result and the palm vein comparison result according to a set palm print comparison weight and a palm vein comparison weight to obtain a decision fusion result;
the palm print comparison weight is determined according to the distribution of the palm print comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set; the palm vein comparison weight is determined according to the distribution of palm vein comparison scores of a palm vein training sample set and the rejection rate of the palm vein training sample set;
And judging whether the two groups of biological characteristic data belong to the same person according to the decision fusion result.
Further, the palm print comparison weight is determined according to the distribution of the comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set, and includes:
the palm print training samples in the palm print training sample set are compared pairwise to obtain a palm print comparison score set; wherein the palm print comparison score set comprises palm print comparison scores of the same person and palm print comparison scores of different persons;
normalizing the palm print comparison score set according to a non-overlapping region of the distribution of the palm print comparison scores of the same person and the distribution of the palm print comparison scores of different persons in the palm print comparison score set;
calculating the palm print comparison weight according to the normalized palm print comparison score set and the rejection rate of the palm print training sample set;
and/or;
the palm vein comparison weight is determined according to the distribution of the comparison scores of the palm vein training sample set and the rejection rate of the palm vein training sample set, and comprises the following steps:
performing pairwise comparison on the palm vein training samples in the palm vein training sample set to obtain a palm vein comparison score set; wherein the palm vein comparison score set comprises palm vein comparison scores of the same person and palm vein comparison scores of different persons;
Normalizing the palm vein comparison score set according to a non-overlapping region of the distribution of the palm vein comparison scores of the same person and the distribution of the palm vein comparison scores of different persons in the palm vein comparison score set;
and calculating the palm vein comparison weight according to the normalized palm vein comparison score set and the rejection rate of the palm vein training sample set.
Further, normalizing the palm print comparison score set according to a non-overlapping region of a distribution of the palm print comparison scores of the same person and a distribution of the palm print comparison scores of different persons in the palm print comparison score set includes:
calculating a non-overlapping region of the distribution of the palm print comparison score of the same person and the distribution of the palm print comparison score of different persons as a palm print normalized dimension by the formula (1);
normalizing the palm print comparison score set according to the palm print normalization dimension and by formula (2);
and/or;
normalizing the set of palm vein comparison scores according to a non-overlapping region of a distribution of palm vein comparison scores of the same person and a distribution of palm vein comparison scores of different persons in the set of palm vein comparison scores, comprising:
calculating a non-overlapping region of the distribution of the palm vein comparison score of the same person and the distribution of the palm vein comparison score of different persons as palm vein normalization dimensions by the formula (1);
Normalizing the palm vein comparison score set according to the palm vein normalization dimension and through a formula (2);
wherein:
q(k)=μ a (k)-min a (k)+max r (k)-μ r (k) (1)
k=1, 2, k=1 denotes a palm print recognition modality, k=2 denotes a palm vein recognition modality, q (k) is the palm print normalized dimension/palm vein normalized dimension, max r (k) For the maximum value of the palm print comparison score/palm vein comparison score, mu for different people in the palm print comparison score/palm vein comparison score set r (k) Average value of palm print comparison score/palm vein comparison score for different people in the palm print comparison score/palm vein comparison score, min a (k) For the palm print comparison score/palm vein comparison score for the same person in the palm print comparison score/palm vein comparison score,μ a (k) An average of the palm print comparison score/palm vein comparison score for the same person in the palm print comparison score/palm vein comparison score set;
the i-th palmprint alignment score/palmar vein alignment score in the palmar vein alignment score set before normalization, ++>For the i-th palmprint/palmar vein score of the normalized palmprint/palmar vein score, i=1, 2,3, …, N k ,N k For the total number of palm print score/palm vein score in the palm print score/palm vein score set, j=0 represents +.>Palm print alignment score/palm vein alignment score for different people, j=1 means +.>Palmprint alignment score/palmvein alignment score for the same person.
Further, the calculating the palm print comparison weight according to the normalized palm print comparison score set and the rejection rate of the palm print training sample set includes:
calculating the palmprint comparison weight by the formula (3);
and/or;
the calculating the palm vein comparison weight according to the normalized palm vein comparison score set and the rejection rate of the palm vein training sample set comprises the following steps:
calculating the palm vein comparison weight by the formula (3);
wherein:
w (k) is the palm print comparison weight/the palm vein comparison weight;
FRR k and the rejection rate of the palm print training sample set/palm vein training sample set is determined. Is that
Further, the palm print comparison result is determined according to the palm print comparison score and a set palm print comparison threshold, the palm vein comparison result is determined according to the palm vein comparison score and a set palm vein comparison threshold, and the palm print comparison result and the palm vein comparison result are weighted and voted according to a set palm print comparison weight and a palm vein comparison weight, so as to obtain a decision fusion result, which comprises:
The decision fusion result is obtained through calculation according to the following formula;
wherein, delta is the decision fusion result, delta (k) is the palm print comparison result/palm vein comparison result;
s (k) is the palm print comparison score/palm vein comparison score, and T (k) is the palm print comparison threshold/palm vein comparison threshold.
Further, the acquiring two sets of biometric data, wherein each set of biometric data includes palm print feature data and palm vein feature data, includes:
acquiring a palm image, detecting key points of the palm image, and extracting an interested region from the palm image according to the detected key points;
extracting palm print characteristics and palm vein characteristics of the region of interest to obtain the line information of three main palm prints of the palm and the palm vein characteristic data;
respectively carrying out principal component analysis on the line information of three main palmprints of the palm;
wherein, l represents the serial numbers of three main palmprints of the palm, and x l Representing the respective texture information, z, of three main palmprints of the palm l Representing the characteristics obtained by respectively carrying out principal component analysis on the line information of three main palmprints of the palm, mu l Mean value, W, of each line information representing three main palm prints of palm l Representing the principal component analysis matrix of each of three main palmprints of the palm; mu (mu) l And W is l Training by respective training sets of three main palmprints of the palm;
the characteristic obtained after the main component analysis is respectively carried out on the line information of three main palm lines of the palm is fused through the following formula, so that the palm line characteristic data are obtained;
wherein Z is the palm print characteristic data, P o 、P 1 And P 2 For projection matrix, P o 、P 1 And P 2 Respectively training the three main palmprint training sets of the palm.
Further, the mu is obtained by training as follows l 、W l 、P o 、P 1 And P 2 :
The average value mu of each line information of three main palm lines of the palm is calculated by the following formula l ;
Wherein,n=1, 2,3, …, N being the total number of the texture samples in the training set for each of the three primary palmprints of the palm;
the covariance matrix S of each of three main palmprints of the palm is calculated by the following formula l ;
Respectively carrying out eigenvalue decomposition on covariance matrixes of three main palmprints of the palm through the following formula;
wherein,and->Respectively S l T=0, 1,2, …, d, d is S l Row and column numbers of (a);
according to the set percentage of the main component Selecting the first r eigenvectors with the duty ratio of M% to form the principal component analysis matrix W l ;
Wherein,
the main component analysis is respectively carried out on the line information of three main palm lines of the palm through the following formulas;
wherein,representing the characteristics obtained by respectively carrying out principal component analysis on the line information of three main palmprints of the palm;
the projection matrix P is optimized and calculated by the following formula o 、P 1 And P 2 ;
Wherein, respectively Z 0 、Z 1 、Z 2 Intra-class variance matrix of->Is Z 0 And Z 1 Is a matrix of inter-class variances of (c),is Z 0 And Z 2 Inter-class variance matrix of->Is Z 1 And Z 2 Inter-class variance matrix of (a).
In a second aspect, the present application provides a palmprint palmvein multimodal identity authentication device, the device comprising:
the data acquisition module is used for acquiring two groups of biological characteristic data, wherein each group of biological characteristic data comprises palm print characteristic data and palm vein characteristic data;
the comparison score calculation module is used for comparing the two palm print characteristic data in the two groups of biological characteristic data to obtain a palm print comparison score; comparing the two palm vein feature data in the two groups of biological feature data to obtain palm vein comparison scores;
the decision fusion module is used for determining a palm print comparison result according to the palm print comparison score and a set palm print comparison threshold, determining a palm vein comparison result according to the palm vein comparison score and the set palm vein comparison threshold, and carrying out weighted voting on the palm print comparison result and the palm vein comparison result according to a set palm print comparison weight and a palm vein comparison weight to obtain a decision fusion result;
The palm print comparison weight is determined according to the distribution of the palm print comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set; the palm vein comparison weight is determined according to the distribution of palm vein comparison scores of a palm vein training sample set and the rejection rate of the palm vein training sample set;
and the multi-mode authentication module is used for judging whether the two groups of biological characteristic data belong to the same person according to the decision fusion result.
In a third aspect, the present application provides a computer readable storage medium for palm print palm vein multimodal identity authentication, comprising a memory for storing processor executable instructions which when executed by the processor implement steps comprising the palm print palm vein multimodal identity authentication method of the first aspect.
In a fourth aspect, the present application provides an apparatus for palm print palm vein multimodal identity authentication, comprising at least one processor and a memory storing computer executable instructions, which when executed by the processor implement the steps of the palm print palm vein multimodal identity authentication method of the first aspect.
The application has the following beneficial effects:
The application provides a palm print palm vein multi-mode identity authentication method which is used for solving the palm print palm vein identity authentication problem under different scenes. According to the method, the palm print comparison result and the palm vein comparison result are fused through the set palm print comparison weight and the palm vein comparison weight, and identity authentication is carried out according to the decision fusion result obtained through fusion. The palm print comparison weight/palm vein comparison weight is determined according to the distribution of palm print/palm vein comparison scores of the palm print/palm vein training sample set and the rejection rate of the palm print/palm vein training sample set. The distribution of the palm print/palm vein comparison score can increase the distribution difference of the palm print/palm vein comparison score of the same person and the palm print/palm vein comparison score of different persons, thereby improving the authentication classification effect. The rejection rate may be used to guide the generation of weights such that optimization of weights proceeds toward a target direction that facilitates a reduction in rejection rate.
The decision fusion result after fusion is used for identity authentication, so that the complementation of palmprint and palmvein can be realized, the defects of palmprint recognition and palmvein recognition are well overcome, the accuracy of palmprint palmvein multi-modal recognition is improved, and the result of palmprint palmvein multi-modal identity authentication is superior to the recognition effect of any palmprint and palmvein single modality, and has larger performance improvement. In addition, the method and the device are integrated in a weighted voting mode, the calculation is simple, and the overall speed is not reduced compared with that of the palm print palm vein multi-mode identity authentication method in the prior art.
Drawings
FIG. 1 is a flow chart of a palmprint palmvein multimodal identity authentication method of the present application;
FIG. 2 is a schematic diagram of a pretreatment process of palm print palmar veins;
FIG. 3 is a statistical distribution diagram of palm print/palm vein comparison scores of the same person and palm print/palm vein comparison scores of different persons;
fig. 4 is a schematic diagram of a palmprint palmvein multi-mode identity authentication device of the present application.
Detailed Description
In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the technical solutions of the present application will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present application. The components of the embodiments of the present application, which are generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, as provided in the accompanying drawings, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, are intended to be within the scope of the present application.
The embodiment of the application provides a palmprint palmvein multi-mode identity authentication method, as shown in fig. 1, which comprises the following steps:
s100: two sets of biometric data are acquired, wherein each set of biometric data includes palm print feature data and palm vein feature data.
The purpose of acquiring the two groups of palm print characteristic data and palm vein characteristic data in the step is to perform palm print palm vein multi-mode identity authentication through the two groups of palm print characteristic data and palm vein characteristic data and judge whether the two groups of biological characteristic data come from the same person.
The method for acquiring the palm print feature data and the palm vein feature data is not limited, and palm print feature extraction and palm vein feature extraction can be performed on palm images through methods such as image processing, machine learning or deep learning, so that the palm print feature data and the palm vein feature data are obtained.
S200: comparing the two palm print characteristic data in the two groups of biological characteristic data to obtain a palm print comparison score; and comparing the two palm vein feature data in the two groups of biological feature data to obtain palm vein comparison scores.
The palm print comparison score (or palm vein comparison score) reflects the similarity of the two palm print feature data (or palm vein feature data), and the higher the similarity, the more likely the two palm print feature data (or palm vein feature data) belong to the same person.
The present application is not limited to the manner in which the palm print comparison score (or palm vein comparison score) is calculated, and in some examples, when the palm print comparison score (or palm vein comparison score) is greater, it means that the higher the similarity of the two palm print feature data (or palm vein feature data) is, the more likely that the two palm print feature data (or palm vein feature data) belong to the same person; in other examples, the smaller the palm print alignment score (or palm vein alignment score), the higher the similarity that the two palm print feature data (or palm vein feature data) are represented, the more likely the two palm print feature data (or palm vein feature data) belong to the same person.
For example, the euclidean distance of the two palm print feature data (or palm vein feature data) may be calculated to obtain a palm print comparison score (or palm vein comparison score), and at this time, the smaller the palm print comparison score (or palm vein comparison score), the higher the similarity of the two palm print feature data (or palm vein feature data) is.
The palmprint alignment score and the palmvein alignment score may be denoted as s (k), k=1, 2, k=1 representing a palmprint recognition modality, and k=2 representing a palmvein recognition modality.
S300: and determining a palm print comparison result according to the palm print comparison score and a set palm print comparison threshold, determining a palm vein comparison result according to the palm vein comparison score and the set palm vein comparison threshold, and carrying out weighted voting on the palm print comparison result and the palm vein comparison result according to the set palm print comparison weight and the palm vein comparison weight to obtain a decision fusion result.
One specific implementation mode of the steps is as follows:
taking the palm print comparison as an example (k=1), when the palm print comparison score is larger, the higher the similarity of the two palm print feature data is, if the palm print comparison score s (k) is larger than the palm print comparison threshold T (k), the two palm print feature data are considered to belong to the same person, the palm print comparison result delta (k) is assigned with a mark 1 (or other numerical value), otherwise, the two palm print feature data are considered to belong to different persons, and the palm print comparison result delta (k) is assigned with a mark 0 (or other numerical value). For the palmar vein alignment, the same is true for the case of k=2.
Another specific implementation manner of the step is as follows:
taking the palm print comparison as an example (k=1), when the palm print comparison score is smaller, the similarity of the two palm print characteristic data is higher, if the palm print comparison score s (l) is smaller than the palm print comparison threshold T (k), the two palm print characteristic data are considered to belong to the same person, the palm print comparison result delta (k) is assigned with a mark 1 (or other numerical value), otherwise, the two palm print characteristic data are considered to belong to different persons, and the palm print comparison result delta (k) is assigned with a mark 0 (or other numerical value). For the palmar vein alignment, the same is true for the case of k=2.
And then carrying out decision fusion on the assigned palmprint comparison result and the palmvein comparison result delta (k), wherein the result is actually a weighted voting, and a final decision fusion result is obtained. The decision fusion result is a result of weighted voting on the palm print comparison result and the palm vein comparison result, and reflects the roles of the palm print comparison result and the palm vein comparison result in final authentication and the proportion of the palm print comparison result and the palm vein comparison result in the final authentication.
When the weighted voting is carried out, the palm print comparison result and the palm vein comparison result are respectively provided with respective palm print comparison weights and palm vein comparison weights; the palm print comparison weight is determined according to the distribution of the palm print comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set; the palm vein comparison weight is determined according to the distribution of palm vein comparison scores of the palm vein training sample set and the rejection rate of the palm vein training sample set.
Since the dimension and distribution of the palm print comparison result and the palm vein comparison result are not uniform, and the biological characteristic authentication performance of the palm print and the palm vein is also different, the contribution of the palm print comparison result and the palm vein comparison result to the final decision fusion result is different. The prior art directly connects the palm print comparison result and the palm vein comparison result in series or in parallel, can not realize the complementation of the palm print and the palm vein, can not better solve the defects of palm print identification and palm vein identification, and can not improve the accuracy of palm print and palm vein identification.
In the application, palm print comparison weight and palm vein comparison weight are respectively determined according to the distribution of comparison scores of respective training sample sets and the rejection rate.
Taking palm print recognition as an example, the distribution of palm print comparison scores of a palm print training sample set can reflect the performance of an authentication method used by a palm print recognition mode, and the palm print comparison scores can be unified according to the distribution. Meanwhile, according to the distribution of the palm print comparison scores, the distribution difference of the palm print comparison scores of the same person and the palm print comparison scores of different persons can be increased, and the palm print recognition effect is improved. Palm vein recognition is the same.
The rejection rate is an index for evaluating the performance of the authentication method, and represents the proportion of the same person sample judged to be different persons. The determination of the introduction of the rejection rate into the weight may be used to guide the generation of the weight such that the optimization proceeds towards a target direction that is favorable for reducing the rejection rate.
Therefore, in the application, the palm print comparison weight is determined according to the distribution of the palm print comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set; the palm vein comparison weight is determined according to the distribution of palm vein comparison scores of the palm vein training sample set and the rejection rate of the palm vein training sample set, so that the dimension and the distribution of the palm vein comparison scores/the palm vein comparison scores can be unified, adverse effects of differences of the performance of authentication methods used for palm vein recognition and palm vein recognition on the fused identity authentication can be reduced, and further, better accuracy of the identity authentication can be obtained.
The specific determination mode of the palm print comparison weight is not limited, and the distribution of the palm print comparison score and the true rejection rate of the palm print training sample set are only needed, so that how to use the distribution of the palm print comparison score and the true rejection rate is not limited. For example, the distribution of the palm print alignment score may be a minimum value, an average value, a maximum value, a difference between the maximum value and the minimum value, a difference between the maximum value and the average value, a difference between the average value and the minimum value, and other parameter values reflecting the distribution of the palm print alignment score. The distribution of the above-exemplified palmprint alignment score may also be subjected to some mathematical processing such as inversion, normalization, absolute value, logarithmic operation, etc., and similarly, the rejection rate may be subjected to some mathematical processing. After the distribution of the palm print comparison score and the rejection rate are subjected to mathematical treatment, the distribution of the palm print comparison score and the rejection rate can be subjected to mathematical operations such as addition, subtraction, multiplication, division, exponential operation, logarithmic operation and the like, or some logic operations are performed, so that the palm print comparison weight is obtained. Palm vein comparison weights are the same. And will not be described in detail.
S400: and judging whether the two groups of biological characteristic data belong to the same person according to the decision fusion result.
Generally, if the decision fusion result is greater than the set decision threshold T, it is determined that the two sets of biometric data belong to the same person, otherwise, it is determined that the two sets of biometric data do not belong to the same person.
Or in some possible cases, if the decision fusion result is smaller than the set decision threshold T, judging that the two groups of biometric data belong to the same person, otherwise, judging that the two groups of biometric data do not belong to the same person.
The application provides a palm print palm vein multi-mode identity authentication method which is used for solving the palm print palm vein identity authentication problem under different scenes. According to the method, the palm print comparison result and the palm vein comparison result are fused through the set palm print comparison weight and the palm vein comparison weight, and identity authentication is carried out according to the decision fusion result obtained through fusion. The palm print comparison weight/palm vein comparison weight is determined according to the distribution of palm print/palm vein comparison scores of the palm print/palm vein training sample set and the rejection rate of the palm print/palm vein training sample set. The distribution of the palm print/palm vein comparison score can increase the distribution difference of the palm print/palm vein comparison score of the same person and the palm print/palm vein comparison score of different persons, thereby improving the authentication classification effect. The rejection rate may be used to guide the generation of weights such that optimization of weights proceeds toward a target direction that facilitates a reduction in rejection rate.
The decision fusion result after fusion is used for identity authentication, so that the complementation of palmprint and palmvein can be realized, the defects of palmprint recognition and palmvein recognition are well overcome, the accuracy of palmprint palmvein multi-modal recognition is improved, and the result of palmprint palmvein multi-modal identity authentication is superior to the recognition effect of any palmprint and palmvein single modality, and has larger performance improvement. In addition, the method and the device are integrated in a weighted voting mode, the calculation is simple, and the overall speed is not reduced compared with that of the palm print palm vein multi-mode identity authentication method in the prior art.
In order to more clearly illustrate the effect of the application, the palm print palm vein multi-mode identity authentication method of the application is tested on a test set, and the identity authentication effect is shown in the following table.
Therefore, compared with the palm print recognition and palm vein recognition of a single mode, the palm print palm vein multi-mode identity authentication method has a better identity authentication effect.
As an improvement of the embodiment of the present application, a specific implementation manner of the foregoing S100 is:
s110: and acquiring a palm image, detecting key points of the palm image, and extracting an interested region from the palm image according to the detected key points.
The present step is a preprocessing process for palm print recognition and palm vein recognition, and since the palm areas used for palm print recognition and palm vein recognition are the same, the same method can be used for preprocessing to obtain a region of interest (ROI area).
The specific pretreatment method can be as shown in fig. 2, and comprises the following steps: the palm print image and the palm vein image are acquired simultaneously by using acquisition equipment, namely, the palm image is acquired; then, finger key point detection is carried out, and then, the interested area is intercepted according to the key points, and normalization and texture enhancement can be carried out if necessary.
S120: and carrying out palm print feature extraction and palm vein feature extraction on the region of interest to obtain the line information and palm vein feature data of three main palm prints of the palm.
The method for extracting the palm vein feature data directly can be any method known to those skilled in the art, and the detailed description of the method is omitted. For the two groups of biological characteristic data, two palm vein characteristic data are respectively obtained for subsequent comparison.
For palm print feature extraction, the human palm includes three main palm prints, and the application requires the texture information of the three main palm prints to be used simultaneously, so that the texture information of the three main palm prints of the palm needs to be extracted.
The three main palmprint lines of the palm information lead to higher dimension of palmprint characteristic data, and extra calculated amount and storage space are introduced. Therefore, the feature is subjected to dimension reduction processing by using the principal component analysis method, more remarkable features are extracted from the original line information, the data amount during classification is reduced, and the classification capacity of palm print feature data is improved.
The principal component analysis method is to project high-dimensional data onto a low-dimensional space through linear mapping so that the projected data variance is maximized, thereby making it retain the characteristics of the original data. The process of performing principal component analysis of the present application is as follows S130.
S130: respectively carrying out principal component analysis on the line information of three main palmprints of the palm;
wherein, l represents the serial numbers of three main palmprints of the palm, and x l Representing the respective texture information, z, of three main palmprints of the palm l Representing the characteristics obtained by respectively carrying out principal component analysis on the line information of three main palmprints of the palm, mu l Mean value, W, of each line information representing three main palm prints of palm l Representing the principal component analysis matrix of each of three main palmprints of the palm; mu (mu) l And W is l The training set is trained by respective training sets of three main palmprints of the palm.
The step is used for analyzing principal components of the line information of three main palmprints respectively, and a principal component analysis matrix W is obtained from a training set in the training process l And mean mu l The reasoning process only needs to use W l And mean mu l Texture information x for three main palmprints of palm l The characteristic z after principal component analysis can be obtained by performing operation l 。
S140: the method comprises the steps of respectively carrying out principal component analysis on line information of three main palm lines of a palm through the following formula to obtain characteristic data of the palm lines;
wherein Z is palmprint characteristic data, P o 、P 1 And P 2 For projection matrix, P o 、P 1 And P 2 Respectively training the three main palmprint training sets of the palm.
This step is used to analyze the principal component for three features z l Fusion is carried out, and a projection matrix P is obtained from a training set in the training process o 、P 1 And P 2 The reasoning process only needs to use P o 、P 1 And P 2 For z l And performing operation to obtain palmprint characteristic data.
And respectively extracting the line information of three main palmprints for the two groups of biological characteristic data, and respectively carrying out principal component analysis and sum to obtain two palmprint characteristic data for subsequent comparison.
From the foregoing, mu l 、W l 、P o 、P 1 And P 2 The training method is obtained through a training process, and comprises the following steps:
S1: the average value mu of each line information of three main palm lines of the palm is calculated by the following formula l ;
Wherein,n=1, 2,3, …, N is the training set for the nth texture sample in each of the three main palmprints of the palmTotal number of medium grain samples.
S2: the covariance matrix S of each of three main palmprints of the palm is calculated by the following formula l ;
Wherein the covariance matrix S l Is d.
S3: respectively carrying out eigenvalue decomposition on covariance matrixes of three main palmprints of the palm through the following formula;
wherein,and->Respectively S l T=0, 1,2, …, d, d is S l Is a number of rows and columns.
S4: according to the set percentage of the main componentThe first r eigenvectors with the duty ratio of M percent (such as 95 percent) are selected to form a principal component analysis matrix W l ;
Wherein,
s5: the main component analysis is respectively carried out on the line information of three main palm lines of the palm through the following formulas;
wherein,and the characteristic obtained by respectively carrying out principal component analysis on the line information of three main palmprints of the palm is shown.
S6: the projection matrix P is optimized and calculated by the following formula o 、P 1 And P 2 ;
Wherein, respectively Z 0 、Z 1 、Z 2 Intra-class variance matrix of->Is Z 0 And Z 1 Is a matrix of inter-class variances of (c), Is Z 0 And Z 2 Inter-class variance matrix of->Is Z 1 And Z 2 Inter-class variance matrix of (a).
The projection matrix P is obtained after the above formula is solved o ,P 1 ,P 2 The training task is completed.
As another improvement of the embodiment of the present application, the S300 includes:
obtaining a decision fusion result through calculation according to the following formula;
wherein, delta is the decision fusion result, delta (k) is the palm print comparison result/palm vein comparison result, and w (k) is the palm print comparison weight/palm vein comparison weight;
s (k) is the palm print comparison score/palm vein comparison score, and T (k) is the palm print comparison threshold/palm vein comparison threshold.
The palm print comparison weight/palm vein comparison weight w (k) is obtained through training, and the training process is as follows:
s10: and carrying out pairwise comparison on the palm print training sample set/the palm vein training sample set and the palm print training sample/the palm vein training sample set to obtain a palm print comparison score set/a palm vein comparison score set.
The palm print comparison score set comprises palm print comparison scores of the same person and palm print comparison scores of different persons; the palm vein comparison score set includes palm vein comparison scores of the same person and palm vein comparison scores of different persons.
For a set of palm print training samples, it includes a series of palm print-like training samples from multiple individuals. And comparing palm print training samples of the same person or different persons to obtain palm print comparison scores of the same person and palm print comparison scores of different persons to form a palm print comparison score set. The palm vein training sample set is the same.
For example, toThe i-th palm print score/palm vein score of the palm print score/palm vein score set, k=1 for palm prints, k=2 for palm veins, j=1, 2,3, …, N k ,N k For the total number of palm print score/palm vein score in the palm print score/palm vein score set, j=0 represents +.>To be differentHuman palmar/venous score, j=1 means +.>Palmprint alignment score/palmvein alignment score for the same person.
Comparing the palm print training sample set with the palm print training sample of the same person in the palm vein training sample set to obtain palm print comparison score/palm vein comparison score of the same person, and marking the palm print comparison score/palm vein comparison score as a (k);
comparing the palm print training sample set/palm vein training sample set with palm print training samples/palm vein training samples of different people to obtain palm print comparison score/palm vein comparison score of different people, and marking as r (k);
the palm print alignment score/palm vein alignment score of the same person and the palm print alignment score/palm vein alignment score of different persons substantially satisfy gaussian distribution, and the palm print alignment score/palm vein alignment score of the same person and the palm print alignment score/palm vein alignment score of different persons are counted, the distribution of which is shown in fig. 3.
Wherein max r (k) Maximum value of palm print comparison score/palm vein comparison score for different people;
min r (k) The minimum value of palm print comparison score/palm vein comparison score for different people;
μ r (k) Average value of palm print comparison score/palm vein comparison score for different people;
max a (k) Maximum value of palm print comparison score/palm vein comparison score for the same person;
min a (k) The minimum value of palm print comparison score/palm vein comparison score for the same person;
μ a (k) The average of the palmprint alignment score/palmar vein alignment score for the same person.
Taking palm print identification as an example, it is obvious that the overlapping part of the distribution of the palm print comparison score of the same person and the palm print comparison score of different persons can reflect the performance of the palm print comparison algorithm, and the larger the overlapping part is, the less obvious the difference between the palm print comparison score of the same person and the palm print comparison score of different persons is, and the worse the performance of the algorithm is; on the contrary, the smaller the overlapping part is, the higher the difference degree between the palm print comparison score of the same person and the palm print comparison score of different persons is, the better the algorithm performance is, so that the difference between the palm print comparison score and the palm print comparison score of different persons is increased by reducing the overlapping part of the scores through a multi-mode fusion method.
S20: the palm print alignment score set/palm vein alignment score set is normalized according to the distribution of the palm print alignment score/palm vein alignment score of the same person in the palm print alignment score set and the non-overlapping areas of the distribution of the palm print alignment score/palm vein alignment score of different persons.
According to the method and the device, the palm print comparison score set/palm vein comparison score set is normalized by using the non-overlapping area, the dimension and the distribution of the palm print comparison score/palm vein comparison score can be unified, and the distribution difference of the palm print/palm vein comparison score of the same person and the palm print/palm vein comparison score of different persons is increased, so that a better identity authentication effect is obtained.
The specific implementation manner of the step can be as follows:
s21: calculating a non-overlapping region of the distribution of the palm print comparison score/palm vein comparison score of the same person and the distribution of the palm print comparison score/palm vein comparison score of different persons as a palm print normalization dimension/palm vein normalization dimension by the formula (1);
q(k)=μ a (k)-min a (k)+max r (k)-μ r (k) (1)
the step is used for carrying out normalization operation on the palm print comparison score/palm vein comparison score, the normalization operation needs to take the parameters of the respective non-overlapping areas as normalization dimensions, and the calculation of the normalization dimensions is as shown in the above formula.
Wherein k=1, 2, k=1 represents a palm print recognition mode, k=2 represents a palm vein recognition mode, q (k) is a palm print normalization dimension/palm vein normalization dimension, and max r (k) Is the maximum value of palm print comparison score/palm vein comparison score of different people in palm print comparison score/palm vein comparison score, mu r (k) The average value of palm print comparison score/palm vein comparison score of different people in the palm print comparison score set/palm vein comparison score is min a (k) Is the minimum value of palm print comparison score/palm vein comparison score of the same person in the palm print comparison score/palm vein comparison score, mu a (k) The average value of the palm print comparison score/palm vein comparison score of the same person in the palm print comparison score set/palm vein comparison score set.
The larger q (k) is used for indicating that the larger the distribution phase difference between the palm print/palm vein comparison score of the same person and the palm print/palm vein comparison scores of different persons is, the higher the degree of distinction is, the better the effect is, so that the overlapping area of the distribution can be reduced by normalizing the palm print/palm vein comparison score by the normalization dimension, and further the better identity authentication effect is obtained.
S22: normalizing the palm print comparison score set/palm vein comparison score set according to the palm print normalization dimension/palm vein normalization dimension and through the method (2);
wherein,the i-th palmprint alignment score/palmar vein alignment score in the palmar vein alignment score set before normalization, ++>For the i-th palmprint/palmar vein score of the normalized palmprint/palmar vein score, i=1, 2,3, …, N k ,N k Palm for handPalm print score/total number of palm vein score/palm vein score in the score set, j=0 means +.>Palm print alignment score/palm vein alignment score for different people, j=1 means +.>Palmprint alignment score/palmvein alignment score for the same person.
S30: and calculating the palm print comparison weight/palm vein comparison weight according to the normalized palm print comparison score set/palm vein comparison score set and the true rejection rate of the palm print training sample set/palm vein training sample set.
In one example, the palmprint comparison weight and the palmar vein comparison weight may be calculated by equation (3);
wherein w (k) is palm print comparison weight/palm vein comparison weight;
FRR k the rejection rate of the palm print training sample set/palm vein training sample set is obtained.
The two indexes of the identification authentication method are respectively a False Reject Rate (FRR) and a false recognition rate (FAR), the FRR represents the proportion of the same person, the biological characteristic data of the same person is judged to be different persons, the FAR judges the biological characteristic data of different persons to be the proportion of the same person, and in the performance evaluation of the identification or authentication, the smaller the two indexes are hoped to be better, and the higher the two indexes are, the certain service risk is caused. In the multi-mode fusion method, not only the influence of the score dimension is needed to be considered, but also the difference of the classification performance of different modes can reduce the performance of the fused algorithm.
Therefore, the present application introduces not only the normalized dimension representing the comparison score distribution into the calculation of the weight, but also FRR into the calculation of the weight, and the weight w (k) of each modality is calculated by the above formula.
The embodiment of the application also provides a palmprint palmvein multi-mode identity authentication device, as shown in fig. 4, which comprises:
the data acquisition module 1 is used for acquiring two groups of biological characteristic data, wherein each group of biological characteristic data comprises palm print characteristic data and palm vein characteristic data.
The comparison score calculation module 2 is used for comparing the two palm print characteristic data in the two groups of biological characteristic data to obtain a palm print comparison score; and comparing the two palm vein feature data in the two groups of biological feature data to obtain palm vein comparison scores.
The decision fusion module 3 is configured to determine a palm print comparison result according to the palm print comparison score and a set palm print comparison threshold, determine a palm vein comparison result according to the palm vein comparison score and the set palm vein comparison threshold, and weight and vote the palm print comparison result and the palm vein comparison result according to the set palm print comparison weight and the palm vein comparison weight, so as to obtain a decision fusion result.
The palm print comparison weight is determined according to the distribution of the palm print comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set; the palm vein comparison weight is determined according to the distribution of palm vein comparison scores of the palm vein training sample set and the rejection rate of the palm vein training sample set.
And the multi-mode authentication module 4 is used for judging whether the two groups of biological characteristic data belong to the same person according to the decision fusion result.
The application provides a palm print palm vein multi-mode identity authentication method which is used for solving the palm print palm vein identity authentication problem under different scenes. According to the method, the palm print comparison result and the palm vein comparison result are fused through the set palm print comparison weight and the palm vein comparison weight, and identity authentication is carried out according to the decision fusion result obtained through fusion. The palm print comparison weight/palm vein comparison weight is determined according to the distribution of palm print/palm vein comparison scores of the palm print/palm vein training sample set and the rejection rate of the palm print/palm vein training sample set. The distribution of the palm print/palm vein comparison score can increase the distribution difference of the palm print/palm vein comparison score of the same person and the palm print/palm vein comparison score of different persons, thereby improving the authentication classification effect. The rejection rate may be used to guide the generation of weights such that optimization of weights proceeds toward a target direction that facilitates a reduction in rejection rate.
The decision fusion result after fusion is used for identity authentication, so that the complementation of palmprint and palmvein can be realized, the defects of palmprint recognition and palmvein recognition are well overcome, the accuracy of palmprint palmvein multi-modal recognition is improved, and the result of palmprint palmvein multi-modal identity authentication is superior to the recognition effect of any palmprint and palmvein single modality, and has larger performance improvement. In addition, the method and the device are integrated in a weighted voting mode, the calculation is simple, and the overall speed is not reduced compared with that of the palm print palm vein multi-mode identity authentication method in the prior art.
As an improvement of the embodiment of the present application, the data acquisition module includes:
the interested region extraction unit is used for acquiring a palm image, detecting key points of the palm image, and extracting the interested region from the palm image according to the detected key points.
The feature extraction unit is used for extracting palm print features and palm vein features of the region of interest to obtain line information and palm vein feature data of three main palm prints of the palm.
And the main component analysis unit is used for respectively carrying out main component analysis on the line information of the three main palmprints of the palm.
Wherein, l represents the serial numbers of three main palmprints of the palm, and x l Representing the respective texture information, z, of three main palmprints of the palm l Texture letter representing three main palmprints of palmThe characteristics, mu, obtained by analyzing the principal components of each sample l Mean value, W, of each line information representing three main palm prints of palm l Representing the principal component analysis matrix of each of three main palmprints of the palm; mu (mu) l And W is l The training set is trained by respective training sets of three main palmprints of the palm.
And the fusion unit is used for fusing the characteristics obtained after the main component analysis is respectively carried out on the line information of the three main palm lines of the palm through the following formula to obtain palm line characteristic data.
Wherein Z is palmprint characteristic data, P o 、P 1 And P 2 For projection matrix, P o 、P 1 And P 2 Respectively training the three main palmprint training sets of the palm.
Mu as described above l 、W l 、P o 、P 1 And P 2 The training method is characterized by comprising the following steps of:
the average value calculation module is used for respectively calculating and obtaining the average value mu of each line information of three main palm lines of the palm through the following formulas l 。
Wherein,n=1, 2,3, …, N are the total number of texture samples in the training set for each of the three primary palmprints.
The covariance matrix calculation module is used for calculating the covariance matrix S of each of the three main palmprints of the palm respectively through the following formula l 。
And the characteristic value decomposition module is used for respectively carrying out characteristic value decomposition on covariance matrixes of the three main palmprints of the palm through the following formulas.
Wherein,and->Respectively S l T=0, 1,2, …, d, d is S l Is a number of rows and columns.
A principal component analysis matrix acquisition module for acquiring a principal component according to a set percentage of the principal componentSelecting the first r eigenvectors with the duty ratio of M% to form a principal component analysis matrix W l 。
Wherein,
and the principal component analysis module is used for respectively carrying out principal component analysis on the line information of the three main palm prints of the palm through the following formula.
Wherein,and the characteristic obtained by respectively carrying out principal component analysis on the line information of three main palmprints of the palm is shown.
Projection matrix calculation module for passingThe projection matrix P is obtained by optimizing the following formula o 、P 1 And P 2 。
Wherein, respectively Z 0 、Z 1 、Z 2 Intra-class variance matrix of->Is Z 0 And Z 1 Is a matrix of inter-class variances of (c),is Z 0 And Z 2 Inter-class variance matrix of->Is Z 1 And Z 2 Inter-class variance matrix of (a).
As another improvement of the embodiment of the present application, the aforementioned decision fusion module is configured to:
Obtaining a decision fusion result through calculation according to the following formula;
wherein, delta is the decision fusion result, delta (k) is the palm print comparison result/palm vein comparison result;
s (k) is the palm print comparison score/palm vein comparison score, and T (k) is the palm print comparison threshold/palm vein comparison threshold.
The palm print comparison weight and the palm vein comparison weight are obtained through training of the following modules:
the training data preparation module is used for carrying out pairwise comparison on the palm print training sample and the palm vein training sample in the palm print training sample set and the palm vein training sample set to obtain a palm print comparison score set and a palm vein comparison score set.
The palm print comparison score set comprises palm print comparison scores of the same person and palm print comparison scores of different persons; the palm vein comparison score set includes palm vein comparison scores of the same person and palm vein comparison scores of different persons.
The normalization module is used for normalizing the palm print comparison score set/palm vein comparison score set according to the distribution of the palm print comparison score/palm vein comparison score of the same person in the palm print comparison score set/palm vein comparison score set and the non-overlapping area of the distribution of the palm print comparison score/palm vein comparison score of different persons.
The weight calculation module is used for calculating the palm print comparison weight/palm vein comparison weight according to the normalized palm print comparison score set/palm vein comparison score set and the true rejection rate of the palm print training sample set/palm vein training sample set.
In one example, the normalization module includes:
and a normalized dimension calculation unit for calculating a non-overlapping region of the distribution of the palm print comparison score/palm vein comparison score of the same person and the distribution of the palm print comparison score/palm vein comparison score of different persons as a palm print normalized dimension/palm vein normalized dimension by the formula (1).
q(k)=μ a (k)-min a (k)+max r (k)-μ r (k) (1)
Wherein k=1, 2, k=1 represents a palm print recognition mode, k=2 represents a palm vein recognition mode, q (k) is a palm print normalization dimension/palm vein normalization dimension, and max r (k) Is the maximum value of palm print comparison score/palm vein comparison score of different people in palm print comparison score/palm vein comparison score, mu r (k) Score for palmprintCollect/palm vein comparison score different people in the collection of palm print comparison scores/average of palm vein comparison scores, min a (k) Is the minimum value of palm print comparison score/palm vein comparison score of the same person in the palm print comparison score/palm vein comparison score, mu a (k) The average value of the palm print comparison score/palm vein comparison score of the same person in the palm print comparison score set/palm vein comparison score set.
And the normalization unit is used for normalizing the palm print comparison score set/palm vein comparison score set according to the palm print normalization dimension/palm vein normalization dimension and through the method (2).
Wherein,the i-th palmprint alignment score/palmar vein alignment score in the palmar vein alignment score set before normalization, ++>For the i-th palmprint/palmar vein score of the normalized palmprint/palmar vein score, i=1, 2,3, …, N k ,N k For the total number of palm print score/palm vein score in the palm print score/palm vein score set, j=0 represents +.>Palm print alignment score/palm vein alignment score for different people, j=1 means +.>Palmprint alignment score/palmvein alignment score for the same person.
Correspondingly, the weight calculation module is used for:
calculating palmprint comparison weight and palmvein comparison weight through the method (3);
wherein w (k) is palm print comparison weight/palm vein comparison weight;
FRR k the rejection rate of the palm print training sample set/palm vein training sample set is obtained.
The device provided in this embodiment of the present application has the same implementation principle and technical effects as those of the foregoing method embodiment, and for a brief description, reference may be made to the corresponding content in the foregoing method embodiment where the device embodiment is not mentioned. It will be clear to those skilled in the art that, for convenience and brevity, the specific working procedures of the apparatus and units described above may refer to the corresponding procedures in the above method embodiments, and are not described herein again.
The method according to the above embodiment provided in the present application may implement service logic through a computer program and be recorded on a storage medium, where the storage medium may be read and executed by a computer, so as to implement the effects of the solution described in the method embodiment of the present specification. Accordingly, embodiments of the present application also provide a computer-readable storage medium for palm print palm vein multimodal identity authentication, comprising a memory for storing processor-executable instructions that, when executed by a processor, implement the steps of a palm print palm vein multimodal identity authentication method comprising the foregoing embodiments.
The storage medium may include physical means for storing information, typically by digitizing the information before storing it in an electronic, magnetic, or optical medium. The storage medium may include: means for storing information using electrical energy such as various memories, e.g., RAM, ROM, etc.; devices for storing information using magnetic energy such as hard disk, floppy disk, magnetic tape, magnetic core memory, bubble memory, and USB flash disk; devices for optically storing information, such as CDs or DVDs. Of course, there are other ways of readable storage medium, such as quantum memory, graphene memory, etc.
The above description of the storage medium according to the method embodiment may further include other implementations, and the implementation principle and the generated technical effects of the embodiment are the same as those of the foregoing method embodiment, and specific reference may be made to the description of the related method embodiment, which is not repeated herein.
The embodiment of the application also provides equipment for palm print palm vein multi-mode identity authentication, which can be a single computer or can comprise actual operation devices and the like using one or more of the methods or one or more embodiment devices of the application. The device for palm print palm vein multi-mode identity authentication can comprise at least one processor and a memory for storing computer executable instructions, wherein the steps of any one or more of the palm print palm vein multi-mode identity authentication methods are realized when the processor executes the instructions.
The above description of the apparatus according to the method embodiment may further include other implementations, and the implementation principle and the generated technical effects of the embodiment are the same as those of the foregoing method embodiment, and specific reference may be made to the description of the related method embodiment, which is not repeated herein.
Finally, it should be noted that: the foregoing examples are merely specific embodiments of the present application, and are not intended to limit the scope of the present application, but the present application is not limited thereto, and those skilled in the art will appreciate that while the foregoing examples are described in detail, the present application is not limited thereto. Any person skilled in the art may modify or easily conceive of the technical solution described in the foregoing embodiments, or make equivalent substitutions for some of the technical features within the technical scope of the disclosure of the present application; such modifications, changes or substitutions do not depart from the spirit and scope of the corresponding technical solutions. Are intended to be encompassed within the scope of this application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A palmprint palmvein multi-mode identity authentication method, the method comprising:
acquiring two groups of biological characteristic data, wherein each group of biological characteristic data comprises palm print characteristic data and palm vein characteristic data;
comparing the two palm print characteristic data in the two groups of biological characteristic data to obtain a palm print comparison score; comparing the two palm vein feature data in the two groups of biological feature data to obtain palm vein comparison scores;
Determining a palm print comparison result according to the palm print comparison score and a set palm print comparison threshold, determining a palm vein comparison result according to the palm vein comparison score and the set palm vein comparison threshold, and carrying out weighted voting on the palm print comparison result and the palm vein comparison result according to a set palm print comparison weight and a palm vein comparison weight to obtain a decision fusion result;
the palm print comparison weight is determined according to the distribution of the palm print comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set; the palm vein comparison weight is determined according to the distribution of palm vein comparison scores of a palm vein training sample set and the rejection rate of the palm vein training sample set;
and judging whether the two groups of biological characteristic data belong to the same person according to the decision fusion result.
2. The palmprint palmvein multi-modal identity authentication method of claim 1, wherein the palmprint comparison weight is determined according to a distribution of comparison scores of a palmprint training sample set and a rejection rate of the palmprint training sample set, comprising:
the palm print training samples in the palm print training sample set are compared pairwise to obtain a palm print comparison score set; wherein the palm print comparison score set comprises palm print comparison scores of the same person and palm print comparison scores of different persons;
Normalizing the palm print comparison score set according to a non-overlapping region of the distribution of the palm print comparison scores of the same person and the distribution of the palm print comparison scores of different persons in the palm print comparison score set;
calculating the palm print comparison weight according to the normalized palm print comparison score set and the rejection rate of the palm print training sample set;
and/or;
the palm vein comparison weight is determined according to the distribution of the comparison scores of the palm vein training sample set and the rejection rate of the palm vein training sample set, and comprises the following steps:
performing pairwise comparison on the palm vein training samples in the palm vein training sample set to obtain a palm vein comparison score set; wherein the palm vein comparison score set comprises palm vein comparison scores of the same person and palm vein comparison scores of different persons;
normalizing the palm vein comparison score set according to a non-overlapping region of the distribution of the palm vein comparison scores of the same person and the distribution of the palm vein comparison scores of different persons in the palm vein comparison score set;
and calculating the palm vein comparison weight according to the normalized palm vein comparison score set and the rejection rate of the palm vein training sample set.
3. The palm print palmar vein multi-modality identity authentication method of claim 2, wherein the normalizing the palm print comparison score set according to a non-overlapping area of a distribution of palm print comparison scores of the same person and a distribution of palm print comparison scores of different persons in the palm print comparison score set comprises:
Calculating a non-overlapping region of the distribution of the palm print comparison score of the same person and the distribution of the palm print comparison score of different persons as a palm print normalized dimension by the formula (1);
normalizing the palm print comparison score set according to the palm print normalization dimension and by formula (2);
and/or;
normalizing the set of palm vein comparison scores according to a non-overlapping region of a distribution of palm vein comparison scores of the same person and a distribution of palm vein comparison scores of different persons in the set of palm vein comparison scores, comprising:
calculating a non-overlapping region of the distribution of the palm vein comparison score of the same person and the distribution of the palm vein comparison score of different persons as palm vein normalization dimensions by the formula (1);
normalizing the palm vein comparison score set according to the palm vein normalization dimension and through a formula (2);
wherein:
q(k)=μ a (k)-min a (k)+max r (k)-μ r (k) (1)
k=1, 2, k=1 denotes a palm print recognition modality, k=2 denotes a palm vein recognition modality, q (k) is the palm print normalized dimension/palm vein normalized dimension, max r (k) For the maximum value of the palm print comparison score/palm vein comparison score, mu for different people in the palm print comparison score/palm vein comparison score set r (k) Average value of palm print comparison score/palm vein comparison score for different people in the palm print comparison score/palm vein comparison score, min a (k) For the minimum value of the palm print comparison score/palm vein comparison score of the same person in the palm print comparison score/palm vein comparison score, mu a (k) An average of the palm print comparison score/palm vein comparison score for the same person in the palm print comparison score/palm vein comparison score set;
the i-th palmprint alignment score/palmar vein alignment score in the palmar vein alignment score set before normalization, ++>For the i-th palmprint/palmar vein score of the normalized palmprint/palmar vein score, i=1, 2,3, …, N k ,N k For the total number of palm print score/palm vein score in the palm print score/palm vein score set, j=0 represents +.>Palm print alignment score/palm vein alignment score for different people, j=1 means +.>Palmprint alignment score/palmvein alignment score for the same person.
4. The palm print palmar vein multi-mode identity authentication method of claim 3, wherein the calculating the palm print comparison weight according to the normalized palm print comparison score set and the rejection rate of the palm print training sample set comprises:
calculating the palmprint comparison weight by the formula (3);
And/or;
the calculating the palm vein comparison weight according to the normalized palm vein comparison score set and the rejection rate of the palm vein training sample set comprises the following steps:
calculating the palm vein comparison weight by the formula (3);
wherein:
w (k) is the palm print comparison weight/the palm vein comparison weight;
FRR k and the rejection rate of the palm print training sample set/palm vein training sample set is determined.
5. The method for multi-modal identity authentication of palmprint and palmvein as recited in claim 4, wherein the determining a palmprint comparison result according to the palmprint comparison score and a set palmprint comparison threshold, determining a palmvein comparison result according to the palmvein comparison score and a set palmvein comparison threshold, and weighting the palmprint comparison result and the palmvein comparison result according to a set palmprint comparison weight and a palmvein comparison weight, and voting to obtain a decision fusion result, comprises:
the decision fusion result is obtained through calculation according to the following formula;
wherein, delta is the decision fusion result, delta (k) is the palm print comparison result/palm vein comparison result;
s (k) is the palm print comparison score/palm vein comparison score, and T (k) is the palm print comparison threshold/palm vein comparison threshold.
6. The method for multi-modal identity authentication of palm print and palm vein according to any one of claims 1 to 5, wherein the acquiring two sets of biometric data, wherein each set of biometric data includes palm print feature data and palm vein feature data, includes:
acquiring a palm image, detecting key points of the palm image, and extracting an interested region from the palm image according to the detected key points;
extracting palm print characteristics and palm vein characteristics of the region of interest to obtain the line information of three main palm prints of the palm and the palm vein characteristic data;
respectively carrying out principal component analysis on the line information of three main palmprints of the palm;
wherein, l represents the serial numbers of three main palmprints of the palm, and x l Representing the respective texture information, z, of three main palmprints of the palm l Representing the characteristics obtained by respectively carrying out principal component analysis on the line information of three main palmprints of the palm, mu l Mean value, W, of each line information representing three main palm prints of palm l Representing the principal component analysis matrix of each of three main palmprints of the palm; mu (mu) l And W is l Training by respective training sets of three main palmprints of the palm;
the characteristic obtained after the main component analysis is respectively carried out on the line information of three main palm lines of the palm is fused through the following formula, so that the palm line characteristic data are obtained;
Wherein Z is the palm print characteristic data, P o 、P 1 And P 2 For projection matrix, P o 、P 1 And P 2 Respectively training the three main palmprint training sets of the palm.
7. The palmprint palmvein multi-modality identity authentication method of claim 6, wherein the μ is obtained by training the following method l 、W l 、P o 、P 1 And P 2 :
The average value mu of each line information of three main palm lines of the palm is calculated by the following formula l ;
Wherein,n=1, 2,3, …, N being the total number of the texture samples in the training set for each of the three primary palmprints of the palm;
the covariance matrix S of each of three main palmprints of the palm is calculated by the following formula l ;
Respectively carrying out eigenvalue decomposition on covariance matrixes of three main palmprints of the palm through the following formula;
wherein,and->Respectively S l T=0, 1,2, …, d, d is S l Row and column numbers of (a);
according to the set percentage of the main componentSelecting the first r eigenvectors with the duty ratio of M% to form the principal component analysis matrix W l ;
Wherein,
the main component analysis is respectively carried out on the line information of three main palm lines of the palm through the following formulas;
wherein, Representing the characteristics obtained by respectively carrying out principal component analysis on the line information of three main palmprints of the palm;
the projection matrix P is optimized and calculated by the following formula o 、P 1 And P 2 ;
Wherein, respectively Z 0 、Z 1 、Z 2 Intra-class variance matrix of->Is Z 0 And Z 1 Is a matrix of inter-class variances of (c),is Z 0 And Z 2 Inter-class variance matrix of->Is Z 1 And Z 2 Inter-class variance matrix of (a).
8. A palmprint palmvein multi-modality identity authentication device, the device comprising:
the data acquisition module is used for acquiring two groups of biological characteristic data, wherein each group of biological characteristic data comprises palm print characteristic data and palm vein characteristic data;
the comparison score calculation module is used for comparing the two palm print characteristic data in the two groups of biological characteristic data to obtain a palm print comparison score; comparing the two palm vein feature data in the two groups of biological feature data to obtain palm vein comparison scores;
the decision fusion module is used for determining a palm print comparison result according to the palm print comparison score and a set palm print comparison threshold, determining a palm vein comparison result according to the palm vein comparison score and the set palm vein comparison threshold, and carrying out weighted voting on the palm print comparison result and the palm vein comparison result according to a set palm print comparison weight and a palm vein comparison weight to obtain a decision fusion result;
The palm print comparison weight is determined according to the distribution of the palm print comparison scores of the palm print training sample set and the rejection rate of the palm print training sample set; the palm vein comparison weight is determined according to the distribution of palm vein comparison scores of a palm vein training sample set and the rejection rate of the palm vein training sample set;
and the multi-mode authentication module is used for judging whether the two groups of biological characteristic data belong to the same person according to the decision fusion result.
9. A computer readable storage medium for palm print palm vein multimodal identity authentication, comprising a memory for storing processor executable instructions which when executed by the processor implement the steps comprising the palm print palm vein multimodal identity authentication method of any of claims 1-7.
10. An apparatus for palm print palm vein multimodal identity authentication, comprising at least one processor and a memory storing computer executable instructions which when executed by the processor implement the steps of the palm print palm vein multimodal identity authentication method of any of claims 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210914591.8A CN117542086A (en) | 2022-08-01 | 2022-08-01 | Palm print palm vein multi-mode identity authentication method, device, storage medium and equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210914591.8A CN117542086A (en) | 2022-08-01 | 2022-08-01 | Palm print palm vein multi-mode identity authentication method, device, storage medium and equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN117542086A true CN117542086A (en) | 2024-02-09 |
Family
ID=89784752
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210914591.8A Pending CN117542086A (en) | 2022-08-01 | 2022-08-01 | Palm print palm vein multi-mode identity authentication method, device, storage medium and equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117542086A (en) |
-
2022
- 2022-08-01 CN CN202210914591.8A patent/CN117542086A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Bhattacharyya et al. | Biometric authentication: A review | |
Bharadi et al. | Off-line signature recognition systems | |
Sharma et al. | Two-stage quality adaptive fingerprint image enhancement using Fuzzy C-means clustering based fingerprint quality analysis | |
CN113469143A (en) | Finger vein image identification method based on neural network learning | |
Kodituwakku | Biometric authentication: A review | |
Shawkat et al. | The new hand geometry system and automatic identification | |
Hammouche et al. | Score level fusion of major and minor finger knuckle patterns based symmetric sum-based rules for person authentication | |
KivutiNjeru et al. | Comparative analysis of minutiae based fingerprint matching algorithms | |
Attallah et al. | Fusing palmprint, finger-knuckle-print for Bi-modal recognition system based on LBP and BSIF | |
CN117542086A (en) | Palm print palm vein multi-mode identity authentication method, device, storage medium and equipment | |
JP7315884B2 (en) | Authentication method, authentication program, and information processing device | |
CN109165640B (en) | Hand back vein identification method and identification system based on bit plane internal block mutual information | |
Chen et al. | Fingerprint Quality Assessment based on Texture and Geometric Features. | |
Wu et al. | Fusion recognition of palmprint and palm vein based on modal correlation | |
Di Martino et al. | A statistical approach to reliability estimation for fingerprint recognition | |
Marasco et al. | Accounting for demographic differentials in forensic error rate assessment of latent prints via covariate-specific ROC regression | |
CN116563958A (en) | Multi-mode identity authentication method, device, storage medium and equipment | |
WO2023143077A1 (en) | Multimodal identity authentication method and apparatus, storage medium, and device for multimodal identity authentication | |
da Silva Vasconcelos et al. | Estimation of Fingerprint Image Quality Based on Neighborhood Strengthness Homogeneity | |
Marasco et al. | Demographic-Adapted ROC Curve for Assessing Automated Matching of Latent Fingerprints | |
CN116563904A (en) | Multi-mode living body detection method, device, storage medium and equipment | |
Tada et al. | Fingerprint image quality estimation using a fuzzy inference system | |
Mubeen | Biometric Authentication: Past, Present, and Future Perspectives | |
Raghavendra et al. | Exploring the Frontiers of Security: Comprehensive Biometric Security Solution with Multiple Modalities Enhanced by Machine Learning | |
Olsen | Fingerprint image quality: predicting biometric performance |
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 |