CN106548134A - GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine - Google Patents
GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine Download PDFInfo
- Publication number
- CN106548134A CN106548134A CN201610905400.6A CN201610905400A CN106548134A CN 106548134 A CN106548134 A CN 106548134A CN 201610905400 A CN201610905400 A CN 201610905400A CN 106548134 A CN106548134 A CN 106548134A
- Authority
- CN
- China
- Prior art keywords
- vena metacarpea
- palmmprint
- svm
- normalization
- matching
- 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.)
- Withdrawn
Links
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06V10/30—Noise filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- 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
- G06V40/70—Multimodal biometrics, e.g. combining information from different biometric modalities
-
- 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)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Human Computer Interaction (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine, it is related to a kind of living things feature recognition algorithm, first, with polyteny core principle component analysis (MKPCA) to palmprint image tensor dimensionality reduction, characteristic vector is projected and is extracted on tensor subspace, and the COS distance calculated between characteristic vector obtains palmprint match scores;Then vena metacarpea extracts the local SIFT feature of vena metacarpea using Scale invariant features transform (SIFT) algorithm, calculate the similarity criteria of the Euclidean distance as characteristic point in two width images of characteristic point, by the normalized of the matching similarity fraction to two mode, fusion obtains new characteristic vector;Classification and Identification is carried out after GA Model Parameter Optimizations with SVM finally.The method compensate for the finiteness of single modal characteristics information, both make use of the matching scoring information after normalization the characteristics of easily classify, and has played the stronger advantage of svm classifier ability after GA optimizations again, has effectively increased the discrimination of system.
Description
Technical field
The present invention relates to a kind of biological feather recognition method, more particularly to a kind of GA optimization SVM and normalization combine
Palmmprint and vena metacarpea fusion identification method.
Background technology
With continuous growth of the information-intensive society to personal identification and regulatory requirement, living things feature recognition is than traditional body
Part authentication method is more safe, secrecy and convenience, at the same have be difficult to forget, it is difficult forge or stolen, carry-on " carrying " and with
When the advantages of can use everywhere.Though single mode living creature characteristic recognition system is ripe, have that characteristic information is limited, certification is unstable
The features such as.Multi-modal biological characteristic technology of identification reduces the impact of these unfavorable factors, and tradition is based on normalized fusion side
Method is relatively easy due to the fusion rule for adopting, and does not make full use of the information after normalization, and less in view of matching fraction
The particularity of distribution, final fusion recognition accuracy rate improve limited;Fusion method based on grader is not for normalized
During scoring information design grader, it is also difficult to reach satisfied classifying quality.On this basis by the method for two kinds of different advantages
Through optimizing integration, more preferable classifying quality can be obtained.
The content of the invention
It is an object of the invention to provide a kind of GA optimizes palmmprint and the vena metacarpea fusion recognition that SVM and normalization combine
Method, the method is a kind of model parameter optimizing, normalization and the biological feather recognition method that combines of Multiple Classifier Fusion, from
The limitation of single method after multi-modal improvement, is further compensated for, the performance of biological recognition system can have been effectively improved, with non-
Often good recognition effect.
The purpose of the present invention is achieved through the following technical solutions:
The method first carries out characteristic matching to palmmprint and vena metacarpea respectively, and the matching score normalization that each grader is given is arrived
Identical numerical intervals, are converted into new characteristic vector;Then classification knowledge is carried out after the parameter optimization with GA to svm classifier model
Not, the realization of the method includes following step:
Step one:Feature extraction is carried out to palmmprint and vena metacarpea image respectively first;
Step 2:The similarity criteria being each adapted to is utilized respectively to palmmprint and vena metacarpea characteristics of image carries out characteristic matching, obtains
To respective matching fraction, it is normalized to matching fraction with specific normalized function, transforms into new feature
Vector;
Step 3:Svm classifier model parameter is optimized by GA, obtains the σ of optimum penalty factor and gaussian kernel function
Value;
Step 4:The new feature vector after normalized is merged in fraction level with the sorting parameter model of optimization.
I.e.:Multi-modal biological characteristic recognition methods of the present invention, is melted using model parameter optimizing, normalization fusion and grader
The multi-modal biological characteristic identification that conjunction combines.The present invention, in fraction level convergence strategy i.e., is selected suitable respectively using characteristic of division
Put the palms together before one the feature interpretation and matching process of line and vena metacarpea, can more accurately be matched fraction, then by appropriate
Blending algorithm carries out joint identification, test result indicate that the method can obtain reasonable recognition effect.
Advantages of the present invention with effect is:
1. the present invention compensate for the limitation of single mode characteristic information using palmmprint and vena metacarpea fusion recognition.
2. Multiple Classifier Fusion and method for normalizing are combined by the present invention, and carry out parameter optimization to disaggregated model with GA,
Substantially increase accuracy of identification.
Specific embodiment
1. adopting carries out characteristic matching to palmmprint and vena metacarpea respectively, first palmmprint and vena metacarpea image is eliminated or is reduced
The impact of the unfavorable factors such as noise present in image, rotation and translation, and image is positioned and is normalized to improve knowledge
The robustness of other algorithm.(1) palmprint image is trained first, and calculating the transformed covariance matrix of nuclear matrix carries out feature decomposition,
Obtain the corresponding characteristic vector of top n eigenvalue of maximum.Palmprint image feature set is projected on tensor subspace, palmmprint is obtained
Characteristic tensor, then training image and test image are projected to feature that low-dimensional is obtained on MKPCA tensor subspaces respectively
Tensor, the COS distance y calculated between characteristic tensor carry out palmprint match, and multiple samples just have multiple similarity measurements, obtain
One group of matching fraction;(2) and then remove vena metacarpea major part background using threshold method to scheme vein ROI
As carrying out medium filtering and gaussian filtering, the local SIFT for extracting vena metacarpea using Scale invariant features transform (SIFT) algorithm is special
Levy, the Euclidean distance of characteristic point is calculated as the similarity criteria of characteristic point in two width images, be similarly obtained one group of matching fraction.Certain specific function can normalize the matching fraction that different classifications device is provided, such as z-score and tanh
Function, mean () and sd () are represented respectively and are averaged and ask mean square deviation computing here:
The parameter optimization of svm classifier model is carried out using genetic algorithm.First, to svm classifier model penalty factor and Gauss
Kernel function σ value carries out binary coding, and randomly generates initialization population;When the combination to SVM parameters is optimized,
During code Design, kernel functional parameter and wrong penalty factor adopt binary coding, coding to be respectively binary system in span
Binary coding combination is just obtained individual chromosome gene string by string.Svm classifier is trained with a part of training sample set data
Device model, and the discrimination RR of test sample collection data is calculated with the SVM classifier for training, what SVM was to be completed is to find most
Two class samples are made a distinction by excellent Optimal Separating Hyperplane.
2. GA optimizations svm classifier model algorithm is as follows:
Step 1:Find and cause functionStructural risk minimizationAnd b, function is converted into:
Step 2:Set up Largrange functions::Wherein;By
Relation between primal problem and dual problem solution is known, if a* is the optimal solution of dual problem,:
Obtaining optimal separating hyper plane is:;Problem to be optimized can be expressed as: ;
Step 3:Using kernel function, then the original optimization problem of Nonlinear Support Vector Machines be converted into:
Step 4:According to cross-validation method principle, discrimination RR reflects the Generalization Ability of SVM models to a certain extent and divides
Class ability, therefore:
Step 5:Crossing-over rate and aberration rate take 0.50 and 0.05 respectively, and in gene string (chromosome), C and σ values adopt binary system
Coding, putting in order as C and σ in chromosome, if their digit is respectively n1 、n2 , then the length of chromosome is n1 、n2
, search space is.For C, by calculating C=10C-1Penalty factor is obtained, for σ, by calculating σ=(σ+1)/200
σ values are obtained, then characteristic vector is brought into svm classifier model;
Step 6:The fitness Fitness=RR of each gene string is constructed according to this can;Then judge the stopping criterion of genetic algorithm
Whether meet, stop calculating if satisfaction, export optimized parameter, otherwise, then the operations such as selection, intersection and variation are performed to produce
Raw a new generation population, and start the heredity of a new generation.
Shown according to actual effect:The algorithm can reach convergence faster, and about iteration can reach most for 10 times or so
Excellent nicety of grading such that it is able to obtain very high discrimination, and it is time-consuming and few.
Claims (1)
1.GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine, it is characterised in that the method is first divided
It is other that characteristic matching is carried out to palmmprint and vena metacarpea, the matching score normalization that each grader is given to identical numerical intervals,
It is converted into new characteristic vector;Then Classification and Identification, the realization of the method are carried out after the parameter optimization with GA to svm classifier model
Including following step:
Step one:Feature extraction is carried out to palmmprint and vena metacarpea image respectively first;
Step 2:The similarity criteria being each adapted to is utilized respectively to palmmprint and vena metacarpea characteristics of image carries out characteristic matching, obtains
To respective matching fraction, it is normalized to matching fraction with specific normalized function, transforms into new feature
Vector;
Step 3:Svm classifier model parameter is optimized by GA, obtains the σ of optimum penalty factor and gaussian kernel function
Value;
Step 4:The new feature vector after normalized is merged in fraction level with the sorting parameter model of optimization.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610905400.6A CN106548134A (en) | 2016-10-17 | 2016-10-17 | GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610905400.6A CN106548134A (en) | 2016-10-17 | 2016-10-17 | GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106548134A true CN106548134A (en) | 2017-03-29 |
Family
ID=58369114
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610905400.6A Withdrawn CN106548134A (en) | 2016-10-17 | 2016-10-17 | GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106548134A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107403161A (en) * | 2017-07-31 | 2017-11-28 | 歌尔科技有限公司 | Biological feather recognition method and device |
CN111191065A (en) * | 2019-12-18 | 2020-05-22 | 海尔优家智能科技(北京)有限公司 | Homologous image determining method and device |
CN111310851A (en) * | 2020-03-03 | 2020-06-19 | 四川大学华西第二医院 | Artificial intelligence ultrasonic auxiliary system and application thereof |
CN112269880A (en) * | 2020-11-04 | 2021-01-26 | 吾征智能技术(北京)有限公司 | Sweet text classification matching system based on linear function |
CN113822308A (en) * | 2020-06-20 | 2021-12-21 | 北京眼神智能科技有限公司 | Comparison score fusion method, device, medium and equipment for multi-modal biological recognition |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101551857A (en) * | 2009-05-13 | 2009-10-07 | 西南交通大学 | High-precise palm-print identifying arithmetic based on single matching fractional layer combination |
CN101901336A (en) * | 2010-06-11 | 2010-12-01 | 哈尔滨工程大学 | Fingerprint and finger vein bimodal recognition decision level fusion method |
CN101957913A (en) * | 2010-10-19 | 2011-01-26 | 北京邮电大学 | Information fusion technology-based fingerprint identification method and device |
CN102542258A (en) * | 2011-12-16 | 2012-07-04 | 天津理工大学 | Imaging device based on finger biometric information and multimoding identity recognition method |
CN103208007A (en) * | 2013-03-19 | 2013-07-17 | 湖北微驾技术有限公司 | Face recognition method based on support vector machine and genetic algorithm |
CN104112125A (en) * | 2014-07-24 | 2014-10-22 | 大连大学 | Method for identity recognition based on palm print and finger crease feature fusion |
CN106022218A (en) * | 2016-05-06 | 2016-10-12 | 浙江工业大学 | Palm print palm vein image layer fusion method based on wavelet transformation and Gabor filter |
-
2016
- 2016-10-17 CN CN201610905400.6A patent/CN106548134A/en not_active Withdrawn
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101551857A (en) * | 2009-05-13 | 2009-10-07 | 西南交通大学 | High-precise palm-print identifying arithmetic based on single matching fractional layer combination |
CN101901336A (en) * | 2010-06-11 | 2010-12-01 | 哈尔滨工程大学 | Fingerprint and finger vein bimodal recognition decision level fusion method |
CN101957913A (en) * | 2010-10-19 | 2011-01-26 | 北京邮电大学 | Information fusion technology-based fingerprint identification method and device |
CN102542258A (en) * | 2011-12-16 | 2012-07-04 | 天津理工大学 | Imaging device based on finger biometric information and multimoding identity recognition method |
CN103208007A (en) * | 2013-03-19 | 2013-07-17 | 湖北微驾技术有限公司 | Face recognition method based on support vector machine and genetic algorithm |
CN104112125A (en) * | 2014-07-24 | 2014-10-22 | 大连大学 | Method for identity recognition based on palm print and finger crease feature fusion |
CN106022218A (en) * | 2016-05-06 | 2016-10-12 | 浙江工业大学 | Palm print palm vein image layer fusion method based on wavelet transformation and Gabor filter |
Non-Patent Citations (1)
Title |
---|
陈东莉: "基于GA和KNN的SVM决策树分类方法研究", 《计算机与数字工程》 * |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107403161A (en) * | 2017-07-31 | 2017-11-28 | 歌尔科技有限公司 | Biological feather recognition method and device |
WO2019024350A1 (en) * | 2017-07-31 | 2019-02-07 | 歌尔科技有限公司 | Biometric recognition method and apparatus |
CN107403161B (en) * | 2017-07-31 | 2019-07-05 | 歌尔科技有限公司 | Biological feather recognition method and device |
CN111191065A (en) * | 2019-12-18 | 2020-05-22 | 海尔优家智能科技(北京)有限公司 | Homologous image determining method and device |
CN111191065B (en) * | 2019-12-18 | 2023-10-31 | 海尔优家智能科技(北京)有限公司 | Homologous image determining method and device |
CN111310851A (en) * | 2020-03-03 | 2020-06-19 | 四川大学华西第二医院 | Artificial intelligence ultrasonic auxiliary system and application thereof |
CN113822308A (en) * | 2020-06-20 | 2021-12-21 | 北京眼神智能科技有限公司 | Comparison score fusion method, device, medium and equipment for multi-modal biological recognition |
CN113822308B (en) * | 2020-06-20 | 2024-04-05 | 北京眼神智能科技有限公司 | Multi-mode biological recognition comparison score fusion method, device, medium and equipment |
CN112269880A (en) * | 2020-11-04 | 2021-01-26 | 吾征智能技术(北京)有限公司 | Sweet text classification matching system based on linear function |
CN112269880B (en) * | 2020-11-04 | 2024-02-09 | 吾征智能技术(北京)有限公司 | Sweet text classification matching system based on linear function |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Yang et al. | Finger vein recognition based on (2D) 2 PCA and metric learning | |
CN106548134A (en) | GA optimizes palmmprint and the vena metacarpea fusion identification method that SVM and normalization combine | |
Kang et al. | From noise to feature: Exploiting intensity distribution as a novel soft biometric trait for finger vein recognition | |
Kokkinos | Highly accurate boundary detection and grouping | |
Khalil-Hani et al. | A convolutional neural network approach for face verification | |
CN108776774A (en) | A kind of human facial expression recognition method based on complexity categorization of perception algorithm | |
CN105138968A (en) | Face authentication method and device | |
CN104036254A (en) | Face recognition method | |
Kokkinos et al. | Bottom-up & top-down object detection using primal sketch features and graphical models | |
CN104021384A (en) | Face recognition method and device | |
CN112488992B (en) | Method, medium and electronic equipment for judging mutation state of epidermal growth factor receptor | |
CN112597812A (en) | Finger vein identification method and system based on convolutional neural network and SIFT algorithm | |
CN106971180A (en) | A kind of micro- expression recognition method based on the sparse transfer learning of voice dictionary | |
Zhang et al. | Fused one-vs-all mid-level features for fine-grained visual categorization | |
Kawulok et al. | Skin detection using spatial analysis with adaptive seed | |
Warrell et al. | Labelfaces: Parsing facial features by multiclass labeling with an epitome prior | |
CN112507804A (en) | Identity recognition method based on hand image | |
Park | A feature selection method using hierarchical clustering | |
CN108154186A (en) | One mode recognition methods and device | |
Wijaya et al. | Pornographic image recognition based on skin probability and eigenporn of skin ROIs images | |
Zhou et al. | Real-time gender recognition based on eigen-features selection from facial images | |
Gona et al. | Multimodal biometric reorganization system using deep learning convolutional neural network | |
CN113920573B (en) | Face change decoupling relativity relationship verification method based on counterstudy | |
CN113887509B (en) | Rapid multi-modal video face recognition method based on image set | |
Zeng et al. | Local discriminant training and global optimization for convolutional neural network based handwritten Chinese character recognition |
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 | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20170329 |
|
WW01 | Invention patent application withdrawn after publication |