fusion by biometrics

81
1 Fusion by Fusion by Biometrics Biometrics 主主主 主主主主 主主主 :、 主主主 主主主主 主主主 :、 主主主主 主主主 主主主主 主主主

Upload: aldan

Post on 16-Jan-2016

33 views

Category:

Documents


0 download

DESCRIPTION

Fusion by Biometrics. 主講人:李 佳明、陳明暘 指導教授:林維暘. Outline. Introduction Score normalization methods Fusion methods Experiment Results Conclusion. Reference Paper. A. K. Jain, K. Nandakumar, and A. Ross, “ Score normalization in multimodal biometric systems," Pattern Recognition , 2005. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Fusion by Biometrics

11

Fusion by BiometricsFusion by Biometrics

主講人:李佳明、陳明暘主講人:李佳明、陳明暘指導教授:林維暘指導教授:林維暘

Page 2: Fusion by Biometrics

22

OutlineOutline

IntroductionIntroduction Score normalization methodsScore normalization methods Fusion methodsFusion methods Experiment ResultsExperiment Results ConclusionConclusion

Page 3: Fusion by Biometrics

33

Reference PaperReference Paper

A. K. Jain, K. Nandakumar, and A. Ross, “Score normalization in multimodal biometric systems," Pattern Recognition , 2005.

Page 4: Fusion by Biometrics

44

Why score normalization ?Why score normalization ?

1. The matching scores at the output of the individual matchers may not be homogeneous.

2. The outputs of the individual matchers need not be on the same numerical scale (range).

3. The matching scores at the output of the matchers may follow different statistical distributions.

Page 5: Fusion by Biometrics

55

Why score normalization ?Why score normalization ? Score normalization refers to changing the location

and scale parameters of the matching score distributions at the output of the individual matchers, so that the matching scores of different matchers are transformed into a common domain.

Page 6: Fusion by Biometrics

66

Score normalizationScore normalization

When the parameters used for normalization are determined using a fixed training set, it is referred to as fixed score normalization.

In adaptive score normalization, the normalization parameters are estimated based on the current feature vector.

Page 7: Fusion by Biometrics

77

A good normalization scheme

Robustness refers to insensitivity to the presence of outliers.

Efficiency refers to the proximity of the obtained estimate to the optimal estimate when the distribution of the data is known.

Page 8: Fusion by Biometrics

88

Normalization Techniques

1. Min-max 2. Decimal scaling 3. z-score 4. Median and MAD 5. Double sigmoid function 6. tanh-estimators

Page 9: Fusion by Biometrics

99

1. Min-max normalization Min-max normalization is best suited for the case

where the bounds (maximum and minimum values) of the scores produced by a matcher are known.

We usually shift the minimum and maximum scores to 0 and 1.

xk : the kth matching score before normalization xk’ : the kth matching score after normalization

Page 10: Fusion by Biometrics

1010

1. Min-max normalization This method is not robust (i.e., the method is highly

sensitive to outliers in the data used for estimation). Min-max normalization retains the original

distribution of scores except for a scaling factor and transforms all the scores into a common range [0, 1].

Page 11: Fusion by Biometrics

1111

2. Decimal scaling For example, if one matcher has scores in

the range [0, 1] and the other has scores in the range [0, 1000], the following normalization could be applied.

The problems with this approach are lack of robustness.

Page 12: Fusion by Biometrics

1212

3. z-score The most commonly used score normalization

technique is the z-score that is calculated using the arithmetic mean and standard deviation of the given data.

Both mean and standard deviation are sensitive to outliers and, hence, this method is not robust.

Z-score normalization does not guarantee a common numerical range for the normalized scores of the different matchers.

If the input scores are not Gaussian distributed, z-score normalization does not retain the input distribution at the output.

Page 13: Fusion by Biometrics

1313

3. z-score

Page 14: Fusion by Biometrics

1414

4. Median and MAD Robust : The median and median absolute

deviation (MAD) are insensitive to outliers and the points in the extreme tails of the distribution.

This normalization technique does not retain the input distribution and does not transform the scores into a common numerical range.

Page 15: Fusion by Biometrics

1515

4. Median and MAD

Page 16: Fusion by Biometrics

1616

5. Double sigmoid function

The normalized score is given by

where m is the reference operating point and s1 and s2 denote the left and right edges of the region

Page 17: Fusion by Biometrics

1717

5. Double sigmoid function

where the scores in the [0, 300] range are mapped to the [0, 1] range using m = 200, s1 = 20 and s2 = 30.

Generally, m is chosen to be some value falling in the region of overlap between the genuine and impostor score distribution, and s1 and s2 are made equal to the extent of overlap between the two distributions toward the left and right of m, respectively.

Page 18: Fusion by Biometrics

1818

5. Double sigmoid function

Page 19: Fusion by Biometrics

1919

6. tanh-estimators The tanh-estimators introduced by Hampel et al. are

robust and highly efficient. The normalization is given by

where μGH and σGH are the mean and standard deviation estimates, respectively, of the genuine score distribution as given by Hampel estimators

Hampel estimatorsare based on the following influence ( )-function:

Page 20: Fusion by Biometrics

2020

6. tanh-estimators

Page 21: Fusion by Biometrics

2121

Summary of Normalization Techniques

Page 22: Fusion by Biometrics

2222

Experimental Results Database of 100 users with three modalities. Each user having five biometric templates for each

modality.

Page 23: Fusion by Biometrics

2323

Experimental Results

Page 24: Fusion by Biometrics

2424

Experimental Results

Page 25: Fusion by Biometrics

2525

Experimental Results

Page 26: Fusion by Biometrics

2626

Experimental Results

Page 27: Fusion by Biometrics

2727

Experimental Results

Page 28: Fusion by Biometrics

2828

Experimental Results

Page 29: Fusion by Biometrics

2929

Experimental Results

Page 30: Fusion by Biometrics

3030

Feature Level FusionFeature Level Fusion

“Biometric A” feature vectors : X “Biometric B” feature vectors : Y

Normalization -> X’ , Y’ Dimension reduction Combine two vector -> Z’ = { X’ , Y’ }

Page 31: Fusion by Biometrics

3131

Reference PaperReference Paper A. Ross and R. Govindarajan. “Feature Level

Fusion Using Hand and Face Biometrics.” In Proc. SPIE Conf. on Biometric Technology for Human Identication II, volume 5779, pages 196-204,Orlando, 2005.

SON, B. and LEE, Y. “The Fusion of Two User-SON, B. and LEE, Y. “The Fusion of Two User-friendly Biometric Modalities: Iris and Face”, friendly Biometric Modalities: Iris and Face”, IEICE Transactions on Information and Systems , IEICE Transactions on Information and Systems , 2006.2006.

Page 32: Fusion by Biometrics

3232

Fusion in feature and matching Fusion in feature and matching levellevel

Normalization method : median and MADNormalization method : median and MAD Dimension reduction method : PCA , LDA Matching score fusion method : sum rule

Consider feature vectors {Xi , Yi} and {Xj , Yj} obtained at two different time instances i and j. Fusion in feature level -> { Zi , Zj } Let sX and sY be the normalized match

(Euclidean distance) scores generated by comparing Xi with Xj and Yi with Yj , respectively.

smatch = (sX + sY)/2 be the fused match score obtained using the simple sum rule.

Page 33: Fusion by Biometrics

3333

Experimentation A set of 500 face images and hand images were

acquired from 100 users (5 biometric samples per user per biometric)

Each face image was decomposed into its component R, G, B channels. Further, the grayscale rendition of the color image - I - was also computed.

The Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA) were performed on these component images (i.e., R, G, B, I) in order to extract representational features.

Page 34: Fusion by Biometrics

3434

Experimental Results

Page 35: Fusion by Biometrics

3535

Experimental Results

Page 36: Fusion by Biometrics

3636

The Fusion of Two User-friendly The Fusion of Two User-friendly Biometric Modalities: Iris and FaceBiometric Modalities: Iris and Face

Page 37: Fusion by Biometrics

3737

Fusion methodFusion method Dimension reduction method :

Wavelet Transform LDA

Page 38: Fusion by Biometrics

3838

Experimentation

Face Databases: IISFace : We sampled frontal face images of 100

subjects from the IIS face database. Each subject has 10 images with varying expressions.

Iris Databases: Iris1 : This data set consists of 1000 iris images

acquired from 100 individuals. (good quality images)

Iris2 : The Iris2 database consists of 1000 iris images containing some bad quality ones acquired from 100 individuals.

Page 39: Fusion by Biometrics

3939

Experimental Results

Page 40: Fusion by Biometrics

4040

Experimentation Face Databases:

ORLFace : The ORL data set consists of 400 frontal faces: 10 tightly cropped images of 40 subjects with variations in poses, illuminations, facial expressions and accessories.

Iris Databases: Iris3 : The Iris3 database is composed of 400

good quality images sampled from the Iris1 database to combine with the ORLFace database.

Iris4 : The Iris4 is composed of 400 iris images containing some bad quality ones sampled from the Iris2 database to combine with the ORLFace.

Page 41: Fusion by Biometrics

4141

Experimental Results

Page 42: Fusion by Biometrics

4242

Reference PaperReference Paper

M. Indovina, U. Uludag, R. Snelick, A. M. Indovina, U. Uludag, R. Snelick, A. Mink and A. Jain, "Multimodal Biometric Mink and A. Jain, "Multimodal Biometric Authentication Methods: A COTS Authentication Methods: A COTS Approach", Approach", Proc. MMUA 2003, Workshop Proc. MMUA 2003, Workshop on Multimodal User Authenticationon Multimodal User Authentication, pp. , pp. 99-106, Santa Barbara, CA, December 99-106, Santa Barbara, CA, December 11-12, 2003. 11-12, 2003.

Page 43: Fusion by Biometrics

4343

abstract

We examine the performance of multimodal biometric authentication systems using Commercial Off-the-Shelf (COTS) fingerprint and face biometrics.

It introduce novel methods of fusion and normalization that improve accuracy still further through population analysis.

Page 44: Fusion by Biometrics

4444

Normalization methods

a matcher score as s from the set S of all scores for that matcher and the corresponding normalized score as n.

Min-MaxMin-Max maps the scores to the [0, 1] range. maps the scores to the [0, 1] range.

Page 45: Fusion by Biometrics

4545

Normalization methods

Z-scoreZ-score transforms the scores to a distribution with transforms the scores to a distribution with

mean of 0 and standard deviation of 1. mean of 0 and standard deviation of 1.

TanhTanh robust statistical techniques. maps the scores to the (0, 1) range.

Page 46: Fusion by Biometrics

4646

Normalization methods

AdaptiveAdaptive Using an adaptive normalization procedure

that aims to increase the separation of the genuine and impostor distributions.

Page 47: Fusion by Biometrics

4747

Normalization methods

Two-Quadrics composed of 2 quadratic segments that

change concavity at c.

Page 48: Fusion by Biometrics

4848

Normalization methods LogisticLogistic

logistic function The general shape of the curve is similar to Two-

Quadrics

f(0) is equal to the constant Δ, which is selected to be a small value (0.01 in this study).

Page 49: Fusion by Biometrics

4949

Normalization methods

Quadric-Line-Quadric The overlapped zone, w, is left unchanged

while the other regions are mapped with two quadratic function segments.

Page 50: Fusion by Biometrics

5050

Fusion methodsFusion methods

Simple Sum Scores for an individual are summed.

Min Score Choose the minimum of an individual’s

scores.

Max Score Choose the maximum of an individual’s

scores.

Page 51: Fusion by Biometrics

5151

Fusion methodsFusion methods

Matcher Weighting Matcher weighting-based fusion makes use of

the Equal Error Rate (EER).

the weights are inversely proportional to the corresponding errors.

The fused score is calculated as

Page 52: Fusion by Biometrics

5252

Fusion methodsFusion methods

EER if the score distributions overlap, the FAR and

FRR intersect at a certain point. The value of the FAR and the FRR at this point, which is of course the same for both of them, is called the Equal Error Rate (EER).

Page 53: Fusion by Biometrics

5353

Fusion methodsFusion methods User Weighting

This method applies weights to individual matchers differently for every user (individual).

The calculation of these user-dependent weights make use of the wolf-lamb concept introduced by Doddington, et al.

Page 54: Fusion by Biometrics

5454

Fusion methodsFusion methods

We assume that for every (i, m) pair, the mean and standard deviation of the associated genuine and impostor distributions are known.

use the d-prime metric as a measure of the separation of these two distributions.

Page 55: Fusion by Biometrics

5555

Experiment ResultsExperiment Results

The best EER values in individual columns are indicated with bold typeface; the best EER values in individual rows are indicated with a star (*) symbol.

Page 56: Fusion by Biometrics

5656

Experiment ResultsExperiment Results

Normalization: varied, fusion: Simple Normalization: varied, fusion: Simple Sum Sum

Page 57: Fusion by Biometrics

5757

Experiment ResultsExperiment Results

Normalization: varied, fusion: Min ScoreNormalization: varied, fusion: Min Score

Page 58: Fusion by Biometrics

5858

Experiment ResultsExperiment Results

Normalization: varied, fusion: Max ScoreNormalization: varied, fusion: Max Score

Page 59: Fusion by Biometrics

5959

Experiment ResultsExperiment Results

Normalization: varied, fusion: Matcher Normalization: varied, fusion: Matcher weightweight

Page 60: Fusion by Biometrics

6060

Experiment ResultsExperiment Results

Normalization: varied, fusion: User Normalization: varied, fusion: User weightweight

Page 61: Fusion by Biometrics

6161

Experiment ResultsExperiment Results

Normalization: Min-Max, fusion: variedNormalization: Min-Max, fusion: varied

Page 62: Fusion by Biometrics

6262

Experiment ResultsExperiment Results

Normalization: Z-score, fusion: variedNormalization: Z-score, fusion: varied

Page 63: Fusion by Biometrics

6363

Experiment ResultsExperiment Results

Normalization: Tanh, fusion: variedNormalization: Tanh, fusion: varied

Page 64: Fusion by Biometrics

6464

Experiment ResultsExperiment Results

Normalization: Normalization: Quadric-Line-Quadric, , fusion: variedfusion: varied

Page 65: Fusion by Biometrics

6565

Experiment ResultsExperiment Results

Normalization: Normalization: Quadric-Line-Quadric, , fusion: Simple Sumfusion: Simple Sum

Page 66: Fusion by Biometrics

6666

Reference PaperReference Paper

Y. Wang, T. Tan and A. K. Jain, Y. Wang, T. Tan and A. K. Jain, "Combining Face and Iris Biometrics for "Combining Face and Iris Biometrics for Identity Verification", Identity Verification", Proc. of 4th Int'l Proc. of 4th Int'l Conf. on Audio- and Video-Based Conf. on Audio- and Video-Based Biometric Person Authentication Biometric Person Authentication (AVBPA)(AVBPA), pp. 805-813, Guildford, UK, , pp. 805-813, Guildford, UK, June 9-11, 2003. June 9-11, 2003.

Page 67: Fusion by Biometrics

6767

abstractabstract

Combining Face and Iris to improve Combining Face and Iris to improve verification performance.verification performance.

Fusing in scores level.Fusing in scores level. We use two different strategies.

The first strategy is to compute either an unweighted or weighted sum .

The second strategy is to treat the matching distances of face and iris classifiers as a two-dimensional feature vector and to use a classifier such as Fisher’s discriminant analysis.

Page 68: Fusion by Biometrics

6868

Fusion methodsFusion methods

Weighted Sum RuleWeighted Sum Rule

Page 69: Fusion by Biometrics

6969

Fusion methodsFusion methods

Fisher Discriminant AnalysisFisher Discriminant Analysis we treat the face and iris matcher outputs x1 we treat the face and iris matcher outputs x1

and x2 as a feature vector X = (x1 ,x2).and x2 as a feature vector X = (x1 ,x2).

Decision RuleDecision Rule

Page 70: Fusion by Biometrics

7070

Experiment ResultsExperiment Results

Distribution of matching distancesDistribution of matching distances

Page 71: Fusion by Biometrics

7171

Experiment ResultsExperiment Results

Total error rateTotal error rate

Page 72: Fusion by Biometrics

7272

Reference PaperReference Paper

Michalis Petrakos, Jon Atli Benediktsson, and Ioannis Kanellopoulos,” The Effect of Classifier Agreement on the Accuracy of the Combined Classifier in Decision Level Fusion”, IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 39, NO. 11, NOVEMBER 2001

Page 73: Fusion by Biometrics

7373

abstractabstract Agreement among classifiers can inhibit the

gains obtained regardless of the method used to combine them. In this work, the level of agreement between different classifiers used in remote sensing is assessed based on statistical measures.

A study is performed in which an image is classified by several methods with different degrees of agreement between them.

As stated previously, the LOP and the LOGP are widely used decision fusion approaches.

Page 74: Fusion by Biometrics

7474

Fusion methodsFusion methods

LOP (Linear Opinion Pool)LOP (Linear Opinion Pool)

LOGP (Logarithmic Opinion Pool)LOGP (Logarithmic Opinion Pool)

Page 75: Fusion by Biometrics

7575

Fusion methodsFusion methods

Then we can combine LOP/LOGP and Then we can combine LOP/LOGP and neural network.neural network.

Their combination schemes were considered to be two stage processes: statistical classifiers in stage one, and a single neural network in stage two.

Page 76: Fusion by Biometrics

7676

Fusion methodsFusion methods

Their suggested voting schemes were: Majority Voting: When the majority of the

individual agree on the classification of a sample, the sample is classified to that class.

Complete Agreement: When all the individual source-specific classifiers agree on the classification of a sample, the sample is classified to that class.

CONSNN-NN: A neural network and a LOGP classifier are trained separately on all the data.

Page 77: Fusion by Biometrics

7777

Experiment ResultsExperiment Results

DatabaseDatabase

Page 78: Fusion by Biometrics

7878

Experiment ResultsExperiment Results The minimum distance (MD) computes the

Mahalanobis distance with every one of the 12 land-use classes for each sample, and assigns the sample to the class with the smallest distance.

Linear discriminant analysis (LDA) assigns a sample to the class with the maximum posterior probability given that all classes have a common (pooled) covariance matrix.

Quadratic discriminant analysis (QDA) assigns a sample to the class with the maximum posterior probability. In this case, each class has its own covariance matrix.

A conjugate gradient neural network (CGNN) with 13 inputs, one hidden layer containing 20 neurons, and 12 outputs.

Page 79: Fusion by Biometrics

7979

Experiment ResultsExperiment Results

Page 80: Fusion by Biometrics

8080

Experiment ResultsExperiment Results

Page 81: Fusion by Biometrics

8181

ConclusionConclusion

Some fusion methods are the best in a biometric system, but they maybe not the best in other biometric system.

The effects of different score normalization techniques in a multimodal biometric system have been studied.