Download - 1 Biometrics and Cryptography -- Finger Biometric CPSC 4600/5600 Biometric and Cryptography University of Tennessee at Chattanooga.

11

Biometrics and Biometrics and Cryptography -- Finger Cryptography -- Finger

Biometric Biometric CPSC 4600/5600 Biometric and CPSC 4600/5600 Biometric and

CryptographyCryptography

University of Tennessee at University of Tennessee at ChattanoogaChattanooga

22

Fingerprint Identification Fingerprint Identification

• Among all the biometric techniques, Among all the biometric techniques, fingerprint-based identification is the fingerprint-based identification is the oldest method which has been oldest method which has been successfully used in numerous successfully used in numerous applications. applications.

• Fingerprinting was first created by Dr. Fingerprinting was first created by Dr. Henry Fault, a British surgeon.Henry Fault, a British surgeon.

• Everyone is known to have unique, Everyone is known to have unique, immutable fingerprints. immutable fingerprints.

• A fingerprint is made of a series of ridges A fingerprint is made of a series of ridges and valleys on the surface of the finger. and valleys on the surface of the finger.

33

Fingerprint IdentificationFingerprint Identification

• The uniqueness of a The uniqueness of a fingerprint can be fingerprint can be determined by the pattern determined by the pattern of of ridgesridges and and valleysvalleys as as well as the minutiae points. well as the minutiae points.

• Minutiae Minutiae points are local points are local ridge characteristics that ridge characteristics that occur at either a ridge occur at either a ridge bifurcation or a ridge bifurcation or a ridge ending.ending.

44

Fingerprint ReadersFingerprint Readers

55

Fingerprint BasicsFingerprint Basics

• A fingerprint has A fingerprint has many identification many identification and classification and classification basics basics

66

Fingerprint Basics Fingerprint Basics (minutiae)(minutiae)

Bifurcation Ridge ending

dotDouble bifurcation

77


Opposed bifurcation Island (short ridge)

Hook (spur) Lake (enclosure)

88


Ridge crossing

trifurcation

Opposed bifurcation/ridge ending)

Bridge

99

Fingerprint BasicsFingerprint Basics

• How many How many different ridge different ridge characteristics can characteristics can you see?you see?

1010

Fingerprint IdentificationsFingerprint Identifications

• A single rolled fingerprint may have as many A single rolled fingerprint may have as many as 100 or more identification points that can as 100 or more identification points that can be used for identification purposes. be used for identification purposes.

• There is no exact size requirement as the There is no exact size requirement as the number of points found on a fingerprint number of points found on a fingerprint impression depend on the location of the print. impression depend on the location of the print.

• As an example the area immediately As an example the area immediately surrounding a delta will probably contain more surrounding a delta will probably contain more points per square millimeter than the area points per square millimeter than the area near the tip of the finger which tends to not near the tip of the finger which tends to not have that many points. have that many points.

1111

Schematic – data storage and Schematic – data storage and processing in finger-scan systemsprocessing in finger-scan systems

1212

Schematic – data storage and Schematic – data storage and processing in finger-scan systemsprocessing in finger-scan systems

1313

General Model for Fingerprint Authentication

Raw data Extracted features template

Authentication decision

Data collection Signal

proc.

matching storage

Match score

decision Application

1414

Fingerprint ClassificationFingerprint Classification

• Large volumesLarge volumes of fingerprints are collected and of fingerprints are collected and stored everyday in applications such as stored everyday in applications such as forensics, access control, and driver license forensics, access control, and driver license registration. registration.

• An automatic recognition of people based on An automatic recognition of people based on fingerprints requires that the input fingerprint be fingerprints requires that the input fingerprint be matched with matched with a large number of fingerprintsa large number of fingerprints in a in a database (FBI database contains approximately database (FBI database contains approximately 70 million fingerprints!). 70 million fingerprints!).

• Classifying these fingerprints can Classifying these fingerprints can reduce the reduce the search time and computational complexitysearch time and computational complexity, so , so that the input fingerprint is required to be that the input fingerprint is required to be matched only with a subset of the fingerprints in matched only with a subset of the fingerprints in the database. the database.

1515

Fingerprint ClassificationFingerprint Classification• Some fingerprint identification systems

use manual classification followed by automatic minutiae matching;

• Automating the classification process would improve its speed and cost-effectiveness.

• PCASYS is to build a prototype classifier that separates fingerprints into basic pattern-level classes known as arch, left loop, right loop, scar, tented arch, and whorl.

1616


Arch Left loopRight loop

1717


Scar Tented arch Whorl

1818

• The loop is by far the most common The loop is by far the most common type of fingerprints.type of fingerprints.

• The human population has The human population has fingerprints in the following fingerprints in the following percentages:percentages:– Loop – 65%Loop – 65%– Whorl -- 30%Whorl -- 30%– Arch -- 5%Arch -- 5%


1919

Minutiae DetectionMinutiae Detection• Human fingerprints are Human fingerprints are uniqueunique to each person, to each person,

certifying the person's identity. certifying the person's identity. • Because Because straightforward matching straightforward matching between the between the

unknown and known fingerprint patterns unknown and known fingerprint patterns is highly is highly sensitive to errors sensitive to errors (e.g. various noises, damaged (e.g. various noises, damaged fingerprint areas, or the finger being placed in fingerprint areas, or the finger being placed in different areas of fingerprint scanner window and different areas of fingerprint scanner window and with different orientation angles, finger with different orientation angles, finger deformation during the scanning procedure etc.). deformation during the scanning procedure etc.).

• Modern techniques focus on extracting minutiae Modern techniques focus on extracting minutiae points points (points where capillary lines have branches (points where capillary lines have branches or ends) from the fingerprint image, and check or ends) from the fingerprint image, and check matching between the sets of fingerprint features. matching between the sets of fingerprint features.

2020

Minutiae Detection -- Minutiae Detection -- PreprocessingPreprocessing• Image ProcessingImage Processing

– Capture the fingerprint images and process them through a Capture the fingerprint images and process them through a series of image processing algorithms to obtain a clear series of image processing algorithms to obtain a clear unambiguous skeletal image of the original gray tone unambiguous skeletal image of the original gray tone impression, clarifying smudged areas, removing extraneous impression, clarifying smudged areas, removing extraneous artifacts and healing most scars, cuts and breaks. artifacts and healing most scars, cuts and breaks.

Original image Undesirable features marked Final image

2121

Minutiae DetectionMinutiae Detection

• Two fingerprints have been compared using discrete features called minutiae.

• These features include points in a finger's friction skin where ridges end (called a ridge ending) or split (called a ridge bifurcation).

• There are on the order of 100 minutiae on a tenprint. Minutiae: bifurcation (square marker)

and ridge ending (circle marker).

2222

• Feature Detection for Feature Detection for MatchingMatchingRidge endsRidge ends and and bifurcationsbifurcations (minutiae) within the skeletal (minutiae) within the skeletal image are identified and image are identified and encoded, providing critical encoded, providing critical placement, orientation and placement, orientation and linkage information for the linkage information for the fingerprint matching process.fingerprint matching process.


2323

Minutiae DetectionMinutiae Detection• The location of each minutia is represented by a

coordinate location within the fingerprint's image from an origin in the bottom left corner of the image.

• Minutiae orientation is represented in degrees, with zero degrees pointing horizontal and to the right, and increasing degrees proceeding counter-clockwise.

A. standard angle, B. FBI/IAFIS angle

2424


• A good reliable fingerprint processing A good reliable fingerprint processing technique requires sophisticated algorithms for technique requires sophisticated algorithms for reliable processing of the fingerprint image:reliable processing of the fingerprint image:– noise elimination, noise elimination, – minutiae extraction, minutiae extraction, – rotation and translation-tolerant fingerprint rotation and translation-tolerant fingerprint

matching. matching.

• At the same time, the algorithms must be as At the same time, the algorithms must be as fast as possible for comfortable use in fast as possible for comfortable use in applications with large number of users. It applications with large number of users. It must also be able to fit into a microchip.must also be able to fit into a microchip.

2525

Minutiae Detection – Extraction Minutiae Detection – Extraction ProcessProcess

2626

Latent FingerprintsLatent Fingerprints

• In addition to tenprints, there is a smaller population of fingerprints also important to the FBI.

• These are fingerprints captured at crime scenes that can be used as evidence in solving criminal cases.

• Unlike tenprints, which have been captured in a relatively controlled environment for the expressed purpose of identification, crime scene fingerprints are by nature incidentally left behind.

• They are often invisible to the eye without some type of chemical processing or dusting.

• It is for this reason that they have been traditionally called latent fingerprints.

2727

Latent FingerprintsLatent Fingerprints• Typically, only a portion of the finger is present in the latent,

the surface on which the latent was imprinted is unpredictable, and the clarity of friction skin details are often blurred or occluded.

• All this leads to fingerprints of significantly lesser quality than typical tenprints.

• While there are 100 minutiae on a tenprint, there may be only a dozen on a latent.

2828


• Due to the poor conditions of latent fingerprints, today's fingerprint technology operates poorly when presented a latent fingerprint image. It is extremely difficult for the automated system to accurately classify latent fingerprints and reliably locate the minutiae in the image.

• Consequently, human fingerprint experts, called latent examiners, must analyze and manually mark up each latent fingerprint in preparation for matching.

2929


•FBI and NIST collaboratively developed a specialized workstation called the Universal Latent Workstation (ULW).

•FBI has chosen to distribute the ULW freely upon request.

3030

• The fingerprint matcher compares data from the The fingerprint matcher compares data from the input search print against all appropriate records input search print against all appropriate records in the database to determine if a probable match in the database to determine if a probable match exists. exists.

• Minutiae relationships, one to another are Minutiae relationships, one to another are compared. Not as locations within an X-Y co-compared. Not as locations within an X-Y co-ordinate framework, but as linked relationships ordinate framework, but as linked relationships within a global context. within a global context.

Compare

Stored imageLive image

Fingerprint MatchingFingerprint Matching

3131

• Each template comprises a multiplicity of Each template comprises a multiplicity of information chunks, every information chunk information chunks, every information chunk representing a minutia and comprising a representing a minutia and comprising a sitesite, a , a minutia slantminutia slant and a and a neighborhoodneighborhood. .

• Each site is represented by two coordinates. [ l = Each site is represented by two coordinates. [ l = (x,y)](x,y)]

• The neighborhood comprises of positional The neighborhood comprises of positional parameters with respect to a chosen minutia for a parameters with respect to a chosen minutia for a predetermined figure of neighbor minutiae. In single predetermined figure of neighbor minutiae. In single embodiment, a neighborhood border is drown about embodiment, a neighborhood border is drown about the chosen minutia and neighbor minutiae are the chosen minutia and neighbor minutiae are chosen from the enclosed region. [ theta]chosen from the enclosed region. [ theta]

• A live template is compared to a stored measured A live template is compared to a stored measured template chunk-by-chunk. A chunk from the template chunk-by-chunk. A chunk from the template is loaded in a random access memory template is loaded in a random access memory (RAM). (RAM).


3232


• The site, minutia slant and neighborhood of the The site, minutia slant and neighborhood of the reference information chunkreference information chunk are compared with are compared with the site, minutia slant and neighborhood of the the site, minutia slant and neighborhood of the stored template ( latent) information chunkstored template ( latent) information chunk by by information chunk. information chunk.

• The neighborhoods are compared by comparing The neighborhoods are compared by comparing every positional argument. If every positional every positional argument. If every positional parameters match, the neighbors match. If a parameters match, the neighbors match. If a predetermined figure of neighbor matches is met, predetermined figure of neighbor matches is met, the neighborhoods match.the neighborhoods match.

• If the matching rate of all information chunks is If the matching rate of all information chunks is equivalent to or superior to the predetermined equivalent to or superior to the predetermined information chunk rate, the live template information chunk rate, the live template matchesmatches the stored (latent) template. the stored (latent) template.

3333

Characteristics of Fingerprint Characteristics of Fingerprint TechnologyTechnology• Biometric (Fingerprint) StrengthsBiometric (Fingerprint) Strengths

– Finger tip most mature measureFinger tip most mature measure– Accepted reliabilityAccepted reliability– High quality imagesHigh quality images– Small physical sizeSmall physical size– Low costLow cost– Low False Acceptance Rate (FAR)Low False Acceptance Rate (FAR)– Small template (less than 500 bytes)Small template (less than 500 bytes)

3434

Characteristics of Fingerprint Characteristics of Fingerprint TechnologyTechnology• Biometric (Fingerprint weaknesses)Biometric (Fingerprint weaknesses)

– Requires careful enrollmentRequires careful enrollment– Potential high False Reject Rate (FRR) Potential high False Reject Rate (FRR)

due to:due to:•Pressing too hard, scarring, misalignment, Pressing too hard, scarring, misalignment,

dirtdirt

– Vendor incompatibilityVendor incompatibility– Cultural issuesCultural issues

•Physical contact requirement a negative in Physical contact requirement a negative in JapanJapan

•Perceived privacy issues with North AmericaPerceived privacy issues with North America

3535

Fake Finger DetectionFake Finger Detection

• As any other authentication technique, fingerprint As any other authentication technique, fingerprint recognition is not totally spoof-proof. recognition is not totally spoof-proof.

• The main potential threats for fingerprint-based The main potential threats for fingerprint-based systems are: systems are: – attacking the communication channels, including replay attacking the communication channels, including replay

attacks on the channel between the sensor and the rest attacks on the channel between the sensor and the rest of the system; of the system;

– attacking specific software modules (e.g. replacing the attacking specific software modules (e.g. replacing the feature extractor or the matcher with a Trojan horse); feature extractor or the matcher with a Trojan horse);

– attacking the database of enrolled templates; attacking the database of enrolled templates; – presenting fake fingers to the sensor.presenting fake fingers to the sensor.

3636


• The feasibility of the last type of attack has been The feasibility of the last type of attack has been reported by some researchers: they showed that it reported by some researchers: they showed that it is actually possible to spoof some fingerprint is actually possible to spoof some fingerprint recognition systems with well-made fake fingertips, recognition systems with well-made fake fingertips, created with the collaboration of the fingerprint created with the collaboration of the fingerprint owner or from a latent fingerprint: in the latter owner or from a latent fingerprint: in the latter case the procedure is more difficult but still case the procedure is more difficult but still possible.possible.

3737


• Based on the analysis of skin distortionBased on the analysis of skin distortion..– The user is required to move his finger while pressing it against The user is required to move his finger while pressing it against

the scanner surface, thus deliberately exaggerating the skin the scanner surface, thus deliberately exaggerating the skin distortion. distortion.

– When a real finger moves on a scanner surface, it produces a When a real finger moves on a scanner surface, it produces a significant amount of distortion, which can be observed to be significant amount of distortion, which can be observed to be quite different from that produced by fake fingers. quite different from that produced by fake fingers.

– Usually fake fingers are more rigid than skin, then the distortion Usually fake fingers are more rigid than skin, then the distortion is definitely lower; even if highly elastic materials are used, it is definitely lower; even if highly elastic materials are used, it seems very difficult to precisely emulate the specific way a real seems very difficult to precisely emulate the specific way a real finger is distorted, because the behavior is related to the way finger is distorted, because the behavior is related to the way the external skin is anchored to the underlying derma and the external skin is anchored to the underlying derma and influenced by the position and shape of the finger bone. influenced by the position and shape of the finger bone.

• Based on odor analysis. – Electronic noses are used with the aim of detecting the odor of Electronic noses are used with the aim of detecting the odor of

those materials that are typically used to create fake fingers those materials that are typically used to create fake fingers (e.g. silicone or gelatin). (e.g. silicone or gelatin).

3838

Advance of Fingerprint Advance of Fingerprint TechnologyTechnology

• As fingerprint technology matures, As fingerprint technology matures, variations in the technology also increase variations in the technology also increase including:including:– Optical – finger is scanned on a platen ( glass, Optical – finger is scanned on a platen ( glass,

plastic or coasted glass/plastic).plastic or coasted glass/plastic).– Silicon – uses a silicon chip to read the Silicon – uses a silicon chip to read the

capacitance value of the fingerprint. capacitance value of the fingerprint. – Ultrasound – requires a large scanning device. Ultrasound – requires a large scanning device.

It is appealing because it can better permeate It is appealing because it can better permeate dirt. dirt.

3939

• The matching accuracy of a biometrics-based The matching accuracy of a biometrics-based authentication system relies on the stability authentication system relies on the stability (permanence) of the biometric data associated (permanence) of the biometric data associated with an individual over time. with an individual over time.

• In reality, however, the biometric data acquired In reality, however, the biometric data acquired from an individual is susceptible to changes from an individual is susceptible to changes introduced due to improper interaction with the introduced due to improper interaction with the sensor (e.g., partial fingerprints, change in pose sensor (e.g., partial fingerprints, change in pose during face-image acquisition), modifications in during face-image acquisition), modifications in sensor characteristics (e.g., optical vs. solid-state sensor characteristics (e.g., optical vs. solid-state fingerprint sensor), variations in environmental fingerprint sensor), variations in environmental factors (e.g., dry weather resulting in faint factors (e.g., dry weather resulting in faint fingerprints) and temporary alterations in the fingerprints) and temporary alterations in the biometric trait itself (e.g., cuts/scars on biometric trait itself (e.g., cuts/scars on fingerprints).fingerprints).

Change of Fingerprint data

4040

• In other words, the biometric In other words, the biometric measurements tend to have a large measurements tend to have a large intra-class variability.intra-class variability.

• Thus, it is possible for the stored Thus, it is possible for the stored template data to be significantly template data to be significantly different from those obtained during different from those obtained during authentication, resulting in an authentication, resulting in an inferior performance (higher false inferior performance (higher false rejects) of the biometric system.rejects) of the biometric system.

Change of Fingerprint data

4141

Evaluation of Fingerprint Evaluation of Fingerprint TechnologyTechnology

• There are two categories of fingerprint There are two categories of fingerprint matching techniques: matching techniques: minutiae-basedminutiae-based and and correlation basedcorrelation based. . – Minutiae-based techniques first find Minutiae-based techniques first find

minutiae points and then map their minutiae points and then map their relative placement on the finger. relative placement on the finger.

– The correlation-based method is able to The correlation-based method is able to overcome some of the difficulties of the overcome some of the difficulties of the minutiae-based approach. minutiae-based approach.

4242


– Minutiae-based processing has problems Minutiae-based processing has problems including:including:• In real life you would have impressions made at In real life you would have impressions made at

separate times and subject to different pressure separate times and subject to different pressure distortions. distortions.

•On the average, many of these images are On the average, many of these images are relatively clean and clear, however, in many of relatively clean and clear, however, in many of the actually crime scenes, prints are anything the actually crime scenes, prints are anything but clear. but clear.

•There are cases where it is not easy to have a There are cases where it is not easy to have a core pattern and a delta but only a latent that core pattern and a delta but only a latent that could be a fingertip, palm or even foot could be a fingertip, palm or even foot impression impression

•The method does not take into account the The method does not take into account the global pattern of ridges and furrows. global pattern of ridges and furrows.

4343

– Fingerprint matching based on minutiae Fingerprint matching based on minutiae has problems in matching different sized has problems in matching different sized (unregistered) minutiae patterns. (unregistered) minutiae patterns.

– Local ridge structures can not be Local ridge structures can not be completely characterized by minutiae. completely characterized by minutiae.

– The solution is to find an alternate The solution is to find an alternate representation of fingerprints which representation of fingerprints which captures more local information and captures more local information and yields a fixed length code for the yields a fixed length code for the fingerprint. fingerprint.


4444

– Correlation-based processing has its Correlation-based processing has its own problems including:own problems including:

•Correlation-based techniques require Correlation-based techniques require the precise location of a registration the precise location of a registration point point

•It is also affected by image translation It is also affected by image translation and rotation. and rotation.


4545

Hands-On Lab of Finger Hands-On Lab of Finger BiometricBiometric

1.1. Download and install NIST Fingerprint Download and install NIST Fingerprint Image Software 2Image Software 2

2.2. Test and Demo Command PCASYS, Test and Demo Command PCASYS, MINDTCT, NFIQ and BOZORTH3MINDTCT, NFIQ and BOZORTH3

3.3. PCASYS (PACSYSX) and MINDTCT are PCASYS (PACSYSX) and MINDTCT are available in NIST Biometric Image available in NIST Biometric Image Software. Software.

4.4. You may need Perforce to download NBIS You may need Perforce to download NBIS software. software.