Introduction to BiometricsOutlineSlide 3Some Topics in BiometricsOverviewFingerprint IdentificationFingerprint Identification (Concluded))Hand GeometryFace Location and RetrievalMulti-BiometricsIntroduction to BiometricsDr. Bhavani ThuraisinghamThe University of Texas at DallasLecture #6Guest Lecture + Some Topics in BiometricsSeptember 12, 2005OutlineGuest LectureSome Topics in BiometricsIntroduction to BiometricsGuest LectureImage Feature Extraction and Annotation September 12, 2005Some Topics in BiometricsReference-http://biometrics.cse.msu.edu/info.html-Papers published on the web by researchers at Michigan State UniversityOverviewFingerprint IdentificationHand GeometryFace LocationMulti-BiometricsOverviewBiometrics is the automatic identification of a person based on his/her physiological and behavioral characteristicsVerification vs. Identification-Am I whom I claim I am?-Who am I?Applications-Criminal identification, ATMs, Cellular Phones, Smart cards, PCs, E-Commerce, Automobiles (biometrics replacing car keys)Fingerprint IdentificationFinger-print matching-Two categories: Minutiae based, Correlation basedMinutiae-based techniques-First find the minutiae points and then map their relative placement on the finger-Issues: difficult to extract minutiae points if fingerprinting is of low qualityCorrelation-based techniques-Spatial correlation of regions-Issues: Affected by Image translationFingerprint Identification (Concluded))Finger-print classification-Classify the fingerprints so that search time is reduced-Form groups of fingerprints; Classification is obtained by matching with pre-specified types of finger-prints-When a new finger-print arrives try and place it into a group-Classification based data mining/machine learning algorithms such as K-Nearest NeighborFingerprint Image Enhancement-Algorithms to enhance the finger-print-This is expected to facilitate finger-print matching-Makes it less difficult to extract minutiae from fingerprintsHand GeometryUses geometric shape of hand for authenticating user’s identity-Combine various individual features of hand for effective verificationHuman hand is not in general unique (not the case with fingerprints)Reason one may want to use hands instead of finger-prints is to ensure privacySome pros and cons-Hand geometry gives better privacy-But hand geometry is not unique; therefore may be used for verification-Not suitable for identificationFace Location and RetrievalProblem-Given an arbitrary black and white still image, find the location and size of every human face it containsApplications-First step in automatic face recognition-Image database indexing-Search by content for surveillance systemsMulti-BiometricsIntegrating Faces and Fingerprints for Identification-Single biometric may not be effective-Integrate multiple biometrics such as fingerprints and facesFingerprint, Face and Speech-Better to use a third biometric and that is speechChallenge-What is an effective combination of
View Full Document
Unlocking...