Biometrics Final Review questions and answers

Biometrics use of physical or behavioral properties of human beings for automatic identity recognition Qualities of physical or behavioral properties used in biometrics 1) something that is characteristic only to me 2) not something that I know (password) 3) not something that i have (key) Necessities of a biometric system 1) data that we process must result from measurement of a living subject 2) biometric sensors must deliver authentic biometric samples 3) we need presentation attack detection to call the system a biometric system 4) not a human expert who decides but applied methodologies 5) typical requirements: uncontrolled environent, high processing speed, repeatability, predictability Positive Recognition A sample represents the subject known to the system (i.e. already registered) Negative Recognition A sample represents the subject unknown to the system (not yet registered) Classic Authentication Types 1) verification 2) identification Verification Confirmation of the identity claim 1:1 comparison A class authentication type New Authentication Types (due to Biometrics) 1) Negative Authentication (includes negative identification and negative verification) 2) Elimination of "Multiple Identities" Identification Searching for a match 1:N commparison a classic authentication type Negative Identification and Negative Verification Negative Identification: I'm not a member of the group X Negative Verification: I'm not the subject X Observation domain Includes biometirc characterisitcs Examples: Face appearance, hand shape, handwritten signature Biometric Characteristics a single physical or behavior property of an individual that we use for biometric recognition Observation Result includes biometric sample Examples: face 3D image, hand 2D image, time series representing handwritten signature Biometric Sample Analog or digital representation of biometric characteristics Feature Domain Includes biometric features Examples: distance between characteristic points in a face image, length and/or width of fingers, length of the signature Biometric Features Numbers or labels extracted from a biometric sample, representation of a sample (typically much shorter than a sample) Template Domain Includes biometric template buometric template may contain some auxiliary data necessary to perform the biometric comparison Examples: set of distances a mong chracteristic points in a face image, lengths and widths of fingers, set of geometrical properties of a signature Biometric Timeline/Steps p.13 of Intro Slides Biometric Mode Combination of biometric characteristics, a sensor type, and a processing method Biometric modes related to physical properites (physical properties = our appearance or anatomical properties) - Momentary observation and immediate biometric samples - Static measurement, time relations not used in recognition Biometric modes related to behavioral properties (behavioral properties = our behavior due to habitss, training, genetics, or mixture of those) - Observation of a conscious action performed by the subject - Dynamic measurement (in time), time relations used in recogntiion Physical biometric modes 1) Commercially used: - finger: fingerprints, finger veins - hand: geometry, palm veins, palm print - face: 2D/3D face images, face tracking - eyes: iris, veins (retina recognition) 2) Emerging, possible to be used in practice - finger: 3D geometry, thermal data, tissue structure under nails - hand: thermal data - face: thermal data - ear: geometry, thermal data, ear prints - other: DNA, scent Behavioral Biometric Modes 1) Commercially used: - handwriting: signature (particularly on-line, i.e., on graphical tablet) - voice: speaker recognition - hand: keystrokes dynamics 2) Emerging, possible to be used in practice: - handwriting: use of any legible text (not just signature) - gait: dynamics measured by mobile devices, silhouette movement measured at the distance - eye: pupil dynamics, eyeball dynamics - evoked potentials: EEG (brain) and ECG (heart) signals - other: lip movement (often merged with speaker recognition) Market Division of Biometric Modes FIngerprint (31%) Iris (16%) Face (15%) Voice (13%) Vein (10%) Signature (4%) Other (11%) Market Perspective of Biometric Modes Revenue is continously and significantly increasing and expected to continue to increase in coming years) Both in Mobile Biometric Revenue and Global Airport Biometric Automaton Revenue Expected Properties of Biometrics 1) Information Capacity 2) Stability over Time 3) User Acceptance 4) Resistance to presentation attacks 5) Amenable to implement Information Capacity Requirement: diversification of subjects 1) genotype and genetic penetrance 2) Identical twins: 1 in 240 births - .8% of regnition errors if we use only genotype-based properties 3) We have no formal proof for uniqueness of the biometric properties - our beliefs based only on experiments ex: DNA (100% dependence on genotype) Iris (estimated close to 0% dependence on genotype) Stability over Time Requirement: correct identity recognition after some long time 1) Resistance to health conditions 2) Template aging (ong