03 - General biometric facts
Overview
- Description:: continuation about general things on biometric systems
Enrollment vs Recognition
- Enrollment: capture and processing of user biometric data for use by system in subsequent authentication operations (gallery)
- Recognition: capture and processing of user biometric data in order to rendere an authentication decision based on the outcome of a matching process of the stored to current template.
Phases
A Conventional sample is what we capture (signal) such as heartbeat while using a smartwatch. This phase is called Acquisition. Feature acquisition is what we are interested in. These features are collected in Feature Vector (identity) is stored in a Template. The template will be in the template archive.
Modules
- Sensor module: where biometric data are caught
- Feature extraction module: where a set of main characteristic is extracted from acquired data
- Matching module: extracted feature are matched with stored template to return one or more matching stores
- Decision module: who pass and who not
Users and experiments
-
cooperative: the user is interested in being recognized
-
non-cooperative: the opposite
-
public/private: users of the system are customers of the entity installing the system
-
used/non used: frequency of use of the biometric system
-
aware/not aware: the user is aware or not of the recognition process
-
experiments could be executed in:
- controlled environment: capture settings can be controlled. Capture could be repeated.
- uncontrolled/undercontrolled environment: it cannot. So template could present several levels of distortion. Templates can be rejected, but capture cannot be repeated
Verification vs Identification
-
verification: the user claims an identity, and the system checks if it matches the identity
-
identification: no claims from the user, the system determines who the user is through gallery
-
open set: the system determines if probe belongs to a subject in the gallery G
- possible error: reject a probe belonging to an enrolled subject
-
closed set: all probes belong to enrolled subjects
- possible error: return the wrong identity
Feature / traits
- universality: the trait must be owned by any person
- uniqueness: any pair of people should be different according to the biometric trait
- permanence: trait should not be change in time
- collectability: should be measurable by some sensor
- acceptability: involved people should not have any objection to allowing collection / measurement of the trait
Good traits: fingerprints biometry, eye biometry - iris and retina, face biometry (photo, infrared), ear biometry, hand biometry, signature biometry (signature recognition) still and dynamic, keys typing, voice biometry, DNA
Verification vs authenticaion
if a system asks for an implicit or an explicit claim, then it’s verification when there is only one stored subject, then it’s verification
when a system store more identities, we cannot say whether is authentication or verification
if it’s up to the system to identify whether the template probed matches the stored template → identification
if a system asks for a password or pincode or whatever and then for a fingerprint → verification