Major Groups of Biometric Identifiers

Biometric identifiers are divided into two major groups, depending on whether they are static (over a long period of time) or dynamic.

 

Static identification techniques are based on the constant physiological parameters of a human, including:

 

• Fingerprints (these identifiers are the core of the most widespread, convenient and efficient biometric technology);
• 2D Face (facial images captured from photos and video;
• 3D Face;
• Retina (rarely used as an identifier);
• Iris (the spread of this technology is governed by patent restrictions);
• Palm, hand or fingerprint geometry (a few market segments);
• Face and hand thermography (not very popular technologies);
• Veinscan (this technology is constantly gaining in popularity; the obstacles are expensive scanners);
• DNA (used for specialized analysis applications);
• Body odor (as of today, there are no automated body odor identification systems);
• Ear shape (as of today, there are no automated ear shape identification systems).

 

Dynamic identification techniques are based on analysis of human behavioral parameters, i.e. when a human performs any action. Dynamic identification techniques are an inferior form to static ones, as far as accuracy and efficiency are concerned, and are often used as supplementary identification methods.

 

Dynamic identifiers are as follows:

 

• Signature
• Keystroke
• Voice
• Lip movements
• Gait
• Handwriting

 

Biometric market segments based on biometric identifiers, according to Acuity Market Intelligence, 2007.

 

General Operating Principles

 

Systems based on biometrics perform four basic operations:

 

• Enrollment – data from a physiological or behavioral identifier is transformed into a digital format and registered on the system;
• Extraction – means extracting distinctive features from a captured biometric identifier;
• Matching –newly captured and previously stored biometric credentials are compared for matching;
• Decision - the system decides whether the above credentials match or not.

 

The match or non-match decision can be transmitted to other related systems (access control, information security, etc.).

 

Identification vs. Verification

 

Biometric parameters can be compared in two modes. During identification, a “one-to-many” comparison is performed: the newly produced identifier is compared against all previously enrolled identifiers. It could be said that during identification the system seeks to answer the “who are you?” question by analyzing the entire biometric database, containing all previously enrolled biometric records.

 

During verification data on two specific identifiers are compared for matching (“one-to-one” comparison). An example is comparing data on the newly produced identifier with data stored in a card memory; in this case both the card and the biometric identifier have to be produced to be successfully identified. When operating in the verification mode, the system answers the “is it really your claimed identity?” question.

 

Systems operating in the verification mode are often automated. Identification systems can also be automated (forming a list of possible candidates for matching, with the final decision taken by the expert).

 

To accelerate user recognition, biometric systems can be supplemented with additional identifiers (such as a PIN-code). In this case during identification the system performs a comparison not with the entire database but only with the part of it related to the said additional identifier.

 

Identification in Law Enforcement and Civil Identification Systems

 

Traditionally, law enforcement bodies have pioneered the use of biometrics. In the long run, biometric technologies have covered a vast majority of daily applications – from information security and access control systems to retail, financial, educational and healthcare applications.

 

Law enforcement and civil applications of biometrics differ greatly. The key difference is that civil identification systems use digitized biometric credentials instead of original images, and restoration of the original biometric image from its template is impossible. This ensures biometric data protection and considerably speeds up the operation of civil identification systems: digitized biometric templates are more compact, which is crucial with a large number of users.

The original fingerprint images cannot be restored from its digital template.

 

A comparative analysis of applying biometrics in law enforcement and civil identification applications is presented in the table that follows.

Comparative Features	Law Enforcement Applications	Civil Identification Applications
Nature of Identification	Obligatory (for specific social groups)	Voluntary (at user’s discretion)
Identifiers Used	Fingerprints (all 10 fingers involved), palmprints and photos	Fingerprints (1-2 fingerprints), iris, face, vein scan, hand geometry, etc.
Ways to Obtain Biometric Credentials	Impressions of fingerprints on paper, glass and other surfaces	Only by way of electronic scanning
Actions with Identifiers	Original fingerprint and palmprint images are stored in tenprint cards	Images are converted into digital templates
Biometric Credential Retrieval from the Database	Available	Not available: there is no restoration of the original biometric image from its template
Processing Biometric Credentials	Two-phase approach: 1) the system automatically searches for a number of credentials similar to the biometric credential in question; 2) the expert selects the required credential from a list of candidates	Search is performed automatically in one-to-one or one-to-many identification modes
Personnel Qualification	Highly qualified personnel	Basic computer skills are sufficient
Number of Transactions	Hundreds or thousands of matches per week (for nationwide systems)	Tens of thousands of matches per day (for nationwide systems)
Number of Enrollment/Identification Workstations	From 10 to several hundreds	From several hundreds to several thousands
Scope	The scope is defined by the size of social groups to be enrolled/identified	Varied – from small to nationwide systems
Application Areas	Application areas are restricted to those designed to combat crime, terrorism and illegal immigration	Application areas vary greatly, including voter systems, transportation security, banking and finance applications, social security and many more spheres