Other Free Encyclopedias » Online Encyclopedia » Encyclopedia - Featured Articles » Contributed Topics from A-E

Digital Biometrics

system person physiological based

Forouzan Golshani
Department of Computer Science & Engineering
Wright State University, Dayton, OH, USA

Definition: Digital biometrics refers to measurements on physiological or behavioral characteristics of a person for determining or verifying the identity of a person.

Biometrics deals with the science of identification (or verification) of a person based on some physiological, behavioral, or genetic characteristics. Digital biometrics refers to measurements on physiological or behavioral characteristics of a person, generally obtained through automated technological means, for determining or verifying the identity of a person. Physiological biometrics is the data gathered from direct measurement of the human body, obtained by means of various procedures and algorithms that can uniquely identify the individual. Examples include: fingerprint, hand geometry, iris, retinal, vein, voice, and facial imaging. Behavioral biometrics, however, is measured by analyzing a specific set of actions of a person. These may include how a person talks, signs their name, walks, writes, or types on a keyboard. Work in the field of multimedia directly contributes to computational biometrics, both physiological and behavioral. Face recognition techniques and retina scan analysis are examples of contributions to the physiological biometrics. Gait analysis, where the person recognized by their activities, for instance manner of walking, is an example of where multimedia techniques play a crucial role.

From another perspective, biometric technologies can be categorized into active and passive techniques. Active biometrics include: fingerprint, hand geometry, retina scanning, and signature recognition technologies. Examples of passive biometrics procedures are: voice recognition, iris recognition, and facial recognition. The first two are somewhat limited but the last, facial imaging, can be truly passive.

Biometric technologies are becoming the foundation of a large number of highly secure identification and human verification solutions. In the identification mode (one to many), the biometric system compares the given individual to all individuals in the database and ranks the top possible matches. In a verification mode (one to one), the system compares the given individual with who that individual says they are and provides a yes/no answer. Identification is generally associated with surveillance, whereas a clear application of verification is access control.

The need for biometrics-based systems can be found in the government sector (federal, state and local), the military, and commercial applications. Many enterprise-wide network security infrastructures are already benefiting immensely from multimedia-based biometric technologies, including secure electronic banking and other financial transactions, retail sales, law enforcement, and government IDs. Biometric-based authentication applications include access control for workstation, network, and domain, data protection, application logon, and remote access to resources.

The performance of a monomodal biometric system depends greatly on the ability of the sensor to capture data and also on the features of the biometric trait captured. Any irregularity in the capture phase reflects on the performance metrics of the system and might greatly increase the error rates. There are several issues regarding the biometrics being used. Not all of the biometric traits (physiological or behavioral) are universal or unique-enough. Some are variant with time (aging), cannot be quantified well, are not acceptable by all people (intrusive), can be faked to fool the system, and some are not robust enough. Noisy data also creates a variance in matching correctly. The success of any biometric verification system relies on the properties of the trait on which the system is based. Usually these peculiarities result in compromised success.

Incorporating multiple biometric traits for authentication can alleviate most of these problems. Multimodal systems are expected to be more reliable and that is due to the fact that multiple sensors are used to catch different biometric traits. The system thus has multiple pieces of evidence of identity (or lack of it). These systems can be configured to match levels of security based on application. Due to the requirement of several biometrics, it will be difficult for intruders to fake all traits and violate the integrity of the system. Information fusion, i.e., how to integrate the information received from individual modalities, is the key to success of these systems. We will review several techniques used in various systems.

Digital Camera Image Processing - CFA Basics, Image Processing for Color Estimation, Camera Image Compression, Video-Demosaicking, CFA Image Indexing [next] [back] Digital Archiving - Conclusion

User Comments

Your email address will be altered so spam harvesting bots can't read it easily.
Hide my email completely instead?

Cancel or