Pattern Recognition for Biometrics and Bioinformatics

Abstract

24.1 Biometrics Biometrics are the personal or physical characteristics of a person. These biometric identities are usually used for identification or verification. Biometric recognition systems are increasingly being deployed as a more natural, more secure, and more efficient means than the conventional password-based method for the recognition of people. Many biometric verification systems have been developed for global security. A biometric system may operate in either the verification or identification mode. The verification mode authenticates an individual's identity by comparing the individ-ual with his/her own template(s) (Am I whom I claim I am?). It conducts one-to-one comparison. The identification mode recognizes an individual by searching the entire template database for a match (Who am I?). It conducts one-to-many comparisons. Biometrics are usually classified into physiological biometrics and behavioral bio-metrics. Physiological biometrics use biometric characteristics that do not change with time. Some examples of these biometrics are fingerprint, face, facial thermo-gram, eye, eye's iris, eye's retina scan, ear, palmprint, footprint, palm, palm vein, hand vein, hand geometry, and DNA. Signature is also known to be unique to every individual. Behavioral biometrics are dynamic characteristics that change over time. For recognition purpose, one has to record at a certain time duration, depending on the Nyquist theorem. Examples of such biometrics are speech, keystroke, signature, gesture, and gait. Both types of biometrics can be fused for some complex systems. Biometric cues such as fingerprints, voice, face, and signature are specific to an individual and characterizes that individual. Verification using fingerprints is the most widely used, as the fingerprint of an individual is unique [31]. The simplest, most pervasive in society, and least obtrusive biometric measure is that of human speech. Speech is unique for each individual. Typically, the biometric identifiers are scanned and processed in an appropriate algorithm to extract a feature vector, which is stored as a template in registration. Several companies such as Identix sell high-accuracy face recognition software with databases of more than 1,000 people. Face recognition is fast but not extremely