LIMITATIONS OF BIOMETRIC SYSTEM

In recent years, biometric systems have been successfully deployed in a number of real-world applications with some biometrics offering reasonably good overall performance. However, even the most advanced biometric systems to date are facing numerous problems, some inherent to the type of data and some of them inherent to system design. In particular, biometric systems generally suffer in person authentication process due to the factors listed below.

Noise can be defined as unwanted data without meaning associated with the data. Noisy data is one of the common problems of biometric systems. Noise can be included in the biometric data during acquisition due to defective, improperly maintained or outdated sensors, due to the failure of providing noise free biometric data acquisition environment or simply produced as an unwanted by-product of other activities. For example, capturing voice biometric data in a noisy environment (i.e. during heavy rain, etc.) will result in a noisy voice signal enrolment. While an undesirable trend, the recognition accuracy of a biometric system could be sensitive to the quality of the biometric data. Developing better algorithms, which can adapt to less than perfect input data and training the biometric system on varied in quality data may alleviate the problem.

This chapter provides reader with an overview of various biometric notions and terms, specifically physical and behavioral identifiers. Among the biometric identifiers, face, fingerprint, signature, voice and iris are the most used biometric identifiers due to the ease of their availability and their recognition performance. New categories of biometric identifiers, namely soft biometrics and social biometrics, are also introduced.

Biometrics authentication can be subdivided into verification and identification. Due to the nature of the application, the user or developer must decide on the appropriate system architecture. The typical biometric system architecture is introduced next. Along with known modules, new approaches to intelligent decision-making, feature extraction, pattern recognition, and system learning are outlined. The overall trend over recent few years was to compensate for inherent issues with biometric data and system performance through introducing radically new methods based on intelligent information fusion and intelligent pattern recognition, thus creating a notion of intelligent security systems. At the end of the chapter, the potential drawback of biometric unimodal system which serves as the motivation to introduce the multimodal biometric system concept in the context of intelligent security systems has been discussed. Last but not the least, issues of privacy and security are given consideration, with a new approach to biometric template protection—cancellable biometric—identified as one of the highly active research directions in a biometric domain.