Biometric Image Processing

Biometric Image Processing

DOI: 10.4018/978-1-4666-3646-0.ch003
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

In most biometric-based security systems, images of the associated biometric identifiers are used as the input to that system. This chapter discusses various image processing methods and algorithms commonly used for biometric pattern recognition. Efficient and reliable processing of images is essential to achieve good performance of biometric systems. Different appearance-based methods, such as eigenimage and fisherimage, and topological feature-based methods, such as Voronoi diagram-based recognition, are discussed in the context of face, ear, and fingerprint application frameworks. Utilizing cognitive intelligence and adaptive learning methods in both physical and behavioral biometrics are some emerging new directions of biometric pattern recognition. As such, neural networks, fuzzy logic, and cognitive architectures would play a more important role in biometric domain of research. The chapter concludes with discussion of the importance of context-based recognition for behavioral biometrics.
Chapter Preview
Top

2. Appearance-Based Image Processing In Biometrics

From the gamut of research on biometric authentication, we observe that the overwhelming part of biometric data processing is realized by using image processing and pattern recognition methods and algorithms (Soledek, Shmerko, Phillips, Kukharevl, Rogers, & Yanushkevich, 1997). As the mainstream direction of biometric image processing, appearance-based methods extract biometric features from the row image by analyzing appearance of the whole image as an entity or a vector in a high-dimensional image space. Such factors as color scheme, orientation, background, luminance, saturation are being analyzed and processed either pixel by pixel or through projection on subspaces, such as in Principal Component Analysis (PCA) methods. As the most evident examples, we consider face, iris and ear biometrics in this context (Soledek, Shmerko, Phillips, Kukharevl, Rogers, & Yanushkevich, 1997).

To solve the biometric data processing problems, the following main methods are typically employed in literature: digitization, compression, enhancement, segmentation, feature measurement, image representation, image models, design methodology. They are summarized in Figure 1. While some of these methods are used during data pre-processing, and some during pattern recognition and matching, there is high potential of employing more intelligent techniques at all stages, with the goal of optimizing processing and increasing overall security system performance. Some of these approaches are overviewed in the subsequent sections devoted to individual biometrics.

Figure 1.

Biometric image processing methods and algorithms

Complete Chapter List

Search this Book:
Reset