Intelligent Biometric Techniques in Fingerprint and Face Recognition

L.C. Jain, U. Halici, I. Hayashi, S.B. Lee, Shigeyoshi Tsutsui, L.C. Jain, U. Halici, I. Hayashi, S.B. Lee, Shigeyoshi Tsutsui, Shigeyoshi Tsutsui

• Publisher: Routledge
• Year: 2022
• eBook ISBN:
  9781351437714
• Topic: Informatica
• Subtopic: Media digitali

The tremendous world-wide interest in intelligent biometric techniques in fingerprint and face recognition is fueled by the myriad of potential applications, including banking and security systems, and limited only by the imaginations of scientists and engineers. This growing interest poses new challenges to the fields of expert systems, neural networks, fuzzy systems, and evolutionary computing, which offer the advantages of learning abilities and human-like behavior. Authored by a panel of international experts, this book presents a thorough treatment of established and emerging applications and techniques relevant to this field.

Chapter 1: Introduction to Fingerprint Recognition

INTRODUCTION TO FINGERPRINT RECOGNITION

U. Halici

Computer Vision and Artificial Neural Networks Research Lab.
Dept. of Electrical and Electronics Eng.
Middle East Technical University, 06531, Ankara, Turkey
[email protected]

L.C. Jain

Knowledge-Based Intelligent Engineering Systems Centre
University of South Australia, Adelaide
Mawson Lakes S.A. 5095, Australia
[email protected]

A. Erol

Halici Software House, METU Technopark
Middle East Technical University, 06531, Ankara, Turkey
[email protected]

One of the most interesting human abilities is the recognition of objects. Recognition is defined as a process involving perception and associating the resulting information with one or a combination of more than one of its memory contents. Visual perception means deriving information from a specific scene. From the psychological point of view, the actual perception process involves some information processing stages. The formation of the image on the retina is followed by the mental processing of the projected image. The actual process is not yet known, but several models exist.

Scientists commonly aim to design machines that emulate human abilities. However, current results are far from successful. Nevertheless, dividing human abilities to smaller tasks and implementing them reveals promising results.

Biometric systems have been an important area of research in recent years [10], [40]. There are two important utilizations of biometric systems: 1) Authentication or verification of a person’s identity, i.e., a person proves that he is the person who he claims to be and 2) identification in which a person’s identity is sought using the biometric sign available. Any physiological or behavioral characteristics can be used to make personal identification as long as it satisfies the following requirements [10], [32], [40].

1. 1. Universality, which means that every person should have the characteristics;
2. 2. Uniqueness, which indicates that no two persons should be the same in terms of the characteristics;
3. 3. Permanence, which means that the characteristics should be invariant with time;
4. 4. Collectability, which means that the characteristics can be measured quantitatively.

Examples of biometric signs are hand, fingerprint, iris [39], face, and speech [40]. In this book, we have included the most recent advances in fingerprint and face recognition. There are other systems used for identification purposes such as retinal image comparison, voice matching, and DNA matching, but these are currently not widely used.

1 The Use of Fingerprints in Personal Identification and Verification

Fingerprints, which have been used for about 100 years, are the oldest biometric signs of identity. Scientific studies on fingerprints were initiated in the late sixteenth century [35], but the foundations of modern fingerprint identification were established by the studies of Sir F. Galton [19] and E. Henry [25] at the end of nineteenth century. A fingerprint is formed of composite curve segments. The light areas of fingerprints are called ridges while the dark areas are called valleys. Galton’s study introduced the minutiae, which are the local discontinuities in the ridge flow pattern, as discriminating features and showed the uniqueness and permanency of minutiae. Henry’s study examined the global structure of fingerprints and established the famous “Henry System” of fingerprint classification which is an effective method of indexing fingerprints and is still in use in most identification systems. In the early twentieth century, fingerprints were formally accepted as valid signs of identity [35] by law-enforcement agencies. However manual fingerprint identification is tedious, time-consuming, and expensive as it needs to be performed by professional fingerprint experts. Therefore, in 1960, the FBI Home Office (UK) and the Paris Police Department initiated studies on automatic fingerprint identification systems [15].

The study by F. Galton [19] extensively examines the details that reside in fingerprints. Fingerprints were examined morphologically and experiments were carried out on different age groups within different races. Two fundamentally important conclusions were reached by Galton. The first was that a fingerprint of a person is permanent, i.e., it preserves its characteristics and shape from birth to death. The second result was that the fingerprints of individuals are unique. In the light of experimental evidence, it was proven that no two persons have the same fingerprints; even identical twins have different fingerprints despite signs of similarity. These fundamentally important results were the building blocks of research in this field over the last 90 years.

In his work, E.R. Henry examined the fingerprint’s global structure and devised a classification method to partition the large fingerprint databases into five classes [25]. Unlike Galton, Henry did not extensively deal with the exact matching of fingerprints. However, his systematic way of partitioning fingerprint classes was so profound that it has traditionally been used by almost all of the government security forces and other users since then. The names given to these classes are Right Loop (R), Left Loop (L), Whorl (W), Arch (A), and Tented Arch (T) respectively. Samples taken from these classes can be seen in Figure 1.

By using the ideas presented above, fingerprints are partitioned by the Henry Classification and exact matching is carried out by comparing Galton Features.

Figure 1. The five basic Henry classes of fingerprints.

The Galton Features are details formed on the ridges. A ridge can be defined as a single curve segment. The combination of several ridges forms a fingerprint pattern. The small features formed by the crossing and ending of ridges are called minutiae in the fingerprint literature. In his work, Galton defines four characteristics: the beginning and end of ridges, forks, islands, and enclosures. When searching for a new fingerprint in a database, a sufficient number of these minutiae should be located to decide the exact match.

Figure 2. Some basic fingerprint features.

After Henry and Galton, work on fingerprint identification and its specifications was extended and refined. The extended Galton Features [26] are shown in Figure 2. As can be seen, there are several extensions to the original Galton Features, but most are not used in automatic fingerprint identification systems. Instead, in accordance with the FBI representation of fingerprints [15], ridge endings and bifurcations are taken as the distinctive feature of fingerprints. In this method, the location and angle of the feature are taken to represent the fingerprint and used in the matching process.

Together with these, fingerprints contain two special types of features called core and delta points. These points are often referred to as singularity points of a fingerprint. The core point is generally used as a reference point for coding minutiae and defined as the topmost point on the innermost recurving ridge [54]. Example core and delta points are shown in Figure 3.

Figure 3. Core and delta points on a fingerprint.

With the increasing power of computers, automated systems have been developed to automate the tedious manual classification and matching methods of fingerprints. There are two types of fingerprint-based biometric systems in terms of their utilizations:

Automatic Fingerprint Authentication System (AFAS),

Automatic Fingerprint Identification System (AFIS).

The block diagram of a basic AFAS is shown in Figure 4. The input is an identity and a fingerprint image; the output is an answer of Yes or No indicating whether the input image belongs to the person whose identity is provided. The system compares the input image with the one addressed by the identity in the database [31].

Figure 4. Block diagram of a basic fingerprint verification system.

In an AFIS (Figure 5), the input is just a fingerprint and the output is a list of identities of persons that can have the given fingerprint and a score for each identity indicating the similarity between the two fingerprints. It is possible to provide partial identity information to narrow the search space. The system compares the input image with many records in the database [12], [45], [46].

Figure 5. Block diagram of a basic fingerprint identification system.

Fingerprint based biometric systems are usually used for criminal identification and police work but now with the development in AFASs, they are highly utilized in civilian applications such as access control and financial security [35]. Therefore, AFASs are in great demand and a number of commercially available AFASs exist.

2 Fingerprint Identification and Authentication Systems

For each operation shown in Figures 4 and 5, several approaches are proposed i...

Table of contents

• Cover
• Title Page
• Copyright Page
• Series Page
• Preface
• Table of Contents
• Chapter 1 Introduction to Fingerprint Recognition
• Chapter 2 Fingerprint Feature Processing Techniques and Poroscopy
• Chapter 3 Fingerprint Sub-Classification: A Neural Network Approach
• Chapter 4 A Gabor Filter-Based Method for Fingerprint Identification
• Chapter 5 Minutiae Extraction and Filtering from Gray-Scale Images
• Chapter 6 Feature Selective Filtering for Ridge Extraction
• Chapter 7 Introduction to Face Recognition
• Chapter 8 Neural Networks for Face Recognition
• Chapter 9 Face Unit Radial Basis Function Networks
• Chapter 10 Face Recognition from Correspondence Maps
• Chapter 11 Face Recognition by Elastic Bunch Graph Matching
• Chapter 12 Facial Expression Synthesis Using Radial Basis Function Networks
• Chapter 13 Recognition of Facial Expressions and Its Application to Human Computer Interaction
• Index