Biometry

Technology, Trends and Applications

Ricardo A. Ramirez-Mendoza, Jorge de J. Lozoya-Santos, Ricardo Zavala-Yoé, Luz María Alonso-Valerdi, Ruben Morales-Menendez, Belinda Carrión, Pedro Ponce Cruz, Hugo G. Gonzalez-Hernandez, Ricardo A. Ramirez-Mendoza, Jorge de J. Lozoya-Santos, Ricardo Zavala-Yoé, Luz María Alonso-Valerdi, Ruben Morales-Menendez, Belinda Carrión, Pedro Ponce Cruz, Hugo G. Gonzalez-Hernandez

• Publisher: CRC Press
• Year: 2022
• Print ISBN: 9780367702472
• eBook ISBN: 9781000626087
• Edition: 1
• Topic: Computer Science
• Subtopic: Biotechnology in Medicine
• Index: Computer Science

Biometrics provide quantitative representations of human features, physiological and behavioral. This book is a compilation of biometric technologies developed by various research groups in Tecnologico de Monterrey, Mexico. It provides a summary of biometric systems as a whole, explaining the principles behind physiological and behavioral biometrics and exploring different types of commercial and experimental technologies and current and future applications in the fields of security, military, criminology, healthcare education, business, and marketing.

Examples of biometric systems using brain signals or electroencephalography (EEG) are given. Mobile and home EEG use in children's natural environments is covered. At the same time, some examples focus on the relevance of such technology in monitoring epileptic encephalopathies in children.

Using reliable physiological signal acquisition techniques, functional Human Machine Interfaces (HMI) and Brain-Computer Interfaces (BCI) become possible. This is the case of an HMI used for assistive navigation systems, controlled via voice commands, head, and eye movements. A detailed description of the BCI framework is presented, and applications of user-centered BCIs, oriented towards rehabilitation, human performance, and treatment monitoring are explored.

Massive data acquisition also plays an essential role in the evolution of biometric systems. Machine learning, deep learning, and Artificial Intelligence (AI) are crucial allies here. They allow the construction of models that can aid in early diagnosis, seizure detection, and data-centered medical decisions. Such techniques will eventually lead to a more concise understanding of humans.

1 Current and Future Biometrics: Technology and Applications

Jorge de J Lozoya-Santos,* Mauricio A Ramírez-Moreno, Gladys G Diaz-Armas, Luis F Acosta-Soto, Milton O Candela Leal, Rafael Abrego-Ramos and Ricardo A Ramirez-Mendoza

School of Engineering and Sciences, Tecnologico de Monterrey, Monterrey, Mexico

* Corresponding author: [email protected]

1.1 Introduction

This chapter will explore biometric definitions, types, commercial and experimental technology as well as current and potential applications. Biometrics considers the means of identifying, authenticating and modeling individuals in a reliable way through the use of unique physiological and behavioral characteristics. The current methods, apparatus and techniques are summarized for each biometric physical body measure (physiological) to gait (behavioral) passing by finger print and facial recognition, speech and human-machine interactions. Biometrics has been an important component in different applications such as security, military aspects and criminology. Biometric systems appear to be mandatory in some applications becoming an essential in healthcare, smart cities and industrial businesses. Some seminal applications in education, neuromarketing and arts could lead all the efforts for biometrics to become an every day asset for the increasing concern for health and human performances in the future world.

1.2 Biometrics

Biometrics is the discipline and techniques involving the recognition, modeling and understanding of the individuals based on their distinctive behavioral, physiological and cognitive characteristics, rather than using something known or possessed, [281, 172].

The utilized techniques in biometrics cover data acquisition, data analysis and characterization, theoretical - heuristic modeling, computational simulation and visualization techniques, [140, 539, 212]. Classical applications have been focused on agriculture, [87], environmental science, biology, [521, 492], and medical sciences, [393].

A biometric system is a pattern recognition system that establishes the authenticity of a specific physiological or behavioral characteristic possessed by a user, [385]. However, digital transformation pushes biometric systems as data acquisition systems with application to the description, prediction, and imitation of human behavior for several purposes as personalization of services and remote body monitoring [308, 65]. Typically, a biometric system is used for: (a) Logical Access Control; (b) Physical Access Control; (c) Time and Attendance; (d) Law Enforcement; and (e) Surveillance. Novel applications are consumer behavior [539], sports [42], and workforce wellness [98, 441].

The fields of biometric applications are product differentiators, and, forensic and government services. Mainly the goal of these systems is user/person identification, [530]. Several biometric traits can be used, [445]. Biometric Commercial applications focus on authenticating users to give them some kind of privilege or access, [574].

1.2.1 Physiological biometrics

Physiological biometrics refer to physical measurements of the human body and thus they vary from person to person, [269]. Most of the times it is necessary to use a specialized sensor/hardware to perform the data acquisition, and its likelihood for reliable human identification is very high, [149]. A non exhaustive list of research results evidences the scientific community interest in these biometrics: ear, [396], electro-encephalgraph, [357], face, [301], face thermography, [215], fingerprint, [82], [251], hand geometry, [266], iris, [325], and wrist vein, [421], Figure 1.1. A region-based classification, [530], consists of: (a) hand region; (b) facial region; (c) ocular region; and (d) medico-chemical phenomena, Figure 1.1. In Table 1.1, there are some examples of the most used physiological biometrics and how well they perform comparing different requirements, [149].

TABLE 1.1: Physiological biometrics, [149].

Characteristic	Universality	Distinctiveness	Collectabillity	Performance	Acceptability
Ear	Medium	Medium	Medium	Medium	High
Hand Geometry	Medium	Medium	High	Medium	Medium
Fingerprint	Medium	High	Medium	High	Medium
Face	High	Low	High	Low	High
Iris	High	High	Medium	High	Low
Voice	Medium	Low	Medium	Low	High

FIGURE 1.1: Main biometric traits, [281], [530], [574], [232]. In the biometric literature, these characteristics are referred to as traits, indicators, identifiers.

1.2.2 Behavioral biometrics

Behavioral biometrics are the behavioral characteristics that relate to the pattern of people doing something (authorship, body dynamics, performance and dynamics of things operated/used by humans), [574], [269], Figure 1.1. Behavioral biometrics have advantages over physiological biometric technologies. They can be collected non-obtrusively, often do not require any special hardware and they are very cost effective. Most of the behavioural biometrics have been shown to provide sufficiently high accuracy identity verification, [574], although performance is generally lower than with physiological biometrics, [381]. Extensive research efforts during the years gone by have shown this topic as an open issue with opportunities and challenges: behaviour profile, [23], driving style, [442], [106], gait, [66], [107], handwriting, [43], human computer interaction, [146], keystroke dynamics [202], [168], mouse dynamics, [14], network traffic, [21], social network interaction, [507], walking sound, [67], and wearable dynamics, [17].

Five categories have been proposed in [574]: (a) authorship-based (signature, text, drawing, painting), (b) Human Computer Interaction (HCI)-based (keystroke, mouse, change of software screens), (c) HCI-Performance-based (network traffic, registry access, storage activity), (d) motor-skill based (gait, handwriting), and (e) pure behavioural (performance analysis of how an individual operates a thing).

Human Activity Recognition (HAR) provides accurate and opportune information on activities and behaviors of persons [286]. It uses behavioral biometrics. The recognition of human activities is key for several fields like medical, military, and security applications. However, recognizing activities requires specificity according to the field of application to provide useful feedback to the generated information end user. The HAR can be divided in those activities performed by a person in spaces or using devices or systems, Figure 1.2.

FIGURE 1.2: Classification of behavioural biometrics using the HAR from [286] (italic bold words) in the context of the use of spaces and devices or systems (bold words).

Using the human activity classification using devices or systems, a set of behavioural biometrics that can be acquired when using mobile devices, non-mobile devices with embedded computer (personal computer or systems such as TV or infotainment systems), microvehicles or fitness equipment and a vehicle. It shows that mobile and non-mobile devices, can deliver almost the same biometrics, Figure 1.3, and smart mobile devices (tablet, smartphone) and smart vehicles or microvehicles can deliver motor-skilled and performance-based biometrics. So, it is possible to get behavioural biometrics from everybody since the use of the Internet of Things technology in products and equipment is very common and continuous on a daily basis, [106].

FIGURE 1.3: Classification of behavioural biometrics using systems or devices according to the required hardware to be acquired and the purely behavioural property. Modified from, [308]. Bold words were classified in [574] as pure behavioural biometrics.

A taxonomy to classify the biometrics from a sensorial point of view proposes a classification based on human senses: hearing, sight, smell, touch and metadata, [381]. This classification approach can be used for that presented in Figures 1.1–1.3 since a human, when using at least one sense, identifying a person for both type of biometrics, physiological or behavioural, depends on their availability. A summary of all behavioural biometrics where the human activity classifies those based on the creation of expression, device or system use and space use links the HAR application, Figure 1.4.

FIGURE 1.4: Behavioural biometrics.

1.3 Technology

This section breaks down the technology of biometric systems: sensors, methods, and the typical setup. Commercial technology presents the available devices and equipment for enabling biometric applications.

1.3.1 Sensors

A biometric system is composed of four parts [126, 233]: (a) a sensor module to obtain the biometric data, (b) a feature extraction module to select the information of interest, (c) a matching and decision-making module that decides if the user’s identity is accepted based on the stored information, and (d) a system database acting as a repository for the biometric information of the authorized users, Figure 1.5.

FIGURE 1.5: Biometric system.

For a biometric system to be reliable, it needs to fulfill certain parameters, including the acceptance of the public (people should be willing to use it), the use of a unique and time-invariant individual characteristic, and the possibility for acquisition and digitization of the human trait without inconvenience to the individual. It is also very important that the chosen characteristic is shown by the whole populati...

Table of contents

• Cover Page
• Title Page
• Copyright Page
• Preface
• Preface
• Table of Contents
• Symbol Description
• 1. Current and Future Biometrics: Technology and Applications
• 2. Analysis of Electrophysiological Activity of the Nervous System: Towards Neural Engineering Applications
• 3. Applications of Machine Learning Classifiers in Epileptic Seizure Detection
• 4. Simultaneous Evaluation of Children Epileptic Encephalopathies with Long-Term EEG via Space-Time Dynamic Entropies
• 5. Mobile and Home Electroencephalography in the Usual Environment of Children
• 6. Health: Human-Machine Interaction, Medical Robotics, Patient Rehabilitation
• 7. APSoC-Based Implementation of an EEG Classifier using Chaotic Descriptors
• Bibliography
• Index