Biometrics Security and Privacy Protection: IEEE Signal Processing Magazine, September 2015

(Excerpt from the guest editorial of the September 2015 Signal Processing Magazine)

Biometrics is the science of recognizing individuals based on their behavioral and biological characteristics such as face, fingerprints, iris, voice, gait, and signature. A typical biometric system may be viewed as a pattern classification system that utilizes advanced signal processing schemes to compare and match biometric data.

Since an individual's biometric data is personal and sensitive, issues related to biometric security and privacy have been raised. The advent of cloud computing technology and personal mobile devices has broadened the application domain of biometrics; however, at the same time, it has brought to the forefront the need for dedicated security technologies to protect biometric data from being misappropriated and used for purposes beyond those intended. Similarly, the use of surveillance systems in public areas presents new challenges with respect to privacy.

The special issue of IEEE Signal Processing Magazine, Sep. 2015, was conceived to champion recent developments in the rapidly evolving field and also to encourage research in new signal processing solutions to security and privacy protection.
The first contribution from Hadid, Evans, Marcel, and Fierrez focuses on the security side of biometrics, providing a gentle introduction to spoofing and countermeasures and a methodology for their assessment. The article also provides a case study in face recognition.
The next contribution discusses how adversarial machine-learning techniques can be harnessed to protect biometric systems from sophisticated attacks. Biggio, Fumera, Russu, Didaci, and Roli argue that security is best delivered with adaptive, security-by-design solutions.
Itkis, Chandar, Fuller, Campbell, and Cunningham report the challenges in designing effective cryptosystems for iris-recognition systems. Their work also illustrates the shortcoming of the more traditional performance metrics used in biometrics and promotes the use of a new entropy metric.
The article by Patel, Ratha, and Chellappa reviews different approaches to cancelable biometric schemes for template protection. The aim of such techniques is to preserve privacy by preventing the theft of biometric templates through the application of noninvertible transforms.
Barni, Droandi, and Lazzeretti describe a different approach to template protection based on cryptographic technology. They illustrate how secure, two-party computation and signal processing in the encrypted domain can be combined to enhance security and protect privacy.
Still on the theme of template protection, Lim, Teoh, and Kim describe their work on biometric feature-type transformation. Such transformations are typically used as a precursor to many forms of biometric cryptosystems that demand specific input formats such as point-set or binary features.
The final article on template protection discusses the practical implications of biometric security and offers a fresh perspective to the problem. Nandakumar and Jain argue that improvements to security and privacy seldom come without degradations to recognition performance and that, consequently, there remains a significant gap between theory and practice.
The special section rounds out with an article by Bustard on the privacy and legal concerns surrounding the collection, storage, and use of personal biometric data. In particular, the article discusses recent European legislation on this issue and its potential impact on the adoption of biometrics technology.

Reference:
Nicholas Evans, Sébastien Marcel, Arun Ross, Andrew Beng Jin Teoh. Biometrics Security and Privacy Protection. IEEE Signal Processing Magazine. 2015, Sep. 17-18