Biometric Presentation Attacks: handcrafted features versus deep learning approaches (BioPAth)
摘要截稿:
全文截稿: 2020-10-31
影响因子: 3.255
期刊难度:
CCF分类: C类
中科院JCR分区:
• 大类 : 计算机科学 - 3区
• 小类 : 计算机:人工智能 - 3区
Overview
In the last decade, biometric technology has been rapidly adopted in a wide range of security applications. This approach to automatic verification of personal identity begins to play a fundamental role in personal, national and international security. Despite this, there are well-founded fears that the technology is vulnerable to spoofing, also known as a presentation attack. For example, fingerprint verification systems can be violated by using fingerprints made of a synthetic material, such as silicone, in which the ridges and valleys of the fingerprints of another individual who has access to the system are imprinted. Iris and face recognition systems can be violated using images or video sequences of the eyes or face of a registered user. Speech recognition systems can be violated through the use of repeated, synthesized or converted speech.
In recent years there has been a considerable effort to develop spoof countermeasures or presentation attack detection (PAD) technology to protect biometric systems from fraud. A PAD method can improve the security level of biometric recognition systems. Most of the PAD methods proposed are based on the use of handcrafted features, designed by an in-depth knowledge of designers. An alternative approach based on deep learning approach is also possible.
This special issue is expected to present original papers describing the very latest developments in spoofing and countermeasures.
What are the approaches to the state of the art?
What are the advantages and what are the limits of handcrafted features and deep learning approaches?
Is an auto-adaptive approach possible?
How much do these systems integrate with the corresponding match systems?
The focus of the special issue includes, but is not limited to the following topics related to spoofing and countermeasures:
Adversarial biometric recognition;
Spoof detection based on deep learning;
Spoof detection based on handcrafted features;
Attack transferability in biometric applications;
Design of robust forgery detectors;
Vulnerability analysis of previously unconsidered spoofing methods;
Advanced methods for standalone countermeasures;
New evaluation protocols, datasets, and performance metrics for the assessment of spoofing and countermeasures;