In addition to an overall improvement of their performance, the widespread deployment of biometric recognition systems has also led to the disclosure of privacy and security concerns related to the use of these sensitive data. In particular, the early common belief that biometric templates are irreversible has been proven wrong. Over the last decade numerous works have studied the challenging problem of how to reconstruct synthetic samples from the stored templates, which match the original biometric samples. Such a process, known as inverse biometrics, poses a severe threat to the privacy offered by biometric systems: not only personal data can be derived from compromised and unprotected templates, but those synthetic samples can be as well used to launch other attacks (e.g., presentation attacks). Due to these serious implications, three different aspects of inverse biometrics have been analysed in the recent past: i. how to generate synthetic samples, ii. how to reconstruct a particular sample from its template, and iii. how to evaluate and counterfeit the aforementioned reconstruction techniques. This chapter summarises the works dealing with these three aspects in the biometric field.

Chapter Contents:

• 6.1 Introduction
• 6.2 Synthetic biometric samples generation
• 6.3 Inverse biometrics methods
• 6.3.1 Template format
• 6.3.2 Similarity scores
• 6.3.3 Similarity score and distance function
• 6.3.4 Feature extraction
• 6.3.5 Summary
• 6.4 Evaluation of inverse biometrics methods
• 6.5 Privacy-related issues and countermeasures
• 6.6 Conclusions
• References

Inspec keywords: data privacy; biometrics (access control)

Other keywords: presentation attacks; biometric recognition systems; privacy; security concerns; sensitive data; biometric templates; inverse biometrics

Subjects: Data security