Face verification put to test: A hill-climbing attack based on the uphill-simplex algorithm
- 1 March 2012
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 23764201,p. 40-45
Abstract
The vulnerabilities of a PCA-based face verification system against a hill-climbing attack using the uphill-simplex algorithm are studied. Experiments are carried out on the face subcorpus of the publicly available BioSecure DB, where the attack has shown a remarkable performance proving the lack of robustness of the tested system to this type of threat. Furthermore, the proposed attacking scheme is not only able to bypass the security of the recognition system, but it is also capable of reconstructing the users face image, with the privacy concerns that this entails. As a possible countermeasure to minimize the effect of the attack, score quantization is applied. This protection method is able to reduce both the success rate and the efficiency of the attack, however it does not completely succeed in preventing a possible intruder from accessing the system. The study also highlights the high adaptation capabilities of the proposed attack which had already been used to break a signature-based verification system.Keywords
This publication has 13 references indexed in Scilit:
- An evaluation of indirect attacks and countermeasures in fingerprint verification systemsPattern Recognition Letters, 2011
- An evaluation of direct attacks using fake fingers generated from ISO templatesPattern Recognition Letters, 2010
- On the vulnerability of face verification systems to hill-climbing attacksPattern Recognition, 2009
- Biometric Template SecurityEURASIP Journal on Advances in Signal Processing, 2008
- New approach for liveness detection in fingerprint scanners based on valley noise analysisJournal of Electronic Imaging, 2008
- Biometrics: A Tool for Information SecurityIEEE Transactions on Information Forensics and Security, 2006
- Images can be regenerated from quantized biometric match score data2004
- Attacks on biometric systems: a case study in fingerprints2004
- Sample images can be independently restored from face recognition templates2004
- Face recognition using eigenfaces2002