Abstract
Efficient sample encryption techniques are investigated for fingervein data. We propose an approach where it suffices to encrypt 0.5% of the sample JPEG2000 bitstream and thereby completely disable biometric recognition. Evaluations
with 5 different recognition schemes on two different datasets reveal that results are stable accross all techniques considered as long as the start of the bitstream is encrypted.
with 5 different recognition schemes on two different datasets reveal that results are stable accross all techniques considered as long as the start of the bitstream is encrypted.
| Originalsprache | Englisch |
|---|---|
| Titel | Proceedings of the 8th International Workshop on Biometrics and Forensics |
| Seitenumfang | 6 |
| Publikationsstatus | Veröffentlicht - Apr. 2020 |
Systematik der Wissenschaftszweige 2012
- 102 Informatik