Privacy preservation for soft biometrics based multimodal recognition system

摘要: There has been a recent proliferation in the utilization of biometric systems as an access granting mechanism for various applications. Researches have demonstrated that overall performance these recognition significantly improves upon inclusion "soft" traits addition to conventional primary modalities. Normal can be subjected wide variety malicious attacks and privacy breaches, which ultimately lead severe jeopardy enrolled users. These security infringements are generally concentrated on central database wherein user's data get stored. The risks associated with soft aided designs comprehensively increase due availability more personalized amount information databases. In our work we addressed this problem proposed preserving framework based by utilizing popular notion differential privacy. We also analyzed scheme real life found preservation technique does not affect utility (performance) basic system. Thus provides system users without compromising its performance.