A Novel Face Hashing Method with Feature Fusion for Biometric Cryptosystems

We present a novel approach to generate cryptographic keys from biometric face data so that their privacy and biometric template can be protected by using helper data schema (HDS). Our method includes three components: feature extraction, feature discretization and key generation. During feature extraction stage, the global features (PCA-transformed) and local features (Gabor wavelet-transformed) of face images are used to produce newly fused feature sets as input feature vectors of generalized PCA in the unitary space so as to achieve superior performance. Then, in the feature discretization stage, a discretization process is introduced to generate a stable binary string from the fused feature vectors. Finally, the stable binary string is protected by helper data schema (HDS) and used as the input parameter of cryptographic key generating algorithms to produce the renewable biometric crypto key