Generating secure cancelable fingerprint templates using local and global features

Biometric template protection is an important issue to be resolved so that practical biometric authentication systems can be deployed. In this paper, we propose a non-invertible transform to perpendicularly project the distances between a pair of minutiae to a circle to generate the features. To achieve good performance, other local features such as relative angles between the minutiae pair, and global features such as orientation, ridge frequency and total number of minutiae of the randomly sampled blocks around each minutia are also utilized. The cancelable templates are eventually generated by the proposed ldquobin-based quantization (BQ)rdquo. Both feature extraction and cancelable template generation are governed by a secret key to guarantee revocability and security. Experimental results conducted on FVC 2002 data set demonstrate the reasonable good performance of the proposed scheme.