Keyed Scalable Minutiae Coding

We propose in this paper a keyed coding method to convert standardized fingerprint minutiae to randomized binary vectors. The geometric relationship among neighboring minutiae is encoded in a scalable way and thus the coding/comparison complexity and biometric performance can be adjusted through modifying the number of neighboring minutiae involved in score vector generation. Each encoded minutia has been self-aligned and has a binary form thus can be individually compared using Hamming distance in high efficiency. Key-controlled permutation and exclusive-or are used to gain diversifiability and security for the encoded minutiae template. Experimental results over the public database FVC2002DB2_A demonstrate high biometric performance by the proposed method. Besides the lightweight security achieved by the proposed method, the binary minutiae codes are easy to combine external cryptographic mechanisms and secure protocols for enhanced protection.