Search

The paper proposes a dual cancellable fingerprint template method based on BioHashing and Fuzzy Commitment. In the new method, firstly, Finger Code extracted from fingerprint image is noninvertibly mapped into a string of binary bits by BioHashing, and then the bit string is bound with a key by Fuzzy Commitment to generate the final template data stored in the database. It can overcome threat of replay attack in the approach of biometric feature transformation, and also avoid latent leakage of original template feature. in addition, as for the poor recognition when the random matrix is public in BioHashing, we also proposes some improvements, including aligning fingerprints with both fingerprint core point and minutiae, and assigning a random point instead of the core point as the reference point of Finger Code. These improvements enhance the ability of fingerprint recognition, and reduce the rate of failure to enroll, and also increase the difficulty of crack for attackers.

The paper proposes a method of extracting a rotation-translation-invariant texture feature from minutiae neighborhood to help matching fingerprint minutiae. The texture feature is extracted by complex filter, and represents pattern of orientation image around minutiae, and is used as one element feature of minutiae. Based on the texture feature, a new and simple texture matching distance differed from the average Euclidean distance is defined. This texture feature is easy integrated into existing minutiae matching algorithm. Experimental test results show that the texture feature is able to distinguish minutiae with different orientation images in neighborhood, and the texture matching distance defined has better discrimination than the average Euclidean distance, and the texture feature has an important role to accelerate the matching speed and improve the matching accuracy when it is integrated into fingerprint matching.