Cancelable Fingerprint Identification And Fingerprint Encryption Based On Minutiae’s Local Information

The extraction of fingerprint features is very important in fingerprint identification system. Generally, the fingerprint features are divided into global ones and local ones. As a typical representative of the local features, the minutiae features process the best distinctiveness, but minutiae extraction is affected by fingerprint image quality and the corresponding extraction algorithm. Besides, the minutiae features themselves only contain limited information of a fingerprint, and it has been proved that the original fingerprint image can by recovered by its minutiae template, thus constitute a latent security risk of the system. This paper introduces minutiae’s local information into the fingerprint features in order to enhance the robustness of the template, and at the same time attempts to preserve the good distinctiveness of the minutiae features.In the first part of this paper’work, the cancelable fingerprint template is constructed by making use of the minutiae’s local information. We conduct the following specific procedures. First calculate the direction differences between every minutia’s direction and the ridge line directions in some selected sampling points in the minutia’s surrounding area. Next combine the direction differences information and a user specified password i.e. the token to produce the corresponding transformation parameters in the cancelable transform. And then the cancelable fingerprint template is generated by transformation functions. Last the matching scheme is designed based on this template. Experiments on FVC2002DB2show that the cancelable template keeps good template distinctiveness and can achieve a good matching performance. We also analyzed the template security by some certain experiments. In the second part, based on the features containing minutiae’s local information, on the processed integer minutia descriptors, we implement the Wrap-around encryption scheme, which is one of the key binding technologies. Then we conduct the encrypted identification and discuss the influences of alignments in the encryption field on the experimental results. Finally, different schemes of score-level fusion are designed and tested, and the experiments show that through these fusion schemes the result can get some improvement to a certain degree.