Research On Cancelable Fuzzy Fingerprint Vault Algorithms Based On PCNN And Chaotic Sequence

Fingerprint identification is one of biometric identification technology, which uses the fingerprints as the main basis of personal identity, and it is the most mature and widely used biometric authentication technology. Fuzzy fingerprint vault algorithms based on fingerprint identification is a modern biometric key algorithm that solves the problems of traditional cryptography by binding the user's fingerprints with his cryptographic key tightly. Therefore, the research on fuzzy vault algorithm based on fingerprint has a very important theoretical significance and practical value.Firstly, based on studying existing fingerprint image thinning algorithm, an algorithm of improved binary fingerprint image thinning using template-based PCNNs is proposed. Besides redesigning the thinning templates, the square-templates and triangle-templates are used together in this approach while the orientation field doesn't need to compute. Results show that this algorithm has many advantages such as complete thinning, quick speed, high thinning rate, and smooth skeleton without spikes; moreover it can also apply to other binary images.Secondly, a new scheme named cancelable fingerprint fuzzy vault based on chaotic sequence is proposed, which changes the original template into transformed template by using transformation function. Constructed by Gaussian function and chaotic sequence in the vault construction phase, the transformation function is one-way and noninvertible. Then, the transformed template is used to construct the vault. The template can be cancelled and replaced by modifying the transformation parameters when compromised. In the vault unlocking phase, the transformed input template is generated when the same transformation is applied to the input fingerprint template. Matching is implemented between the transformed original template and the transformed input template to break the vault and get secret key. Experimental results show that our approach can protect the original template from cross-matching by different transformed fingerprint templates in different applications. Moreover, revocation of transformed template is very easy. This makes a high unlocking complexity for attackers with an acceptable unlocking accuracy for the legal user. By these ways, higher level of security of the vault and secret key is achieved.