An Hybrid Of Fuzzy And K-Means Clustering Based Cancelable Template Generation For Multimodal Biometrics

ABSTRACT

Multi-biometric systems, which incorporate several modality biometric features, provide a benefit over uni-biometric systems in terms of the level of precision and security. Though there are major challenges such as feature fusion and biometric template security, there is a lack of literature regarding cancelable multi-biometric systems. A fingerprint and iris-based biometric cancelable framework that offers template security and revocability are presented in this article. Each modality is calculated and then computed to create a vector of scores. The fusion approach is used on the score level. We ran a preliminary analysis that used the k-means clustering process and based on that we segmented the score spectrum into three zones of concern that would be beneficial to the proposed recognition method. After separating the areas, the fusion is applied using the Fuzzy method to create a cancelable template. The proposed methods were tested on standard biometric databases using two parameters, respectively, FalseAcceptRate (FAR) and FalseRejectRate (FRR). The outcomes have been discovered to be particularly important, as they dramatically show that the proposed fusion methods can outperform the approaches focused on a Gabor-HOG previous process.

FULL TEXT