A new RBFN with modified optimal clustering algorithm for clear and occluded fingerprint identification

Abstract

In this present paper, a Radial Basis Function Network (RBFN) based on Modified Optimal Clustering Algorithm (MOCA) have been developed for clear and occluded fingerprint identification. Unlike conventional OCA technique which only considers intra cluster similarity for performing the desired number of clusters, MOCA combines both intra and inter cluster similarity while grouping such that not only the desired numbers of clusters or groups are formed, but also no misclassification is formed within any group. The approach of using MOCA within Modified RBFN for performing learning and identification of the different fingerprints is effective and efficient. Also the performance evaluation with accuracy, precision, recall and F-score of the classifier are quiet high and the learning time of fingerprints are quite low.