K-Means and FCM Hybrid Clustering for Personal Identification: Ear Biometrics

We present development of personal identification system. In this paper, the system is built on clustering using both k-means (KM) and fuzzy c-means (FCM) algorithms. The case in point is ear biometrics features; for the development of the system, 605 ear images were considered; two independent systems – one based on KM algorithm and the other based on FCM algorithm; both systems follow same flow while implementation that is assort the given image to a particular cluster and display the personal details available after retrieving from the database. Both the methods have resulted in three groups with low variation in number of samples – KM and FCM have shown. The number of samples in group 2 is exactly same for both methods, whereas groups 1 and 3 have a variation in the range of 5–20% for groups 3 and 1, respectively.