HUMAN EAR CLASSIFICATION SCHEME BASED ON LOBULE CHAIN CODE AND HELIX DIMENSIONALITY

Abstract

In this paper, a human ear classification approach which uses the ear image contour was proposed. Basically, the scheme has four classification groups which are based on only two concepts namely: Chain code (Lobed and Lobeless Ear) and Dimensionality of Helix (Big and Small Ear). Using the chain code feature vectors extracted from the ear contour after necessary pre-processing activities, the ear were classified based on the code pattern of the last twenty or fifteen codes which uses the rule of thumb classifier we built. Similarly, the second classification process implored the Helix dimensional features which were a product of the Angular Points and landmark space, using the Square of Sum Difference. Therefore, the experimental result showed about 94.80% and 92.20% accuracy respectively.