Feature Level Fused Ear Biometric System

Abstract

In this paper, a feature-level fusion approach for improving the efficiency of ear image verification is proposed. SIFT is used to extract the features on a ear images at different pose, which is then merged according to a fusion rule to produce a single feature called the Fused Template. The similarity of SIFT extracted features of the live image and the enrolled user template in the database is measured by their Euclidean distance. IITK database is used to validate the performance of the proposed method. Comparing to non-fusion approach the fusion approach performs better in verification.