Orientation angle-based 2D ear recognition system

Abstract

Ear recognition system has recently become a focus in the biometric field, owning to the rich and stable geometrical structure of the ear. This work proposes the use of orientation angle of the geometrical structure of ear outer edge image for human recognition, since using the pixel value could be erroneous due to pixel intensity variations. After necessary image normalization processes, the edge image of the ear and its coordinates are stored. Subsequently, the feature vector which is the orientation angle of the stored edge image is used in our previously proposed classifier and euclidean distance measure. The proposed method is evaluated using the University of Science and Technology Beijing (USTB) ear database, which confirms to some extent the effectiveness of the proposed technique