GEOMETRIC FEATURE EXTRACTION FOR EAR IDENTIFICATION USING ASPECT RATIO PRINCIPLE

ABSTRACT

In the realm of biometrics, ear recognition stands out as a unique approach for human authentication. Unlike other biometric methods such as face and fingerprint recognition, ear recognition offers several merits. The distinctive contour of each person’s ear serves as the primary reason for adopting this technique. We have proposed an ear identification system using aspect ratio of ear image contour for feature extraction. In detail, firstly, the acquired ear image is enhanced to ensure better image going for further processing. The ear image contours are now detected using canny operator, which has been rated as the best in this research work. The ear image is further enhancement using morphological dilation to remove some noisy edges which could have been disconnected. Secondly, the centroid of detected edge is computed by making constructing a bounded rectangle and a subsequently, landmark points are detected which are termed minor and major axis. Employing the aspect ratio principle, the ear image features are extracted and Euclidean distance used for identification of the ear. The experimental result shows an identification accuracy rate of 92.20%, 90.25%, 89.18 and 86.14% respectively with USTB II standard ear Database.

Keywords: Biometrics, System, Ear Recognition, Identification, Enhancement, and eccentricity