Authors: Samik Chakraborty; Madhuchhanda Mitra; Saurabh Pal
Addresses: Department of AEIE, Heritage Institute of Technology, Kolkata, India ' Department of Applied Physics, University of Calcutta, Kolkata, India ' Department of Applied Physics, University of Calcutta, Kolkata, India
Abstract: Ear is a potential biometric parameter which has drawn the attention due to its structural uniqueness and stability over the age, obesity, disease, expression, etc., unlike other common biometric traits. In this work a geometric retrieval algorithm has been proposed for ear-based biometric analysis with occluded image. First the occlusion problem is countered by an empirical data driven technique and then PSO-based optimal features are extracted for comparison that reveals the authenticity of the subject with respect to a stored database. A search of minima from Euclidian distance-based analysis is used for final decision. The proposed system is tested on 50 subjects collected in multiple sessions in laboratory with a good recognition rate superior to similar reported works as indicated in the result section.
Keywords: biometry; ear recognition; occlusion; interpolation; particle swarm optimisation; PSO; Euclidian distance.
International Journal of Biometrics, 2020 Vol.12 No.3, pp.247 - 268
Received: 06 Jun 2019
Accepted: 07 Nov 2019
Published online: 24 Jun 2020 *