Classification of Crease Features from Interdigital and Hypothenar Regions of Palmprint Image for Race Identification using Convolutional Neural Network
Roszaharah Yaacob1, Chok Dong Ooi2, Seng Chun Hoo3, Haidi Ibrahim4, Helmi Hadi5, Nik Fakhuruddin Nik Hassan6
1Roszaharah Yaacob, School of Health Sciences, Health Campus, Universiti Sains Malaysia, Kubang Kerian, Kelantan, Malaysia.
2Helmi Hadi, School of Health Sciences, Health Campus, Universiti Sains Malaysia, Kubang Kerian, Kelantan, Malaysia.
3Nik Fakhuruddin Nik Hassan, School of Health Sciences, Health Campus, Universiti Sains Malaysia, Kubang Kerian, Kelantan, Malaysia
4Chok Dong Ooi, School of Electrical and Electronic Engineering, Engineering Campus, Nibong Tebal, Pulau Pinang, Malaysia.
5Seng Chun Hoo, School of Electrical and Electronic Engineering, Engineering Campus, Nibong Tebal, Pulau Pinang, Malaysia.
6Haidi Ibrahim, School of Electrical and Electronic Engineering, Engineering Campus, Nibong Tebal, Pulau Pinang, Malaysia.
Manuscript received on 01 May 2019 | Revised Manuscript received on 15 May 2019 | Manuscript published on 30 May 2019 | PP: 1812-1818 | Volume-8 Issue-7, May 2019 | Retrieval Number: G5877058719/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In this paper, we investigate the feasibility of crease features from interdigital and hypothenar regions of palmprint image to identify race of individual. The classification is done by means of deep learning architecture known as convolutional neural network (CNN). In this research, two square region of interests (ROIs) have been used, corresponding to interdigital and hypothenar regions, as the input data for the CNN classifier. Three sizes of input data have been used. Experiments to select suitable CNN parameters have also been carried out. These parameters are the number of training epoch, activation function, and data augmentation. Results obtained through a four-fold cross validation have shown that variation of input data size would deviate computational complexity and classification performance of the CNN classifier. Besides that, fine tuning on CNN parameters and data augmentation could induce positive effect on classification.
Keyword: Artificial intelligent, biometrics, convolutional neural network, identification, machine learning, palmprint, race identification.
Scope of the Article: Neural Information Processing.