Adegboye Olujoba James1* and Awodoye Olufemi Olayanju2
1Department of Computer Science, The Federal Polytechnic, Ilaro, Ogun State, Nigeria
2Department of Computer Engineering, Ajayi Crowther University, Oyo State, Nigeria
*Corresponding Author: Adegboye Olujoba James, Department of Computer Science, The Federal Polytechnic, Ilaro, Ogun State, Nigeria.
Received: July 26, 2022; Published: September 14, 2022
The paper aimed at developing an enhanced Convolution Neural Network based system to recognize iris features. The enhancement was done using Gravitational Search Algorithm (GSA) due to its advantages over other algorithms. Four hundred and fifty (450) Iris image were acquired from CASIA (vi) iris dataset. Without changing the images, the original iris images were resized to 200 by 200 pixels. Sixty percent of the acquired images (270) were used for training and forty percent (180) were used for testing. Processes such as iris segmentation, normalization, feature extraction and template matching were carried out within. On the basis of recognition accuracy, false positive rate, sensitivity, specificity, and average recognition time, the effectiveness of the existing CNN and GSA-CNN on both trained and recognized iris was evaluated. The effectiveness of the performance metrics was evaluated using a confusion matrix. it was inferred from the performance of both algorithms that the GSA-CNN model gave an increased 2.11% recognition accuracy, 2.22% specificity, 0.74% sensitivity and a decreased FPR of 2.22% over the CNN model at 0.75 threshold value. GSA-CNN outperformed CNN using the said metrics therefore we recommend that our proposed system can be used to handle security challenges in Banks, Schools, the Military, Medicine and any security-threat prone Organizations than CNN.
Keywords: CASIA; Confusion Matrix; Convolution Neural Network; Gravitational Search Algorithm; Iris
Citation: Adegboye Olujoba James and Awodoye Olufemi Olayanju. “An Iris Recognition System Using Enhanced Convolution Neural Network". Acta Scientific Computer Sciences 4.10 (2022): 24-29.
Copyright: © 2022 Adegboye Olujoba James and Awodoye Olufemi Olayanju. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.