A COMPREHENSIVE STUDY OF SOFT COMPUTING TECHNIQUES FOR HUMAN IRIS BIOMETRIC SYSTEMS

Abstract

Biometric identification has advanced significance, with iris recognition emerging as one of the most popular and effective methods. One of the primary challenges in human iris identification and recognition using soft computing techniques include occlusion, noise, and variations in lighting conditions. The study's primary focus is to explore the effectiveness of soft computing methods in processing and analyzing iris images for accurate identification and recognition, with a particular emphasis on classification using Convolutional Neural Networks (CNN). The research utilized a CNN model to extract detailed features from iris images, which leads to accurate classification and recognition results. The proposed method is trained and tested using the Chinese Academy of Sciences Institute of Automation Iris Database Version 4(CASIA-IrisV4) dataset. Images in the (CASIA-IrisV4) dataset undergo preprocessing, including resizing and normalization, to ensure accurate detection of the iris boundary. The use of soft computing techniques improves the model's ability to adapt and handle complex iris patterns, resulting in high accuracy and precision. The recall and F1 scores demonstrate the model's ability to accurately identify positive instances while minimizing false negatives. The suggested model works really well, with an F1 score of 76%, an accuracy rate of 97%, a precision of 85%, and a recall of 74%. This research represents a significant advancement in biometric security systems, highlighting the potential of soft computing and CNN-based approaches to enhance human identification technologies.