Loading…

RETRACTED ARTICLE: An effective deep learning features based integrated framework for iris detection and recognition

In recent years, Iris recognition has emerged as an important and trustworthy biometric model to recognize humans. The application of automatic iris recognition models find useful in different fields namely border control, citizen confirmation, and criminal to commercial products. This paper introdu...

Full description

Saved in:
Bibliographic Details
Published in:Journal of ambient intelligence and humanized computing 2021-03, Vol.12 (3), p.3271-3281
Main Authors: Jayanthi, J., Lydia, E. Laxmi, Krishnaraj, N., Jayasankar, T., Babu, R. Lenin, Suji, R. Adaline
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In recent years, Iris recognition has emerged as an important and trustworthy biometric model to recognize humans. The application of automatic iris recognition models find useful in different fields namely border control, citizen confirmation, and criminal to commercial products. This paper introduces an effective deep learning (DL) based integrated model for precise iris detection, segmentation and recognition. The projected model involves different stages namely preprocessing, detection, segmentation and recognition. Initially, preprocessing of images takes place to improve the quality of the input image using Black Hat filtering, Median filtering and Gamma Correction. Then, Hough Circle Transform model is applied to localize the region of interest, i.e. iris in an effective way. Afterwards, Mask region proposal network with convolution neural network (R-CNN) with Inception v2 model is applied for trustworthy iris recognition and segmentation i.e., recognizing iris/non-iris pixels. For validating the results of the presented model, a detailed simulation takes place using a benchmark CASIA-Iris Thousand dataset and the results are validated interms of detection accuracy. The attained simulation outcome depicted that the projected technique shows maximum recognition accuracy of 99.14% which is superior to other methods such as UniNet.V2, AlexNet, VGGNet, Inception, ResNet and DenseNet models in a significant way.
ISSN:1868-5137
1868-5145
DOI:10.1007/s12652-020-02172-y