Feature extraction using two dimensional (2D) legendre wavelet filter for partial iris recognition
An increasing need for biometrics recognition systems has grown substantially to address the issues of recognition and identification, especially in highly dense areas such as airports, train stations, and financial transactions. Evidence of these can be seen in some airports and also the impl...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/887/1/24p%20DANLAMI%20MUKTAR.pdf http://eprints.uthm.edu.my/887/2/DANLAMI%20MUKTAR%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/887/3/DANLAMI%20MUKTAR%20WATERMARK.pdf http://eprints.uthm.edu.my/887/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | An increasing need for biometrics recognition systems has grown substantially to
address the issues of recognition and identification, especially in highly dense areas
such as airports, train stations, and financial transactions. Evidence of these can be
seen in some airports and also the implementation of these technologies in our mobile
phones. Among the most popular biometric technologies include facial, fingerprints,
and iris recognition. The iris recognition is considered by many researchers to be the
most accurate and reliable form of biometric recognition because iris can neither be
surgically operated with a chance of losing slight nor change due to aging. However,
presently most iris recognition systems available can only recognize iris image with
frontal-looking and high-quality images. Angular image and partially capture image
cannot be authenticated with the existing method of iris recognition. This research
investigates the possibility of developing a technique for recognition partially captured
iris image. The technique is designed to process the iris image at 50%, 25%, 16.5%,
and 12.5% and to find a threshold for a minimum amount of iris region required to
authenticate the individual. The research also developed and implemented two
Dimensional (2D) Legendre wavelet filter for the iris feature extraction. The Legendre
wavelet filter is to enhance the feature extraction technique. Selected iris images from
CASIA, UBIRIS, and MMU database were used to test the accuracy of the introduced
technique. The technique was able to produce recognition accuracy between 70 – 90%
CASIA-interval with 92.25% accuracy, CASIA-distance with 86.25%, UBIRIS with
74.95%, and MMU with 94.45%. |
|---|