Volume 12, Issue 8

Performance Analysis of ANN Optimized FT-PCA Human Palm Features

Author

Bharti Jamwal*1, Er.Ritika2and Prof. Satnam Singh Dub3

Abstract

The aim is to develop and test the performance of an image processing
based Human Palmprint Recognition Application that was developed using
combination of Artificial Neural Networks in combination with Fast Fourier
transforms and Principal Component Analysis (ANN-FTPCA) techniques is
accomplished. After the palmprint images are collected, the preprocessing
of those palmprint images is one of the important phases in palmprint
recognition systems. Instead of the large palmprint image, a small central
sub-image is required for feature extraction. So, in the preprocessing
scheme, the central region of interest (ROI) is extracted from the large
palmprint image. Also sometimes, the palmprint images need
enhancements. During the enrollment phase, multiple samples of palmprints
per user are collected in regular intervals from three different databases, so
each image from each database is different, as the palm's position suffers
from a little bit of rotation and translation in every database. So, the central
ROI generated each time will be different. But as the required area is of
central sub-images generated from the preprocessing phase, it should be all
similar. The developed system is used for human identification based on
palmprints; there is a need for the development of high-quality classification
methods and accurate feature extraction, which is very significant to
execute the system in actual operating environment. The MLP neural
network is used for feature optimization and matching to improve accuracy
and recognition rate as in the speed of the application and the performance
is evaluated on three palmprint databases CASIA, IITD and Poly U.

DOI

https://doi.org/10.62226/ijarst20231132

PAGES : 0 | 0 VIEWS | 0 DOWNLOADS

Download Full Article