The Limitation of Pre-processing Techniques to Enhance the Face Recognition System Based on LBP

Abstract

Most systems are intelligent and the industrial world is moving now towards
technology. Most industrial systems are now computerized and offer a high speed.
However, Face recognition is a biometric system that can identify people from their
faces. For few number of people to be identified, it can be considered as a fast
system. When the number of people grew to be bigger, the system cannot be adopted
in a real-time application because its speed will degrade along with its accuracy.
However, the accuracy can be enhanced using pre-processing techniques but the
time delay is still a challenge. A series of experiments had been done on AT&TORL
database images using Enhanced Face Recognition System (EFRS) that is
adopting a Local Binary Pattern Histogram (LBPH) as feature extraction method.
During these experiments, LBP parameters and pre-processing techniques have been
changing and the results were recorded for accuracy and time for different database
sizes. The accuracy reaches 98.5% but the time delay keep increasing with the
number of people in the database.