Development Of A Principal Component Analysis Based Face Recognition System Using K-Nearest Neighbour Algorithm

ABSTRACT

Recognition is an important task in many organisations ranging from law enforcement to

education. Therefore, organisations are constantly in search of more efficient and error free

ways to carry out the recognition process. Existing methods of recognition can be categorized

as traditional or technology assisted. The aim of this project is to develop a recognition

system that is based on authentication using one physiological trait of an individual.

The proposed recognition system was designed based on unimodal biometrics. The biometric

recognition system uses the face as the only physiological trait. The processes involved in

the unimodal biometric system include data acquisition, biometric image preprocessing,

feature extraction, matching and evaluation using accuracy, false accept rate (FAR) m1d false

reject rate (FRR) as metrics.

The developed system was evaluated using accuracy, false accept rate (FAR), false reject

rate (FRR), with each metric giving a positive result. It has an accuracy of 86 %, false accept

rate (FAR) of 0.11, false reject rate (FRR) of 0.15 m1d an average execution time of 3.60

seconds. The results gotten show that the developed system is efficient and can be

implemented in other types of biometric systems.

This project work designed m1d implemented a face biometric based recognition system. This

was achieved by extracting the features from the face using PCA and classifying the facial

features using K-Nearest Neighbour with Euclidem1 distance (K=l). The developed system

can be used in applications such as attendance taking and access control.