Face recognition using bit-plane feature extraction approach in neutal networks

• Main Author: Ting, Kung Chuang
• Format: Thesis
• Language: English
• Published: Universiti Malaysia Sarawak, UNIMAS 2009
• Subjects: TA Engineering (General). Civil engineering (General)
• Online Access: http://ir.unimas.my/id/eprint/3149/8/Face%20recognition%20using%20bit-plane%20feature%20extraction%20approach...%28fulltext%29.pdf; http://ir.unimas.my/id/eprint/3149/
• Institution: Universiti Malaysia Sarawak

Face recognition technology (FRT) has been studied and researched for over 30 years and nowadays it has variety of potential applications in security and law enforcement. It is extensively researched because of its potential to become a new method of interaction between human and machine without using any keyboard mouse. This will be able to replace conventional methods such as Personal Identity Number (PIN) and token system, which are vulnerable to forgery. This research presents a novel approach of face feature extraction using bit-level information which we called as bit-plane feature extraction method. The face databases used for evaluation are CMU AMP Face Expression Database and Yale Face Database. CMU AMP Face Expression Database consists of 13 subjects and each within 75 images while Yale Face Database only has 15 subjects with 11 images per subject. The research is concluded with proven result that bit-plane feature extraction is a reliable feature for face recognition due to its high recognition rate and ability in impostors’ rejection.