Eigenface based algorithms to regognize noisy and occluded faces

Abstract

Face recognition is one of the remarkable human abilities to recognize people, while building a computer based face recognition system is still an active on-going research. Although the research in face recognition technology has grown gradually for the past five decades, computer based face recognition systems which are commercially available at present are still at infancy stage and have some practical limitations. In this research, two main problems in face recognition system, namely noise and occlusion had been identified for research focus. In this research, Eigenface Technique (EFT) had been employed as a core technique for recognizing faces under noisy and occluded conditions. Extensive experimental runs have been carried out by subjecting faces with various types of noise and variety of random occlusions.