| Face recognition is an important problem in computer vision with many potential applications including surveillance, identity fraud, and human-computer interaction. Considerable progress has been made in face recognition, especially with the development of powerful models of face appearance. The most popular method is the method of Eigenfaces.; Recently, Random Projection (RP) has been proposed as a promising approach for dimensionality reduction. The goal of this thesis is to investigate the application of RP for face recognition. RP has several important properties that make it suitable for face recognition. First, it is data independent. Second, it improves cluster separability. Third, it is robust to noise. Finally, it is computationally faster than other techniques.; We have conducted extensive experiments using three popular face databases. For comparison, we have experimented with the PCA approach. Our experimental results indicate RP is a promising approach for face recognition, especially using ensembles of RPs. |
