Dimensionality Reduction of Extracted Feature Database for Face Recognition System Using Two Dimensional Maximum Margin Criteria
K P Gaikwad, Sagar Arun More · 2010
In statistical pattern recognition, high dimensionality is a major cause of the practical limitations of many pattern recognition technologies. Moreover, it has been observed that a large number of features may actually degrade the performance of classifiers if the number of training samples is small relative to the number of features. This fact, which is referred to as the “peaking phenomenon”, is caused by the “curse of dimensionality”. In this paper solution to this problem is stated. Dimensionality of images after feature extraction for storing feature database is reduced in this paper. The input for the system is images from standard database. Features are extracted of given images using Two Dimensional Maximum Margin Criteria from row as well as column direction.