Optimization of PCA based face recognition models

In this paper, an optimization on principal component analysis (PCA) based subspace methods using mean and standard deviation is proposed. An image I (m x n) is transformed into a matrix S (m x 2). Each row elements of the transformed, matrix S(m x 2) is the mean (V) and the standard deviation (a) of the elements of the corresponding row of the actual image I (m x n). PCA based face recognition methods are then applied on the transformed image. Representation of the facial images in p and a reduces the computational burden in obtaining the. feature vector, which eventually decreases the recognition time. Experimentations carried out during the course of this research have revealed that the proposed optimization on the PCA based subspace methods is competent enough and has better runtime performance when compared to the conventional schemes.