Image Analysis for Face Recognition Using PCA

Abstract

Face is a compound multidimensional construction and requires superior computing techniques for identification in image analysis. In this paper, the appearance-based statistical approach called Principal Component Analysis (PCA) is proposed for decreasing the number of variables in face recognition. PCA is very prominent method as it can reduce the larger dimensionality of data to the lesser one of feature space which can project data accurately. PCA is efficient when the database is of smaller size. In this each image in the trained set is represented as weighted sum of Eigen vectors called Eigen faces. These Eigen vectors are obtained from covariance matrix of a training set. Mathematical approach for formulating the calculation of eigenface has been discussed and the results are also presented. The coding is performed using MATLAB software and GUI (Graphic User Interface)..