Eigenfaces Recognition Technique for Verifying Noisy Facial Images

Abstract

In this paper, the eigenfaces recognition method is used to verify noisy images of human faces. The Principal Component Analysis “PCA” which is based on the implementation of the Karhunen-Loeve “KL” transformation is used to compute the Eigenvectors of the test’s images. The similarity between the preserved test’s faces (as Database record) and the input test’s face is performed by utilizing the Mean-Square-Error “MSE” criterion. Different amount of noises (Gaussian, Uniform, Salt-and-Pepper) have been added on the test’s image face and compared with preserved faces to deduce the verification reliability of the utilized recognition method.