Face Recognition: A Fourier Transform and SVD Based Approach

Authentication based on a person´s face is one of the most stringent measures to secure a place or system. In this report such a method has been proposed that successfully identifies a person. The proposed algorithm also identifies a person with emotions/varied face expressions. This algorithm first enhances the image and then computes its singular value decomposition(SVD) to yield a matrix containing singular values followed by its singular value decomposition which further yields a single numerical value, instead of a matrix, that has been employed to compare images. This report presents a method employing frequency domain transformation followed by Singular Value Decomposition of the test image which is used to compare it with the original image in the database. Experiments are performed with face images of Psychological Image Collection at Stirling(PICS) and the results are also shown.