SHEIFALI GUPTA [1] , O.P.SAHOO [2] , RUPESH GUPTA [3] , AJAY GOEL [4] [1] Department of ECE, Singhania University, Rajasthan, INDIA [2] Department of Electronics & Communication Engineering, N.I.T. Kurukshetra, INDIA [3] Department of Mechanical Engineering, Singhania University, Rajasthan, INDIA [4] Department of Computer Science & Engineering, Singhania University, Rajasthan, INDIA sheifali@yahoo.com , opsahu_reck@yahoo.co.in , , rup_esh100@yahoo.co.in , goelajay1@gmail.com Abstract: In this paper, the performance of subspace LDA for face recognition is evaluated with ORL database using MATLAB. It is shown that as the number of training images per individual increases, success rate also goes on increasing but it also causes increase in processing time because size of training database increases. When the training images per individual are 5 or 6, it gives maximum success rate with optimized performance time. Also there is a proportionately high recognition rate when the eigenface space’s dimension is small (40A60) and it is less when eigenface space’s dimension is large (180A200). When only significant eigen vectors are used in subspace LDA with 5 or 6 training images, then it gives maximum success rate up to 92%. KeyWords: Fisherface, Face recognition, LDA (Linear Discriminant Analysis), PCA (Principal Component Analysis), Euclidean distance. The necessity for personal identification in the fields of private and secure systems made face recognition one of the main fields. The importance of face recognition rises from the fact that a face recognition system does not require the cooperation of the individual while the other systems need such cooperation. Face recognition algorithms try to solve the problem of both verification and identification [1]. When verification is on demand, the face recognition system is given a face image and it is given a claimed identity. The system is expected to either reject or accept the claim. On the other hand, in the identification problem, the system is trained by some images of known individuals and given a test image. It decides which individual the test image belongs to. Much of the work in face recognition by computers has focused on detecting individual features such as the eyes, nose, mouth and head outline, and defining a face model by the position, size, and relationships among these features. Such approaches have proven to depend on the precise features [2]. The Eigenface Method of Turk and Pentland [4] is one of the main methods applied in the literature which is based on the KarhunenALoeve expansion. It is based on the application of Principal Component Analysis to the human faces. It treats the face images as 2AD data, and classifies the face images by projecting them to the eigenface space which is composed of eigenvectors obtained by the variance of the face images. Eigenface recognition derives its name from the German prefix eigen, meaning own or individual. The eigenface approach works well as long as the test image is similar to the training images used for obtaining the eigenfaces. Etemad and Chellappa [3] proposed a method on appliance of Linear/Fisher Discriminant Analysis for the face recognition process. LDA is carried out via scatter matrix analysis. The aim is to find the optimal projection which maximizes between class scatter of the face data and minimizes within class scatter of the face data. As in the case of PCA, where the eigenfaces are calculated by the eigenvalue analysis, the projections of LDA are calculated by the generalized eigenvalue equation. Here an alternative method which combines PCA and LDA is studied. This method consists of two steps; the face image is projected into the eigenface space which is constructed by PCA, and then the eigenface space projected vectors are projected into the LDA classification space to construct a linear Proceedings of the 9th WSEAS International Conference on TELECOMMUNICATIONS and INFORMATICS ISSN: 1790-5117 79 ISBN: 978-954-92600-2-1