A high-speed method of face recognition using the combination of Linear Discriminant Analysis (LDA) and Principal Component Analysis (PCA) that is also known as Fisher's Linear Discriminant(FLD) is presented in this thesis. LDA is used to reduce the dimension of the original face in an image and the area illumination variations are lighten. Once the dimension is reduced, PCA is used to find the linear combination of features that have the most total varience in data. Through this a lot of discriminative information will lost when throwing some components away. This actually can produce in bad result especially when it comes to classification. So, in order to find the combination of features that seperates best between classes, Eigenvectors been used as well. This linear projection method FLD is applied further in feature vectors to get the most discriminating and invariant feature of faces.