Face Recognition Using Discrete Cosine Transform and Fuzzy Linear Discriminant Analysis

In order to solve the problem of uncertainty that occurred in the field of face recognition because of the variations of the facial expressions, illuminations and poses as well, a new feature extraction method for face recognition is proposed in this paper. The method combines the DCT with FLDA. Firstly, DCT is performed on the entire face image to obtain all the components in the frequency domain. Due to the energy compaction, only the lower frequency components will be retained and thus, the dimension reduced features are obtained, then FLDA is employed to extract the most discriminating features. Finally, the nearest neighbor classifier is employed for classification. The results of experiments conducted on Olivetti Research Laboratory (ORL) database show that the proposed method is better than other methods in terms of accurate recognition rate.