Face recognition system based on features extracted from two domains

A face recognition system which represents each image as a superposition of the dominant components in two transform domains is proposed. The Discrete Wavelet Transform (DWT) and the Discrete Cosine Transform (DCT) are the two domains. By the end of the Training mode, each pose in the gallery will have two final matrices. Feature Extraction step in the Training includes transforming the preprocessed image to the DWT domain followed by the DCT. Then, the first feature matrix is obtained by retaining certain number of the DCT coefficients while the rest of the DCT matrix, the Residual (R), is transformed back to the Wavelet domain. Next, the DWT is applied several times to get the other feature matrix. The Classification mode consists of the same sequence of steps as in the Training to calculate the feature matrices. The Euclidean distance measure is used to compute the separation of test matrices and the training ones. Since the features are in two different domains, a voting scheme is utilized to give the final decision which is based on selecting the minimum distance. Two publicly available databases, namely, ORL, and YALE are used to evaluate the performance of the proposed technique. As shown in the results, the system gives higher recognition rates compared with existing approaches. The other two design parameters, the computational complexity and the storage requirements, were also lower.