Face recognition based on Independent Component Analysis on wavelet subband

In this paper a multi-resolution analysis based on Independent Component Analysis (ICA) for face recognition is examined. We extract image features of facial images from various wavelet transforms (Haar, Daubechies, Coiflet, Symlet, Biothogonal and Reverse Biorthogonal) by decomposing face image in subbands 1 to 8. These features are analyzed by ICA and Euclidean distance measure. A series of experiments based on ORL database were then performed to evaluate the performance. The results show that for the entire wavelets, subbands 2 and 3 give the best accuracy and are computationally most efficient. Reverse Biorthogonal and 8^th order Symlet are found to be the best among all. Our experiments also prove that face recognition accuracy using ICA on wavelet subbands is higher than ICA used alone.