Image processing techniques can be used for face recognition applications. The process involves decomposing face images into subbands using discrete wavelet transform. The mid-frequency subband is selected and principal component analysis is applied to extract representational bases. These bases are stored for training images and used to translate probe images into representations which are classified to identify faces by matching with training representations. This approach segments discriminatory facial features to recognize identities despite variations in illumination, pose, expression and other factors.