Face recognition approach based on wavelet transform

The paper presented describes a biometric face identification method based on frontal face images that combines the wavelets theory with the probabilistic neural nets structure. We will use the wavelet transformed domain in order to reduce the dimensionality of the input image, and a radial basis neural net classifier that will perform the identification process. The method developed will be compared against the classical method of Turk and Pentland [1], based on the Karhunen-Loeve transform. Experimental results show better recognition accuracies and reduced computational cost.