Discriminative Zernike and Pseudo Zernike Moments for Face Recognition

Usually magnitude coefficients of some selected orders of ZMs and PZMs have been used as invariant image features. The careful selection of the set of features, with higher discrimination competence, may increase the recognition performance. In this paper, the authors have used a statistical method to estimate the discrimination strength of all the extracted coefficients of ZMs and PZMs whereas for classification, only the coefficients with estimated higher discrimination strength are used in the feature vector. The performance of these selected Discriminative ZMs (DZMs) and Discriminative PZMs (DPZMs) features have been compared to that of their corresponding conventional approaches on YALE, ORL and FERET databases against illumination, expression, scale and pose variations. An extension to these DZMs and DPZMs have been proposed by combining them with PCA and FLD. It has been observed from the exhaustive experimentation that the recognition rate is improved by 2-6%, at reduced dimensions and with less computational complexity, than that of using the successive ZMs and PZMs features