Curvelet based face recognition via dimension reduction

Multiresolution ideas, notably the wavelet transform, have been proved quite useful for analyzing the information content of facial images. Numerous papers and research articles have discussed the application of wavelet transform in face recognition. However, little attention has been paid to the newly developed multiresolution tools (contourlet, curvelet, etc.) despite their improved directional elements and other promising abilities compared to traditional wavelet transform. In this article we introduce the application of digital curvelet transform in conjunction with different dimensionality reduction tools, looking particularly at the problem of facial feature extraction from 2D images. The purpose of this paper is exploratory. We do not claim that the results achieved here are the best possible. Rather, we aim at showing that curvelets can serve as an effective alternative to wavelets as a feature extraction tool. This work can be seen as a stepping stone for further research in this direction. Our methods have been evaluated on well-known databases like ORL, Essex Grimace and Yale face. Curvelet based results have been compared with that achieved using wavelets and other existing techniques to show that curvelets indeed has the potential to supersede wavelet based results.