Fast Face Recognition Technique for Small and Portable Devices

Abstract

Face recognition finds many important applications in many life sectors and in particular in commercial and law enforcement applications. Face recognition systems which are constrained by limited memory size and computational resources, are under deep investigation from researchers. The rapid progress in multimedia hardware technology has motivated us to explore new and practical methods in face recognition that could be implemented and loaded on small and portable devices. In this paper, we describe a new approach for face recognition where device's size and memory are limited. The proposed approach has two main features: (a) it extracts the most important and discriminant features of the face space represented by the MPEG- 7 Fourier Feature Descriptor (FFD) and (b) it applies the Principal Component Analysis (PCA) to it. The proposed method achieves substantial savings in the computation time needed by the recognition system besides keeping the recognition rate as high as possible. Therefore, the proposed method could be considered to be a good candidate that can accommodate the mobile's memory constraints.