A novel method of feature extraction based on local scatter and nonlocal scatter

A new unsupervised discriminant projection for dimensionality reduction of high dimensional data is presented in this paper. The new method is a linear projection based on both the local and nonlocal statistically quantities. The discriminant criterion function be characterized by difference between the nonlocal scatter and the local scatter of feature vector, seeking to find a group of projection axis that simultaneously maximizes the nonlocal scatter and minimizes the local scatter of feature vector. The experimental results on Olivetti Research Laboratory (ORL) face database and AR face database show that the proposed method consistently outperforms locality preserving projection (LPP) and unsupervised discriminant projection (UDP), and even outperforms Fisher linear discriminant analysis.