中文摘要：

提出一种核监督鉴别投影分析方法.首先将训练样本通过一个核函数非线性映射到特征空间,在该特征空间分别计算样本的局部、非局部和类内离散度,设计了一个改进的鉴别准则函数,基于该准则获得一组最优投影轴,使得投影后的样本不仅保留局部邻域信息,而且能够抽取更有利分类的非线性鉴别特征.在Yale人脸数据库上的实验结果表明：文中方法有效且性能优于Fisher线性鉴别分析和非监督鉴别投影分析方法.

英文摘要：

This paper develops a new method called kernel supervised discrirninant projection analysis. It firstly maps the training samples into a feature space via a nonlinear mapping determined by a kernel function, and then respectively computes the local scatter, nonlocal scatter and within - class scatter in feature space, and thus designs an improved discriminant criterion function, and simultaneously we obtain a set of optimal projection axes, which not only make the projected samples preserve the local neighborhood information, but extract the nonlinear diseriminant characteristics for effective classification. The experimental results on Yale face image database show that the proposed method is effective and outperforms the Fisher linear diseriminant analysis and unsupervised discriminant analysis method.