摘要
不相关空间算法是求解不相关鉴别矢量集的快速算法,但是将其应用在人脸识别中将遇到小样本问题,并且算法只是一种线性的特征提取方法。该文提出一种核不相关空间算法,该方法的关键是高维特征空间中不相关空间的计算,对此提出一种简单的计算方法,即根据eigenface中将高阶矩阵计算转化成低阶矩阵计算的思想,将高维特征空间中不相关空间的计算仍归结为标准的特征值分解问题。所提出的算法能够有效地解决小样本问题。在ORL人脸库上的实验结果验证了所提出的算法的可行性和有效性。
Uncorrelated space algorithm is a fast method for the uncorrelated discriminant vectors extraction, but it may encounter the small size samples problem when it is applied to face recognition task. In addition, it is only a linear feature extraction technique. In this paper, kernel uncorrelated space algorithm is proposed. The key of the proposed algorithm is to how to compute the uncorrelated space in the higher dimensional feature space. As to this problem, a very simple and easy method is proposed, which originates from the eigenface that transforms the computation of the high order matrix into the computation of the low order matrix, and then the actual computation of the uncorrelated space in the higher dimensional feature space is reduced to a standard eignenvalue problem. In addition, the proposed algorithm can effectively overcome small size samples problem. The numerical experiments on facial databases of ORL show that the proposed method is effective and feasible.
出处
《电子与信息学报》
EI
CSCD
北大核心
2009年第8期1812-1815,共4页
Journal of Electronics & Information Technology
基金
中国博士后基金(20060400809)
黑龙江省青年科技基金(QC06C022)资助课题
关键词
人脸识别
不相关空间算法
小样本问题
核不相关空间算法
Face recognition
Uncorrelated space algorithm
The small size samples problem
Kernel uncorrelated space algorithm