改进的统计不相关最优鉴别矢量集

该文对统计不相关最优鉴别矢量集算法进行研究,在分析统计不相关最优鉴别矢量集算法的基础上提出了一种改进的方法。该方法在类内散布矩阵的特征空间中求解统计不相关最优鉴别矢量集。为了加快特征抽取速度,利用基于图像鉴别分析的维数压缩方法,对图像数据进行了压缩。在ORL和Yale人脸数据库的数值实验,验证本文所提出的方法的有效性。

This paper presents a research on the algorithm of optimal set of statistically uncorrelated discriminant vectors. An improved algorithm has been proposed on the basis of the analysis of the conventional algorithm of statistical uncorrelated discriminant vectors, which solves the optimal set of statistically uncorrelated discriminant vectors in the eigen space of the within-class scatter matrix Sw . The dimension of images has been reduced using the dimension reduction method based on image discriminant analysis in order to speed the process of feature extraction. The numerical experiments on facial databases of ORL and Yale show the effectiveness of the proposed method.

国家自然科学基金(60072034)中国科学院沈阳自动化研究所机器人学研究室基金(RL200108)江苏省高校自然科学研究计划项目(01KJB52002)江苏省自然科学基金(BK2002001,BK2004058)图像处理与通信实验室开放基金(KJS03038)江苏省高校博士(后)基金资助课题