Abstract: Fisher linear discriminant analysis algorithm based on vector muster is presented in this paper.The original high-dimensional vectors are divided into a set of sub-vectors with low-dimensional.Fisher linear discriminant analysis is adopted based on vector muster.This algorithm can deal with all high-dimensional and small sample size problems.Otherwise，selecting appropriate dimension of sub-vector can extract the optimization feature value of vector.Fisher linear discriminant analysis algorithm based on vector muster is the extension of Fisher linear discriminant analysis and two-dimensional fisher linear discriminant analysis.

Key words: Fisher linear discriminant analysis, between-class scatter matrix, within-class scatter matrix, high-dimensional and small sample size problem

摘要： 提出了一种基于向量组的Fisher线性鉴别分析方法。该方法先将原始的高维向量分割为低维的子向量组，再对向量组运用Fisher线性鉴别分析。这种处理方法，不但能够解决任意高维下的小样本问题，而且通过选择恰当的子向量维数，可以从向量中抽取出最有效的特征值。此外，基于向量组的Fisher线性鉴别分析是Fisher线性鉴别分析和二维Fisher线性鉴别分析的进一步推广。

关键词: Fisher线性鉴别分析, 类间散布矩阵, 类内散布矩阵, 高维小样本问题