Fisher模式识别问题的困难之一是维数问题，低维特征空间的分类问题一般比高维空间的分类问题简单。因此，人们力图将特征空间进行降维，降维的一个基本思路是将d维特征空间投影到一条直线上，形成一维空间-One of the difficulties Fisher pattern recognition problem is the dimension of the problem, classification of low-dimensional feature space is generally simple classification problem is high dimensional space. Therefore, it is trying to reduce the dimension of feature space, a basic idea is to reduce the dimension d-dimensional feature space projected onto a straight line, one-dimensional space