The novel LDA method defines its between-class and within-class scatter matrices based on these representative samples.

• 该新的LDA方法基于定义在这些典型样本上的类间散布矩阵与类内散布矩阵实现特征提取。