摘要： 为了形成稳定有效的签字样本库,利用Fisher鉴别准则即为Rayleigh商这一特点,根据广义Rayleigh商的极值性质,提出了一种如何在样本库的形成过程中采用学习机制的新方法,进而设计出符合系统特点的分类器。经验证该方法具有良好的稳定性与可靠性。

关键词: 小波变换, Fisher鉴别准则, 特征抽取, K-L变换, 阈值

Abstract: In order to form a stable and effective pattern library, the method is developed based on the fact that the Fisher distinction criterion function is a generalized Rayleigh quotient in essence. The extremum properties of generalized Rayleigh quotient are analyzed and a new way is given by studying system of organization in the course of producing pattern library, so a sorter is designed which adapt to the system peculiarity. This algorithm has good stability and reliability to the experiment.

中图分类号: 

• TP391.41