本文主要研究了将不同的判别分析方法应用于光学可变图像的识别。首先，介绍了光学可变图像的发展历程和光学可变图像的特点。同时，介绍了模式识别这门学科的发展以及部分基础理论，在此基础上讨论了在模式识别领域中解决小样本问题(样本数小于样本维数)时，常用的正则化方法，包括quadratic discriminate analysis(QDA), linear discriminate analysis(LDA), regularized discriminate analysis (RDA), leave—one-out covariance matrix estimate(LOOC) and Kullback-Leibler information measure based method(KLIM)。 本论文的目的是将模式识别方法应用于光学可变图像的研究，希望能根据光学可变图像的数据特征，并结合实验结果，比较各个判别分析算法应用于光学可变图像时的优缺点，从而改进判别流程，以得到一种针对普通光学可变图像具有较高识别效率的判别分析方法。

This paper mainly discusses the comparison of discriminant analysis methods applied to Diffractive optically variable image (DOVI). At first, it introduces the progress of DOVI as well as its main character of it. Then the progress of pattern recognition will be depicted and some of its basic theory will be referred. Based on the theory above, it talks about some popular regularization methods in pattern recognition field when dealing with the small sample size(SSS) problem, namely quadratic discriminate analysis(QDA), linear discriminate analysis(LDA), regularized discriminate analysis(RDA), leave—one-out covariance matrix estimate(LOOC), and Kullback-Leibler information measure based method(KLIM). The purpose of this paper is to apply these methods to Diffractive optically variable image, and to find out the best method among them that can be used to classify DOVI data.