节点文献
有效最佳鉴别特征的抽取与维数问题
Effective Extraction of Optimal Discriminant Features and the Dimension Problem
【摘要】 在模式识别领域中 ,基于 Fisher鉴别准则函数的最佳鉴别变换有着重大的影响 .该文首先对 Foley-Samm on最佳鉴别变换、广义最佳鉴别变换与具有不相关性的最佳鉴别变换给出了形式相似的求解公式 ,然后根据 Fukunaga维数定理 ,提出了最佳鉴别变换的有效度的概念 ,最后利用国际上广泛使用的 Concordia UniversityCENPARMI手写体数字数据库做了最佳鉴别变换的有效性实验 ,实验结果表明 ,具有不相关性的最佳鉴别变换是最有效的 .
【Abstract】 Based on Fisher’s discriminant function, optimal sets of discriminant vectors have great influence in the area of pattern recognition. This paper gives out three formula in similar formation to compute Foley Sammon′s optimal set of discriminant vectors, the generalized optimal discriminant vectors and the uncorrelated optimal discriminant vectors. According to Fukunaga′s theorem about the dimension problem, an idea about the efficiency of an optimal set of discirminant vectors is presented. Experiments with Concordia University CENPARMI′s handwritten digit database have been performed. Experimental results show that the uncorrelated optimal set of discriminant features is the most effective.
【Key words】 pattern recognition; feature extraction; discriminant analysis; dimension problem;
- 【文献出处】 计算机学报 ,CHINESE JOURNAL OF COMPUTERS , 编辑部邮箱 ,2000年01期
- 【分类号】TP391.4
- 【被引频次】31
- 【下载频次】235