基于分离度的图象特征提取与识别方法

连石柱

doi:10.11834/jig.19960361

本期目录 | 浏览量 : 0 下载量: 59 CSCD: 0

PDF
导出
分享
收藏
专辑

基于分离度的图象特征提取与识别方法
A Method of Feature Selection and Classification Based on Divergence
1996年1卷第3期页码：196
网络出版：2012-03-20，

纸质出版：1996
DOI： 10.11834/jig.19960361
稿件说明：

移动端阅览

连石柱. 基于分离度的图象特征提取与识别方法[J]. 中国图象图形学报, 1996,1(3):196. DOI： 10.11834/jig.19960361.

A Method of Feature Selection and Classification Based on Divergence[J]. Journal of Image and Graphics, 1996, 1(3): 196. DOI： 10.11834/jig.19960361.

摘要

对于图象中不同类别（以其统计分布函数一表征）的可识别性，分离度的统计定义可做出定量的描述。根据分离度在一对应映射变换下的不变性，应用Karhunen-Loeve变换对两类分布撮识别特征，发现分离度只依赖于特征值最大和最小的两个特征向量。而且，分离度的大小依赖于特征值怀某个定定值的偏差。由此我们提出一个识别模型，使每次分类后的样本集的特征值都趋近于两上定值，从而得到最佳的识别效果。由该模型设计的

Abstract

The separability of two pattern classes can be measured by the divergence for two Gaussian distributions. Because the divergence is invariant under the linear transformation we can extract "good" features for separating two patterns via Karhunen-Loeve transformation. It is shown that the divergence is only dependent on two of n eigenvalues. One property of a "good" dichotomy is that each feature should be effective for classification. Thus

a criterion function is proposed. This algorithm

which is called as sample exchange algorithm

is convergent and it is a reasonable unsupervised clustering method for classification.