摘要
特征抽取是图象识别中的一个最重要、最基本的问题。代数特征抽取是一种有效的特征抽取方法。该文给出了一种新的代数特征抽取方法。首先给出了图象矩阵的特征矩阵的概念，再根据特征矩阵抽取代数特征。该文证明了这样抽取的代数特征具有一些重要的代数和几何不变性。作为一个应用，将此方法用于人脸识别，取得了很好的效果。
The feature extraction is the most important and fundamental problem in image recognition.The algebraic feature extraction is an effective method of the feature extraction.In this paper,a novel algebraic feature extraction method is presented.At first the eigen matrix (EM) of the image is defined,then the EM is used for extracting the algebraic feature.This paper proved that this algebraic feature has some important properties of algebraic and geometric invariance.As an example,this method is used for facial image recognition.The experimental result shows that this method is effective.
出处
《南京理工大学学报》
EI
CAS
CSCD
1997年第5期387-390,共4页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金
国家教委博士点基金
关键词
矩阵
特征抽取
模式识别
人脸
图象识别
matrices (mathematics),feature extraction,pattern recognition
optimal discriminant projection vector
