摘要
分类器设计的好坏对于图像识别效果有着重要影响 .本文基于黄等所提出的识别方法 ,定义了一类更广泛的隶属函数 ,并借助于投影算子理论、子空间理论 ,对所提隶属函数的性质进行了深入的理论分析 ,证明了所提隶属函数所具有的若干良好特性 ,证实了所设计的分类器是二次非线性分类器 .最后 ,设计了从特征抽取到识别的整个流程 ,并应用于人脸识别 .实验结果表明 。
The design of classifier has significant influence upon image recognition.Based on Huang′s recognition methods,class of more comprehensive membership functions are defined in this paper.According to the projection operator theory and the subspace methods,the properties of the membership functions are theoretically analyzed.It is proved that the functions have some good properties and the designed classifiers based on the functions are quadric nonlinear separable.Finally,the whole process from feature extraction to recognition is designed and applied to face recognition.Experimental results show that the proposed methods are effective and feasible.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2000年第11期50-54,共5页
Acta Electronica Sinica
基金
国家自然科学基金! (No .6997 30 2 1 )
