摘要
类似于通过求解图像协方差矩阵特征值所得到的特征脸方法,提出在双三次插值得到图像降维基础上,再进行线性鉴别分析(LDA)的插值脸方法,从而得到鉴别矢量集,实现人脸图像识别.试验结果显示,插值脸方法比普遍采用的特征脸方法效果更好.所提出的思想方法对于人脸识别的理论研究和工程应用具有较大价值.
This paper puts forward to a new method called interpolation faces based on bi-cubic interpolation for dimension reduction of face images followed by linear discriminant analysis method for discriminant vectors, and this research is similar to traditional eigenfaces method. From the satisfactory experiment results and theoretical analysis, the proposed research method is better than eigenfaces. This method is very valuable to theory research and engineering applications of face recognition.
出处
《小型微型计算机系统》
CSCD
北大核心
2005年第7期1246-1250,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金(60072034)资助
留学回国人员资助资金(k206001)资助.
关键词
插值方法
人脸识别
特征脸
interpolation method
face recognition
eigenface
