摘要： 采用插值方法降低人脸样本向量的维数,并分析比较了不同插值方法及其组合的降维效果,从而为选择合适的插值方法及其组合形式奠定了基础,并试验了不同距离分类器的计算结果。试验证明插值脸方法比PCA方法更加有效,研究思路和方法简洁,结果满意,对探索人脸识别技术和基于生物特征鉴别分析的工程应用具有潜在的应用价值。

关键词: 人脸识别, 插值方法, 线性鉴别分析(LDA)

Abstract: This paper uses interpolation methods to reduce the face image dimensions, and analyzes the availability of different interpolationmethods and their fusion, which provides a good base for selecting proper interpolation methods and their combination types, moreover differentdistance classifiers are tested for face recognition. The tests in this paper prove the better performance compared to PCA dimension reductionmethod. The research methodology in this paper is very simple and clear, all of these have great potential value to exploring human face recognitiontechnology and engineering applications.

中图分类号: 

• TP391.41