摘要
本文提出一种新型的人脸识别方法。该方法首先通过二维小波变换提取人脸图像的低频特征,然后采用最近邻凸包分类器对该特征进行分类。二维小波变换是提取图像特征的有效方法之一,在保留原始图像的主要特征的同时,还能够有效降低图像维数;最近邻凸包分类器是一种以测试样本点到各类别训练样本凸包的距离作为相似性度量的分类算法。本文将这两项技术相结合在ORL人脸识别数据库上取得了良好的实验效果。
A new face recognition algorithm based on wavelet transform (WT) and nearest neighbor convex hull (NNCH) classifier is presented in this paper. The nearest neighbor convex null (NNCH) classifier is a novel classification algorithm, in which the distance from point to convex hull is the similarity measurement of nearest neighbor classification algorithm. NNCH classifier attempts to take the convex hull of all training samples of each class as the extended class set, and estimate test samples with the convex ...
出处
《微计算机信息》
北大核心
2008年第4期212-214,共3页
Control & Automation
基金
国家自然基金资助项目(No.60472060
No.60632050)
关键词
小波变换
凸包
最近邻凸包分类
图像识别
人脸识别
wavelet transform
convex hull
nearest neighbor convex hull classifier
image recognition
face recognition
