基于加权平均脸的最近邻算法  

Nearest neighbor algorithm based on weighted average face

作　　者：赵深茂 唐琳[2] 王汝言[3] 

机构地区：[1]四川化工职业技术学院,四川泸州646005 [2]中国电信股份有限公司珠海分公司,福建珠海519002 [3]重庆邮电大学通信学院,重庆400065

出　　处：《数字通信》2009年第3期42-44,49,共4页Digital Communications and Networks

摘　　要：提出了一种基于加权平均脸的最近邻算法。该方法提取与测试样本距离较近的几个样本点,并用提取出的样本点组成加权平均脸。加权平均脸方法扩展了人脸图像的姿态和表情变化,在ORL人脸库和YALE人脸库上的实验表明,该方法提高了人脸样本的识别率。This paper proposes a kind of nearest neighbor algorithm based on weighted average face. This algorithm extracted some points of training samples which is near the test samples, and then the weighted average face was calculated. The algorithm extends the changes of facial expression and posture. Experiments show that this algorithm increases the recognition rate.

关 键 词：最近特征线 最近特征面 加权平均脸 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]