一个新的基于最近邻分类的人脸识别类器

A Novel NNC-based Classifier for Face Reognition

在人脸识别技术中,传统的最近邻分类器(NNC)是最重要的分类器之一,它直接利用测试样本与有训练样本的最小距离进行分类.文章在最近邻分类基础上提出了一个基于最近邻分类人脸识别分类器,利用同一类训练样本的线性组合来表示测度样本,通过计算测试样本到同类训练样本与对应系数乘积的距离进行分类.试验样本来自ORL人脸数据库,试验结果表明文章方案的识别效果优于传统的NNC分类器及其扩展方案CNNC.

For face recognition,the classical nearest neighbor classifier（ NNC） is one of the mosi important classifiers whichdirectly exploits the distance between the test sample and all of the training samples to perform classification.ln this paper,the authors proposed a novel NNC- based classifier for face recognition. In our method,we first used a linear combination of all the training samples in one class to represent the test sample,then performed face recognition using the calculated distance between the test sample and the result of multiplying training samples in one class by the corresponding coefflcint. The adopted test samples are from ORL face database and the experiments show that our method perfoms better than classical nearest neighbor classifier and it＇s extended method,CNNC,for face recognotion.