摘要
主成分分析方法(PCA)是目前广泛应用在人脸等图像识别领域的重要手段。为了更准确地识别人脸的表情信息,有效抽取出图像中对表情识别贡献较大的局部特征,提出了一种类内分块PCA方法对人脸表情进行特征提取。首先对图像进行分块,再对分块得到的所有子图像块利用PCA方法进行鉴别分析,并计算出各类训练样本的子空间,然后计算测试样本到各类子空间的距离,最后输入最近邻分类器得到分类结果。在JAFFE人脸表情库上进行的实验结果表明,使用该方法后获得的识别率优于传统的PCA方法。
Principal component analysis （PCA） is the important technique widely used in the areas of images recognition such as human face. Aiming at recognizing the information of human face expression, extracting the more important local feature for ex- pression recognition, the technique of within-class modular PCA was presented. The images were divided firstly and the PCA method was directly used to the sub-images obtained from the previous step. Then the subspaces for each class of training samples were calculated. Finally, the distances from the tested samples to the subspace were eomputed and the classified recognition was carried by the nearest-neighbor. The results of the experiment in the Japanese female face expression database indicate that the recognition rate of the modular PCA is obviously superior to that of traditional PCA.
出处
《机电工程》
CAS
2009年第7期74-76,共3页
Journal of Mechanical & Electrical Engineering
关键词
主成分分析方法
特征提取
类内分块PCA
人脸表情识别
principal component analysis （PCA）
feature extraction
within-class modular PCA
facial expression recognition （FER）
作者简介
作者简介：龚婷（1980-），女，江西丰城人，主要从事数字图像处理、模式识别方面的研究．E-mail：gongting@ZUSt．edu．cn
通信联系人：胡同森，男，教授，硕士生导师E-mail：hts@zjut．edu．cn
