摘要
本文提出了一种新型的表情组合模板与Gabor小波相结合的面部表情识别方法。借助Gabor小波变换在图像处理领域的优势,克服不同因素给表情识别带来的不利影响,有效的提取与表情变化有关的特征,从分类的角度来构造的表情组合模板。在人脸检测的基础上,利用Gabor小波变换提取各个区域的特征矢量,与待识别表情对应的各个特征区域的特征矢量进行比较,利用欧氏距离选择最小值对应的矢量,返回表情组合模板中,确定待识别表情的类型。实验表明该方法的识别率可达83%,表情识别效果较好。
This paper presents a new method of face recognition, with a combination of facial expression templates and Gabor wavelet, With the aid of Gabor wavelet transformation in the field of image processing, the method discussed could effectively extract charac- teristics of facial expression, overcome the negative impacts of individual differences as well as construct expression templates from the perspective of the classification. First, on the basis of face detection, we use of Gabor wavelet transformation to extract vector in each area. Then compares feature vectors of expression to be recognized in each feature area. After choosing the vector corresponding to the minimum by using Euclidean distance, it will turn back to the expression combination template'and finally identify the type of expression. Experiments show that the identification rate of method is as high as 83% and the effect of expression identification is ex- cellent.
出处
《微计算机信息》
2010年第4期10-12,共3页
Control & Automation
基金
基金申请人:高晓兴
项目名称:基于小波变换的人脸跟踪关键技术研究
基金颁发部门:沈阳市科技局(1071121-2-00)
基金申请人:常桂然
项目名称:博士点基金资助项目
基金颁发部门:教育部(20030145017)
关键词
表情组合模板
欧氏距离
人脸表情识别
特征矢量
Expression Combination Template
Euclidean distance
Facial Expression Recognition
Feature Vector