摘要
论文将Curvelet变换用于人脸表情识别,提出了一种基于Curvelet变换与SVM相结合的人脸表情识别方法。在表情特征提取过程中,还采用了主分量分析方法对Curvelet变换分解后得到的系数特征进行降维处理。分别对JAFFE和Cohn-Kanade表情数据库进行了实验,结果表明该方法可以有效地对人脸表情进行识别,与其他方法比较,采用该文方法得到人脸表情的平均识别率明显更优。
In this paper, curvelet transform is used for facial expression recognition. A method based on curvelet transform and SVM is introduced to facial expression recognition. During expression feature extracting, principal component analysis is also used ',o reduce the di- mension of coefficient features after eurvelet transform decomposition. Conducting experiments on JAFFE and Cohn-Kanade expression data- base respectively, the results show that the method can effectively identify the facial expression. Compared with other methods, the proposed method that gets an average recognition rate of facial expression is significantly better.
出处
《计算机与数字工程》
2013年第7期1165-1168,共4页
Computer & Digital Engineering
基金
云南师范大学青年科学基金项目(编号:201205)资助