图像代数特征在面部表情识别中的应用  

Application of Image Algebraic Character in Facial Expression Recognition

作　　者：杨国亮[1] 王志良[1] 

机构地区：[1]北京科技大学信息工程学院

出　　处：《计算机工程》2006年第2期186-188,共3页Computer Engineering

基　　金：教育部重点科技基金资助项目(11139991)

摘　　要：探讨了图像代数特征在面部表情识别中的应用,首先对面部表情图像进行了分割,得到眼睛和嘴巴区域,然后分别对眼睛和嘴巴区域提取不变矩和奇异值特征向量,并进行Fisher线性判别分析,最后训练了支持向量机分类器。实验结果表明该方法取得了比较好的识别效果。The paper discusses facial expression recognition based on image algebraic characters. First, eyes and mouth are segmented from the facial expression image and variant moments and singular value feature vectors of eyes and mouth are extracted. Then Fisher linear discriminant analysis is used to find a set of optimal feature vectors. Finally, It trains a SVM classifier. The results show that the presented method has a higher accuracy and robust performance.

关 键 词：不变矩 奇异值分解 FISHER线性判别分析 支持向量机 面部表情识别 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]