摘要
为提高痛苦表情识别的准确率,提出一种基于监督保局投影(SLPP)与多核线性混合支持向量机(MKLMSVM)的识别方法。引入先验类标签信息的SLPP获取痛苦表情特征,以解决保局投影方法在未使用先验类标签信息的情况下忽略类内局部结构的问题,并采用MKLMSVM实现痛苦表情的分类。实验结果表明,该方法的识别准确率可达88.56%,明显优于主动外观模型方法,与一般的支持向量机分类相比,可以提升决策函数的可解释性及分类性能。
In order to improve the accuracy rate of pain expression recognition, a method is proposed based on Supervised Locality Preserving Projections(SLPP) and Multiple Kernel Linear Mixture Support Vector Machines(MKLMSVM). The SLPP using prior class label information is adopted for extracting feature of pain expression, which can solve the problem that LPP ignores the within-class local structure without the use of the prior class label information, and then MKLMSVM is employed for recognizing pain expression. Experimental results demonstrate that the accuracy of the proposed approach can reach 88.56%, and is significantly better than the Active Appearance Models(AAM), compared with normal Support Vector Machine(SVM), which can improve the interpretability of decision function and classifier performance.
出处
《计算机工程》
CAS
CSCD
2013年第12期196-199,共4页
Computer Engineering
基金
国家博士点基金资助项目(20090162110057)
湖南省科技计划基金资助项目(2011GK3213)
关键词
痛苦表情识别
监督保局投影
先验类标签
多核支持向量机
多核线性混合
主动外观模型
pain expression recognition
Supervised Locality Preserving Projections(SLPP)
prior class label
Multiple Kernel SupportVector Machines(MKSVM)
multiple kernel linear mixture: Active Appearance Models(AAM)