摘要： 支持向量机可以处理2类问题，通过“一对一”和“一对多”方式能将2类支持向量机扩展为多类支持向量机。提出一种基于两类支持向量机融合的多类支持向量机构成方法。对分类器融合采用极大值法、极小值法、乘积法、均值法、中值法、投票法和各种决策模板融合方法。在日本女性表情数据库JAFFE上应用该方法进行人脸表情识别，结果证明了其有效性。

关键词: 多类支持向量机, 分类器融合, 决策模板, 人脸表情识别

Abstract: Support Vector Machine(SVM) can deal with binary class problem. By using “one against one” and “one against all” approach, binary class SVM can be expanded to multi-class SVM. A construction method for multi-class SVM based on binary class SVMs fusing is proposed. The classifier fusion approaches include Maximum, Minimum, Product, Mean, Median, Major Voting fusion methods and decision template fusion methods. This method is applied to the facial expression recognition for Japanese female expression database JAFFE and proved to be effective.

Key words: multi-class Support Vector Machine(SVM), classifier fusion, decision template, facial expression recognition

中图分类号: 

• TP391.4