Abstract: In the paper，a new approach based on AAM（Active Appearance Model） and RS-SVM（Rough Set theory based attribution reduction and Support Vector Machine） is proposed for face recognition.Firstly，an AAM which based on statistical theory is implemented for facial feature points location.In order to select the necessary features for face recognition，the attribution reduction in rough set theory is used which can effectively reduce the dimensions of features.Secondly，SVM is adopted for classification.Finally，the experiment results show that the algorithm reduces the computing cost of SVM with no difference in classification ability.

Key words: face recognition, Active Appearance Model（AAM）, rough set theory, Support Vector Machine（SVM）, Rough Set theory based attribution reduction and Support Vertor Machine（RS-SVM）

摘要： 提出了一种基于AAM模型和RS-SVM的人脸识别算法。首先，使用一种基于统计学定位的图像定位方法—主动外观模型（AAM），将其应用到人脸特征定位。为了从所有提取的特征中选择出与人脸识别相关的、必要的特征，使用了粗糙集理论（Rough Set）的属性约简算法进行特征选择，有效降低特征维数。然后用支持向量机（SVM）进行分类。实验证明，该方法在不影响识别率的情况下，可以有效降低SVM的运算复杂度。

关键词: 人脸识别, 主动外观模型, 粗糙集理论, 支持向量机, 粗糙集-支持向量机（RS-SVM）