关键词: 加权主成分分析（WPCA）, 粗糙集, 特征选择, 表情分类

Abstract: In view of the question that WPCA method emphasizes information insufficiently and the characteristic dimension extracted excessively high，rough set attribute reduction algorithm with updated WPCA applied in expression features selection is advanced.The weighting principal components analysis’s principle is used.The characteristic weighted sum principal components analysis is unified.A new bidirectional three center Gaussian distribution function is constructed as the weighting function.The image characteristics of each dimension are weighted in order to get characteristic vector，and then the improved rough set properties reduction algorithm is used to filter the obtained feature vector to remove redundant information.Experimental results show that this method is effective.

Key words: Weighted Principal Component Analysis（WPCA）, rough set, features selection, expression classification