【摘要】 对人脸表情图像进行分割得到眉区、眼区和嘴部区域,再对分割出来的表情区域利用高维局部自相关(HLAC)计算特征并得到加权的特征向量,其中加权系数根据心理学中的FACS表情测量理论选取,最后利用近邻中心距离分类器进行表情识别。实验基于CMU-PITTSBURGH表情图像库,在没有增大计算量的前提下相比PCA方法,特征融合(HLAC+WPCA)的方法显著地提高了表情的识别率。更多还原

【Abstract】 The facial expression image were segmented to form the eyebrows,eyes and mouth areas,and these areas were computed with Higher-order Local Auto-Correlations method,through which the Weighted Principal Component Analysis values of these areas were obtained according to facial expression measure system Face Action Coding System(FACS) in psychology.In classification part,minimum-distance classifier was used to recognize different expressions.Based on the CMU-PITTSBURGH AU-Coded Face Expression Image Database,the results showed that the features fusing method was superior to PCA-based method.更多还原