提出了基于特征融合和模糊核判别分析(FKDA)的面部表情识别方法。首先，从每幅人脸图像中手工定

位34个基准点，作为面部表情图像的几何特征，同时采用Gabor小波变换方法对每幅表情图像进行变换，并提取基

准点处的Gabor小波系数值作为表情图像的Gabor特征；其次，利用典型相关分析技术对几何特征和Gabor特征进

行特征融合，作为表情识别的输人特征；然后，利用模糊核判别分析方法进一步提取表情的鉴别特征；最后，采用最

近邻分类器完成表情的分类识别。通过在JAFFE国际表情数据库和Ekman“面部表情图片”数据库上的实验，证实

了所提方法的有效性。-Proposed based on feature fusion and fuzzy kernel discriminant analysis (FKDA) facial expression recognition. First, face images of each piece of hand-set

Bit 34 basis points, as the geometric features of facial expression images, while using Gabor wavelet transform method to transform the images of each piece of expression, and extraction-based

Quasi-point of the Gabor wavelet coefficients, as Gabor features of facial expression image　second, using canonical correlation analysis on the geometric features and Gabor features into

Line feature fusion, as expression recognition of input features　then, using fuzzy kernel discriminant analysis method to extract and further identification features of expression　Finally, the most

Neighbor classifier to complete expression of the classification. International expression by JAFFE database and Ekman "facial image" database on the experiment, confirmed

The proposed method.