中文摘要：

提出了一种结合Gabor变换与神经网络树的面部表情自动分类方法。通过调整Gabor滤波器参数可以建立低信息冗余的Gabor表情特征。与线性判别分析方法相比，本文提出的方法在表情分类应用中更加稳定有效。实验结果表明Gabor特征提取位置的对应性在面部表情自动分类准确性上具有关键作用。

英文摘要：

A facial expression automation classification method is presented by the combined Gabor and neural network tree. The Gabor filter with adjusted parameters achieves lower redundancy. The method is more effective and steady compared with linear discriminate analysis method. Experimental results show that the position correspondence of extracted Gabor feature is crucial for automatic facial expression classification.