摘要
脸部表情的识别分类是一个非常复杂的问题,采用传统的方法很难取得满意的结果。为此,通过Gabor滤波器对人脸部图像进行滤波,提取滤波后图像的统计信息作为表情识别的特征信息,采用多分类器集成的方法对得到的神经网络输出向量进行线性加权集成得到最终的识别结果。实验结果表明了该方法的正确性。
Facial expression recognition is a very complicated topic. The traditional way of recognition can not meet current demands. In this paper, we proposed Gabor filters to process the face image. We used statistic data of the filtered image as the information for the facial expression classification. Finally, we adopted the method of multi-classifier integration to get the output vector. Experiments results have shown good performance in reality.
出处
《长沙交通学院学报》
2005年第2期70-74,共5页
Journal of Changsha Communications University
基金
湖南省自然科学基金资助项目(99JJY20058)
