一种基于Gabor小波特征的人脸表情识别新方法  

New Approach for Facial Expression Recognition Based on Gabor Features

作　　者：罗飞[1] 王国胤[1,2] 杨勇[1,2] 

机构地区：[1]重庆邮电大学计算机科学与技术研究所,重庆400065 [2]西南交通大学信息科学与技术学院,成都610031

出　　处：《计算机科学》2009年第1期181-183,215,共4页Computer Science

基　　金：重庆市自然科学基金项目(CSTC2007BB2445);重庆市计算机网络与通信重点实验室开放基金项目“情感识别的关键技术研究”的资助

摘　　要：近来,表情识别成为人机交互研究的热点。将Gabor小波变换与2DPCA结合提出了一种表情识别的新方法。首先对静态灰度表情图片进行预处理,然后对其进行Gabor小波变换,通过2DPCA进行降维,根据Gabor不同尺度不同方向的变换结果训练不同的分类器,由校验集得到分类器权值,通过隶属度函数将各个分类结果模糊化,实现了分类器集成和表情特征数据的融合。实验证明了Gabor小波与2DPCA结合在表情识别中的有效性,以及基于Gabor小波模糊分类器集成的方法能够进一步提高识别率。In recent years, facial expression analysis is becoming a research hotspot in human-computer interaction. A novel facial expression recognition method was proposed based on the Gabor wavelet transformation and two-dimensional principal componentanalysis （2DPCA）. Features of expression images were extracted with Gabor wavelet and 2DPCA. The classifiers on different scales and orientations of Gabor wavelet were designed. The weight of each classifier was discussed. The classification result of each classifier was fuzzed with membership function and then integrated with ensemble method. Experiment results show the efficiency of combining Gabor wavelet and 2DPCA in facial expression recognition. The correct recognition rate is further improved with the proposed classification method.

关 键 词：表情识别 表情特征提取 GABOR小波 2DPCA 模糊分类 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP391.41[自动化与计算机技术—控制科学与工程]