疲劳驾驶面部表情识别算法  

Facial expression recognition algorithm for fatigue driving

作　　者：杨秋芬[1] 桂卫华[1] 周书仁[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083

出　　处：《计算机应用研究》2008年第10期3039-3041,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(50374079);国家"973"计划资助项目(2002cb12203)

摘　　要：针对疲劳驾驶的六种表情,提出几何规范化结合Gabor滤波提取表情特征,使用支持向量机对疲劳驾驶的面部表情分类识别的系统。首先对视频图像预处理进行几何规范化,利用二维Gabor核函数构造最优滤波器48个,获取48个面部表情特征点,最后利用支持向量机进行面部表情分类识别。实验结果表明径向基函数的SVM性能最好。This paper proposed an expression recognition system based on geometry standardization, Gabor wavelet filter and support vector machine （SVM） according to six face expression of fatigue driving. The system ensured the geometry standardization of video. Then it constituted forty-eight optimization filters according to 2D-Gabor kernel functions to acquire forty-eight expressional feature point inside the latency rectangle region. It was applied to recognizing facial expression using SVM. Experimental results demonstrate that RBF the radial basis function（RBF） SVM has better performance than other SVMs on human facial expression recognition, and prove that the algorithm is efficient and feasible.

关 键 词：表情识别 加博滤波 核函数 支持向量机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]