基于Gabor特征和Adaboost的人脸表情识别

摘要

为了改菩人脸表情的识别率,提高分类器的性能,通过提取人脸表情图像的Gabor特征,再结合Adaboost算法,从而进行人脸表情的识别(facial expression recognition,FER).利用Gabor滤波器是人脸表情特征提取的一个重要手段,Adaboost算法则将一系列的弱分类器组合,最终生成一个强分类器.对表情识别这个多类识别问题,采取1:1的办法来解决,总共产生k(k-1)/2(k为总类别数)个强分类器,将多个强分类器进行级联实现人脸表情的多类分类.实验结果表明,相对于其他识别方法如MVBoost算法等,这种方法的识别准确率有很大的提高.%In order to improve the recognition rate of facial expression and enhance the performance of classifier,an approach is proposed to recognize facial expression using Gabor feature combined Adaboost algorithm.Gabor filter is one of the most important methods to extract features, weak classifiers would be constructed by Adaboost algorithm to generate a strong classifier.To solve the multi-class classification problem, we designed classifier by one-to-one mode,so the number of strong classifiers of Adaboost was k(k-1)/2 (k,number of categories).Finally, all strong classifiers were cascaded, Gabor features were feed into these classifiers and facial expression classification can be recognized.Experiment resuks showed that the recognition rate of Gabor plus Adaboost algorithm is significantly higher than that of other methods such as MVBoost algorithm.