摘要
Gabor特征可以用来描述图像纹理信息,在表情识别中Gabor小波变换具有很好的识别性但是存在着冗余度高和特征维数高的问题。对于此文中给出了一种基于人脸T型区域Gabor小波变换的表情识别方法。该方法首先提取经过预处理之后的表情图像的Gabor特征,然后抽取人脸T型区域的Gabor特征,使用主成分分析法(Principal Component Analysis,PCA)进行降维。实验在JAFFE表情库中进行,在实验中通过更改构建Gabor滤波器时的各项参数来提取不同的人脸表情特征,改变了多种降维维度和采用了不同的划分方式和比例对测试集训练集进行划分,处理后得到的特征向量在支持向量机(Support Vector Machine,SVM)的不同核函数下进行分类。实验结果证明了该方法具有很好的鲁棒性和较好的识别性能。
Gabor features can be used to describe image texture information.Gabor wavelet transform has good recognition performance in facial expression recognition,but it has the problems of high redundancy and high feature dimension.in this paper,an expression recognition method based on face T-zone Gabor wavelet transform is presented.First,extract the Gabor feature of the pre-processed expression image,and then extract the T-zone face Gabor feature and uses Principal Component Analysis(PCA)for dimensionality reduction.The experiment was performed in the JAFFE expression library.In the experiment,different facial expression features were extracted by changing the parameters of the Gabor filter and various dimension reduction dimensions were selected.Besides,the test set and the training set are divided by different division methods and proportions.After that the feature vectors obtained after processing are classified by different kernel functions of Support Vector Machine(SVM).The experimental results show that the proposed method has good robustness and good recognition performance.
作者
段晓珊
王坤侠
DUAN Xiaoshan;WANG Kunxia(School of Electronics and Information Engineering,Anhui Jianzhu University,Hefei 230601,China)
出处
《安徽建筑大学学报》
2019年第6期102-107,共6页
Journal of Anhui Jianzhu University
基金
佛山市科技创新团队项目(可穿戴设备研发创新团队)(2015IT100095S)
安徽省自然科学基金(1708085MF167)。
作者简介
段晓珊(1995-),女,硕士研究生,研究方向为情感计算、机器学习。
