摘要
针对传统的基于均匀采样提取面部表情特征点方法没有考虑面部不同区域对表情识别贡献大小的问题,提出了一种局部非均匀采样(LNUS)特征点和Gabor小波相结合的面部表情特征提取方法.该方法不仅提取了表情图像中局部关键特征点而且兼顾了整体信息,其利用主成分分析(PCA)和线性判别分析(LDA)方法进行特征降维,最后用支持向量机(SVM)进行表情识别.实验结果表明:所提方法不仅识别率更高,而且对光照和姿态变化鲁棒性强,能实时控制智能轮椅的运动.
Aiming at the problem that tradition method based on uniform sampling does not consider the contribution of different regions of the facial expression recognition when extracts the facial feature points,the facial expression characterized extraction method that combine the local non-uniform sampling(LNUS)feature points with Gabor wavelet was proposed.The local key points were extracted in the expression image and the overall information was considered in the method,and a two-stage method principal component analysis(PCA)and linear discriminant analysis(LDA)were used to extract feature for feature dimension reduction.Finally support vector machine(SVM)classifier was adapted to recognition expression.Experimental results demonstrate that this method not only has higher recognition rate compared with the conventional methods based on whole or partial sampling Gabor expression feature extraction,but also has strong robustness to illumination and postures change,and can real-time control the movement of the intelligent wheelchair.
出处
《华中科技大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2013年第S1期305-308,共4页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(51075240)
关键词
智能轮椅控制
面部表情识别
局部非均匀采样
GABOR小波
主成分分析
支持向量机
wheelchair control
facial expression recognition
local inhomogeneous sampling
Gabor wavelets
principal component analysis
support vector machine