摘要
针对人脸表情时空域特征信息的有效提取问题,提出一种多频域LBP-TOP与最大间隔球形支持向量机相结合的人脸表情识别算法。采用小波分解增强序列各帧的图像信息,对同频率的子图像序列提取分块改进的LBP-TOP特征,采用最大间隔球形支持向量机进行样本的训练及识别。实验结果证明,该方法能有效提取运动的表情特征,识别率高,同时符合实时性要求。
According to the problem of effective extraction of facial expression information in space-time domain,one kind facial recognition method based on multi-frequency Local Binary Patterns from Three Orthogonal Panels(LBP-TOP) features and Maximal-margin Sphericalstructured Support Vector Machine(MSSVM) is proposed.It adapts wavelet decomposition to enhance information of each frame in image sequence.It extracts improved LBP-TOP features of sub-images on the same frequency.MSSVM is applied for sample training and recognition.Experimental result indicates that,this method can extract movement expression feature more effectively,as well as recognition rate is better,and it meets the requirement of real-time.
出处
《计算机工程》
CAS
CSCD
北大核心
2010年第15期176-178,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60673190)
关键词
模式识别
人脸表情识别
小波分解
二元局部模式
最大间隔球形支持向量机
pattern recognition
facial expression recognition
wavelet decomposition
local binary pattern
Maximal-margin Spherical-structured Support Vector Machine(MSSVM)