摘要
提出一种基于特征点LBP信息的表情识别方法。在分析了表情识别中的LBP特征之后,选择含有丰富表情信息的上半脸眼部周围和下半脸嘴部周围的特征点,计算每个特征点邻域的LBP信息作为表情特征进行表情识别。实验表明,基于特征点LBP信息的方法不需要对人脸进行预配准,较传统的LBP特征更有利于表情识别的实现。
An facial expression recognition method is proposed based on the Local Binary Pattern (LBP) of feature-points.First, the LBP feature in facial expression recognition is presented.Then the feature-points around the eyes of upper face and the mouth of lower face is fixed which hold rich expression information.And the LBP map of the neighbor field of each feature point is computed as expression feature for facial expression recognition.Experimental results show that,the face normalization is not necessary by using the proposed method,which can improve the facial expression recognition.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第31期138-139,150,共3页
Computer Engineering and Applications
基金
国家自然科学基金No.60873163~~
关键词
特征提取
局部二值模式
局部二值模式直方图序列
表情识别
feature extraction
local binary pattern(LBP)
histogram sequence of LBP
facial expression recognition