人脸表情的LBP特征分析

为了有效提取面部表情特征，提出了一种新的基于LBP（局部二值模式）特征的人脸表情识别特征提取方法。首先用均值方差法对表情图像进行灰度规一化，通过对图像进行积分投影，定位出眉毛、眼睛、鼻和嘴巴这些关键特征点，进而划分出各特征部件所在子区域，然后对子区域进行分块，提取各个子区域的分块LBP直方图特征。为了验证所提出的方法的合理性，最后在JAFFE表情库上进行了实验，结果表明提出的方法能够有效地描述表情的特征。

Facial expression analysis using LBP features

In order to effectively extract facial expression feature,a novel facial feature extraction approach for facial expression recognition based on Local Binary Pattern（LBP） is proposed in the paper.Firstlyf,acial expression images＇gray level is normalized with the average-variance method.By doing integral projection,some critical facial feature points are located,such as eyebrow,eye,nose and mouth.Then sub-regions belong to each facial component are partitioned.And then facial expression features are presented with LBP histograms of each sub-region,which is divided into several blocks.To validate the rationality of the method proposed,experiments are implemented on JAFEE（Japanese female facial expression database） database.The results illustrate that the method proposed is effective to represent facial expression feature.