基于局部区域LBP特征的人脸识别研究

传统的LBP方法往往分散特征区域所表达的信息，提出一种基于局部区域LBP特征（Local Binary Pattern，局部二值模式）提取的人脸识别方法。用积分投影法在表情图像上定位出眉毛、眼睛、鼻子和嘴巴这些特征点的位置，根据这些特征点的位置确定这些特征部件所在子区域，然后对这些子区域进行不同的分块，提取各子区域的局部LBP直方图，按顺序把它们连结成一个特征直方图序列。实验结果表明：局部区域LBP方法与传统的LBP方法相比有很强的人脸表情特征识别能力，能够更准确地提取面部局部区域的表情信息。

Human face recognition based on the local LBP features

The traditional LBP method often disperses the facial expression information of feature area. A method of face recognition is proposed,which is based on LBP features of local area. We can determine the position of the feature points such as eyebrows, eyes, nose and mouth by using integral projection method and determine the facial points＇ region according to these feature points; then divide the region into blocks, extract each block＇s LBP histogram,and construct histogram feature sequence in good order. The experimental results show that the LBP is better than the classical LBP in the facial expression recognition and it extracts the facial expression information in local area more accurately.