基于区域改进LBP的三维人脸识别  

3D face recognition method based on regional enhanced local binary pattern

作　　者：吕士文 达飞鹏[1,2] 邓星[1,2] 

机构地区：[1]东南大学自动化学院,南京210096 [2]东南大学复杂工程系统测量与控制教育部重点实验室,南京210096

出　　处：《东南大学学报（自然科学版）》2015年第4期678-682,共5页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(51175081;51475092;61405034);教育部博士点基金资助项目(20130092110027)

摘　　要：针对三维人脸识别中的表情问题,提出一种基于区域改进局部二值模式(LBP)的三维人脸识别算法.首先将预处理后的三维点云转化为深度图并进行归一化处理;然后根据表情对人脸的影响,利用二元掩膜提取人脸的刚性、半刚性和非刚性区域;对每个局部区域,计算其改进LBP特征,并用等价模式进行表征;最后使用稀疏表示分类器(SRC)对单个局部区域进行识别实验,并使用带权重的稀疏表示分类器(W-SRC)对刚性和半刚性区域进行决策级融合,给出最终识别结果.在FRGC v2.0人脸数据库上的实验结果表明,该方法具有较好的鲁棒性和较高的识别精度.Aiming at the problem of facial expression, a 3D face recognition method based on re- gional enhanced local binary pattern（LBP） is proposed. First, the depth image converted from the preprocessed 3D point clouds is normalized. Then, rigid region, semi-rigid region and non-rigid re- gion are extracted by binary masks according to the influence of different expressions on faces. The feature represented by the uniform pattern of enhanced LBP （eLBP） is calculated in every local re- gion. Finally, for single region, classification experiments with sparse representation classifier （SRC） are conducted. Score-level fusion with weighted SRC （W-SRC） for rigid region and semi- rigid region is also tested and compared. The experiments on FRGC v2.0 database demonstrate that the proposed method is robust and efficient.

关 键 词：三维人脸识别 改进LBP 深度图 稀疏表示分类器 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]