基于LDP和LTP模型的人脸识别技术探究  

Research on Face Recognition Based on LBP and LTP Model

作　　者：姜洪溪[1] 

机构地区：[1]湖北文理学院数学与计算机科学学院,湖北襄阳441053

出　　处：《科技通报》2015年第3期192-197,共6页Bulletin of Science and Technology

基　　金：国家自然科学基金青年项目(编号61201247)

摘　　要：表面的纹理特征常被用于目标的识别。这种特征被广泛的应用于各个领域,比如基于纹理特征的人脸识别系统、监视、身份识别等。局部二值模式纹理特征方法(LBP)是最成功的基于纹理特征的人脸识别方法。由于LBP的巨大成功,出现了许多LBP的变异方法,比如多层LBP、中心对称LBP、主LBP等。本文的主要目的在于研究各种LBP的变异方法在人脸识别中的有用性。同时本文还评估了各种实际问题,比如光照变化、旋转、人脸图像表示的变化等在不同LBP人脸识别方法中的影响。基于实际人脸图像数据库的验证表明局部纹理模式(LTP)和局部衍生模式(LDP)具有更好的识别性能。Texture is the surface property that is used to identify and recognize objects. This property iswidely used in many applications including texture- based face recognition systems, surveillance,identity verification and so on. The Local binary pattern （LBP） texture method is most successful for facerecognition. Owing to the great success of LBP, recently many models, which are variants of LBP havebeen proposed for texture analysis, such as multivariate local binary pattern, centre symmetric localbinary pattern, dominant local binary pattern, advanced local binary pattern, local texture pattern （LTP）and local derivative pattern （LDP）. In this scenario, it is essential to review, whether LBP or theirderivatives perform better for face recognition. The real-time challenges such as illumination changes,rotations, angle variations and facial expression variations are evaluated by different LBP-based models.Experiments were conducted on face databases. The results show that LDP and LTP perform much betterthan the other LBP-based models.

关 键 词：人脸识别 局部二值模式 局部纹理模式 局部衍生模式 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]