基于改进局部方向模式特征的人脸识别算法  

FACE RECOGNITION ALGORITHM BASED ON IMPROVING LOCAL DIRECTIONAL PATTERN FEATURES

作　　者：樊利军[1] 魏昊[1] 田柏林[1] 

机构地区：[1]北京工业职业技术学院信息工程系,北京100042

出　　处：《计算机应用与软件》2015年第9期166-169,234,共5页Computer Applications and Software

基　　金：北京市属高等学校创新平台(PXM2014_014225_000020)

摘　　要：为了提高人脸识别率和效率,提出一种改进局部方向模式特征的人脸识别算法。首先将人脸图像分割成若干不重叠的子块,采用改进局部方向模式算法提取每个子块特征,然后对所有子块的特征进行连接,构成人脸图像的特征向量,最后采用最小二乘支持向机对人脸图像进行识别。在多个人脸库上进行仿真实验,结果表明,该算法获得了比传统算法更高的人脸识别率,而且加快了运行时间,较好地满足人脸识别实时性要求。In order to improve face recognition rate and efficiency, we proposed a novel face recognition algorithm which improves the fea- tures of local directional pattern. First, the algorithm divides the face image into non-overlapping sub-blocks, and extracts the feature of each block by improving local direction mode algorithm, and then connects the features of all sub-blocks to form feature vectors of face image, fi- nally, it uses least square support vector machine to recognise the face image. Simulation experiment is carried out on various face databases, results show that the proposed algorithm achieves higher face recognition rate than traditional algorithms, and speeds up the operation time as well, it well meets the real-time requirement of face recognition.

关 键 词：人脸识别 局部方向模式 特征提取 最小二乘支持向机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]