摘要
针对视频序列中的人脸识别问题,提出一种基于局部二值模式(LBP)和多类组LASSO算法的人脸识别方法。首先,将脸部区域分成若干个块区域,对每个块区域计算其超完备局部二值模式直方图。然后,对传统最小绝对收缩和选择算子(LASSO)进行改进,形成多类组LASSO算法,使其能够从LBP直方图中选择出一个能够同时辨别所有类的稀疏表示的特征组。最后,通过支持向量机(SVM)进行人脸识别。实验结果表明,提出的方法能够对无约束性视频序列中人脸进行准确识别。
For the issues of the face recognition problem in video sequences, a face recognition scheme based on local binary pattem （LBP） and multi-class group LASSO algorithm is proposed. Firstly, the face region is divided into several regions, and each region is calculated by its super complete local binary pattern histogram. Secondly, the traditional least absolute shrinkage and selection oper- ator （LASSO） is improved to form a multi-class group LASSO algorithm, so that it can select feature group which can identify all classes from the LBP histogram. Finally, face recognition is performed by support vector machine （SVM）. Experimental results show that the proposed scheme can accurately identify the human face in an unconstrained video sequence.
出处
《微型电脑应用》
2016年第9期15-17,80,共4页
Microcomputer Applications
基金
新疆维吾尔自治区高校科研计划青年教师科研启动基金项目(No.XJEDU2014S074)
关键词
人脸识别
多类组LASSO
局部二值模式
稀疏表示
Face Recognition
Multi-class Group Lasso
Local Binary Pattern
Sparse Representation
