| 基于判别式字典的正则化稀疏表示人脸识别算法 |
| |
| 引用本文： | 陆振宇,张铃华,何珏杉.基于判别式字典的正则化稀疏表示人脸识别算法[J].南京气象学院学报,2015,7(6):519-524. |
| |
| 作者姓名： | 陆振宇 张铃华 何珏杉 |
| |
| 作者单位： | 南京信息工程大学电子与信息学院, 南京, 210044;南京信息工程大学江苏省大气环境与装备技术协同创新中心, 南京, 210044;南京信息工程大学电子与信息学院, 南京, 210044;南京信息工程大学电子与信息学院, 南京, 210044 |
| |
| 基金项目： | 国家自然科学基金(61473334);江苏省高校优势学科建设工程项目 |
| |
| 摘 要： | 为了克服非约束性(光照、表情变化)条件下会大大降低人脸识别率的缺陷,提出一种基于Fisher判别准则的正则化稀疏表示人脸识别算法.首先将人脸图像经过Gabor滤波器滤波得到Gabor幅值图像,提取其统一化的局部二进制直方图,然后利用Fisher判别准则学习得到新的字典,最后通过正则化的稀疏表示判断测试图像所属类.利用AR数据库的数据进行实验的结果表明,与SRC、FDDL、RSC识别算法相比,本文算法在非约束性条件下具有最佳的识别率.
|
| 关 键 词： | 人脸识别 正则化的稀疏表示 统一化的局部二进制模式 Gabor滤波 学习字典 |
| 收稿时间： | 2015/9/7 0:00:00 |
Face recognition algorithm based on discriminative dictionary learning and regularized robust coding |
| |
| LU Zhenyu,ZHANG Linghua and HE Jueshan.Face recognition algorithm based on discriminative dictionary learning and regularized robust coding[J].Journal of Nanjing Institute of Meteorology,2015,7(6):519-524. |
| |
| Authors: | LU Zhenyu ZHANG Linghua and HE Jueshan |
| |
| Institution: | School of Electronic & Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044;Jiangsu Collaborative Innovation Center on Atmospheric Environment and Equipment, Nanjing University of Information Science & Technology, Nanjing 210044;School of Electronic & Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044;School of Electronic & Information Engineering, Nanjing University of Information Science & Technology, Nanjing 210044 |
| |
| Abstract: | To address the reduced face recognition accuracy in uncontrolled conditions such as the change of illumination,countenance or posture,a face recognition algorithm was proposed based on discriminative dictionary learning and regularized robust coding.Firstly,a face image is filtered by the Gabor filter to obtain the Gabor amplitude images,and the uniform local binary histogram is extracted.Then the Fisher criterion is used to gain a new discriminative dictionary,finally the regularized sparse representation is employed to test and classify the image.The experimental results based on AR face database show that the proposed algorithm has the highest face recognition rate in the existing uncontrolled environments,compared with algorithms such as Sparse Representation based Classifier,Fisher Discrimination Dictionary Learning,and Robust Sparse Coding for face recognition. |
| |
| Keywords: | face recognition regularized sparse representation uniform local binary pattern Gabor filtering dictionary learning |
|
| 点击此处可从《南京气象学院学报》浏览原始摘要信息 |
| 点击此处可从《南京气象学院学报》下载免费的PDF全文 |
|
