激光与光电子学进展, 2015, 52 (3): 031001, 网络出版: 2015-02-10
基于NSCT和仿生模式的人脸图像识别方法 下载: 536次
Face Image Recognition Method Based on the NSCT and Bionic Pattern
图像处理 非下采样Contourlet 变换 仿生模式识别 人脸识别 高维空间覆盖 image processing nonsubsampled Contourlet transform bionic pattern recognition face recognition high-dimensional space cover
摘要
针对现有的人脸图像识别算法准确度不高的问题,提出了一种基于非下采样Contourlet 变换(NSCT)和仿生模式的人脸图像识别的方法。对人脸图像进行NSCT 分解,并将分解后的各系数矩阵转化为能量特征,利用仿生模式识别算法实现对人脸图像的识别。使用UMSIT、Yale 和ORL 人脸库进行实验,且设计了无拒识和有拒识两组方案,实验结果表明:与传统方法相比,利用基于非下采样Contourlet 变换和仿生模式的人脸图像识别的方法能够获得更高的正确率,而有拒识的方案能够获得更好的综合性能。
Abstract
Concerning the low accuracy of classical face recognition algorithm,a face image recognition method based on the nonsubsampled Contourlet transform (NSCT) and bionic pattern is proposed. A face image is decomposed with NSCT and the decomposed coefficients are converted into energy features, then the face image is recognized by using the bionic pattern recognition algorithm. The UMIST,Yale and ORL face databases are used in the simulation experiment and two schemes- non- rejection and rejection are designed for the experiment. The experimental results show that compared with classical methods, the face image recognition method based on the NSCT and bionic pattern is superior to others, and the scheme of rejection shows a better comprehensive performance.
周亮基, 李庆武, 霍冠英, 朱国庆. 基于NSCT和仿生模式的人脸图像识别方法[J]. 激光与光电子学进展, 2015, 52(3): 031001. Zhou Liangji, Li Qingwu, Huo Guanying, Zhu Guoqing. Face Image Recognition Method Based on the NSCT and Bionic Pattern[J]. Laser & Optoelectronics Progress, 2015, 52(3): 031001.