摘要
针对人脸图像的非线性特点,将基于核方法的核独立分量分析算法用于提取人脸图像特征.为避免多类支持向量机出现不可识别域,提出基于二叉树思想的层次支持向量机算法,用于多类人脸识别.将层次支持向量机和核独立分量分析算法相结合进行人脸识别,首先对人脸图像进行预处理和主成分分析法降维;然后运用核独立分量分析算法估算出独立基影像,从而得到人脸特征;最后将人脸特征输入层次支持向量机进行分类识别.在ORL人脸库上的仿真结果表明该算法较好地兼顾了识别率和运行速率.
Considering the nonlinear characteristic of face images, kernel independent component analysis (KICA) based on kernel method was applied in face feature extraction. To avoid non-recognition areas, Hierarchical Support Vector Machine was proposed based on binary tree to deal with multi-class classification. The algorithm of face recognition based on Hierarchical SVM and KICA was presented in this paper. Firstly, the image pre-processing was needed and the dimension of image was reduced by principal component analysis (PCA). Secondly, the independent basis was estimated through KICA algorithm. Then the features of face images can be obtained by mapping. Finally, face features were put into a support vector machine for categorization identification. The results of simulations in ORL face database show that this method is efficient in terms of recognition rate and running speed.
作者
曹未丰
CAO Wei-feng(School of Automation,Nanjing Institute of Technology,Nanjing 211167,China)
出处
《南京工程学院学报（自然科学版）》
2019年第3期76-79,共4页
Journal of Nanjing Institute of Technology(Natural Science Edition)
关键词
人脸识别
支持向量机
核独立分量分析
主分量分析
face recognition
support vector machine
kernel independent component analysis
principal component analysis
