欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
主成份分析（PC A ）算法是特征提取的重要方法之一,由于其本身没有提供更多的分类信息,直接在其上进行识别效果往往并不理想.为了提取PCA特征值中有利于识别的特征信息,提出一种带权稀疏PCA算法.它利用基本PCA算法实现去噪功能,利用Lagrange乘子方法求得使PC A特征空间中类内距离最小,类间距离最大的一组权值,并利用稀疏PC A （S PC A ）算法解决维数约简和保留小特征值对应的特征向量所含的分类信息.在公开人脸数据库上对该算法进行测试,实验结果表明该算法不仅运行速度快,而且有较高的正确识别率.
英文摘要：
The principal component analysis（PCA） is one of the important methods for feature extraction , but it can＇t provided more classification information by itself .In order to pick up feature information in favor of recognition from PCA eigenvector ,a weight sparse principal component analysis is proposed in the paper .It achieves image de-noising function by using primitive PCA algorithm ,acquires the group of weight values which are able to maximize within-class distance and minimize between-class distance in PCA feature space by utilizing Lagrange multiplier ,and finishes dimension reduction by using sparse PCA （SPCA ） to retain effectively some classification information of eigenvectors with little eigenvalue . In the end ,the proposed algorithm is tested on an all-known public face database .The experiment results indicate the proposed algorithm has not only faster running speed but also better rate of recognition .
同期刊论文项目
同项目期刊论文
Stability analysis for the generalized Hopfield neural networks with multi-level activation function
期刊信息
位置：