摘要
针对主成分分析(PCA)算法中存在不能提取非线性特征的问题,提出了利用KPCA提取图像特征,最近邻法分类的人脸识别改进方法。基于ORL数据库的相关实验表明,这样的系统能够取得比传统PCA更好的识别性能。
To solve the problems that Principal Component Analysis (PCA) can not extract nonlinear character, this paper presents an image recognition method. This method makes use of Kernel-PCA to extract the feature information of images and the NN( Nearest Neighbor) is selected to recognize the character. The experiment based on ORL-Database showed that this method could get better performance than traditional PCA methods.
出处
《河北科技师范学院学报》
CAS
2008年第3期45-48,62,共5页
Journal of Hebei Normal University of Science & Technology
基金
宁夏大学自然科学基金资助项目(项目编号:ZR200716)