一种基于核主成分分析的人脸图像去噪
首发时间:2010-01-20
摘要:本文讨论了一种基于核主成分分析的人脸图像去噪方法。首先利用核主成分分析,对训练样本做特征提取,舍弃在特征空间中投影方差小的特征,得到主分量。然后利用主分量按照以特征空间中重构误差最小为原则的降噪算法对模式进行重构,从而达到人脸图像降噪的目的。对算法的实现表明,在最优核参数的条件下,噪声的类型、噪声源的多少对降噪效果影响不大。从而证明了该方法的可操作性和有效性。
关键词: Kernel PCA 主成分分析 核函数
For information in English, please click here
A Method for Face Denoising Based on Analysis of KPCA Features
Abstract:In this paper, a face denoising method based on Analysis of Kernel PCA feature is discussed. First of all, the analysis of KPCA features is used to make the training sample featured extraction and then discards the characteristic which had small variance in the feature space. Second, the principle that bases the main component according to restructure the error to be smallest in the feature space of reducing noise restructures the pattern, thus it achieves the goal which reduced the noise of a imaging person face. According to realize the algorithm. It indicated that the effect of reducing noise in which the most superior nuclear parameters\
Keywords: Kernel Principal Component Analysis Principal Component Analysis Nucleus Function
基金:
论文图表:
引用
No.3924249221312639****
同行评议
共计0人参与
勘误表
一种基于核主成分分析的人脸图像去噪
评论
全部评论0/1000