摘要
提出了一种基于主分量分析(PCA)和上下截集模糊Kohonen聚类网络(UDSFKCN)的、无监督的、不同时相的和卫星影像的像素级变化检测新算法。将PCA和UDSFKCN两种方法结合,并将它应用于不同时相的卫星影像的变化检测。该方法结合每个像素的邻域信息,利用PCA,产生每个像素对应的基于邻域信息的特征向量;又将变化区域检测问题转化为两类间的分类问题;然后利用UDSFKCN对每个像素所对应的特征向量进行变化类与未变化类的聚类,得到像素级的变化区域的检测图。实验结果表明,与传统方法相比,对于高斯和斑点噪声,本文算法具有更高的检测准确性和抗噪性能。
An unsupervised change detection algorithm for multi-temporal satellite images based on principal component analysis(PCA) and up-down-set fuzzy Kohonen clustering network(UDSFKCN) is proposed in this paper.This method generated eigenvector corresponding to every pixel combining itself with its neighbours using principal component analysis.At the same time,the detection of the changed pixel in a region was converted into the classification between two groups,changed group and unchanged group.Since every pixel was described as one eigenvector,to obtain a changed map of the changed region at pixel level,up-down-set fuzzy Kohonen clustering network was applied to classify all the eigenvectors into changed ones and unchanged ones.Experimental results demonstrate that the proposed method has higher accuracy and stability than traditional algorithms against Gaussian and speckle noise.
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2011年第3期461-464,共4页
Journal of Optoelectronics·Laser
基金
科技部国际科技合作项目(2009DFA12870)
教育部促进与美大地区科研合作与高层次人才培养项目