测绘学报

• 学术论文 •    

基于非局部均值加权的动态模糊Fisher分类器的遥感图像变化检测

辛芳芳1,焦李成2,王桂婷2   

  1. 1. 西安微电子技术研究所 研发部
    2. 西安电子科技大学
  • 收稿日期:2011-04-12 修回日期:2011-08-16 出版日期:2012-08-25 发布日期:2012-08-25
  • 通讯作者: 辛芳芳

Change Detection in Multitemporal Remote Sensing Images based on Dynamic Fuzzy Fisher Classifier and Non Local Mean Weighted Method

  • Received:2011-04-12 Revised:2011-08-16 Online:2012-08-25 Published:2012-08-25

摘要: 提出一种新的变化检测算法,利用改进的动态模糊Fisher分类器,通过对多时相图像的联合直方图进行分类得到变化区域。在此基础上,根据图像空间关系对待检测点进行非局部均值加权,并以一定比例选取可靠性高的数据先进行标类,增加了数据的可分性和算法的可靠性。根据更新后的样本动态调整待检测点权重及分类器参数,直到所有点判别完毕为止。本算法不受参数模型限制,不受差异算子影响并充分利用了图像的空间与时间信息。真实遥感数据结果表明本算法提高了检测精度。

Abstract: A novel change detection approach based on the dynamic fuzzy Fisher classifier for multitemporal remote sensing images is proposed in this paper, which detects the changes with the joint histogram. To increase the separability of the unlabeled pixels, a non-local mean weighted method is used to introduce the spatial information. The unlabeled pixels are labeled with a predefined probability based on their predictive values. The weights of the unlabeled pixels and the parameters of the dynamic classifier are adjusted according to the updated samples until all the pixels are classified. The proposed method is distribution free, context-sensitive and not affected by the comparison operators. Experimental results on real multitemporal remote sensing images confirm the effectiveness of the proposed technique.