| 利用纹理融合与广义高斯模型的高分辨率SAR影像变化检测 |
| |
| 引用本文： | 刘本强,赵争,盛玉婷,张忠芳.利用纹理融合与广义高斯模型的高分辨率SAR影像变化检测[J].测绘工程,2018(6):19-25. |
| |
| 作者姓名： | 刘本强 赵争 盛玉婷 张忠芳 |
| |
| 作者单位： | 山东科技大学 测绘科学与工程学院,山东 青岛266590;中国测绘科学研究院,北京100830
中国测绘科学研究院,北京,100830
北京航天世景信息技术有限公司,北京,100089 |
| |
| 基金项目： | 国家基础测绘科技计划(2018KJ0103)，国家重点研发计划(2017YFB0503004)，中国科学院创新交叉团队项目(2016Q1634)，中国测绘科学研究院基本科研业务费项目(7771808) |
| |
| 摘 要： | 为了充分利用高分辨率SAR影像的纹理特征,提出一种纹理信息融合与广义高斯模型相结合的SAR影像变化检测方法。通过灰度共生矩阵计算影像的纹理特征进而构造纹理差异影像,利用离散平稳小波变换,融合灰度差异影像和纹理差异影像。然后利用广义高斯模型进行统计建模,估计融合后差异影像上变化类和未变化类的概率分布,利用KI阈值准则获取最佳分割阈值,实现多时相SAR影像的非监督变化检测。选取两组TerraSAR-X数据进行实验,结果表明融合纹理信息与广义高斯模型的变化检测方法可行,其中融合逆差距纹理信息的检测性能最优。
|
| 关 键 词： | 合成孔径雷达 变化检测 纹理 小波变换 KI算法 广义高斯模型 synthetic aperture radar(SAR) change detection texture wavelet transform Kittler-Illingw orth(KI)algorithm generalized Gaussian model |
High resolution SAR image change detection based on texture fusion and generalized gaussian model |
| |
| LIU Benqiang,ZHAO Zheng,SHENG Yuting,ZHANG Zhongfang.High resolution SAR image change detection based on texture fusion and generalized gaussian model[J].Engineering of Surveying and Mapping,2018(6):19-25. |
| |
| Authors: | LIU Benqiang ZHAO Zheng SHENG Yuting ZHANG Zhongfang |
| |
| Abstract: | In order to make full use of high resolution SAR image texture feature,this paper proposes a method for SAR image change detection combined with texture information fusion and generalized Gaussian model.First,obtain image texture feature through calculating of gray level co-occurrence matrix and then construct texture difference image.Second,fuse gray difference image and texture difference image by discrete stationary wavelet transform.Finally,estimate the probability distribution of unchanged pixel and changed pixel in fusion difference image by generalized Gaussian model and achieve unsupervised change detection on multi-temporal SAR images by using KI threshold criterion to obtain the best segmentation threshold.This paper uses two images sets of TerraSAR-X to experiment and the result shows that it is feasible for change detection combined with texture information fusion and the generalized Gaussian model, which can get the best performance by fusing texture feature of inverse difference moment. |
| |
| Keywords: | |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
