| 引用本文: | 高欧阳， 牛朝阳， 刘伟， 张浩波， 邹玮琦， 张雅歌. TOPSIS融合极化信息的复杂背景下滑坡检测方法[J]. 雷达科学与技术, 2022, 20(6): 643-650.[点击复制] |
| GAO Ouyang, NIU Chaoyang, LIU Wei, ZHANG Haobo, ZOU Weiqi, ZHANG Yage. A Method for Detecting Landslides in Complex Backgrounds Using TOPSIS to Fuse Polarization Information[J]. Radar Science and Technology, 2022, 20(6): 643-650.[点击复制] |
|
| 摘要: |
| 如何在具有多种地物类型的复杂背景下准确地识别滑坡区域，同时较好地抑制其他地物造成的虚警，是基于极化合成孔径雷达（PolSAR）数据进行滑坡检测的关键问题。针对此问题，提出一种利用优劣解距离法（TOPSIS）融合极化信息的滑坡检测方法。首先通过Yamaguchi分解、H/A/α分解和相关系数计算得到8个极化特征参数，然后对这些参数进行归一化和同趋势化处理，同时利用层次分析法（AHP）优化设计TOPSIS的权重，在此基础上精确计算每个像素与滑坡的相似度，经阈值分割和形态学处理得到疑似滑坡区域；进一步采用滑坡前、后单极化数据的相干图来剔除疑似滑坡区域中的虚警，有效实现精准检测复杂背景下滑坡区域的目的。实验以深圳“12·20”滑坡和日本北海道“9·6”滑坡为例，结果表明，与现有方法相比，本文方法在有效抑制虚警率的情况下将滑坡检测正确率提升了10%以上。 |
| 关键词: 极化合成孔径雷达 滑坡检测 优劣解距离法 散射特性 极化特征参数 |
| DOI：DOI:10.3969/j.issn.1672-2337.2022.06.007 |
| 分类号:TN957.52 |
| 基金项目:国家自然科学基金（No.61302160， 41301481） |
|
| A Method for Detecting Landslides in Complex Backgrounds Using TOPSIS to Fuse Polarization Information |
|
GAO Ouyang, NIU Chaoyang, LIU Wei, ZHANG Haobo, ZOU Weiqi, ZHANG Yage
|
|
School of Data and Target Engineering, PLA Strategic Support Force Information Engineering University, Zhengzhou 450001, China
|
| Abstract: |
| How to accurately identify the landslide area under a complex background with multiple ground features while better suppressing false alarms caused by other ground features is a vital issue for landslide detection based on PolSAR data. Aiming at this problem, a landslide detection method using TOPSIS to fuse polarization information is proposed. Firstly, eight characteristic polarization parameters are obtained through Yamaguchi decomposition, H/A/α decomposition, and correlation coefficient calculation. Then normalization and homo-trend processing are performed on these parameters. At the same time, AHP is used to optimize the weight of TOPSIS. On this basis, the similarity between each pixel and the landslide is accurately calculated, and the suspected landslide area is obtained by threshold segmentation and morphological processing. The coherence map of the single-polarization data before and after the landslide is further used to eliminate the false alarms in the suspected landslide area and effectively realize the purpose of accurately detecting the landslide area with complex background. The experiment takes Shenzhen “12·20” landslide and Hokkaido “9·6” landslide as examples. The results show that this method can effectively suppress the false alarm rate and improve the accuracy of landslide detection by more than 10% compared with the existing methods. |
| Key words: polarimetric synthetic aperture radar landslide detection technique for order preference by similarity to ideal solution scattering characteristics polarization characteristic parameter |
