利用遥感影像偏移追踪的地震滑坡监测  

Monitoring Earthquake-triggered Landslide Using Optical Image Offset-tracking Algorithm

作　　者：郭晓丹[1] 查显杰[1] 黄金水[1] 

机构地区：[1]中国科学技术大学地球和空间科学学院,合肥230026

出　　处：《遥感信息》2016年第3期56-60,共5页Remote Sensing Information

基　　金：教育部留学回国人员科研启动基金(WF2080000021);国家自然科学基金(91014005;40774045;41374037)

摘　　要：针对现有的滑坡监测方法往往难以准确、及时地获取地表位移信息这一问题,通过对传统时域互相关算法的改进,提出了一种基于光学遥感影像子像素偏移追踪的地震滑坡监测方法。利用时域互相关算法只能获取整像素偏移,而双线性插值和高斯回归算法则可以获取子像素偏移。该文将时域互相关、双线性插值和高斯回归三种算法进行有机结合,可以有效提高偏移估算精度。理论测试结果显示:该方法可以从光学遥感影像对中获取地表位移信息,理论精度基本达到千分之几个像素水平。本文还利用该方法获取了芦山地震滑坡位移场,结果与唐川老师等得到的滑坡分布具有很好的一致性。Earthquake-triggered landslide is a serious natural disaster. The existing methods form onitoring landslide areoften difficult to obtain the surface displacem entinformation accurately and timely. This paper proposed a method form onitoring the earthquake-triggered landslide from optical remote sensing images by improving the traditional cross-correlationin time-dom ain. Using the cross-correlation in time-domain can only get the offset of pixel level, whereas using the bilinear interpolation and Gaussian regression algorithms could get the offset of sub-pixel level. By com bining the cross-correlation intim e-domain, bilinear interpolation and Gaussian regression algorithms, we can effectively im prove the offset estim ationaccuracy. Theoretical test results show th at the proposed method can acquire surface displacement information from the opticalremote sensing image pairs, and the theoretical accuracy can reach a few thousandths of a pixel. This article also used thepresented method to acquire the displacem ent field of Lushan earthquake-triggered landslide, which is consistent with thelandslide distribution by Prof. Tang Chuan.

关 键 词：地震滑坡 地表位移 光学遥感影像 子像素偏移追踪 滑坡分布 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]