基于局部拉普拉斯算子的灾后建筑物损毁检测

Damage Buildings Detection of Post-disaster Based on Local Laplacian Operators

灾后建筑物损毁检测有助于灾后的迅速救助,因此,开展对高分辨率遥感影像灾后建筑物损毁检测的研究具有重要意义。本文提出了一种基于局部拉普拉斯算子的高分辨率遥感影像灾后建筑物损毁检测的方法。首先,利用局部拉普拉斯算子将灾后高分辨率遥感影像分为平滑表面类和粗糙表面类,然后,根据灾害前建筑物数字线划图提取灾害后未损毁建筑物得到检测结果。为验证本文方法的有效性,采用两组纽约市2012年飓风后的高分辨率遥感影像进行实验。结果表明,本文提出的方法能取得超过92%的总体精度。

The detection of damaged buildings in disaster area is helpful to the rapid relief and reaction of the disaster. Therefore,it is of great significance to study the damage detection of high-resolution remote sensing images. In this paper,a method of detecting high-resolution remote sensing images of damaged buildings based on local Laplacian is proposed. Firstly,the local Laplacian operator is used to divide the post-disaster high-resolution remote sensing image into smooth surface and rough surface. Then,the damage is obtained after the disaster is extracted from the building＇s digital line drawing. In order to verify the effectiveness of this method,two high-resolution remote sensing images of New York City after 2012 hurricane were used. There sults show that the proposed method can achieve more than 92% over all accuracy.