局部空间自相关统计洪水灾害影像分析

Analysis of floods image by using local spatial autocorrelation statistics

针对洪水灾害分析在速度、准确性和及时性等方面的需求,该文在研究空间自相关分析的基础上,提出了一种基于局部自相关统计的洪水灾害影像分析方法。首先,对影像进行掩膜处理,去除云层干扰;其次,采用局部空间统计的方法对影像进行统计分析;然后,通过密度分割的方法提取水体,将影像分为水体和陆地两类;最后,将3幅影像分类的结果进行空间叠加分析,分析洪水灾害影响情况。以2013年嫩江流域3个时期的影像为实验数据,设计了仿真实验。实验结果表明,该方法可以较准确地对大面积洪水影响区域进行分析。

Floods disaster is a kind of common natural disasters and it has great destructive power for people＇s lives and property. The study of floods affected areas is significant for disaster damage assess- ment, reconstruction, etc. For the analysis of floods disaster, this paper proposed a method to analyze floods based on local autocorrelation statistics according to the study of spatial autocorrelation. First, masking the images and removing the interferences of clouds on the images. Next, using the local autocor- relation statistics to analyze three images. Then, extracting water by density segmentation method and di- viding every image into two bodies of water and land. Finally, overlaying the three classification results and analyzing the disaster situations of floods. An experiment was designed by using the three period images of Nenjiang basin in 2013. Experimental results showed that this method could analyze the large affected areas relatively and accurately.