摘要
中国青海省玉树自治州结古镇附近于2014年4月14日7时左右发生了Ms 7.1级地震。利用灾前灾后两个不同时期的高分辨率影像,以结古镇南部地区为例,探讨基于e Cognition平台,选用多尺度分割和最邻近法,对灾前灾后遥感影像进行各类地物信息提取,分析研究区内各地物的动态变化。研究表明:通过面向对象分类的e Cognition软件提取的各地物信息颜色信息丰富,形状大小特征明显,总体分类精度达到90%以上,分类效果良好。并根据提取成果进行统计分析,快速提供受灾地区地震前后整体动态变化情况,为各级政府部门迅速开展地质灾害抢险救援与监测提供支撑。
An Ms 7.1 strong earthquake occurred in Jiegu of Yushu autonomous prefecture, on April 14, 2014, 7am. This paper, take the southern part of the Jiegu town as an example, explores the dynamic changes of land with high resolution remote sensing image by using the method of multistate segmentation and nearest neighbor based on eCognition. The results show the object - oriented classi- fication of the eCognition software to extract the information of land are rich information of color, obvious information of shape and size, the accuracy of the overall classification more than 90%. At same time, with statistical analysis of the extraction results, quickly providing the dynamic changes of the affected area before and after the earthquake, and providing the important basis for geological disaster rescue and monitoring.
出处
《测绘与空间地理信息》
2016年第2期31-35,共5页
Geomatics & Spatial Information Technology
基金
四川省应急测绘与防灾减灾工程技术研究中心开放基金项目(K2015B001)资助
