|
|||||
|
|
| 基于最优特征空间的震后损毁建筑物信息提取研究 | |
| 引用本文： | 陈晋,陈文凯,窦爱霞,李雯,孙艳萍.基于最优特征空间的震后损毁建筑物信息提取研究[J].地震,2019,39(3):71-83. |
| 作者姓名： | 陈晋 陈文凯 窦爱霞 李雯 孙艳萍 |
| 作者单位： | 1.中国地震局兰州地震研究所, 甘肃 兰州 730000; 2.中国地震局地震预测研究所, 北京 100036 |
| 基金项目： | “十三五”国家重点研发计划项目(2017YFB0504104); 甘肃省科技支撑项目(1504FKCA065); 中国地震局地震预测研究所基本科研业务专项(2015IESLZ06) |
| 摘 要： | 基于传统面向对象方法, 提出了一种基于最优特征空间的损毁建筑物信息提取方法。 采用ESP(Estimate of Scale Parameter)工具对图像进行最优尺度分割, 之后通过选取样本, 计算各类地物距离矩阵和最小分离距离寻求最优特征空间, 最后运用最优特征空间对震后损毁建筑物影像进行提取实验, 在QuickBird影像中提取总体精度达到了83.1％, Kappa系数达到了0.813, 在无人机影像中提取总体精度为92.9％, Kappa系数达到了0.940。 本文建立的提取方法与传统分类决策树方法相比, 其提取精度和效率都有较大提高, 在损毁建筑物信息提取方面具有较好的推广价值。 |
| 关 键 词： | 最优特征空间 损毁建筑物 ESP |
| 收稿时间： | 2018-06-05 |
Information Extraction of Damaged Buildings after Earthquake Based on Optimal Feature Space |
|
| CHEN Jin,CHEN Wen-kai,DOU Ai-xia,LI Wen,SUN Yan-ping.Information Extraction of Damaged Buildings after Earthquake Based on Optimal Feature Space[J].Earthquake,2019,39(3):71-83. | |
| Authors: | CHEN Jin CHEN Wen-kai DOU Ai-xia LI Wen SUN Yan-ping |
| Institution: | 1.Institute of Lanzhou Earthquake Research, China Earthquake Administration, Gansu Lanzhou 730000, China; 2.Institute of Earthquake Science, China Earthquake Administration, Beijing 100036, China |
| Abstract: | This paper proposes an information extraction method which is based on traditional object-oriented method and the optimal feature space. In this method, images are segmented at an optimal scale by using ESP (estimate of scale parameter) tools. Then, distance matrix and minimum separation distance between various types of objects and ground are calculated, through the selected samples, so as to seek the optimal feature space. Finally, the post-earthquake images of damaged buildings are extracted according to the obtained optimal feature space. For the QuickBird image, the total precision of extraction is 83.1％ and the Kappa coefficient is 0.813. For the UAV(unmanned aerial vehicle) image, the total precision of extraction is 92.9％ and the Kappa coefficient is 0.940. Compared with the traditional object-oriented method, the precision and efficiency of the extraction method are greatly improved, which shows that this method has good popularization value in the information extraction of damaged buildings. |
| Keywords: | Optimal feature space Damaged buildings ESP |
| 点击此处可从《地震》浏览原始摘要信息 | |
| 点击此处可从《地震》下载免费的PDF全文 | |