地震研究

基于遥感影像的震后避难空间快速提取模型研究——以2021年云南漾濞M_S6.4地震为例

(1.云南省地震局,云南昆明 650224; 2.中国地震局工程力学研究所,黑龙江哈尔滨 150080; 3.云南大学地球科学学院,云南昆明 650031)

关键词：无人机影像; 避难空间; 面向对象分类; GIS栅格化; 漾濞M_S6.4地震

Study on the Rapid Extraction Model of Post-earthquake Shelter Sites Based on Remote Sensing Images:[WT4"]A Case Study of the Yangbi M_S6.4 Earthquake

A Case Study of the Yangbi M_S6.4 Earthquake]

(1.Yunnan Earthquake Agency,Kunming 650224,Yunnan,China)(2.Institute of Engineering Mechanics,China Earthquake Administration,Harbin 150080,Heilongjiang,China)(3.School of Earth Science,Yunnan University,Kunming 650031,Yunnan,China)

Keywords：UAV images; shelter sites; object-oriented classification; GIS rasterization; the Yangbi M_S6.4 Earthquake

DOI: 10.20015/j.cnki.ISSN1000-0666.2023.0013

备注

摘要

全文

图/表

参考文献

震后避难空间是居民遭遇地震时紧急疏散、避难、临时生活的重要区域。以无人机影像为基础,采用影像面向对象分类与GIS栅格化的分析方法,构建震后避难空间评价指标,建立基于遥感影像的震后避难空间快速提取模型,并以2021年云南漾濞MS6.4地震为例,将避难空间提取的结果与震后居民实际选取的避难空间进行比较。结果表明:模型共计提取可用避难空间70个,根据目标函数F得到最优避难空间5个,其中每个避难空间在漾濞MS6.4地震中实际帐篷数量分别为72、55、54、30、44顶,模型计算结果与实际结果匹配。

Emergency shelter sites are important for the earthquake-hit residents to evacuate and take refuge after an earthquake.Based on UAV remote sensing images,we use the object-oriented classification method and the GIS rasterization method to analyze the UAV images,thus to decide the evaluation indexes for post-earthquake shelter sites,and to build the rapid extraction model of post-earthquake shelter sites.We use our model to extract shelter sites in Yangbi County,and obtain 70 available shelters.And according to the objective function F,we obtain 5 optimal shelters.We compare our results with the actual shelter sites chosen by the local residents after the Yangbi MS6.4 Earthquake.The number of tents put up on 5 shelter sites after the Yangbi MS6.4 Earthquake was 72,55,54,30 and 44,respectively.The calculated results from the model match with the actual situation.

引言
1 研究方法及技术路线
2 数据来源与处理
3 结果
4 对比分析
5 结论

图1 基于遥感影像的震后避难空间快速提取方法流程图<br/>Fig.1 The flow chart of fast extraction method for post-earthquake shelters based on remote sensing images

图1 基于遥感影像的震后避难空间快速提取方法流程图
Fig.1 The flow chart of fast extraction method for post-earthquake shelters based on remote sensing images

图2 K邻近分类法示意图<br/>Fig.2 The schematic diagram of the K-Nearest Neighbor method

图2 K邻近分类法示意图
Fig.2 The schematic diagram of the K-Nearest Neighbor method

图3 震后漾濞县城GF7号卫星影像(a)、无人机航拍影像(b)以及 DSM数字表面高层模型(c)<br/>Fig.3 The GF7 satellite image(a),the UAV image(b),and the DSM image(c) of Yangbi County after the Yangbi MS6.4 earthquake

图3 震后漾濞县城GF7号卫星影像(a)、无人机航拍影像(b)以及 DSM数字表面高层模型(c)
Fig.3 The GF7 satellite image(a),the UAV image(b),and the DSM image(c) of Yangbi County after the Yangbi MS6.4 earthquake

图4 研究区原始影像(a)、分割尺度与合并尺度分别为65、95的分割结果(b)<br/>Fig.4 The original image(a)and the result from the segmentation scale 65 and the combined segmentation scale 95(b)

图4 研究区原始影像(a)、分割尺度与合并尺度分别为65、95的分割结果(b)
Fig.4 The original image(a)and the result from the segmentation scale 65 and the combined segmentation scale 95(b)

图5 基于目视解译的地物分布<br/>Fig.5 Ground-feature distribution based on visual interpretation

图5 基于目视解译的地物分布
Fig.5 Ground-feature distribution based on visual interpretation

表1 研究区地物提取规则<br/>Tab.1 Rules of ground object extraction

表1 研究区地物提取规则
Tab.1 Rules of ground object extraction

图6 多尺度分割结果(a)及基于规则分类结果(b)<br/>Fig.6 Multi-scale segmentation results(a)and rule-based classification results(b)

图6 多尺度分割结果(a)及基于规则分类结果(b)
Fig.6 Multi-scale segmentation results(a)and rule-based classification results(b)

表2 漾濞县城地物提取精度评价结果<br/>Tab.2 The accuracy evaluation of the extracted ground objects in Yangbi County

表2 漾濞县城地物提取精度评价结果
Tab.2 The accuracy evaluation of the extracted ground objects in Yangbi County

图7 漾濞县城无人机航拍影像灰度图(a)、遥感影像多特征融合的地物提取结果(b)及基于ArcGIS修正后的地物矢量化结果(c)<br/>Fig.7 The UAV grayscale image(a),the ground objects extracted from multi-feature fusion of the remote sensing image(b)and the vectorization of the ground objects based on ArcGIS in Yangbi County(c)

图7 漾濞县城无人机航拍影像灰度图(a)、遥感影像多特征融合的地物提取结果(b)及基于ArcGIS修正后的地物矢量化结果(c)
Fig.7 The UAV grayscale image(a),the ground objects extracted from multi-feature fusion of the remote sensing image(b)and the vectorization of the ground objects based on ArcGIS in Yangbi County(c)

图8 研究区空地(a)、建筑物(c)、建筑物结构(e)、道路(g)、楼层数(i)、帐篷(k)矢量数据; 空地(b)、建筑物(d)、建筑物结构(f)、道路(h)、楼层数(j)、帐篷(l)栅格化数据<br/>Fig.8 The vector data of the open space(a),buildings(c),the architectural structure(e),roads(g),number of floors(i),tents(k)and the rasterized data of the open space(b),buildings(d),the architectural structure(f),roads(h),number of floors(j),tents(l)

图8 研究区空地(a)、建筑物(c)、建筑物结构(e)、道路(g)、楼层数(i)、帐篷(k)矢量数据; 空地(b)、建筑物(d)、建筑物结构(f)、道路(h)、楼层数(j)、帐篷(l)栅格化数据
Fig.8 The vector data of the open space(a),buildings(c),the architectural structure(e),roads(g),number of floors(i),tents(k)and the rasterized data of the open space(b),buildings(d),the architectural structure(f),roads(h),number of floors(j),tents(l)