论文部分内容阅读
2010年1月12日海地发生了特大地震,本文使用GeoEye-1影像和机载激光雷达数据(光探测和测距)对震后太子港地区建筑物以及破损建筑物的碎片瓦砾出图。方法是基于影像和雷达数据,综合光谱、结构和高度信息,采用面向对象土地覆盖分类技术。总体分类精度约为87%,而建筑物和瓦砾接近80%。在城中心一个2km2范围内,与人工筛选出的200个实际损毁建筑相比,此方法达到90%的准确度,并且建立了覆盖太子港地区约30km2共55000栋建筑物的三维建筑模型。我们发现,大多数受损建筑物是城市规划区的混凝土和砖体结构,而受损较小的则是庇护所和临时用金属片屋顶搭建的房屋。研究表明,融合光学图像和激光雷达数据可以有效地反映由地震造成的城区破坏性质、严重程度、受害范围和受损模式,而这些地区人口稠密,比如太子港。
On January 12, 2010, a massive earthquake hit Haiti. This article uses the GeoEye-1 imagery and airborne lidar data (light detection and ranging) to map debris debris in Port-au-Prince buildings and damaged buildings in Port-au-Prince. The approach is based on image and radar data, combining spectral, structural and elevation information with object-oriented land cover classification techniques. The overall classification accuracy is about 87%, while the building and rubble are close to 80%. Within a 2km2 radius of the city center, this method achieves 90% accuracy compared with the manually selected 200 physically damaged buildings, and a three-dimensional building model covering 55,000 buildings of about 30km2 in Port-au-Prince area has been established. We found that most of the damaged buildings were concrete and brick structures in the urban planning area, while those less damaged were shelters and temporary metal roofing. Research shows that the fusion of optical images and lidar data can effectively reflect the devastating nature, severity, victimization and damage patterns of urban areas caused by earthquakes, which are densely populated, such as Port-au-Prince.