科研之友
微信
新浪微博
Facebook
分享链接
摘要
On January 12, 2010, the Port-au-Prince, Haiti's capital, was devastated by an earthquake. Registering 7.0 on the Richter scale, it was the biggest earthquake in its history and killed or injured twenty-five hundred thousand people. This earthquake caused an extensive destruction of infrastructures and relief action after disaster could not perform immediately. A successive recovery planning after the disaster is required from the determination of location and extent of the damage. The change detection using multi-date satellite images is one of the efficient tools and a prerequisite for quick analysis and assessments of damage attributed to disaster over a wide area. In this study, it has adopted object-based change detection technique to assess building damage using high resolution satellite images acquired over different dates.
Our study is divided into three parts. First part is the automatic building extraction based on image segmentation for the prompt relief action. The second step is the accurate building digitizing of the multidate images for the precise relief action, although it is labor-intensive and time-consuming. On last part, building damages, which are categorized into intact, half damaged and completely damaged, are detected automatically using segment matching and destruction size analysis. Although Haiti, one of the poorest nations, has very complicated and mixed building structures and rooftops in urban area, the generalized building damage assessment was proposed. The results can be advantageous for recovery planning and urban disaster management.
- 出版日期2012-10
