高分一号汶川极震区滑坡提取研究

选择汶川地震极震区的高分一号卫星影像，通过面向对象的分析技术提取滑坡信息；采用多尺度分割算法，结合高分影像和滑坡特点将以往经验式参数选取方法进行优化，分析极震区滑坡的特征，选择合适的特征参数，构建分类规则，实现滑坡的识别与提取。滑坡灾害信息的提取结果采用野外实际调查的滑坡点进行精度评价，滑坡提取总精度为84%，表明利用高分一号高分辨率卫星数据可以较好地提取滑坡灾害信息，基本满足滑坡灾害识别的要求。

Research on Landslides Extraction Based on the Wenchuan Earthquake in GF-1 Remote Sensing Image

Earthquake induced landslide is the most common and destructive type of earthquake, so it is of great significance to quickly and accurately obtain the distribution of landslides after the earthquake. The traditional method of artificial interpretation of remote sensing is of low efficiency. In this paper, we select the GF-1 image of the Wenchuan earthquake zone and extract the landslide information by using the object-oriented method. Multi-scale segmentation is used to set the different segmentation parameters. The suitable parameters are selected according to the features of spectrum, texture and geometry, and the classification rules are constructed to extract the landslide area. Through the comparison of landslide with landslide points by actual investigating, the accuracy of object-oriented classification method for 84％. The result shows that GF-1 data can be used to extract information of landslide hazard and meet the basic requirements of landslide hazard identification.