快鸟影像用于泥石流冲毁区自动提取

摘要

Aiming at the automatic extraction problem of debris flow destroyed areas,this study discusses an automatic extraction method of debris flow destroyed area based on QuickBird high resolution remote sensing imagery,and proposes an automatic extraction model of multi-information combination.Firstly,use a tasseled cap transformation for the original reflectance images to calculate moisture component and greenness component,and then use the difference of them to get the initial debris flow destroyed area image;Secondly,use automatic extraction technology to get threshold for separating suspected debris flow destroyed image from other surface features,and then get candidate debris flow destroyed image;Finally,extract the water and sparse vegetation areas to perform mask operation on candidate debris flow destroyed image for getting debris flow destroyed area,and perform post-processing for small spots in the debris flow destroyed area to get relatively accurate debris flow destroyed area.The experimental results show that the proposed model is feasible and effective,and achieves good results;the extraction accuracy of small debris flow is not high,which needs to be strengthened.%针对泥石流冲毁区难以自动提取的问题，该文基于快鸟（QuickBird）高分辨率影像，探讨了泥石流冲毁区自动提取方法，提出一种多信息结合自动提取模型。该方法首先对原始反射率影像做缨帽变换，利用湿度分量和绿度分量的差得到初始泥石流冲毁区影像；其次利用阈值自动提取技术将疑似泥石流冲毁区区域与其他地物分离，得到泥石流冲毁区候选区域；最后提取出水体信息和植被稀少地区，对泥石流冲毁区候选区域进行掩膜运算，得到泥石流冲毁区，并对泥石流冲毁区中的小斑块做后处理，得到相对准确的泥石流冲毁区分布范围。实验结果表明，此泥石流冲毁区自动提取模型是可行和有效的，取得了较好的结果；但对小斑块泥石流提取的精度不高，需要加强研究。