基于光谱、空间和形态特征的面向对象滑坡识别

发布时间：2018-04-23 01:13

本文选题：面向对象 + 遥感影像　；参考：《遥感技术与应用》2017年05期

【摘要】：地震经常诱发数量众多,覆盖范围广的滑坡并造成极大危害,因此,需要对大范围的滑坡灾害进行快速评估。随着遥感影像分辨率的提高,面向对象分类方法在这方面的应用比传统的目视解译和基于像素的方法更具优势。但是,目前基于面向对象方法的滑坡识别研究还相对较少,而且通常针对小范围的研究区。基于SPOT5 2.5m多光谱影像,提出一种综合光谱、空间、地形和形态特征的面向对象滑坡自动识别方法,并应用于较大范围研究区。结果表明:面向对象滑坡自动识别方法能将研究区内95%的滑坡识别出来,综合考虑滑坡的过度提取与遗漏提取情况,滑坡提取质量为74.04%,效果较好,能够快速、有效地识别大范围的滑坡。该方法可以应用于对地震或强降雨引起的大范围滑坡灾害进行快速评估,为灾后应急救援和恢复重建工作提供参考。
[Abstract]:Earthquakes often induce a large number of landslides with wide coverage and cause great harm. Therefore, it is necessary to evaluate landslide disasters quickly. With the improvement of the resolution of remote sensing image, the application of object oriented classification is more advantageous than the traditional visual interpretation and pixel based method. However, the research of landslide recognition based on object-oriented method is relatively few, and usually for small research areas. Based on SPOT5 2.5m multispectral image, an object-oriented automatic landslide recognition method is proposed, which integrates spectral, spatial, terrain and morphological features, and is applied to a wide range of research areas. The results show that the object-oriented automatic identification method can identify 95% of landslides in the study area. Considering the over-extraction and omission of landslide, the quality of landslide extraction is 74.04, the effect is good and the landslide can be quickly extracted. Identify large areas of landslides effectively. This method can be applied to the rapid assessment of large-scale landslide disasters caused by earthquakes or heavy rainfall, and provide a reference for emergency rescue and recovery and reconstruction after disasters.
【作者单位】：北京师范大学环境演变与自然灾害教育部重点实验室;民政部—教育部减灾与应急管理研究院;
【基金】：“十二五”科技支撑计划项目(2012BAK10B03) 国家自然科学基金项目(41271544)
【分类号】：P642.22

【相似文献】