基于DEM和遥感影像的区域黄土滑坡体识别方法研究

发布时间：2018-02-03 18:35

本文关键词： 黄土滑坡可视性 SEaTH算法人工神经网络多层感知器　出处：《地理与地理信息科学》2017年04期 　论文类型：期刊论文

【摘要】：应用定量地貌学的原理和方法,对天水市南北两山已有黄土滑坡进行了自动识别研究。首先,引入地形可视性辅助遥感影像目视解译,以选择滑坡和非滑坡样本;其次,综合滑坡的地形特征和遥感特征,通过采取SEaTH算法遴选出能有效区分滑坡与非滑坡的特征差异指标建立滑坡有效识别指标体系;最后,基于滑坡有效识别指标特征集进行面向对象分割,通过建立多层感知器的滑坡自动识别模型对分割后的对象进行识别,并作精度评价。结果表明:1)针对该研究区的已有黄土滑坡,粗糙度、高程变异、切割度、起伏度、坡度、可视性6种地形指标具有识别价值;2)该模型对滑坡的识别精度达71.03%,非滑坡的识别精度达92.02%,较好地识别出了该区域的滑坡范围。
[Abstract]:Based on the principle and method of quantitative geomorphology, the automatic identification of loess landslide in the north and south of Tianshui City has been carried out. Firstly, terrain visibility is introduced to assist the visual interpretation of remote sensing images. To select landslide and non-landslide samples; Secondly, the landform feature and remote sensing feature of landslide are synthesized, and the effective identification index system of landslide is established by adopting SEaTH algorithm to select the characteristic difference index which can effectively distinguish landslide from non-landslide. Finally, object oriented segmentation is carried out based on the feature set of effective landslide recognition index, and the segmented objects are identified by establishing a multi-layer perceptron model for automatic identification of landslide. The results show that the loess landslide, roughness, elevation variation, cutting degree, fluctuation degree, slope and visibility of the study area are valuable for recognition. 2) the recognition accuracy of the model for landslide is 71.03 and that of non-landslide is 92.02. The range of landslide in this area is well recognized.
【作者单位】：兰州大学资源环境学院;西安测绘总站;
【基金】：国际科技合作专项项目(2013DFE23030) 甘肃省民生计划项目(1503FCME006)
【分类号】：P208;P237;P642.22
【正文快照】： 0引言滑坡是当今世界可造成巨大经济损失的自然灾害之一,已经成为自然灾害中第三大致死灾害。甘肃省天水市位于秦岭山地和黄土高原过渡地带,地质环境差,其南北两山的黄土滑坡灾害频发,且滑坡数量大、规模大,因其紧邻市区,对群众生活和社会生产影响巨大。因此,本文选取天水市南

【参考文献】