| The Mayang red bed area in Hunan is very likely to cause landslide disasters in hilly areas,with the peculiarity of stratigraphic lithology and frequent heavy rainfall,resulting in significant loss of life and property.Quickly obtaining the location and range of landslide points is an urgent need for landslide emergency investigation and post-disaster assessment.With the rapid development of high-resolution domestic satellites,landslides with more scattered and smaller scale can be effectively identified from the images.In this paper,Mayang County is selected as the research area.For the application of remote sensing technology in the red layer rainfall landslide,the landslide points in the work area are identified and combed,and the development characteristics of the landslide and the remote sensing image characteristics of the landslide are comprehensively analyzed.The optical remote sensing GF-2 satellite better than 1 meter was selected to classify and describe the types of landslides in mountainous areas,roadside slopes,and river bank slopes,and select representative experimental points as experimental objects.For the experimental points,the information extraction method based on pixels and the object-based information extraction method are used to carry out research on information extraction of landslides.After pixel-based sample selection,the classification algorithms of maximum likelihood and support vector machine are used to extract the landslide information;the optimal segmentation scale is selected through the object-oriented optimal scale segmentation tool,and the features are selected based on the information gain method Factor,the eigenvalue distribution characteristics of the image feature factors,select the membership function to establish the feature factor rule set,and extract the landslide information.The extraction results of the two information extraction methods show that the different types of landslides in the study area based on the object-oriented extraction method compared with the pixel-based extraction method,the user accuracy is increased by 40%,and the maximum error is reduced 1.5.Comprehensive research shows that using object-oriented information extraction method on high-resolution satellite imagery,it is feasible to extract rapid and effective information on sudden rainfall landslides in the red layer area of Hunan Province,which can be used for macro investigation and post-disaster geological disasters.Provide technical method reference for evaluation and emergency investigation. |
