Information Extraction Of Landslide Geological Hazards After The Earthquake Based On The Technology Of Remote Sensing And 3D

The earthquake induced a large number of secondary geological disasters, mainly landslides, landslides, mud rock flow and so on.These secondary geological disasters make people's lives and property suffered huge losses.The active period of the secondary geological hazards will continue for quite a long time.The landslides which had happened caused a lot of accumulation, the accumulation of a lot of debris will further breed debris flow.The landslide information can be extracted qualitatively and quantitatively in the first time, which is especially important for disaster prevention and reduction, and disaster management.So it is a significant measure to extract the information of landslide.Using remote sensing technology in landslide hazard identification for many years, in recent years, with the development of satellite technology, the remote sensing image resolution is also getting higher and higher, the original pixel oriented remote sensing image analysis method has been unable to meet the demand. However, with the development of remote sensing technology and information extraction technology, especially the appearance of object-oriented classification method, the accuracy and efficiency of the information identification of landslide disaster are improved.This paper chose Wenchuan County Ying Zhen Xiu as the research area, in the field research study based on using multi-source remote sensing data(2008 aerial image, spatial resolution of 0.5 meters, high in 2013, a remote sensing image, multi spectral resolution of 5 meters, 2 meters resolution panchromatic wave band), image preprocessing; based on digital photogrammetry technology to produce high precision DEM; the establishment of different scales in the research area of 3D visualization scene; based on the based on supervised classification, support vector machine(SVM) method, based on the rules of the multi-scale segmentation method, based on dimensional analysis technique of landslides in the study area was extracted, and the landslide extraction results for accuracy assessment. The content of this research and the main research results are as follows:(1)the morphological characteristics and image characteristics of the landslide disaster are analyzed from the aspects of hue, texture and geometry, in the process of research.(2)Processing multi-source remote sensing data andregistrating of aerial images and GF-1wich had fused accuratly promised the landslide information extraction accuracy and comparison in different years of multi-source remote sensing image extraction of landslide.Based on digital photogrammetry method to produce high precision DEM, and edit DEM.Using DEM calibration of remote sensing images to generate DOM, making 3D remote sensing images, to achieve three-dimensional visualization of the study area. This three-dimensional visualization model can reflect the terrain and topography of the study area objectively and realistically, and the interpretation of landslide information will be transformed from 2D to 3D, which can improve the accuracy and speed of the landslide extraction.(3)The reasonable training samples are set up, on the basis of the establishment of the remote sensing interpretation signs of landslide,and the support vector machine method is used to classify the objects in the study area. Combined with the results of field research, a certain number of samples were randomly selected, and the classification results were evaluated by using the confusion matrix.(4)The object properties characteristic of landslide are selected, and setted the threshold of the false positive regions of landslide, the final determination of the optimal segmentation parameters,in the analysis of the three kinds of image segmentation method, multi-scale segmentation parameters under the premise,. The number and area of landslide are extracted by multi-scale segmentation. The number and area of the landslide, the landslide, the landslide, and the landslide are extracted. Based on error analysis and accuracy evaluation.(5) landslides were extracted by using the 3D analysis, 3D change detection in DEM are solved based on the variation of elevation. The study area difference before the earthquake after the earthquake the DEM data set, DEM difference threshold, change detection results for statistical analysis. Three dimensional display of the results of the test can be intuitive and macro to show changes in the regional scope. To select the six Bay of the research and reflection show as a typical example, the corresponding calculation method is adopted to realize the quantitative extraction of the landslide.In this paper, three kinds of methods are used to extract the landslide from the qualitative to quantitative, and the corresponding accuracy assessment is carried out. Based on the rule of multi-scale segmentation and DEM 3D change analysis, the results of the landslide are obtained.