Research On Road Fluctuation And Landslide Information Extraction Method Based On SAR Image

Surface fluctuation information describes the height variation of the topographic surface and plays a major role in geological mapping and civil engineering.As one of the natural disasters with large destructive power and high frequency,landslide has caused great harm to people’s property and life safety.China’s geological environment is complex,especially in the southwestern region,with developed water systems and frequent earthquakes.It is the main cause of landslides.The landslide information extraction is an urgent and far-reaching task.From these two aspects,this paper studies the road undulation and landslide information extraction method based on SAR image.The main research contents are as follows:(1)This paper proposes a more comprehensive method for extracting surface fluctuation information using InSAR technology,and gives a complete process for extracting fluctuation information.During the experiment,the urban and mountain images were selected to study the application of the above algorithm in different ground conditions.Comparing the experimental results with the field survey and Google Earth data proves the reliability of the algorithm.(2)This paper establishes the error model of the fluctuation information extraction,summarizes the source of the error,analyzes and statistics the relationship between the main parameters and the error transfer coefficient in several models.Finally,the error improvement scheme is proposed.(3)For the baseline parameters that have the greatest influence on the fluctuation information extraction error.In this paper,a base-line estimation method based on the reference points of the coherent coefficient graph is proposed to solve the problem that the base-line estimation method based on the control points with the highest accuracy cannot set up the control points in some mountainous areas.(4)Unsupervised classification of polarimetric SAR images.In this paper,the boundary feature,regional feature,statistical feature and the polarization feature processed by freeman-durden & H/α/A polarization decomposition technique are combined to construct the energy functional and solve the classification problem of multifeature contour fuzzy model.The fuzzy membership function family is added to the model to improve the accuracy and the idea of hierarchy is used for reference to solve the constraint problem.Finally,the real SAR image classification results are compared with several popular classification algorithms to verify the accuracy and reliability of mothod.(5)Taking several landslides in Maoxian County of Sichuan Province as an example,it reveals that the ground objects near the landslide area before and after the landslide have not changed,and the types of ground objects in the landslide area have changed from low-lying jungle to bare soil.Applying the above-mentioned full-polarization SAR image classification algorithm,the two images are preprocessed and divided into nine categories.The change detection method was used to extract the landslide area.Finally,the jinsu transmission channel and guizhou oil pipeline are taken as examples to verify the accuracy of the algorithm and the practicability of landslide information extraction.