Research On Feature Extraction Method Of Deformation Disaster Based On S-SAR

Landslide deformation disasters such as landslides,collapses,and mudslides have occurred in succession in recent years,posing serious threats to people's lives,property,and construction safety.Traditional contact-point monitoring technologies are increasingly failing to meet the new situation of China's economic and social development.The urgent need for serious safety accidents and the ability to improve disaster prevention,mitigation and relief.With all-weather radar slope of microwave remote sensing technology advantages of its long-range capability and a wide range of surveillance throughout the day,the judgment is more conducive to predict deformation field and deformation trends of landslide,has gradually become a research focus and direction of development of deformation disaster monitoring technology.At present,there are mature slope radar systems at home and abroad.More than 100 mines around the world are equipped with slope radars,and the data sources are extensive.However,there are many problems with the early warning models and early warning methods of existing slope radars.The early warning model still uses the single curve analysis method of the traditional point monitoring technology,and does not make full use of the area point group data to construct the model.The early warning method has not yet exerted the space-time continuity advantage of multi-scale telemetry data.The early warning stays in the time dimension landslide time prediction,lacks the effective space dimension landslide orientation and landslide area warning forecast method,and is equipped with advanced remote sensing area monitoring methods and a large number of points.Group data,but the situation is helpless.In view of this situation,this paper takes the actual monitoring data of S-SAR type slope radar jointly developed by China Academy of Safety and Production Science and Institute of Electronics of Chinese Academy of Sciences as the object,relying on the central-level public welfare research institute " large field of view slope The basic research project of radar monitoring key technology and test verification(2017JBKY05),from the inversion method of early warning model parameters,the fusion of multi-scale remote sensing data fusion technology in different frequency bands such as laser,optics and microwave,and the rock slope early warning model The gradual and in-depth study is carried out,and the time-space analysis and early warning model based on the fusion of the area point group telemetry data is proposed.The parameter inversion algorithm,information fusion algorithm and time series data space analysis algorithm are developed for the model,which can be used for high landslide.The risky rock slope is used to predict the landslide time,the location of the deformation disaster and the deformation area prediction.The specific research contents and innovations are as follows:1.Proposed a parameter inversion method for the warning model of the slope radar radar with information in SAR imagesThe warning content of the slope radar includes the location warning for the occurrence of the deformation disaster,and also includes the area warning for the deformation disaster.The traditional point monitoring method can predict the location of the deformation disaster,but the area of the deformation disaster cannot be accurately predicted.Aiming at the above problems and making full use of the advantages of slope radar data,a method of a priori information of the slope radar monitoring image combined with the geostatistical inversion early warning model deformation parameter is proposed.The parameters that can be used for early warning in the S-SAR slope radar image are analyzed.Then,according to the zero-baseline and step-frequency continuous wave imaging interference theory,the ideal state of the radar image pixel is derived.The point target simulation and the early warning performance parameter verification experiment are carried out,and the radar image,coherence map and deformation map of the slope are discovered.The intrinsic link between the response characteristics and the state of the slope.Taking the monitoring of homogeneous buildings in a city as an example,the method of inversion of synergistic deformation using geostatistical Kriging method and coherent image features is described.Provide more area continuity information in addition to the deformation variable and deformation rate for the early warning model.2.A multi-scale telemetry point group information method for slope radar surface data fusion is proposed.Two-dimensional radar slope surface domain data with the diagnostic interpretation difficult,you need to be mapped to the three-dimensional space.The existing three-dimensional mapping method still cannot meet the needs of the early warning model.The scale effect is not considered,and the mapping position accuracy verification method is lacking.At present,there is no overall scheme for measuring the accuracy of slope deformation measurement in the three-dimensional high-precision close-range and distant view.Aiming at the above problems,has designed a controlled scene data fusion and deformation processes comparative experiment carried out close-range experimental program to improve the integration of a controlled three-dimensional mapping point target location accuracy and error check in open pit mining prospects Slope Mapping correction method.3,proposed rock slope before sliding warning model based on domain-aware adaptive fusion system environmentAiming at the problem of lack of effective early warning model for slope radar area data and point group time series data,based on the integrated trailer monitoring system of an open pit mine,planning the functions of each component in the fusion system,using 3D modeling DEM data to control the occurrence of landslide Slope characteristics.The slope radar data is mapped to the 3D model,and the high-order Laplacian variation is introduced to calculate the continuous deformation of the slope surface.The selection scheme of the timing PS point is improved,and high-quality deformation,cumulative deformation,deformation rate and deformation area are obtained.The curve fitting,prediction and classification methods are combined with the tangential angle method and the speed reciprocal method in the field of landslide warning.Finally,a set of early warning model for predicting the location of landslide,the area of deformation and the time of occurrence of landslide is proposed.The national-level major landslide emergency monitoring site is gradually applied.