Research On Edge Detection Method Of Seismic Data

Edge detection of seismic data,especially fault detection,is very important in seismic processing.Seismic data processing is the premise of interpretation,the corresponding seismic data edge(such as faults,sharp out)detection method is also increasingly important.Fault is not only the boundary of oil and gas field,but also the passage of oil and gas migration,so fault detection has important guiding significance for oil and gas exploration and development.The edge detection method has been used in seismic exploration for many years,and has a good application effect in detecting the edge of underground geologic body in seismic data.But the edge detection method is particularly sensitive to noise and needs noise suppression before detecting the seismic data with poor quality and low signal-to-noise ratio.In this paper,we study the edge preserving filtering of seismic data by using the anisotropic diffusion filtering method,and then use the Gauss-Laplace operator with edge attribute information to detect the edge of seismic data.Anisotropic diffusion filtering is developed from the application of partial differential equations to image denoising.The diffusion equation in physics is the basis of this method.Anisotropic diffusion filtering uses diffusion tensor to preserve edge filtering,but diffusion tensor is based on structure tensor.In this paper,the gradient and the second derivative are combined to construct the structure tensor,and the concrete construction method is given.The eigenvalues and eigenvectors of structure tensor can represent local stratigraphic structure information well.The estimation and eigenvalue detection of local structure information of seismic data using eigenvalues and eigenvectors are discussed emphatically,and the discontinuous geological boundary such as fault and river can be detected with confidence degree.Two specific methods of constructing diffusion tensor using structure tensor are also discussed and the differences between the two methods are compared.Finally,the results of the Marmousi2 model and the actual seismic data are used to illustrate the effect of this method in the denoising of seismic data.For edge detection of seismic data,in order to reduce the influence of noise of seismic data on edge detection operators,anisotropic diffusion filter is first used to preserve edge filtering of seismic data before detection.This paper focuses on the method theory of Gauss-Laplace operator and Canny operator.On the basis of analyzing the advantages and disadvantages of Gauss-Laplace operator,the Gauss-Laplace operator with edge attribute is deduced.Finally,the synthetic model data and the actual seismic data are used for fault detection and good results are obtained.In this paper,the anisotropic diffusion filtering method and the Gauss-Laplace operator with edge attributes are used to detect faults in both the model data and the actual seismic data,and satisfactory results are obtained.