Simultaneous erratic noise attenuation and seismic data reconstruction based on structure-oriented space-varying median filter under the FISTA framework
Limited by the physical environment,acquisition cost,and other factors,field data always suffer from irregularly missing traces.In order to provide complete data for subsequent seismic data processing and imaging,it is inevitable to reconstruct the missing seismic traces.The interpolation method based on sparse transform is widely used for seismic data reconstruction,but the traditional interpolation method based on sparse transform is very sensitive to erratic noise,and the reconstruction accuracy of this method is seriously affected when erratic noise exists in the seismic data.To overcome this problem,we propose a interpolation method that simultaneously suppresses erratic noise and reconstructs missing seismic traces with high accuracy by introducing the structure-oriented median filter technique in the framework of the Fast Iterative Shrinkage-Thresholding Algorithm(FISTA).The method performs median filtering of the seismic signal along its local dip direction in each iteration of FISTA to attenuate the erratic noise and maximize the protection of the effective signal from damage.Benefit from this,the proposed method can perform high-precision reconstruction on under-sampled seismic data with erratic noise.Both model and field data tests show that the proposed method is superior to the conventional sparsity-promoting interpolation method in terms of interpolation accuracy and stability when erratic noise is present in the seismic data.
Seismic data reconstructionSparsity promotingFast Iterative Shrinkage-Thresholding Algorithm(FISTA)Structure-oriented median filterErratic noise
万方数据
维普
