Simultaneous reconstruction and denoising of seismic data based on rank reduction and sparsity constraints
Field seismic data contain various random noise and irregular channel missing.Their simultaneous reconstruction and de-noising is necessary for subsequent data processing.Currently,most simultaneous reconstruction and denoising methods only use a sin-gle sparsity or rank reduction constraint.The sparsity constraint exhibits high efficiency but lacks adaptability to various data.In con-trast,the rank reduction constraint can adapt to various data but shows a high computational cost.To take a full advantage of different constraints,this study proposed a method for simultaneous reconstruction and denoising of seismic data based on combined constraints.This method regards projection onto convex sets(POCS)based on Fourier transform as the sparsity constraint,and damped multichan-nel singular spectrum analysis(DMSSA)as the rank reduction constraint.It employs the truncated singular value decomposition(TS-VD)algorithm and the exponential threshold equation,fully utilizing the high computational efficiency of the sparsity constraint and the strong adaptability of the rank reduction constraint.As indicated by the processing results of theoretical and field data,this method based on combined constraints can consider and utilize the spatio-temporal correlations of seismic data,achieving higher signal-to-noise ratios via fewer iterations compared to methods based on a single constraint.
万方数据
维普
