An iterative half thresholding method for seismic data interpolation

In this paper, we introduce an analysis-based iterative half thresholding algorithm to interpolate the seismic data. Different from the conventional methods, this approach takes advantages of the nonconvex sparsity constraint, namely the L1/2-norm constraint. The redundant Fourier transform is employed from a spectral compressive sensing perspective. The experimental results shows that double zero-padding in spatial direction of seismic data can further enhance the reconstruction quality, allowing for the improved interpolation performance and the increased computational cost and storage requirement; zero-padding in temporal does nothing to improve the reconstruction performance.