Comparison of pursuit algorithms for seismic data interpolation imposing sparseness

Recently, technical results from the theory of Compressive Sensing have been applied to the Fourier class of algorithms for solving the ever-increasing seismic data interpolation problem while handling potentially irregularly sampled geometries. The method we here propose makes use of the so-called Conjugate Gradient Pursuit with the Stagewise Selection Strategy, a novel general purpose algorithm for sparse representation, which brings advantages in terms of computational costs when compared to the well-known Matching Pursuit, while not affecting accuracy. The effectiveness and efficiency of the derived interpolation method is proved and compared to the performances of the Matching Pursuit-based interpolation method when applied on a real dataset showing, in turns, intrinsic sampling irregularity and heavy manual trace decimation.