Convolutional sparse coding for noise attenuation in seismic data

We have developed convolutional sparse coding (CSC) to attenuate noise in seismic data. CSC gives a data-driven set of basis functions whose coefficients form distribution. The attenuation method by can be divided into the training and denoising phases. Seismic data with relatively high signal-to-noise ratio are chosen for get learned functions. Then, we use all (or subset) random or coherent Numerical experiments on synthetic show that learn shifted invariant filters, which reduce redundancy filters traditional sparse-coding method. achieves good performance when noisy better similar but noiseless set. numerical results from field test indicate effectively suppress complex By excluding features, our further separate ground roll.