A method for increasing similarity between a first seismic dataset and a second seismic dataset from at least one seismic survey of a subsurface. The first and second seismic datasets are migrated to a dip angle image domain. The migrated first and second seismic datasets are used in the dip angle image domain to calculate a set of decimating weights to be applied to the first seismic dataset and the second seismic dataset to maximize a similarity between the first seismic dataset and the second seismic dataset. The decimated weights are applied to the first and second seismic datasets, and an image of the subsurface is generated using the first seismic dataset and the second seismic dataset following application of the decimated weights.