next up previous [pdf]

Next: Acknowledgments Up: Liu etc.: 1-D time-varying Previous: Processing of field data


We have proposed and tested the time-varying median filter (TVMF) for attenuating prestack random, spike-like noise. We used a signal-to-noise ratio estimation to design the reference filter window and time-varying filter windows, in which the filter can adapt filter-window length to meet the intensity of the random, spike-like noise. The ability of this method to eliminate random noise while protecting desired signal further attests to the strength of the method. Our experiments show that in field data the TVMF can eliminate random noise enough to enhance the continuity of events. Spectral analysis also shows that the TVMF can effectively suppress random, spike-like noise in the whole frequency band. Comparison of different methods shows that the TVMF is more effective than the stationary median filter for eliminating spike-like noise, and it can enhance results of band-pass filtering by attenuating spike-like noise within the pass band.