摘要

In the application of multilevel median filter to seismic data processing, it is more effective to use a longer wavelength in removing the random noises, but the longer wavelength has a more damaging effect to the data, and is detrimental in preserving the details of the signals. This paper brings forward a new fussy nesting multilevel median filter based on the multilevel median filter concept. The new filter uses a long wavelength for the filter when removing the random noises, and uses a short wavelength when preserving the details of the signals. Therefore, we can remove the random noises yet are still able to preserve the details of the signals. We have had very good results from both model analyses and actually processing the data.