Seismic exploration signal random noise suppression method based on extrusion short-time Fourier transform

Abstract

The invention discloses a seismic exploration signal random noise suppression method based on extrusion short-time Fourier transform. The method comprises the following steps of firstly, carrying outshort-time Fourier transform on each signal in a high-dimensional seismic data volume, performing extrusion operation on the short-time Fourier transform coefficient greater than a set threshold value1 to obtain an extrusion short-time Fourier transform coefficient, then, using the positions of the extrusion short-time Fourier transform coefficients of the adjacent channels as constraints to perform threshold processing on a current channel extrusion short-time Fourier transform coefficient, and finally, reconstructing the processed short-time Fourier transform coefficient to obtain a seismicexploration signal after random noise suppression. Basic data with a high signal-to-noise ratio is provided for subsequent seismic data processing.

Description

technical field [0001] The invention belongs to the technical field of exploration geophysics, and in particular relates to a method for suppressing random noise of seismic exploration signals by using squeezed short-time Fourier transform. Background technique [0002] Random noise mainly comes from ground microseisms, instruments, and scattered waves caused by medium inhomogeneity. It appears as chaotic vibrations in seismic records and has a wide frequency band. At the same time, due to complex surface and terrain conditions, such as mountains, loess plateaus, rugged seabeds, deserts, Gobi, and multi-stage superimposed structures, various random noises interfere with deep seismic records. Random noise interference will reduce the signal-to-noise ratio and quality of seismic records, and have a great impact on subsequent seismic signal processing and interpretation. Therefore, in order to apply the collected seismic data more fully and accurately, it is necessary to supp...

AI Technical Summary

Problems solved by technology

[0003] The existing random noise suppression methods for seismic exploration signals include Fourier domain filtering methods, frequency-wavenumber domain filtering methods, time-frequency domain transform filtering methods and other two-dimensional wavelet transform filtering methods. For filtering methods, the broadband seismic signal and random noise overlap each other in the frequency domain, and the two cannot be completely separated; at the same time, the distribution of random noise in the time domain is not completely uniform, and it is not complete to directly suppress the noise in the frequency domain. Signal damage is relatively large; for frequency-wavenumber domain filtering methods, broadband seismic signals and random noise overlap each other in the frequency-wavenumber domain, which is less than that of Fourier domain filtering methods, but they cannot completely combine Complete separation; at the same time, the distribution of random noise in the time-space domain is not completely uniform, and the suppression of noise directly in the frequency-wavenumber domain is not complete and the signal damage is greater

For the time-frequency domain transformation and filtering methods, the calculation efficiency is low and the effect of suppressing noise depends on the time-frequency transformation tool. The resolution and aggregation degree of the time-frequency transformation have a great influence on the filtering

For other two-dimensional wavelet transform filtering methods, the amount of calculation is large, and the resolution of various transforms is not high due to the constraint of the uncertainty principle, so the coefficients of the signal are not focused in the transform domain

Images