Abstract:

To make the solution of the traditional Wiener filtering method with the least mean square criterion more accurate and stable, we proposed a least mean square criterion with regularization idea based on the derivative operator constrains in this paper. Under this criterion, Wiener filter coefficients and its frequency response were derived. The filter result can be controlled by different regularization function values in different frequencies. The analyses of theoretical and practical seismic data show that this method is better than the traditional Wiener filter in the denoising. And the method can reduce the losing of  the energy of effective signal. In view of the foregoing it shows that the regular terms is reasonable and effective. Finally, the expression of the regularization function is built; and its purpose is convenient to compute the regularization function value. And the possibility of improved filtering is discussed by smoothing amplitude spectrum.

Key words: seismic prospecting, random noise, Wiener filtering, regularized method