GCV Regularization Wiener Filtering And Its Application On Noise Attenuation For Seismic Data

Noise suppression is the main research topic of modern signal processing. Seismic data are the principal sources of information of Geological prospecting &Oil exploration, However, Artificial collecting Seismic data always under the influence of stochastic noise. The signal to noise ratio (SNR) of Seismic data may reduce when the energy of stochastic noise is too big. Signal to noise ratio is the basis of Seismic data processing, most of the signal processing methods can be very good for suppressing random noise with the condition of high SNR. But when the signal to noise ratio is reduced to a certain extent, most of filtering methods result deterioration. Therefore, the research of suppressing strong stochastic noise filtering of Seismic data can cut down the cost of Seismic prospecting and adapt to the development.Wiener filtering is an optimal linear filter based on Minimum Mean Square Error criterion. It can apply to stationary random process whether continuous or discrete, scalar or vector. Under the condition of high SNR, it is widely used in signal processing, object recognition, image processing, etc. However, when used wiener filtering with strong noise in Seismic data processing, the noise interference causes problem of ill-conditioned matrix when solving the wiener-hoff equation, which makes filtering results appear problems of noise suppression incomplete, amplitude attenuation etc. For the problems above, understanding the regularization method can solve the ill- conditioned matrix problem, we propose GCV regularization wiener filtering method. The objective functions of this new method composed by two parts, one is the objective function of wiener filtering, the other is regularization objective function controlled by regularization factor, regularization objective function is the derivative of the Wiener filtering objective function composition. Theoretical derivation work out filter factor expression, coupled with GCV optimal regularization methods to select regularization factor, so as to find out the optimal filter factor, realize GCV regularization wiener filtering processing. This method can handle strong random noise background Seismic data, achieve noise attenuation.According to the improvement algorithm, we do a lot of simulation experiments. Firstly, we apply GCV regularization wiener filtering to simulate signal processing of different kinds of signal, such as single-axis signal, multi-axis signal, multi-frequency signal and fault signal, then analysis of the results in time-domain and theirs spectrum images. To deal with the above situations, this new method can effectively suppress stochastic noise and maintain signal amplitude good. Comparing experimental results with the results of wiener filtering, it proves that the GCV regularization wiener filtering has a good effect, better adaptability to the Wiener filtering. Finally this method applies to the practical Seismic data processing, the results show that the GCV regularization wiener filtering processing the practical Seismic data can effectively attenuate stochastic noise, improve the output useful signal, increase the resolution and SNR of output signal.