Methods Of Random Noise Removing In Seismic Exploration Data

Signal processing technique of seismic exploration developing for a long time until today, the important task of the technique is focused on the improving resolution ratio of seismic exploration. The noise with useful signal is an essential factor in affecting precision of seismic signal processing, as long as the development of seismic exploration, the target of exploration is increasingly focused on the deep area. But, because the target layer is deeper, seismic wave from hypocenter of ground layer by transmit and dispersion in the medium for a long way, and the attenuation of the seismic signal by the inelasticity effect of the medium, the energy of reflected signal becomes weaker. In the random noise background, the reflection seismic signal sometimes is hit and miss and sometimes is submerged. When the reflection seismic signal is submerged in the random noise, the reflection event is difficult to be tracked, exploration task is hard. So on the base of keeping the resolution ratio, how to remove the noise and get the useful signal from strong noise, is a problem attended by people.Fist of all; explain the noise of the seismic data. Broadly speaking, seismic signal noise is divided into two parts, random noise and coherent noise. Different noise has different character. Different character noise should be processed by different denoising method. Removing random noise is the key content of this paper. Because of the complexity of the ground layer, lots of random noise contained in seismic data such as slight shock and background interference. The noise distributes abroad, strongly impact the signal noise ratio of the seismic data. In the processing of the seismic data, generally, methods of removing random noise is as follows, polynomial fitting, one dimension filter, f - x predictive filter,τ-p transform, K-L transform and so on, each method has the effect, but has itsown localization. Once the signal noise ratio of the seismic data is low, existing method hardly possible pick valid signal. If this problem is solved, it will stimulate the development of productivity and bring large economic benefit. In order to remove the random noise of the seismic data, this paper proposes two solution methods.In denoising domain, wavelet gains much attention because of its favorable time-frequency character, and initiates the method of nonlinear denoising. In this paper, we introduce the origin of wavelet, describe the basic theory of wavelet transform, and explain the concept, the idea and the conclusion of wavelet transform. Including continue wavelet transform theory, dyadic wavelet, wavelet frame theory, multi-resolutions analysis and so on. Introducing several common denoising method of wavelet transform, for example, wavelet shrinkage arithmetic, correlation arithmetic, projection arithmetic, with analysis and comparing.We study the scheme based on singular value decomposition and wavelet threshold method. Explaining the basic theory of singular value decomposition and improving it. The improved scheme using the chaotic oscillator detect technology, detecting the position and the speed of the event which submerged in the noise, factitious moving the event, gaining the best relativity, and then doing SVD, to remove the noise. Algebra features of characters can be effectively extracted by singular value decomposition, the wavelet transform is kinds of denoising methods on finite time and frequency, the two methods have their own excellence in denoising. We combine the two methods, in order to remove the random noise in the seismic exploration data. The simulation result indicates the algorithm is effect even for the low SNR data. The signal can be shown clearly with hardly any noise. So the arithmetic based on singular value decomposition and wavelet threshold method which processing random noise of seismic exploration data is feasible and has research value.We study the hyperbolic filter algorithm. In the former multi-channel wiener filter algorithm, how to satisfy the desired output is a problem hardly to resolve. For this reason, in the denoising of seismic data, multi-channel wiener filter algorithm hardly achieves ideal effect. In this paper, we use chaotic oscillator detect technology, sufficiently resolve the problem. The detail method is as follows: using the selected chaotic oscillator, we successfully detect the position and the speed of the event which submerged in the noise; as follows, we use the position and the speed as the desired output of the multi-channel wiener filter algorithm. We combine the two techniques, named hyperbolic filter. The simulation demonstrates that this method of signal enhancement could clearly show the position of events and the Ricker wavelets submerged in the noise are resumed well. We compared seismic data corrupted in noise with the filtering result from the wavelets and Fourier spectrum. It was concluded that removing random noise of seismic data by hyperbolic filter is feasible. At the same time, comparing with the one channel wiener filter algorithm and theτ?p transform, hyperbolic filter is better.