Research On Wavelet Denoising Method For Seismic Image Based On Genetic Algorithm

The seismic material processing for comprehensive geological study provides the high quality of seismic data, so as to meet the geology objectives better. Material processing is the link of data acquisition and data interpretation. Processing quality affects the accuracy of geological objectives, the level of construction, the authenticity of reservoir's prediction and the credibility of fluid's detection. With the formation of earthquake's digital image, the signal is often accompanied with random noise. If noise energy is too much, it will affect the final result image and bring disadvantage effects to explain result. Therefore, there is significance to filter the image noise.The seismic material have three requirements :High signal-to-noise ratio, High resolution, and High-fidelity, Signal-to-noise ratio is the foundation, improving the SNR is the first assignment of seismic material processing. The common filtering method of earthquake image is average filtering, median filtering etc. These methods have strong ability to reduce the noise, but they will make blurring the linear texture of the earthquake image and they can't solve embarrassment between fuzzy edge and the noise removal.Wavelet analysis can satisfy all sorts of denoising requirements, such as lowpass, qualcomm, wave, random noise removal and so on. Comparison with the traditional's noise removal method, it has the incomparable advantages and it is the internationally recognized high-tech of signal and information processing areas. In the application of image denoising, people pay close attention to the method.Wavelet threshold denoising uses the stratified denoising, it has improved the embarrassment between fuzzy edge and the noise removal to a certain extent. Hard threshold and soft thresholding function are two of the most popular thresholding function of the wavelet threshold denoising algorithm, hard thresholding function's discontinuous results in the apparent noise of the denoising signal; Soft thresholding function has good continuity, but there is still constant deviation of estimating wavelet coefficients and noise signal wavelet coefficients.For the faults of traditional's wavelet hard denoising method and soft threshold denoising method, this paper puts forward a method based on genetic algorithm of wavelet threshold denoising. Genetic algorithm is a kind of adaptive algorithms of global optimization probability searching. Genetic algorithm provides general framework to solve complex system optimization problem, it does not depend on the specific domain and has strong robustness to the problem species.The proposed algorithm can make full use of the image itself characteristic and the advantage of multi-scale wavelet transform, combination of global optimization searching characteristics of the genetic algorithm, it uses the genetic algorithm to solve each scales optimal threshold value and makes the threshold value adjust adaptively according to multi-scale wavelet multiresolution characteristics. The algorithm was applied to the seismic interpretation system of oilfield exploration domain and restrained the random noise in the earthquake image, achieved a good effect of image denoising.