Threshold Selection And Nise Location In Seismic Image Denoising Based On Curvelet

With the geological condition of seismic prospecting increasingly complicated, the noises arising in the seismic data are more and more serious. The random noise and linear noise are especially common in the seismic data, which severely influence the quality of the seismic signal. Consequently, the attenuation of the noise should be firstly performed before interpretation of seismic data. The noise attenuation of seismic data is the vital step for seismic oil exploration. The quality of noise attenuation and signal reconstruction impose direct influences on follow-up links of seismic prospecting.This paper firstly reviews the development and nature of Curvelet as well as the rapid realization of its discrete algorithm. Compared to wavelet transform (WT), Curvelet transform features better sparseness in seismic signal. Thereby, it is further applicable to noise attenuation of seismic data.The third chapter of the paper mainly introduces attenuating the random noise of seismic signal by Curvelet threshold method. In this chapter, an improved threshold calculation method "Adaptive Scale Threshold" is proposed. Firstly experiments of linear equidirectional axial model and single shot record are performed. The denoising effect of the two models with random noise are compared with traditional threshold method and the threshold method mentioned in this paper. The experimental result shows that better visual effects and higher signal to noise ratio could be obtained by the random noise attenuated with adaptive scale threshold and the minimum loss of signal is realized. According to the follow-up contrast experiments of treatment effects by using median filter and wavelet transform and Curvelet adaptive scale threshold method, they explain the advantageous of superiority of Curvelet adaptive scale threshold method over dealing with random noise. Ultimately, Curvelet adaptive scale threshold is adopted to deal with the seismic data actually. According to the comparison it to the treatment result of data center, it indicates that the new method put forward by the paper has better treatment effect.The fourth chapter of the paper mainly introduces the method of using Curvelet transform to attenuate the linear noise in the seismic signal. This paper puts forward a new method for automatically extracting the linear noise in the seismic signal. Firstly, analyze simulated seismic profile and put forward the method of using OTSU to perform threshold treatment of the squared model of each direction matrix so as to extract the direction matrix with linear interference, and analyze the feasibility of the method through analysis of the attenuation effect. Then attenuate the linear noise in the actual seismic data by using the method put forward on the simulated seismic profile. Finally, comparatively analyze the method put forward by this paper, FK filtering and wavelet transform to explain the advantages of the method put forward by this paper over attenuation of linear noise in the seismic signal.