Research On Noise Suppression Of Seismic Data Based On Curvelet Transform

In the process of seismic signal processing,noise suppression can effectively improve the signal-to-noise ratio of seismic data,but it often distort the original data.In order to improve the contradiction between signal-to-noise ratio and signal fidelity in the field of denoising,it is urgent to research and explore more effective denoising methods.This paper studies the attenuation of seismic data noise based on the Curvelet transformation theory.An adaptive local threshold denoising method is proposed to overcome the shortcomings of the conventional soft-hard thresholding method based on Curvelet transform for overshooting the effective signal.An improved method is proposed to solve the problem of the existence of the loss of the effective signal in the separation of surface waves by the conventional Curvelet transform.It determines the conventional method as the Curvelet coefficient in the direction of the surface wave to perform threshold processing to extract the effective reflected wave signal mixed with the surface wave.A joint denoising technique has been proposed for the suppression of random noise in the conventional meandering threshold method and the phenomenon of partial artifact noise suppression—combined denoising based on Curvelet transform and Two-dimensional Empirical Mode Decomposition(BEMD).First,by testing the theoretical model and the actual data,the traditional global threshold is compared with the adaptive local threshold method to suppress the quality of random noise.Using the improved surface wave separation method to process the noise attenuation of the theoretical model and the actual data,and compare it with the result of the traditional Curvelet transform to separate surface waves.The single Curvelet transform method and the combined denoising method proposed in this paper were used to denoise the theoretical model and the actual seismic record respectively in order to compare and analyze their respective denoising effects.Experimental results show: The adaptive local threshold method proposed in this paper can strengthen the protection of weak signals in seismic data and achieve the purpose of amplitude reduction and denoising of seismic data.The improved surface wave separation method in this paper can retain more effective signals than before the improvement,and the separation of surface waves is more thorough.The joint method proposed in this paper combines the advantages of BEMD and Curvelet transform.It can optimize the conventional Curvelet threshold denoising method and has more advantages than a single Curvelet threshold method,regardless of visual effects,signal-to-noise ratio and fidelity.