The Research Of Suppressing Random Noise Via CEEMD And Local Similarity

As an important part of seismic exploration,the selection and optimization of processing methods are the key factors to determine the quality of data after processing.Different processing methods aim at different interference,as for random noise removal method,the current commonly used methods are mainly f-x EMD filtering,median filtering,nonlocal mean filtering,etc.,however,they have their own limitations.In this paper,it proposes a random noise removal method based on the characteristics of random noisec to suppress it,which is the combination of CEEMD and Local Similarity,and introduces a new type of denoising threshold method in the process that overcoming the limitations of the conventional threshold.Conventional random noise removal mainly uses f-x EMD filtering,but the limitations of this method is that it's only applicable to the cases with horizontal events,it's not suitable for the cases with dip or bend events,because while removing noise,this method will also remove the effective signal.Thus a new method based on the Complementary Ensemble Empirical Mode Decomposition(CEEMD)and Local Similarity is proposed for random noise suppressing,and with the help of a new threshold method,it can effectively overcome the disadvantages of traditional methods,which will cause damage to the effective signal in the process of random noise removal.This approach will firstly use CEEMD decomposition for seismic data,and according to the noise level to set an adaptive threshold and apply the new denoising threshold for each IMF,then reconstruct the signal.Conventional processing generally ends like it mentioned above,ignoring the residue of effective signal in removed noise section,this paper based on the above and with help of local Similarity,it retrieve the effective signal from the noise section.Compared with the traditional method,this paper improves the fidelity of the processed data effectively while improving the SNR.