Shearlet Transform Based On Adaptive Context Model For Seismic Signal Enhancement

Seismic exploration,as a main means of modern geophysical exploration,is also one of the most important methods for exploration of hydrocarbon resources.Along with the increasing demand for oil-gas resources of production and life,the exploitation of abundant unconventional hydrocarbon reservoir has become a topic issue.As an important means of unconventional oil-gas exploration,microseismic monitoring technology makes the collected data have weak effective signal energy and low signal-to-noise ratio because of the special source generation and reception mechanism.Thus,it brings great difficulties to data processing.The Tarim Basin,which is covered by the desert,is the largest oil and gas reserve base in China,also is the most important seismic exploration area currently.On account of the special geographical environment,in the seismic data collected in desert area,the energy of valid signal is weak and the valid signal almost obscured by a large number of low-frequency random noise.Also,there is an overlap frequency band between the noise and effective signal,which is not conducive to the suppression of low frequency noise.In face of the complicated random noise interference,how to improve the signal-to-noise ratio of seismic data is a difficult problem.In order to exploit and utilize the resources of various strata more conveniently,it is necessary to have a higher quality of seismic records.So it is of great significance to suppress random noise and enhance effective seismic signals in seismic data for the exploration of oil-gas resources.Over the years,in the field of seismic exploration and signal processing,domestic and international experts and scholars have made unremitting efforts to suppress random noises in seismic data and enhance effective seismic signals.Shearlet transform,as a new multiscale geometric analysis method,has excellent characteristics in multiscale,multi-direction and anisotropisation.Compared with other multiscale geometric analysis methods,Shearlet transform has more sensitive direction and its sparse representation ability approaches to optimality.The shearlet atom has a simple mathematical structure and a high operation rate.It is easier to realize the discrete form,and it is the only transformation method that can be used to unified process continuous and discrete data in a multiscale domain.Considering the excellent characteristics of Shearlet transform,it can be applied to process seismic data to achieve the purpose of filtering random noise.In the seismic data processing,the Shearlet transform can decompose the seismic data into different scale layers and different directions.The scale layers vary from coarse to fine,and the relative difference of frequency distribution for data is also refined.Therefore,it is possible to make use of the frequency difference between effective signal and noise of seismic data to suppress noise in transform domain.Meanwhile,the Shearlet coefficient has a larger value when the direction of shearlet basis function and seismic signal is consistent,and when the direction difference between them is great,its coefficient is relatively small.Thus,it can also achieve the purpose of separation of signal and noise for seismic data to use the direction difference between them in Shearlet domain.According to the multi-scale and multi-directional characteristics of Shearlet transform,it can accurately extract effective signals and suppress random noise by shrinking the sparse coefficients in a threshold method.In this paper,by studying the innate characteristics of Shearlet transform,we propose a seismic signal enhancement technology.Considering the uniform threshold which is chose by traditional Shearlet transform method will produce excessive cutting of Shearlet coefficient and result in the loss of effective information.It is not conducive to suppress random noise and extract weak effect signal in the low SNR conditions.Accordingly,we propose a Shearlet denoising algorithm based on an adaptive Context model in this paper.The established model can enhance the recognition of effective signals to a certain extent and design the Shearlet coefficients easily.Through the Context partition model,the Shearlet coefficient of each scale in each direction with similar energy will be into the same group.Then,estimating the threshold of coefficients in each energy group and shrinking coefficients,it can achieve enhancement of effective seismic signals and suppression of noise under the condition of low signal-to-noise ratio.In order to verify feasibility and effectiveness of the proposed algorithm,we applied it to surface microseismic data and desert seismic data processing.From the simulation and practical data processing results,it is obviously that the Shearlet denoising algorithm based on adaptive Context model has great advantages in random noise suppression and effective signal amplitude maintenance.