Seismic random noise attenuation via 3D block matching

Abstract

The lack of signal to noise ratio increases the final errors of seismic interpretation. In the present study, we apply a new non-local transform domain method called “3 Dimensional Block Matching (3DBM)” for seismic random noise attenuation. Basically, 3DBM uses the similarities through the data for retrieving the amplitude of signal in a specific point in the f–x domain, and because of this, it is able to preserve discontinuities in the data such as fractures and faults. 3DBM considers each seismic profile as an image and thus it can be applied to both pre-stack and post-stack seismic data. It uses the block matching clustering method to gather similar blocks contained in 2D data into 3D groups in order to enhance the level of correlation in each 3D array. By applying a 2D transform and 1D transform (instead of a 3D transform) on each array, we can effectively attenuate the noise by shrinkage of the transform coefficients. The subsequent inverse 2D transform and inverse 1D transform yield estimates of all matched blocks. Finally, the random noise attenuated data is computed using the weighted average of all block estimates. We applied 3DBM on both synthetic and real pre-stack and post-stack seismic data and compared it with a Curvelet transform based denoising method which is one of the most powerful methods in this area. The results show that 3DBM method eventuates in higher signal to noise ratio, lower execution time and higher visual quality.