The Research Of Seismic Random Noise Attenuation Based On Hessian Matrix

With the deepening of oil and gas exploration and development,the requirements of seismic data acquisition and processing technology at home and abroad are also increasing.It is important to remove random noise,improve signal-to-noise ratio and extract weak signal from strong noise background for accurate imaging of complex structures.Considering that traditional denoising methods can not achieve ideal denoising results in complex areas,this paper introduces the scale space theory and proposes a multi-scale random noise suppression method based on Hessian matrix.Hessian matrix is widely used in image processing,but there is no precedent in seismic data processing.In this paper,the application of Hessian matrix in suppressing random noise of 2-D and 3-D seismic data is studied.In the two-dimensional data,the effective signals with different frequency components are regarded as combinations of different curvature"curves" of multiple scales,so that the signal-noise separation can be realized by using the good performance of Hessian matrix in curve detection.Furthermore,the algorithm is extended to three-dimensional seismic data by utilizing the spatiotemporal continuity of effective signals in three-dimensional data volume,and the random noise suppression of three-dimensional seismic data is realized.Compared with the traditional method,this method is not limited by the dip angle of strata,so it can process seismic data in complex areas and keep the boundary of structures better.Two-dimensional and three-dimensional models and real data are used to validate the proposed method and compare it with the traditional F-X deconvolution method and the denoising results of commercial software.The results show that the multi-scale random noise suppression method based on Hessian matrix can achieve good denoising results for two-dimensional pre-stack and post-stack data and three-dimensional data volume,and can suppress random noise at the same time.Enough to maintain the integrity of effective signals,taking into account the signal-to-noise ratio and fidelity of seismic data.