CN-121995487-A - Node data random noise suppression method, device and medium based on extrusion transformation

CN121995487ACN 121995487 ACN121995487 ACN 121995487ACN-121995487-A

Abstract

The invention provides a node data random noise suppression method, device and medium based on extrusion transformation, and belongs to the field of oil and gas geophysical exploration. The method comprises the steps of inputting passive source data and original seismic data, extracting random noise from the passive source data, counting the distribution situation of the random noise, simulating the random noise according to the distribution situation of the random noise, counting the distribution situation of the simulated random noise in an extrusion transformation domain, and denoising the original seismic data according to the distribution situation of the random noise in the extrusion transformation domain to obtain denoised seismic data. The method can remove a large amount of random noise in the node data, and provides a basis for processing the subsequent seismic data.

Inventors

WANG XIAOPIN
DONG QIANQIAN

Assignees

中国石油化工股份有限公司
中石化石油物探技术研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20241104

Claims (10)

1. The node data random noise suppression method based on extrusion transformation is characterized by comprising the following steps of: S100, inputting passive source data and original seismic data; s200, extracting random noise from passive source data, and counting the distribution situation of the random noise; s300, simulating random noise according to the distribution situation of the random noise, and counting the distribution situation of the simulated random noise in the extrusion transformation domain; S400, denoising the original seismic data according to the distribution condition of random noise in the extrusion transformation domain, and obtaining denoised seismic data.
2. The method of claim 1, wherein the statistical random noise distribution in S200 comprises: Assuming that random noise is subject to Gaussian distribution and the mean value is 0, establishing data of 5 hours windows respectively recorded as v 1 ,v 2 ,v 3 ,v 4 ,v 5 , sequentially extracting the 5 sample data according to the time sequence of node data respectively, wherein the length of the time window is 100 sampling points, and recording: Where v is the algebraic sum of the random noise of the truncated 5 hour window and v i represents the random noise of the i-th hour window; if |v| <0.001, then the 5 sets of data are considered to be mean 0 and the statistics of the variance of the random noise are performed according to equation (1): Wherein, i represents the ith time window, t represents the sampling point sequence number of the noise signal intercepted by the time window; thus, the distribution of the statistical random noise compliance is: where s is the value of the random noise.
3. The method of claim 2, wherein the step of simulating the random noise according to the distribution of the random noise and counting the distribution of the simulated random noise in the squeeze transformation domain in S300 comprises the steps of: S310, according to the length N of the original seismic data, randomly giving N random numbers, and simulating random noise by adopting a formula (2); s320, converting the simulated random noise into an extrusion transformation domain; s330, the distribution of the random noise in the extrusion transformation domain is statistically simulated.
4. A method according to claim 3, wherein S320 the step of converting the simulated random noise into the squeeze domain comprises: the simulated random noise is converted into the squeeze transformation domain by adopting the formula (3), v s =ST(v) (3) Where v is random noise data, v s is data after random noise extrusion transformation, and ST is an extrusion transformation operator.
5. The method of claim 4, wherein the statistically modeled random noise distribution in the squeeze domain in S330 comprises: assuming that the random noise extrusion transformed data obeys gaussian distribution, the new distribution is noted as: Wherein, the I, j are index numbers of pixel points in the image, M, N are the number of the total sample points in the longitudinal direction and the transverse direction of the image, and v s is data after random noise extrusion transformation.
6. The method of claim 5, wherein the denoising processing is performed on the original seismic data according to the distribution of random noise in the extrusion transform domain in S400 to obtain denoised seismic data, and the specific operations include: step 410, converting the original seismic data into a squeeze conversion domain to obtain squeeze converted seismic data; And 420, calculating the probability of each pixel value belonging to random noise of the seismic data after extrusion transformation by adopting a formula (4), setting the value of the corresponding synchronous extrusion transformation domain to be 0 when the probability is more than or equal to 60%, and returning to a time domain after all pixel operations are completed to obtain the seismic data after denoising.
7. Node data random noise suppression device based on extrusion transformation, characterized by comprising: an input unit for inputting passive source data and raw seismic data; the first statistics unit is used for extracting random noise from the passive source data and counting the distribution situation of the random noise; The second statistical unit is used for simulating the random noise according to the distribution situation of the random noise and counting the distribution situation of the simulated random noise in the extrusion transformation domain; and the noise suppression unit is used for denoising the original seismic data according to the distribution condition of random noise in the extrusion transformation domain to obtain denoised seismic data.
8. The apparatus of claim 7, wherein the second statistical unit comprises: the random noise simulation subunit is used for randomly giving N random numbers according to the length N of the original seismic data and simulating random noise; A conversion subunit for converting the simulated random noise into an extrusion transform domain; And the statistics subunit is used for counting the distribution condition of the simulated random noise in the extrusion transformation domain.
9. The apparatus of claim 8, wherein the conversion subunit is configured to convert the simulated random noise into a squeeze domain, specifically performing the following operations: The simulated random noise is converted to the squeeze domain using the following formula: v s =ST(v) Where v is random noise data, v s is data after random noise extrusion transformation, and ST is an extrusion transformation operator.
10. A computer-readable storage medium storing at least one program executable by a computer, the at least one program when executed by the computer causing the computer to perform the steps in the extrusion transformation-based node material random noise suppression method as recited in any one of claims 1-6.

Description

Node data random noise suppression method, device and medium based on extrusion transformation Technical Field The invention belongs to the field of oil and gas geophysical exploration, and particularly relates to a node data random noise suppression method, device and medium based on extrusion transformation. Background Noise in seismic data has an interference effect on accurate seismic analysis and interpretation, and the conventional seismic data denoising method comprises methods such as mean value filtering, median filtering, wavelet denoising, singular Value Decomposition (SVD) denoising and the like, wherein the mean value filtering is a simple denoising method, noise is reduced by calculating the average value of the neighborhood around a data point by using a sliding window, and the method is suitable for noise with zero mean value such as Gaussian noise. The median filtering is a nonlinear filtering method, the neighborhood around the data point is ordered, the median value is selected as the denoised numerical value, and the median filtering has a good noise removing effect on abnormal values such as salt and pepper noise. The wavelet denoising method utilizes the multi-scale analysis characteristic of wavelet transformation to decompose the signal into sub-bands with different frequencies, then carries out threshold processing on each sub-band, gradually filters low-amplitude noise, and can effectively reduce background noise. Singular Value Decomposition (SVD) denoising is a linear algebraic technique, which can decompose an original seismic data matrix into singular values, left singular vectors and right singular vectors, and can suppress noise and recover main signals of data by retaining larger singular values for reconstruction. The extrusion transformation is a multi-scale analysis method, which can extract the detail information and texture characteristics of the image, and a method for denoising in the synchronous extrusion transformation domain is also studied at present, for example, J.Kang et al propose an image denoising method based on the extrusion transformation and a non-local mean filter, which performs noise removal by converting the image into the extrusion domain and further reduces the noise by combining with the non-local mean filter. Lin et al propose an image denoising method based on extrusion transformation, which uses a multi-scale filter bank based on wavelet transformation to extract a contractible representation of an image, has good performance in noise detection and denoising, and can maintain details and texture characteristics of the image. However, these methods are not practical in many cases, and the noise information is assumed to be subject to some analysis. Disclosure of Invention The invention aims to solve the problems in the prior art and provides a node data random noise suppression method, device and medium based on extrusion transformation. The invention is realized by the following technical scheme: in a first aspect of the present invention, there is provided a node data random noise suppression method based on extrusion transformation, comprising: S100, inputting passive source data and original seismic data; s200, extracting random noise from passive source data, and counting the distribution situation of the random noise; s300, simulating random noise according to the distribution situation of the random noise, and counting the distribution situation of the simulated random noise in the extrusion transformation domain; S400, denoising the original seismic data according to the distribution condition of random noise in the extrusion transformation domain, and obtaining denoised seismic data. The invention further improves that: in S200, the statistics of the random noise distribution includes: Assuming that random noise is subject to Gaussian distribution and the mean value is 0, establishing data of 5 hours windows respectively recorded as v 1,v2,v3,v4,v5, sequentially extracting the 5 sample data according to the time sequence of node data respectively, wherein the length of the time window is 100 sampling points, and recording: Where v is the algebraic sum of the random noise of the truncated 5 hour window and v i represents the random noise of the i-th hour window; if |v| <0.001, then the 5 sets of data are considered to be mean 0 and the statistics of the variance of the random noise are performed according to equation (1): Wherein, i represents the ith time window, t represents the sampling point sequence number of the noise signal intercepted by the time window; thus, the distribution of the statistical random noise compliance is: where s is the value of the random noise. The invention further improves that: in S300, random noise is simulated according to the distribution condition of random noise, and the distribution condition of the simulated random noise in the extrusion transformation domain is counted, and the specific operation c