Image-guided seismic pre-stack gather denoising method

Abstract

The invention provides an image-guided seismic pre-stack trace gather denoising method. The image-guided seismic pre-stack trace gather denoising method comprises the steps of 1, inputting seismic pre-stack trace gather data needing to be denoised; 2, calculating an amplitude quantile; 3, calculating structure guiding information and a continuity coefficient; step 4, applying bilateral filtering along the feature vector direction corresponding to the maximum feature value; and step 5, outputting the seismic pre-stack gather after image guide denoising. According to the image-guided seismic pre-stack gather denoising method, a filtering algorithm in image processing is introduced into seismic data pre-stack gather processing; the method not only has a good construction-oriented edge-preserving filtering function, but also does not need priori boundary information.

Description

Technical Field

[0001] The invention relates to the technical field of geophysical processing, and in particular to an image-guided seismic pre-stack gather denoising method. Background Art

[0002] Prestack inversion requires dividing the data gather into several parts and stacking the parts, which greatly reduces the signal-to-noise ratio of each part stacking. Some random noise existing in the local offset will be more prominent, resulting in unstable prestack inversion results. At present, the prestack data gather noise is generally removed by smoothing filtering, median filtering or prestack random noise attenuation, but these methods often cause the loss of useful signals, and the boundary information of seismic phase axis is poorly protected, which seriously affects the effect of reservoir prediction.

[0003] Smoothing filtering is the simplest and most basic method for denoising seismic pre-stack gathers. Smoothing filtering divides an image into two parts: the result and the residual. The smoothed result is the large-scale feature (Structure) in the original image, which is called structure or signal in seismic processing. The residual is the small-scale feature (Texture) in the original image, which is called noise or error in seismic processing. The simplest one is rectangular smoothing, which directly averages the surrounding points. The commonly used one is Guassian smoothing, which weighted averages the surrounding points. However, this method not only has no edge preservation function, but also does not have the function of structure guidance.

[0004] In the Chinese patent application with application number: CN202011085246.5, a method for pre-stack well-seismic joint intelligent denoising is involved, including: 1. Data segmentation of the forward modeling data set on the well and forming all the segmented sample data into a sample training data set; 2. Randomly extracting part of the data from the sample training data set, and applying the dictionary learning algorithm of overcomplete dictionary and sparse coefficient iterative learning to obtain an overcomplete dictionary; 3. Performing data segmentation of the same size on the seismic pre-stack data set to obtain a pre-stack data set of predicted sample data; 4. Randomly extracting part of the sample data from the predicted sample data set, using the overcomplete dictionary to obtain the sparse representation coefficient of each sample data, establishing a quadratic target function, and solving to obtain a pre-stack denoised data set; 5. Traversing all data sets according to steps three and four to complete denoising. The invention achieves the goal of denoising by reconstructing the effective signal with amplitude preservation. The pre-stack data set denoising by the algorithm not only improves the signal-to-noise ratio, but also takes into account the AVO trend, which is beneficial to the inversion of pre-stack elastic parameters.

[0005] In the Chinese patent application with application number: CN202010371098.7, a method for amplitude-preserving denoising of prestack gathers based on the characteristics of signal variation with offset is involved. The method for amplitude-preserving denoising of prestack gathers based on the characteristics of signal variation with offset includes: step 1, generating prestack gathers based on the original acquisition records; step 2: reconstructing the signal of the prestack gathers according to the offset; step 3: predicting and removing noise from the prestack gather data; step 4: backfilling the noise to obtain amplitude-preserving denoised prestack gathers. The method for amplitude-preserving denoising of prestack gathers based on the characteristics of signal variation with offset can effectively remove random noise, partial oblique interference and abnormal amplitude values ​​without losing effective signals, avoid the defects of existing methods, complete the optimization correction of prestack gathers, do not change the AVO law of the original gathers, and the AVO law is stronger, which lays a good data foundation for subsequent prestack inversion.

[0006] In the Chinese patent application with application number: CN202011148879.6, a method and device for optimizing the processing of pre-stack gather data along a river in a fluvial reservoir is involved, which provides pre-stack gather data; describes the spatial distribution morphology of the river channel, and applies it to the pre-stack gather data to obtain pre-stack gather data along the river channel; performs denoising on the pre-stack gather data along the river channel to obtain denoised pre-stack gather data along the river channel; corrects the amplitude of the denoised pre-stack gather data along the river channel to obtain corrected gather data; obtains angle gather data by changing the angle of the corrected gather data; and pre-processes the angle gather data to obtain flattened corrected gather data. This invention can improve the accuracy and efficiency of pre-stack seismic inversion of fluvial reservoirs by describing the spatial distribution morphology of the river channel and applying it to the pre-stack gather data to obtain pre-stack gather data along the river channel, and further performs denoising, flattening correction and other steps.

[0007] In the Chinese patent application with application number: CN201210187263.9, a method for denoising, correcting and superimposing CRP gathers is involved, which is characterized by comprising the following steps: a. superimposing the initial CRP gathers to obtain the reference traces, and performing preliminary denoising on the initial CRP gathers, the preliminary denoising including frequency domain filtering, wavelet transform, median filtering or mean filtering; b. setting a sliding time window, and obtaining the correction amount according to the reference traces within the time window, and then correcting the initial CRP gathers according to the correction amount to obtain a first-level CRP gather; c. taking the first-level CRP gathers as the initial CRP gathers, repeating steps a and b in sequence to obtain a second-level CRP gather; d. applying SVD denoising to the second-level CRP gathers to obtain a third-level CRP gather; e. superimposing the data of the third-level CRP gathers to obtain a result profile. The present invention can not only correct the CRP gather event to a horizontal level, thereby greatly improving its signal-to-noise ratio, but also can enhance the continuity of the event of the stacked profile, thereby improving the signal-to-noise ratio and strengthening weak signals.

[0008] The above existing technologies are all significantly different from the present invention and fail to solve the technical problem we want to solve. Therefore, we have invented a new image-guided seismic pre-stack gather denoising method. Summary of the invention

[0009] The object of the present invention is to provide an image-guided seismic pre-stack gather denoising method which has a good structure-oriented edge-preserving filtering function and does not require prior boundary information.

[0010] The object of the present invention can be achieved by the following technical measures: an image-guided seismic pre-stack gather denoising method, the image-guided seismic pre-stack gather denoising method comprising:

[0011] Step 1: Input the seismic pre-stack gather data to be denoised;

[0012] Step 2: Calculate the amplitude quantile;

[0013] Step 3: Calculate the structure-oriented information and continuity coefficient;

[0014] Step 4: Apply bilateral filtering along the direction of the eigenvector corresponding to the maximum eigenvalue;

[0015] Step 5: Output the seismic pre-stack gathers after image-guided denoising.

[0016] The purpose of the present invention can also be achieved by the following technical measures:

[0017] In step 1, the common imaging point seismic pre-stack gather data to be denoised is input.

[0018] The image-guided seismic pre-stack gather denoising method further comprises, after step 1, performing distortion removal of shallow far-offset gathers.

[0019] When performing distortion removal of shallow far-offset gathers, double-pass reflection travel time-offset data pairs are used to form removal lines to perform distortion removal of common imaging point prestack gathers.

[0020] In step 2, the amplitudes of the seismic pre-stack gather data are sorted from small to large.

[0021] In step 3, the structural tensor of the seismic data of the common imaging points is calculated, also called the second-order cumulant matrix, which is an expression of the partial derivatives of the matrix. In the field of image processing research, this method is mainly used for gradient calculation, edge detection or information similarity analysis.

[0022] In step 3, the structure tensor is expressed as formula (1):

[0023]

[0024] Among them, the matrix element I x and I y are the partial derivatives of seismic data in the horizontal and vertical directions respectively.

[0025] In step 3, the structure tensor matrix S is subjected to eigenvalue decomposition, and two eigenvalues ​​and eigenvectors can be obtained respectively. The eigenvectors represent the direction of the spatial structure of the seismic data, and the eigenvalues ​​indicate the magnitude of the gradient structure along the direction of their eigenvectors. The combination of eigenvalues ​​and eigenvectors can fully describe the spatial structural information of the seismic data. The eigenvalues ​​and eigenvectors of the structure tensor matrix S are calculated, and the maximum eigenvalue corresponds to the direction of the event axis.

[0026] In step 3, the structural tensor is calculated from the seismic pre-stack gather data, which can be simply understood as finding the directional derivatives of the three-dimensional pre-stack gather data body. The partial derivatives in each direction form a matrix, which is decomposed by eigenvalues ​​to obtain the continuity coefficient. The continuity coefficient is added to the bilateral filtering algorithm in step 4 to play a directional guiding role.

[0027] In step 4, bilateral filtering is an image algorithm that can smooth image noise while protecting the boundary information of image mutations.

[0028] In step 4, the filter composition of bilateral filtering is shown in formula (2):

[0029]

[0030] Where: Wp is the energy normalization factor; is a spatial Gaussian smoothing function; is a step function of the change of spatial amplitude value, indicating the spatial mutation information; q and p are weighting coefficients in the horizontal and vertical directions of space respectively; I q and I p are the partial derivatives in the horizontal and vertical directions of space respectively; BF is the filter.

[0031] In step 5, bilateral filtering is applied to output the image-guided denoised seismic pre-stack gathers. The original common imaging point gathers are subjected to bilateral filtering along the eigenvector direction corresponding to the maximum eigenvalue in step 4 to obtain the denoised common imaging point seismic pre-stack gathers.

[0032] Compared with the prior art, the image-guided seismic pre-stack gather denoising method of the present invention has the following advantages and positive effects:

[0033] The present invention provides an image-guided seismic pre-stack gather denoising method, which has a good structural-oriented edge-preserving filtering function and does not require a priori boundary information, so as to achieve the purpose of removing random noise in pre-stack gathers with amplitude preservation. The combined eigenvalue and eigenvector can fully describe the spatial structural information of seismic data, and the structural information can guide bilateral filtering to smooth seismic data. This smoothing is completely driven by pre-stack gather data, and the seismic pre-stack gather data is processed adaptively with good amplitude preservation. The image-guided seismic pre-stack gather denoising method is a brand-new seismic-oriented edge-preserving denoising method, which introduces the filtering algorithm in image processing into the processing of pre-stack gathers of seismic data; the method not only has a good structural-oriented edge-preserving filtering function, but also does not require a priori boundary information. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 A flowchart of a specific embodiment of the image-guided seismic pre-stack gather denoising method of the present invention;

[0035] Figure 2 It is a schematic diagram of the application effect of the seismic pre-stack gather denoising method guided by shallow noise image in the seismic pre-stack gather in an embodiment of the present invention;

[0036] Figure 3 Schematic diagram of a seismic pre-stack gather that needs to be denoised for mid-deep noise in an embodiment of the present invention;

[0037] Figure 4 Schematic diagram of seismic pre-stack gathers after the image-guided denoising method for mid-deep noise in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0039] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations and / or combinations thereof.

[0040] like Figure 1 As shown, Figure 1 The flowchart of the image-guided seismic pre-stack gather denoising method of the present invention is as follows. The image-guided seismic pre-stack gather denoising method comprises:

[0041] Step S1: inputting the seismic pre-stack gather data to be denoised;

[0042] Step S2: performing distortion removal of shallow far-offset gathers;

[0043] Step S3: Calculate the amplitude quantile;

[0044] Step S4: calculating structure orientation information and continuity coefficient;

[0045] Step S5: applying bilateral filtering along the direction of the eigenvector corresponding to the maximum eigenvalue;

[0046] Step S6: Output the seismic pre-stack gathers after image-guided denoising.

[0047] This method introduces the filtering algorithm in image processing into seismic data processing. It not only has a good structure-oriented edge-preserving filtering function, but also does not require prior boundary information.

[0048] The following are several specific embodiments of the present invention.

[0049] Example 1

[0050] In a specific embodiment 1 of the present invention, the image-guided seismic pre-stack gather denoising method includes the following steps:

[0051] Step S1: inputting the seismic pre-stack gather data to be denoised: that is, inputting the common imaging point seismic pre-stack gather data to be denoised.

[0052] Step S2: Perform distortion removal of shallow far-offset gathers; Performing distortion removal of shallow far-offset gathers is beneficial to improving the denoising effect. Shallow far-offset gathers generally have distortion. Double-pass reflection travel time-offset data pairs are used to form a removal line to perform distortion removal of common imaging point pre-stack gathers.

[0053] Step S3: Calculate the amplitude quantile: Sort the amplitudes of the seismic pre-stack gather data from small to large, such as the ninth quantile is the amplitude at the 90% position.

[0054] Step S4: Calculating structure-oriented information and continuity coefficient. Calculating structure-oriented information and continuity coefficient specifically includes:

[0055] The structure tensor of the seismic gather data of the common imaging point is also called the second-order cumulant matrix, which is the expression of the partial derivative of the matrix. In the research field of image processing, this method is mainly used for gradient calculation, edge detection or information similarity analysis. The structure tensor is expressed as formula (1):

[0056]

[0057] Among them, the matrix element I x and I y are the partial derivatives of seismic data in the horizontal and vertical directions respectively.

[0058] By performing eigenvalue decomposition on the structure tensor matrix S, two eigenvalues ​​and eigenvectors can be obtained respectively. The eigenvectors represent the direction of the spatial structure of the seismic data, and the eigenvalues ​​indicate the size of the gradient structure along the direction of their eigenvectors.

[0059] The combination of eigenvalues ​​and eigenvectors can fully describe the spatial structure information of seismic data. The eigenvalues ​​and eigenvectors of the structure tensor matrix S are calculated, and the maximum eigenvalue corresponds to the direction of the event axis.

[0060] Calculating the structural tensor from the seismic pre-stack gather data can be simply understood as finding the directional derivative of the three-dimensional pre-stack gather data body. The partial derivatives in each direction form a matrix, which is decomposed into eigenvalues ​​to obtain the continuity coefficient. The continuity coefficient is added to the bilateral filtering algorithm in step S5 to play a directional guiding role.

[0061] Step S5: Apply bilateral filtering along the direction of the eigenvector corresponding to the maximum eigenvalue. Bilateral filtering specifically includes:

[0062] Bilateral filtering is an image algorithm that can smooth image noise while protecting boundary information such as image mutations. The filter composition is shown in formula (2):

[0063]

[0064] Where: Wp is the energy normalization factor; is a spatial Gaussian smoothing function; is a step function of the change of spatial amplitude value, indicating the spatial mutation information; q and p are weighting coefficients in the horizontal and vertical directions of space respectively; I q and I p are the partial derivatives in the horizontal and vertical directions of space respectively; BF is the filter.

[0065] Step S6: Apply bilateral filtering to output the seismic pre-stack gathers after image-guided denoising. This method has a good structure-oriented edge-preserving filtering function and does not require prior boundary information, so as to achieve the purpose of removing random noise in pre-stack gathers with amplitude preservation.

[0066] The original common imaging point gathers are subjected to bilateral filtering along the eigenvector direction corresponding to the maximum eigenvalue in step 5 to obtain the denoised common imaging point seismic pre-stack gathers.

[0067] This image-guided seismic prestack gather denoising method has good structure-oriented edge-preserving filtering function, and does not require prior boundary information, so as to achieve the purpose of removing random noise in prestack gathers with amplitude preservation. The combined eigenvalue and eigenvector can fully describe the spatial structure information of seismic data. With the structural information, bilateral filtering can be guided to smooth seismic data. This smoothing is completely driven by prestack gather data, and it processes seismic prestack gather data adaptively with good amplitude preservation.

[0068] Example 2

[0069] In the specific embodiment 2 of the present invention, Figure 2 This is a schematic diagram of the application effect of the image-guided seismic prestack gather denoising method provided by the present invention for shallow noise in seismic prestack gathers. The pictures are before denoising, after denoising and the removed noise. The method has a good structural guidance edge-preserving filtering function for shallow random noise in seismic prestack gathers. The removed noise has no effective signal and has good amplitude preservation.

[0070] Example 3

[0071] In specific embodiment 3 of the present invention, Figure 3 It is a seismic pre-stack gather that needs to be denoised. There is residual random noise in the middle and deep layers of the seismic profile. Figure 4 It is a seismic pre-stack gather after image-guided denoising. This method has better structure-oriented edge-preserving filtering function for the mid-deep random noise in seismic pre-stack gathers, and does not require prior boundary information, so as to achieve the purpose of amplitude-preserving removal of random noise in pre-stack gathers.

[0072] The beneficial effects of the present invention are as follows: first, the present invention is a brand-new seismic-oriented edge-preserving denoising method, which introduces the filtering algorithm in image processing into the pre-stack gather processing of seismic data; second, the present invention not only has a good structural-oriented edge-preserving filtering function, but also does not require a priori boundary information.

[0073] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions recorded in the aforementioned embodiments or replace some of the technical features therein with equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0074] Except for the technical features described in the specification, all other technical features are known technologies to those skilled in the art.

Claims

1. An image-guided seismic pre-stack gather denoising method, characterized in that: The image-guided seismic pre-stack gather denoising method includes: Step 1: Input the seismic pre-stack gather data to be denoised; Step 2: Calculate the amplitude quantile; Step 3: Calculate the structure-oriented information and continuity coefficient; Step 4: Apply bilateral filtering along the direction of the eigenvector corresponding to the maximum eigenvalue; Step 5: Output the seismic pre-stack gathers after image-guided denoising.

2. The image-guided seismic pre-stack gather denoising method according to claim 1, characterized in that: In step 1, the common imaging point seismic pre-stack gather data to be denoised is input.

3. The image-guided seismic pre-stack gather denoising method according to claim 1, characterized in that: The image-guided seismic pre-stack gather denoising method further comprises, after step 1, performing distortion removal of shallow far-offset gathers.

4. The image-guided seismic pre-stack gather denoising method according to claim 3, characterized in that: When performing distortion removal of shallow far-offset gathers, double-pass reflection travel time-offset data pairs are used to form removal lines to perform distortion removal of common imaging point prestack gathers.

5. The image-guided seismic pre-stack gather denoising method according to claim 1, characterized in that: In step 2, the amplitudes of the seismic pre-stack gather data are sorted from small to large.

6. The image-guided seismic pre-stack gather denoising method according to claim 1, characterized in that: In step 3, the structural tensor of the seismic data of the common imaging points is calculated, also called the second-order cumulant matrix, which is an expression of the partial derivatives of the matrix. In the field of image processing research, this method is mainly used for gradient calculation, edge detection or information similarity analysis.

7. The image-guided seismic pre-stack gather denoising method according to claim 6, characterized in that: In step 3, the structure tensor is expressed as formula (1): Among them, the matrix element I x and I y are the partial derivatives of seismic data in the horizontal and vertical directions respectively.

8. The image-guided seismic pre-stack gather denoising method according to claim 7, characterized in that: In step 3, the structure tensor matrix S is subjected to eigenvalue decomposition, and two eigenvalues ​​and eigenvectors can be obtained respectively. The eigenvectors represent the direction of the spatial structure of the seismic data, and the eigenvalues ​​indicate the magnitude of the gradient structure along the direction of their eigenvectors. The combination of eigenvalues ​​and eigenvectors can fully describe the spatial structural information of the seismic data. The eigenvalues ​​and eigenvectors of the structure tensor matrix S are calculated, and the maximum eigenvalue corresponds to the direction of the event axis.

9. The image-guided seismic pre-stack gather denoising method according to claim 8, characterized in that: In step 3, the structural tensor is calculated from the seismic pre-stack gather data, which can be simply understood as finding the directional derivatives of the three-dimensional pre-stack gather data body. The partial derivatives in each direction form a matrix, which is decomposed by eigenvalues ​​to obtain the continuity coefficient. The continuity coefficient is added to the bilateral filtering algorithm in step 4 to play a directional guiding role.

10. The image-guided seismic pre-stack gather denoising method according to claim 1, characterized in that: In step 4, bilateral filtering is an image algorithm that can smooth image noise while protecting the boundary information of image mutations.

11. The image-guided seismic pre-stack gather denoising method according to claim 10, characterized in that: In step 4, the filter composition of bilateral filtering is shown in formula (2): Where: Wp is the energy normalization factor; is a spatial Gaussian smoothing function; is a step function of the change of spatial amplitude value, indicating the spatial mutation information; q and p are weighting coefficients in the horizontal and vertical directions of space respectively; I q and I p are the partial derivatives in the horizontal and vertical directions of space respectively; BF is the filter.

12. The image-guided seismic pre-stack gather denoising method according to claim 1, characterized in that: In step 5, bilateral filtering is applied to output the image-guided denoised seismic pre-stack gathers. The original common imaging point gathers are subjected to bilateral filtering along the eigenvector direction corresponding to the maximum eigenvalue in step 4 to obtain the denoised common imaging point seismic pre-stack gathers.

Images