Seismic data denoising method and device, storage medium, electronic equipment and product
By combining the transform learning method of group sparse constraints and the dictionary learning method of near-end gradients, the problem of complex and low accuracy of the denoising algorithm in seismic data processing is solved, and efficient and accurate seismic data denoising is achieved.
Patent Information
- Application Number
- CN202510130430.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-05
- Publication Date
- 2025-05-13
AI Technical Summary
In the process of seismic exploration data processing, the existing denoising method algorithm is complex and has low accuracy, making it difficult to effectively remove background noise in complex seismic targets, resulting in difficulty in attenuating and identifying seismic wave signals.
The denoising method is adopted, which combines the transform learning method based on group sparse constraints and the dictionary learning method of near-end gradients, and the denoised data is obtained by noise suppressing and restoring the seismic data.
It reduces the complexity of the denoising algorithm, improves the denoising efficiency and accuracy of seismic data, and can effectively remove noise and retain the energy of seismic signals.
Smart Images
Figure CN119986805A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of data processing technology, and in particular, to a method, device, storage medium, electronic equipment and product for denoising seismic data. Background Art
[0002] In the process of seismic exploration data processing, a more important goal is to improve the signal-to-noise ratio of coal three-dimensional seismic data. In the prior art, it is generally done by identifying redundant data in seismic data and then using the data's own structural characteristics to perform denoising. However, with the continuous deepening of exploration and development, the seismic targets being explored are more complex, and various factors such as external conditions, construction factors, and instruments produce different background noises, resulting in the attenuation of seismic transmission signals during the propagation of seismic waves, which results in the relatively weak energy of the reflected signal received by the surface detector. For the identification of weak signals, the denoising methods in the prior art have high algorithm complexity and low accuracy. Summary of the invention
[0003] The purpose of the present disclosure is to provide a seismic data denoising method, device, storage medium, electronic device and product to solve the above technical problems.
[0004] In order to achieve the above object, according to a first aspect of an embodiment of the present disclosure, a method for denoising seismic data is provided, the method comprising:
[0005] Acquiring first seismic data requiring noise suppression;
[0006] Performing noise suppression on the first seismic data by a transformation learning method based on group sparsity constraints, and applying a dictionary learning method based on proximal gradients to the noise suppression to obtain second seismic data after noise suppression;
[0007] Perform seismic data recovery on the second seismic data to obtain third seismic data; wherein the third seismic data includes a denoising result, and one or more of removed noise data and energy loss data.
[0008] Optionally, the performing noise suppression on the first seismic data by using a transformation learning method based on group sparsity constraints comprises:
[0009] Acquire a target training sample required by the transformation learning method according to the first seismic data; the target training sample includes a plurality of local blocks;
[0010] The multiple local blocks in the target training sample are sparsely represented in the same transform domain.
[0011] Optionally, the step of acquiring target training samples required by the transformation learning method according to the first seismic data includes:
[0012] Acquire original training samples in the first seismic data;
[0013] A block extraction operator is used to extract local blocks in the original training sample in an overlapping manner, and vectorization is performed, and the result is used as a column vector of the target training sample.
[0014] Optionally, the sparsely representing the multiple local blocks in the target training sample in the same transform domain includes:
[0015] Obtaining multiple training sets by applying a grouping operator to the target training samples;
[0016] In the same transform domain, the sparse coding of each set of data in the training set is subject to an infinite zero norm regularization constraint.
[0017] Optionally, applying the proximal gradient-based dictionary learning method to the noise suppression comprises:
[0018] Obtaining a dictionary matrix and a sparse representation vector in the noise suppression by a dictionary learning method based on proximal gradient;
[0019] The sparsely approximated seismic data is represented by a setting formula as the second seismic data.
[0020] Optionally, the setting formula includes:
[0021]
[0022] Where M is the two-dimensional seismic data; the matrix D represents the dictionary obtained through dictionary learning; y i,j represents the (i,j)th local block in the 2D seismic data; A block extraction operator for the (i, j)th data in the two-dimensional seismic data; Extract operators for blocks P is a set of data blocks, which includes all pixel points of the two-dimensional seismic data; τ p is a diagonal matrix; Dx i,j is the sparse representation vector at position (i, j) in the two-dimensional data.
[0023] Optionally, performing seismic data recovery on the second seismic data includes:
[0024] Recovering different sets of data blocks in the second seismic data in a non-overlapping block manner to obtain multiple data recovery results;
[0025] Obtaining an average value of the multiple data recovery results;
[0026] The average value is used as the third seismic data.
[0027] According to a second aspect of an embodiment of the present disclosure, a device for denoising seismic data is provided, the device comprising:
[0028] An acquisition module, used for acquiring first seismic data requiring noise suppression;
[0029] A noise suppression module, used for performing noise suppression on the first seismic data by a transformation learning method based on group sparsity constraints, and applying a dictionary learning method based on proximal gradient to the noise suppression process to obtain second seismic data after noise suppression;
[0030] A data recovery module is used to perform seismic data recovery on the second seismic data to obtain third seismic data; wherein the third seismic data includes a denoising result and one or more of removed noise data and energy loss data.
[0031] According to a third aspect of an embodiment of the present disclosure, a non-temporary computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in any one of the first aspect of the embodiment of the present disclosure are implemented.
[0032] According to a fourth aspect of an embodiment of the present disclosure, there is provided an electronic device, including:
[0033] a memory having a computer program stored thereon;
[0034] A processor is used to execute the computer program in the memory to implement the steps of any one of the methods described in the first aspect of the embodiments of the present disclosure.
[0035] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps of any one of the methods described in the first aspect of an embodiment of the present disclosure.
[0036] In the above technical solution, first seismic data that needs to be noise suppressed is obtained, noise is suppressed on the first seismic data by a transformation learning method based on group sparsity constraints, and a dictionary learning method based on proximal gradient is applied to the noise suppression to obtain second seismic data after noise suppression, and seismic data recovery is performed on the second seismic data to obtain third seismic data, wherein the third seismic data includes a denoising result, and one or more of the removed noise data and energy loss data. Through the above technical solution, the transformation learning method based on group sparsity constraints and the dictionary learning method based on proximal gradient are combined and applied to the denoising process of seismic data, and the transformation learning method and the dictionary learning method reduce the complexity of the algorithm, and the dictionary learning process has good convergence, thereby improving the denoising efficiency and accuracy of seismic data.
[0037] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings:
[0039] Figure 1 is a flow chart of a method for denoising seismic data according to an exemplary embodiment;
[0040] Figure 2 is a flow chart of a method for denoising seismic data according to an exemplary embodiment;
[0041] Figure 3 is a flow chart of a method for denoising seismic data according to an exemplary embodiment;
[0042] Figure 4 is a flow chart of a method for denoising seismic data according to an exemplary embodiment;
[0043] Figure 5 is a schematic diagram of seismic data after denoising by a dictionary learning method based on proximal gradient according to an exemplary embodiment;
[0044] Figure 6 is a schematic diagram of noise data removed by a dictionary learning method based on proximal gradient according to an exemplary embodiment;
[0045] Figure 7 is a schematic diagram of post-stack seismic data according to an exemplary embodiment;
[0046] Figure 8is a schematic diagram of seismic data after denoising by a transformation learning method based on group sparsity constraints according to an exemplary embodiment;
[0047] Fig. 9 is a schematic diagram of noise data removed by a transformation learning method based on group sparsity constraints according to an exemplary embodiment;
[0048] Fig.10 is a schematic diagram of seismic data after denoising by a dictionary learning method based on proximal gradient according to an exemplary embodiment;
[0049] Fig.11 is a schematic diagram of noise data removed by a dictionary learning method based on proximal gradient according to an exemplary embodiment;
[0050] Fig.12 is a block diagram of a seismic data denoising device according to an exemplary embodiment;
[0051] Fig.13 is a block diagram of an electronic device 1300 according to an exemplary embodiment. DETAILED DESCRIPTION
[0052] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.
[0053] It should be noted that all actions of acquiring signals, information or data in the present disclosure are carried out in compliance with the relevant data protection laws and policies of the country where the device is located and with the authorization given by the owner of the corresponding device.
[0054] Figure 1 is a flow chart of a method for denoising seismic data according to an exemplary embodiment. Figure 1 As shown, the method includes the following steps.
[0055] Step S11, obtaining the first seismic data that needs to be noise suppressed.
[0056] It is understandable that in the process of seismic exploration data acquisition, there are unconventional geological bodies that are deeply buried and small in scale. Such situations cause the seismic waves excited by the surface source to propagate and scatter in the medium for a long time, and the inelastic effect of the medium causes the attenuation of the seismic reflection signal. The seismic signal received on the ground contains a variety of disturbances. Among these disturbances, the disturbance that hinders the identification and tracking of effective waves is called noise. Noise will affect the processing and analysis of later seismic data. Correspondingly, the detection of weak signals of deep seismic will also be greatly disturbed. During the propagation process, the seismic signal is affected by the interference of near-surface noise and the absorption attenuation of deep strata, causing the energy to weaken and the high-frequency component to be severely attenuated, resulting in a small reception and identification reflection coefficient in the process of high-density seismic exploration, as well as the problem of high-frequency weak reflection signals. Therefore, in order to obtain seismic signals with higher accuracy, it is necessary to suppress the noise of seismic data.
[0057] Step S12, noise suppression is performed on the first seismic data by a transformation learning method based on group sparsity constraints, and a dictionary learning method based on proximal gradient is applied to the noise suppression to obtain second seismic data after noise suppression.
[0058] Among them, group sparsity refers to the property that there is local similarity between local weak features of seismic data, and the collection of highly similar blocks has similar sparse representation patterns in the same transform domain. The transformation learning method based on group sparsity constraints is to jointly extract local sparsity and self-similarity within the block through a unified mathematical model, classify and transform the seismic data, and apply it to seismic data noise removal; solving the dictionary learning problem through the proximal gradient method is to use a non-overlapping blocking strategy in seismic data noise suppression after training the dictionary. The dictionary learning method based on the proximal gradient can use the extrapolation method to alternately update the dictionary and the sparse representation system, and the dictionary training efficiency is relatively high.
[0059] Step S13, performing seismic data recovery on the second seismic data to obtain third seismic data; wherein the third seismic data includes a denoising result, and one or more of removed noise data and energy loss data.
[0060] Among them, seismic data recovery adopts non-overlapping block partitioning. Compared with the overlapping block partitioning method, non-overlapping block partitioning processes each block independently, can be calculated in parallel, and does not require additional calculations to process the overlapping parts between blocks, thereby reducing the calculation complexity and improving the data processing speed.
[0061] Figure 2is a flowchart of a method for denoising seismic data according to an exemplary embodiment. The noise suppression of the first seismic data by a transformation learning method based on group sparsity constraints described in step S12 may include the following steps:
[0062] Step S121, obtaining a target training sample required by the transformation learning method according to the first seismic data; the target training sample includes a plurality of local blocks.
[0063] For example, the method for obtaining the target training sample may include: obtaining the original training sample in the first seismic data, extracting overlapping local blocks in the original training sample using a block extraction operator, and vectorizing the result as a column vector of the target training sample.
[0064] Step S122: sparsely represent the multiple local blocks in the target training sample in the same transform domain.
[0065] Optionally, step S122 may include:
[0066] (1) obtaining multiple training sets by applying a grouping operator to the target training sample;
[0067] (2) In the same transform domain, the sparse coding of each set of data in the training set is subject to an infinite zero norm regularization constraint.
[0068] For example, the target training sample can be represented by z, and the grouping operator can be represented by V i , the grouping operator V i Acting on the target training sample z, the vector set in z is rearranged to form multiple training sets G i , where the training set G i It can be expressed as
[0069] It is understandable that group sparsity can be defined as Among them, X(j,:) represents the jth row in the matrix X. Since l ∞,0 The norm returns the maximum absolute value of all elements in each row, and the l0 norm counts the number of non-zero elements. ∞,0 The norm returns the number of nonzero rows in the matrix X, so l ∞,0 The norm can be used as a constraint term in the mathematical model of group sparsity.
[0070] Optionally, in the same transform domain, the mathematical optimization model in the transform learning method based on group sparsity constraints can be:
[0071]
[0072] in,‖·‖F represents the Frobenius norm of the matrix, λ is the regularization parameter, G i Represents multiple training sets, which are sets of similar blocks extracted from z; W i represents the transformation matrix, X i Represents a sparse matrix. After grouping, in the same transformation domain, the sparse matrix of each group of data is Add infinite zero norm l ∞,0 The regularized constraint, where the sparse matrix Each column has the same support set, which is expressed as supp(x) = {j:x j ≠0}, where x represents the matrix X i A column of j Represents the j-th element of column x.
[0073] For example, in a possible implementation, the seismic data denoising method may include the following steps:
[0074] (1) Input the noise observation data y, the noise standard deviation δ, the constant τ, the number of iterations L and the initial transformation matrix W 0 .
[0075] (2) loop over the seismic data set, for i = 1:N;
[0076] renew W i =W 0 , δ0=δ;
[0077] Loop over the data set, forl=1:L.
[0078] By formula Update X i ,in and c are empirical parameters;
[0079] By formula Update W i , and re-estimate G i , in is the update matrix;
[0080] like The loop is interrupted; otherwise, the noise standard deviation δ is updated;
[0081] Enter the loop again until the condition is met and the loop ends.
[0082] (3) Through the formula Update the estimated data after denoising And output the denoised data
[0083] in, V i The adjoint operator, V i Used to store the position information of each data block in the two-dimensional data. is a diagonal matrix, its inverse matrix Easier to obtain.
[0084] Figure 3 is a flow chart of a method for denoising seismic data according to an exemplary embodiment. Figure 3 As shown, the step S12 of applying the dictionary learning method based on the proximal gradient to the noise suppression may include the following steps:
[0085] Step S123, obtaining a dictionary matrix and a sparse representation vector in the noise suppression by using a dictionary learning method based on proximal gradient;
[0086] Step S124, using a setting formula to represent the sparsely approximated seismic data as the second seismic data.
[0087] Optionally, the setting formula may include:
[0088]
[0089] Where M is the two-dimensional seismic data; the matrix D represents the dictionary obtained through dictionary learning; y i,j represents the (i,j)th local block in the 2D seismic data; A block extraction operator for the (i, j)th data in the two-dimensional seismic data; Extract operators for blocks The adjoint operator of ; P is a set of data blocks, which includes all pixel points of the two-dimensional seismic data; τ p is a diagonal matrix; Dx i,j is the sparse representation vector at position (i, j) in the two-dimensional data.
[0090] It can be understood that in this denoising method, it is necessary to solve the following dictionary learning problem based on the proximal gradient method:
[0091]
[0092] For example, the solution process may include the following steps:
[0093] (1) Input sample matrix y, parameter λ>0, (D -1 ,X -1 )=(D 0 ,X 0 );
[0094] (2) looping the seismic data set WHILE k≤K;
[0095] By formula Among them, ‖XX T ‖ F and ‖D T D‖ F is the Lipschitz constant of the gradient;
[0096] It should be noted that the weight ω k The algorithm is widely used in the fast iterative threshold method, so it can be expressed by formula (2-4) to determine the formula (2-3) The weight of
[0097] in, The settings can be: τ<1;
[0098] set up By formula Get D k ;in, Representatives The Euclidean mapping of
[0099] set up By formula Get X k ;in, Represents the soft threshold operator.
[0100] During the calculation process, if F(X k ,D k )>F(D k-1 ,X k-1 ), update D by the above formulas (2-5) and (2-6) k With X k ,in,
[0101] When the stop condition is met, stop iterating and output (X k ,D k ); Output the dictionary matrix and sparse representation matrix (X) when the loop condition is met k ,D k ).
[0102] It is understandable that after obtaining the dictionary matrix and the sparse representation matrix, the above formula can be used Represents a sparse approximation of seismic data.
[0103] Where M is the two-dimensional seismic data; the matrix D represents the dictionary obtained through dictionary learning; y i,j represents the (i,j)th local block in the 2D seismic data; A block extraction operator for the (i, j)th data in the two-dimensional seismic data; Extract operators for blocks The adjoint operator of ; P is a set of data blocks, which includes all pixel points of the two-dimensional seismic data; τ p is a diagonal matrix; Dx i,j is the sparse representation vector at position (i, j) in the two-dimensional data.
[0104] Figure 4 is a flow chart of a method for denoising seismic data according to an exemplary embodiment. Figure 4 As shown, the recovery of the second seismic data in step S13 may include the following steps:
[0105] Step S131, recovering different sets of data blocks in the second seismic data in a non-overlapping block manner to obtain multiple data recovery results;
[0106] Step S132, obtaining an average value of the multiple data recovery results;
[0107] Step S133, taking the average value as the third seismic data.
[0108] For example, in a possible implementation, noisy seismic data is obtained by adding random noise to simulated seismic data. In the denoising process of seismic data, the size of the selected training set is 100×20000, the size of the dictionary trained in the dictionary learning method based on the proximal gradient is 100×200, and the data is blocked by 7 non-overlapping blocking strategies. For example, each non-overlapping blocking method can be a middle block size of 10×10, and the left and right boundary blocks and upper and lower boundary blocks are respectively selected to be smaller than 10×10. After the blocking is completed, each data block is denoised by applying the YALL1 (Your Algorithms for L1) algorithm and the above formula (2-1), and the 7 denoising results are restored by the non-overlapping recovery strategy to obtain 7 data recovery results, which can be recorded as M1, M2, M3, M4, M5, M6 and M7. The final seismic data noise suppression result is the average value of the seven recovery results, that is,
[0109] Figure 5 is a schematic diagram of seismic data after denoising by a dictionary learning method based on proximal gradient according to an exemplary embodiment, Figure 6is a schematic diagram of noise data removed by a dictionary learning method based on proximal gradient according to an exemplary embodiment. Figure 5 and Figure 6 It can be concluded that the dictionary learning method based on proximal gradient has a better effect on suppressing seismic data noise. It can reduce the algorithm redundancy to a certain extent in the denoising process and can reduce human errors.
[0110] In another possible implementation, in order to verify the effectiveness of the denoising method, the noise suppression method based on the transformation learning method based on group sparsity constraints and the dictionary learning method based on proximal gradients is applied to actual seismic data. Figure 7 is a schematic diagram of post-stack seismic data according to an exemplary embodiment. Figure 7 As shown, the data size is 400 × 300. The post-stack seismic data is denoised by a transformation learning method based on group sparsity constraints and a dictionary learning method based on proximal gradients, and the denoised seismic data and the removed noise data can be obtained. Figure 8 is a schematic diagram of seismic data after denoising by a transformation learning method based on group sparsity constraints according to an exemplary embodiment, Fig. 9 is a schematic diagram of noise data removed by a transformation learning method based on group sparsity constraints according to an exemplary embodiment, Fig.10 is a schematic diagram of seismic data after denoising by a dictionary learning method based on proximal gradient according to an exemplary embodiment, Fig.11 FIG. 1 is a schematic diagram of noise data removed by a dictionary learning method based on proximal gradient according to an exemplary embodiment. Figures 7 to 11 It can be seen that the post-stack seismic data is denoised by a transformation learning method based on group sparsity constraints and a dictionary learning method based on proximal gradients, and seismic data is restored in a non-overlapping manner, which can maintain good data in-phase continuity and seismic signal energy.
[0111] In the above technical solution, first seismic data that needs to be noise suppressed is obtained, noise is suppressed on the first seismic data by a transformation learning method based on group sparsity constraints, and a dictionary learning method based on proximal gradient is applied to the noise suppression to obtain second seismic data after noise suppression, and seismic data recovery is performed on the second seismic data to obtain third seismic data, wherein the third seismic data includes a denoising result, and one or more of the removed noise data and energy loss data. Through the above technical solution, the transformation learning method based on group sparsity constraints and the dictionary learning method based on proximal gradient are combined and applied to the denoising process of seismic data, and the transformation learning method and the dictionary learning method reduce the complexity of the algorithm, and the dictionary learning process has good convergence, thereby improving the denoising efficiency and accuracy of seismic data.
[0112] Fig.12 is a block diagram of a seismic data denoising device according to an exemplary embodiment. Fig.12 As shown, the device 1200 includes an acquisition module 1210 , a noise suppression module 1220 and a data recovery module 1230 .
[0113] The acquisition module 1210 is used to acquire first seismic data that needs to be noise suppressed;
[0114] The noise suppression module 1220 is used to suppress the noise of the first seismic data by a transformation learning method based on group sparsity constraints, and apply a dictionary learning method based on proximal gradient to the noise suppression process to obtain the second seismic data after noise suppression;
[0115] The data recovery module 1230 is used to perform seismic data recovery on the second seismic data to obtain third seismic data; wherein the third seismic data includes a denoising result and one or more of removed noise data and energy loss data.
[0116] Optionally, the noise suppression module 1220 includes a first acquisition submodule and a first data processing submodule;
[0117] The first acquisition submodule is used to acquire a target training sample required by the transformation learning method according to the first seismic data; the target training sample includes a plurality of local blocks;
[0118] The first data processing submodule is used to perform sparse representation on the multiple local blocks in the target training sample in the same transform domain.
[0119] Optionally, the first acquisition submodule is further used to:
[0120] Acquire original training samples from the first seismic data;
[0121] The block extraction operator is used to extract local blocks in the original training sample in an overlapping manner, and vectorized, and the result is used as the column vector of the target training sample.
[0122] Optionally, the first data processing submodule is further used to:
[0123] A plurality of training sets are obtained by applying a grouping operator to the target training sample;
[0124] In the same transform domain, the sparse coding of each set of data in the training set is subject to an infinite zero norm regularization constraint.
[0125] Optionally, the noise suppression module 1220 further includes a second acquisition submodule and a second data processing submodule;
[0126] The second acquisition submodule is used to acquire a dictionary matrix and a sparse representation vector in the noise suppression by using a dictionary learning method based on a proximal gradient;
[0127] The second data processing submodule is used to represent the sparse approximate seismic data by setting a formula as the second seismic data.
[0128] Optionally, the setting formula includes:
[0129]
[0130] Where M is the two-dimensional seismic data; the matrix D represents the dictionary obtained through dictionary learning; y i,j represents the (i,j)th local block in the 2D seismic data; A block extraction operator for the (i, j)th data in the two-dimensional seismic data; Extract operators for blocks The adjoint operator of ; P is a set of data blocks, which includes all pixel points of the two-dimensional seismic data; τ p is a diagonal matrix; Dx i,j is the sparse representation vector at position (i, j) in the two-dimensional data.
[0131] Optionally, the data recovery module 1230 includes a recovery submodule, a third acquisition submodule and a determination submodule;
[0132] The recovery submodule is used to recover different sets of data blocks in the second seismic data in a non-overlapping block manner to obtain multiple data recovery results;
[0133] The third acquisition submodule is used to obtain an average value of the multiple data recovery results;
[0134] The determination submodule is used to use the average value as the third seismic data.
[0135] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.
[0136] Fig.13 FIG. 1 is a block diagram of an electronic device 1300 according to an exemplary embodiment. Fig.13 As shown, the electronic device 1300 may include: a processor 1301 and a memory 1302. The electronic device 1300 may also include one or more of a multimedia component 1303, an input / output (I / O) interface 1304, and a communication component 1305.
[0137] The processor 1301 is used to control the overall operation of the electronic device 1300 to complete all or part of the steps in the above-mentioned seismic data denoising method. The memory 1302 is used to store various types of data to support the operation of the electronic device 1300, and these data may include, for example, instructions for any application or method used to operate on the electronic device 1300, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 1302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 1303 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 1302 or sent through the communication component 1305. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 1304 provides an interface between the processor 1301 and other interface modules, and the above-mentioned other interface modules may be keyboards, mice, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 1305 is used for wired or wireless communication between the electronic device 1300 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, etc., or a combination of one or more of them, is not limited here. Therefore, the corresponding communication component 1305 may include: Wi-Fi module, Bluetooth module, NFC module, etc.
[0138] In an exemplary embodiment, the electronic device 1300 can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned seismic data denoising method.
[0139] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by a processor, the steps of the above-mentioned seismic data denoising method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 1302 including program instructions, and the above-mentioned program instructions can be executed by the processor 1301 of the electronic device 1300 to complete the above-mentioned seismic data denoising method.
[0140] In another exemplary embodiment, a computer program product is also provided. The computer program product includes a computer program executable by a programmable device, and the computer program has a code portion for performing the above-mentioned seismic data denoising method when executed by the programmable device.
[0141] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, a variety of simple modifications can be made to the technical solution of the present disclosure, and these simple modifications all fall within the protection scope of the present disclosure.
[0142] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.
[0143] In addition, various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.
Claims
1. A method for denoising seismic data, characterized in that: The method comprises: Acquiring first seismic data requiring noise suppression; Performing noise suppression on the first seismic data by a transformation learning method based on group sparsity constraints, and applying a dictionary learning method based on proximal gradients to the noise suppression to obtain second seismic data after noise suppression; Perform seismic data recovery on the second seismic data to obtain third seismic data; wherein the third seismic data includes a denoising result, and one or more of removed noise data and energy loss data.
2. The method for denoising seismic data according to claim 1, characterized in that: The noise suppression of the first seismic data by a transformation learning method based on group sparsity constraints includes: Acquire a target training sample required by the transformation learning method according to the first seismic data; the target training sample includes a plurality of local blocks; The multiple local blocks in the target training sample are sparsely represented in the same transform domain.
3. The method for denoising seismic data according to claim 2, characterized in that: The step of acquiring target training samples required by the transformation learning method according to the first seismic data includes: Acquire original training samples in the first seismic data; A block extraction operator is used to extract local blocks in the original training sample in an overlapping manner, and vectorization is performed, and the result is used as a column vector of the target training sample.
4. The method for denoising seismic data according to claim 2, characterized in that: The sparsely representing the multiple local blocks in the target training sample in the same transform domain includes: Obtaining multiple training sets by applying a grouping operator to the target training samples; In the same transform domain, the sparse coding of each set of data in the training set is subject to an infinite zero norm regularization constraint.
5. The method for denoising seismic data according to claim 1, characterized in that: The applying the dictionary learning method based on proximal gradient to the noise suppression comprises: Obtaining a dictionary matrix and a sparse representation vector in the noise suppression by a dictionary learning method based on proximal gradient; The sparsely approximated seismic data is represented by a setting formula as the second seismic data.
6. The method for denoising seismic data according to claim 5, characterized in that: The setting formula includes: Where M is the two-dimensional seismic data; the matrix D represents the dictionary obtained through dictionary learning; y i,j represents the (i,j)th local block in the 2D seismic data; A block extraction operator for the (i, j)th data in the two-dimensional seismic data; Extract operators for blocks P is a set of data blocks, which includes all pixel points of the two-dimensional seismic data; τ p is a diagonal matrix; Dx i,j is the sparse representation vector at position (i, j) in the two-dimensional data.
7. The method for denoising seismic data according to claim 1, characterized in that: The recovering the second seismic data includes: Recovering different sets of data blocks in the second seismic data in a non-overlapping block manner to obtain multiple data recovery results; Obtaining an average value of the multiple data recovery results; The average value is used as the third seismic data.
8. A denoising device for seismic data, characterized in that: The device comprises: An acquisition module, used for acquiring first seismic data requiring noise suppression; A noise suppression module, used for performing noise suppression on the first seismic data by a transformation learning method based on group sparsity constraints, and applying a dictionary learning method based on proximal gradient to the noise suppression process to obtain second seismic data after noise suppression; A data recovery module is used to perform seismic data recovery on the second seismic data to obtain third seismic data; wherein the third seismic data includes a denoising result and one or more of removed noise data and energy loss data.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.
10. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the steps of the method according to any one of claims 1 to 7.
11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.