Seismic data denoising method and system

Abstract

The invention discloses a seismic data denoising method and system. The method comprises the following steps: carrying out normalization processing on original seismic data; decomposing the data into sub-band coefficients of multiple scales and directions by applying curvelet transform, and reducing the influence of a boundary effect and a pseudo-Gibbs phenomenon through a cyclic translation technology; performing multi-scale adaptive block complex domain threshold processing on the sub-band coefficient after curvelet transformation, and adjusting a threshold according to signal distribution and noise level, so as to remove noise and protect effective signals; a fast non-local mean denoising algorithm is applied to the data after inverse transformation, the efficiency is improved by optimizing the calculation process, and the residual noise is further suppressed; weighted smoothing is carried out in the main direction by using a directional smoothing diffusion algorithm, and excessive smoothing of signal details is avoided; and outputting the denoised seismic data. The method has the advantages that the denoising effect of the seismic data is effectively improved, meanwhile, signal details are kept, and the method is suitable for processing and analyzing the seismic data.

Description

Technical Field

[0001] The present invention relates to the field of seismic data processing technology, and in particular to a seismic data denoising method and system based on an improved multi-scale adaptive block complex domain threshold curvelet transform and a directional smooth diffusion algorithm, which aims to improve the signal quality of seismic data, remove noise, protect effective signals, and improve the denoising effect. Background Art

[0002] In the process of oil and gas exploration and geological research, the acquisition and processing of seismic data are crucial. However, seismic data are often interfered by various noises during the acquisition process, including environmental noise, instrument noise, and random noise. These noises will mask weak reflection signals, thus affecting subsequent seismic imaging and data analysis.

[0003] At present, the commonly used seismic data denoising methods include Fourier transform, wavelet transform and curvelet transform, etc. Although these methods have achieved certain results in denoising, there are still some problems, especially the balance between signal detail protection and noise removal.

[0004] For example, due to the lack of translation invariance, the curvelet transform is prone to produce pseudo-Gibbs phenomenon at signal boundaries and discontinuities, affecting data quality; while traditional denoising methods such as wavelet transform and non-local mean denoising method (FNLM) are less effective in dealing with complex noise and have high computational complexity.

[0005] Therefore, proposing an efficient and accurate seismic data denoising method that can effectively solve the problems in the existing technology is still a technical problem that needs to be solved urgently in the technical field. Summary of the invention

[0006] The present invention aims at the defects of the prior art and provides a method and system for denoising seismic data. The method can effectively solve the problems existing in the prior art, such as pseudo-Gibbs phenomenon, insufficient protection of weak signals, incomplete removal of random noise, and high computational complexity. The present invention can improve the signal-to-noise ratio of seismic data, protect effective signals, and effectively suppress noise.

[0007] In order to achieve the above invention object, the technical solution adopted by the present invention is as follows:

[0008] A seismic data denoising method comprises the following steps:

[0009] S1. Data preprocessing: Normalize the original seismic data to remove possible systematic deviations and provide stable input for subsequent denoising processing.

[0010] S2. Cyclic shift curvelet transform: Apply curvelet transform to seismic data to decompose it into sub-band coefficients of multiple scales and directions. Cyclic shift technology is used to reduce the influence of pseudo-Gibbs phenomenon by multiple shifts and averaging for boundary effects and discontinuities in the data.

[0011] S3, multi-scale adaptive block complex domain threshold processing: Adaptive block complex domain threshold processing is performed on each sub-band coefficient after curvelet transformation based on multi-scale characteristics. The threshold of each scale is adjusted according to the noise level and signal distribution to ensure that the effective signal is protected as much as possible while removing noise.

[0012] S4, Fast Non-Local Mean (FNLM) denoising: The FNLM algorithm is applied to the data after the inverse curvelet transform to further suppress the residual noise by calculating the similarity between image blocks. In view of the high computational complexity of FNLM, the FNLM algorithm is optimized to reduce the amount of search window calculation, thereby improving the computational efficiency.

[0013] S5, Directional Smoothing Diffusion: Directional Smoothing Diffusion algorithm is applied to the denoised data, and the gradient direction information of the data is used to perform weighted smoothing in the main direction. The diffusion process is controlled by regularization parameters to ensure that the signal details are not over-smoothed.

[0014] S6. Result display: Output the denoised seismic data.

[0015] Furthermore, the specific sub-steps of S2 are as follows:

[0016] S21, determining the number of scales and directions of the curvelet transform, and selecting an appropriate curvelet basis function;

[0017] S22. Perform curvelet transform to decompose the seismic data into sub-band coefficients of different scales and directions. The formula is as follows:

[0018]

[0019] in, represents the subband coefficient of the sth scale and the dth direction, where S and D represent the number of scales and directions respectively;

[0020] S23, apply the cyclic translation technique to each subband coefficient To pan left, right, up or down:

[0021]

[0022] Average multiple translation results:

[0023]

[0024] Where N is the number of translations, is the average subband coefficient after removing the pseudo-Gibbs factor, and Δx represents the unit length of each translation during cyclic translation.

[0025] S24, the subband coefficients after cyclic shift The result of the curvelet transform is synthesized as follows:

[0026]

[0027] Among them, IFDCT is the inverse curvelet transform operator, D restored represents the reconstructed seismic data;

[0028] S25. Save the processed seismic data D restored , for use in subsequent steps.

[0029] Furthermore, the specific sub-steps of S3 are as follows:

[0030] S31, the subband coefficients at each scale Divided into multiple small blocks, the formula is as follows:

[0031]

[0032] in, represents the i,jth block in the sth scale.

[0033] S32. Calculate the complexity of each data block to determine whether it contains significant signal features. The formula is as follows:

[0034]

[0035] Among them, μ i,j is the mean of the data block, N is the data block The total number of pixels in Represents the specific value or coefficient of the (x,y)th coordinate in the sth scale and the i,jth block; uses texture features to enhance complexity measurement.

[0036] S33, threshold setting: according to the set threshold T i,j , process the coefficients of each small block:

[0037]

[0038] Among them, α represents the adjustment parameter of the threshold setting, σ represents the standard deviation of noise, and logN represents the logarithm of the number of pixels N in the block, which is used to balance the relationship between the threshold and the data block size;

[0039] For important signal blocks, appropriately lower the threshold T i,j, to protect details; for areas dominated by noise, increase the threshold T i,j , to enhance denoising.

[0040] S34, according to the set threshold T i,j , apply soft or hard thresholding to the coefficients of each small block;

[0041] Soft Thresholding:

[0042]

[0043] The coefficients below the threshold are reduced, and the coefficients above the threshold are retained and adjusted.

[0044] Hard Threshold:

[0045]

[0046] S35, synthesizing the processed subband coefficients of each block to obtain processed curvelet transform coefficients;

[0047] S36. Save the seismic data after the adaptive block complex domain threshold processing.

[0048] Furthermore, the specific sub-steps of S4 are as follows:

[0049] S41, dividing the processed data into multiple image blocks, and calculating the similarity between each block;

[0050] S42. Calculate the weight according to the similarity between blocks. The formula is as follows:

[0051]

[0052] Among them, ω i,j It is block B i and B j The weight between them, h is the adjustment parameter used to control the smoothing strength, d(B i ,B j ) 2 Indicates data block B i and B j The distance or difference measure between .

[0053] S43, performing weighted averaging on the pixel values ​​in each image block, and using the block with higher similarity as a reference to adjust the pixel value of the target block.

[0054] S44, FNLM calculation:

[0055] Search for similar blocks within a local search window:

[0056]

[0057] S45, synthesizing the denoised image blocks back into original data.

[0058] S46. Save the seismic data after FNLM denoising for subsequent processing.

[0059] Furthermore, the specific sub-steps of S5 are as follows:

[0060] S51, perform gradient calculation on the denoised data to obtain the gradient direction of each pixel, the formula is as follows:

[0061]

[0062] Among them, I(x,y) represents the pixel value of the image, and are the horizontal and vertical gradient components respectively.

[0063] Use the gradient operator to calculate the directional information of the data. The formula is as follows:

[0064]

[0065]

[0066] in, Indicates the brightness change rate of the image in the x direction, represents the brightness change rate of the image in the y direction, and θ represents the direction of the image gradient;

[0067] S52. Define the diffusion tensor D according to the gradient direction θ(x,y) of each pixel. The formula is as follows:

[0068]

[0069] Directional weighted diffusion is performed on each pixel, and the formula is as follows:

[0070]

[0071] Here, t is the diffusion time.

[0072] S53, introduce regularization parameters to control the diffusion intensity and avoid excessive smoothing of the signal. Adaptively adjust the parameters in the diffusion process according to the local characteristics of the data.

[0073] S54. Iteratively apply the directional diffusion equation to the entire data set to gradually reduce small noises. The formula is as follows:

[0074]

[0075] Among them, Δt is the time step.

[0076] Iteration stop condition: the gradient change rate meets the threshold condition; the preset number of iterations is reached.

[0077] S55. Save the denoised seismic data after directional smooth diffusion.

[0078] The present invention also discloses a seismic data denoising system, which can be used to implement the above seismic data denoising method, specifically comprising:

[0079] Data preprocessing module: normalize the original seismic data to remove possible systematic deviations and provide stable input for subsequent denoising processing.

[0080] Loop-translation curvelet transform module: Curvelet transform is applied to the normalized seismic data to decompose it into sub-band coefficients of multiple scales and directions, and the cyclic translation technique is used to handle boundary effects and reduce pseudo-Gibbs phenomenon.

[0081] Multi-scale adaptive block complex domain threshold processing module: Adaptive block complex domain threshold processing is performed on the sub-band coefficients after curvelet transformation, and the threshold is adjusted according to the noise level and signal distribution to protect the valid signal and remove noise.

[0082] Fast Non-Local Means (FNLM) Denoising Module: The FNLM algorithm is applied on the inverse curvelet transformed data to suppress residual noise by calculating the similarity between image patches and to optimize the FNLM algorithm to improve computational efficiency.

[0083] Directional smoothing diffusion module: The directional smoothing diffusion algorithm is applied to the denoised data, using the gradient direction information of the data to perform weighted smoothing in the main direction, and the diffusion process is controlled by the regularization parameter to avoid over-smoothing of signal details.

[0084] Result display module: outputs the denoised seismic data and compares it with the original data for quality assessment.

[0085] The present invention also discloses a computer device, including a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned seismic data denoising method when executing the program.

[0086] The present invention also discloses a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned seismic data denoising method is implemented.

[0087] Compared with the prior art, the advantages of the present invention are:

[0088] 1. The present invention effectively removes noise from seismic data and significantly improves the quality of the signal. Experimental results show that the method of the present invention significantly improves the signal-to-noise ratio (SNR) and peak signal-to-noise ratio (PSNR) compared to the traditional method.

[0089] 2. In the curvelet transform stage, the boundary effects and discontinuities in the seismic data are effectively processed through the cyclic translation technology, which reduces the influence of the pseudo-Gibbs phenomenon and ensures that the denoised data is smoother and rich in details.

[0090] 3. The multi-scale adaptive block complex domain threshold processing method can adaptively adjust the threshold according to the noise level and signal distribution of each scale, so that in the denoising process, the effective signal can be protected and the noise can be removed more accurately, thereby improving the denoising accuracy.

[0091] 4. The present invention optimizes the fast non-local mean (FNLM) denoising algorithm, reduces the amount of calculation of the search window, greatly improves the calculation efficiency, overcomes the shortcoming of the high calculation complexity of the FNLM algorithm, and makes the method suitable for real-time processing of large-scale seismic data.

[0092] 5. In the data processing after denoising, a directional smoothing diffusion algorithm is used to perform weighted smoothing using gradient direction information, effectively avoiding excessive smoothing of the signal while retaining important detail information. This technology ensures a balance between signal quality improvement and detail retention.

[0093] 6. The method of the present invention can adapt to different types of seismic data, and can process large-scale complex seismic signals as well as be applied to small-scale high-resolution data processing, and has good versatility and adaptability.

[0094] 7. The present invention improves the signal quality of seismic data by denoising, providing higher quality input for subsequent data analysis, processing and interpretation, thereby providing more accurate and reliable basic data for fields such as seismic exploration and disaster warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] Figure 1 4 is a flow chart of a method for denoising seismic data in an embodiment of the present invention.

[0096] Figure 2 It is the overall picture and local picture of the original seismic data before being processed by the seismic data denoising method in the embodiment of the present invention.

[0097] Figure 3 It is the overall and local images after seismic data is processed using traditional wavelet transform.

[0098] Figure 4It is the overall and local images of seismic data after traditional curvelet processing.

[0099] Figure 5 These are the overall and local diagrams of the final denoising effect of the improved method of the present invention in the embodiment of the present invention. DETAILED DESCRIPTION

[0100] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples.

[0101] like Figure 1 As shown, the present invention provides a method for denoising seismic data, comprising the following steps:

[0102] S1. Data preprocessing: Normalize the original seismic data to remove possible systematic deviations and provide stable input for subsequent denoising processing.

[0103] The specific sub-steps are as follows:

[0104] S11. Data loading: Load the original seismic data from the data source. Check the integrity of the data and handle missing values.

[0105] S12. Removal of systematic deviations: Identify and remove possible systematic deviations by calculating the mean or trend of seismic data. For non-stationary signals, remove trend components by polynomial fitting or high-pass filtering.

[0106] S13. Normalization: Adjust the amplitude range of seismic data to [0,1] or [-1,1] to make subsequent processing more stable. The normalization method uses linear normalization or z-score normalization. For large-scale data sets, ensure that the normalization process does not lose valid signals.

[0107] S14. Data smoothing: Low-pass filter the data to remove low-frequency noise. Use methods such as sliding average, weighted average or Gaussian filtering.

[0108] S15. Save the processed seismic data for use in subsequent steps.

[0109] S2, cyclic translation curvelet transform: This step aims to utilize the multi-scale and multi-directional characteristics of the curvelet transform, combined with the cyclic translation technology, to effectively reduce the pseudo-Gibbs phenomenon and improve the denoising effect of seismic data. The specific sub-steps are as follows:

[0110] S21. Determine the number of scales of the curvelet transform (e.g., 3 scales) so that it can decompose the low-frequency and high-frequency information of the seismic data; determine the number of directions, for example, 8 directions at each scale to capture signal characteristics in different directions; select an appropriate curvelet basis function (e.g., Morlet wavelet or Mexican Hat wavelet) to meet the frequency and direction requirements of the signal.

[0111] S22. Perform curvelet transform: Perform discrete curvelet transform on the seismic data and decompose it into sub-band coefficients of different scales and directions:

[0112]

[0113] in, It represents the subband coefficient of the sth scale and the dth direction, where S and D represent the number of scales and directions respectively.

[0114] Subband coefficient C for each scale s Contains the frequency information corresponding to the scale, the subband coefficients in each direction Indicates the directional characteristics of the signal.

[0115] S23, cyclic shift technology application: for each subband coefficient Apply the circular panning technique:

[0116] Shift the data left or right (or up or down), for example, move △x = 1 pixel to calculate the shifted subband coefficients:

[0117]

[0118] Average multiple translation results:

[0119]

[0120] Where N is the number of translations, is the average subband coefficient after removing the pseudo-Gibbs.

[0121] This operation effectively reduces boundary effects, especially at signal discontinuities, and suppresses pseudo-Gibbs phenomena.

[0122] S24, the subband coefficients after cyclic shift Synthesize into curvelet transform result:

[0123]

[0124] Among them, IFDCT is the inverse curvelet transform operator, D restored Represents the reconstructed seismic data.

[0125] S25. Save the processed seismic data D restored, providing input for subsequent processing (such as fast non-local means denoising). The result has a higher signal-to-noise ratio and clearer boundaries, reducing visual distortion caused by pseudo-Gibbs phenomenon.

[0126] S3, multi-scale adaptive block complex domain threshold processing: Adaptive block complex domain threshold processing is performed on each sub-band coefficient after curvelet transformation based on multi-scale characteristics. The threshold of each scale is adjusted according to the noise level and signal distribution to ensure that the effective signal is protected as much as possible while removing noise.

[0127] The specific sub-steps are as follows:

[0128] S31, block processing: the sub-band coefficients at each scale Divided into several small pieces:

[0129]

[0130] in, It represents the i,jth block in the sth scale. The size of the block can be selected as 4×4, 8×8 or other suitable sizes.

[0131] Each block is regarded as an independent signal unit and threshold processing is performed on each block.

[0132] S32. Calculate block complexity: by calculating the complexity of each data block:

[0133] Using local variance calculation:

[0134]

[0135] Among them, μ i,j is the mean of the data block, N is the total number of pixels in the data block; the complexity metric is enhanced by using texture features (such as gradient strength or frequency domain distribution).

[0136] According to the size of the complexity, it is determined whether the block contains significant signal features or noise.

[0137] S33, threshold setting: according to the set threshold T i,j , process the coefficients of each small block:

[0138]

[0139] For important signal blocks, appropriately lower the threshold T i,j , to protect details; for areas dominated by noise, increase the threshold T i,j , to enhance denoising.

[0140] S34, perform threshold processing: apply soft threshold or hard threshold processing to the coefficients of each small block.

[0141] According to the set threshold T i,j , process the coefficients of each small block:

[0142] Soft Thresholding:

[0143]

[0144] The coefficients below the threshold are reduced, and the coefficients above the threshold are retained and adjusted.

[0145] Hard Threshold:

[0146]

[0147] S35, result synthesis: synthesize the processed sub-band coefficients of each block to obtain the processed curvelet transform coefficients.

[0148] S36. Output the data after threshold processing: save the seismic data after adaptive block complex domain threshold processing.

[0149] S4, Fast Non-Local Mean (FNLM) denoising: Apply the FNLM algorithm to the data after the inverse curvelet transform, and further suppress the residual noise by calculating the similarity between image blocks. In view of the high computational complexity of FNLM, the present invention optimizes it, reduces the amount of search window calculation, and thus improves the computational efficiency.

[0150] The specific sub-steps are as follows:

[0151] S41, block similarity calculation: divide the processed data into multiple image blocks, calculate the similarity between each block, and the similarity measurement includes Euclidean distance, cosine similarity, etc.

[0152] S42. Weight calculation:

[0153] Calculate the weights based on the similarity between blocks:

[0154]

[0155] Among them, ω i,j It is block B i and B j The weight between them, h is the adjustment parameter used to control the smoothing strength.

[0156] S43, weighted average processing: weighted average the pixel values ​​in each image block to reduce the influence of noise. Use the block with higher similarity as a reference to adjust the pixel value of the target block.

[0157] S44, FNLM calculation:

[0158] Limit the search window: Search for similar blocks only within the local search window, rather than in the entire image:

[0159]

[0160] The search window size can be set as required (eg, 15×15 or 21×21).

[0161] Sparsification strategy: Use sparse representation of block features to reduce computational complexity; prune redundant information in block similarity calculation to improve efficiency.

[0162] S45, synthesizing denoising results: synthesizing the denoised image blocks back into original data.

[0163] S46. Output denoising results: save the seismic data after FNLM denoising for subsequent processing.

[0164] S5, Directional Smoothing Diffusion: Directional Smoothing Diffusion algorithm is applied to the denoised data, and the gradient direction information of the data is used to perform weighted smoothing in the main direction. The diffusion process is controlled by regularization parameters to ensure that the signal details are not over-smoothed.

[0165] S51, calculating gradient information: performing gradient calculation on the denoised data to obtain the gradient direction of each pixel.

[0166]

[0167] in, and are the horizontal and vertical gradient components respectively.

[0168] Use gradient operators (such as Sobel operator, Prewitt operator, etc.) to calculate the directional information of the data. The formula is as follows:

[0169]

[0170]

[0171] S52, directional diffusion: According to the gradient direction θ(x,y) of each pixel, define the diffusion tensor D:

[0172]

[0173] This tensor is used to adjust the diffusion strength so that the signal diffuses more in the main direction and less in the perpendicular direction.

[0174] Perform directional weighted diffusion on each pixel:

[0175]

[0176] Here, t is the diffusion time.

[0177] S53, Control diffusion intensity: Introduce regularization parameters to control diffusion intensity and avoid excessive smoothing of signals. Adaptively adjust the parameters in the diffusion process according to the local characteristics of the data.

[0178] S54, smoothing: iteratively apply the directional diffusion equation to the entire data set, gradually reducing small noise:

[0179]

[0180] Among them, Δt is the time step.

[0181] Iteration stop condition: the gradient change rate meets the threshold condition; the preset number of iterations is reached.

[0182] S55, output smoothed data: save the denoised seismic data after directional smooth diffusion.

[0183] S6. Result presentation and quality assessment: Finally, the denoised seismic data is output and compared with the original data. The denoising effect is quantitatively evaluated using indicators such as signal-to-noise ratio (SNR) and peak signal-to-noise ratio (PSNR).

[0184] In order to prove the superiority of the present invention, the present invention has carried out comparative analysis on the experimental results, specifically as follows Figures 2 to 5 As shown:

[0185] Figure 2 The overall structure and local details of the original seismic data are shown. The overall picture shows the overall picture of the seismic data and shows the changing rules of the signal in the time or space dimension. The local picture focuses on the random noise components in the data and their interference with the signal, especially the blurring of weak signal areas. These features provide a clear research goal and necessary data basis for the subsequent denoising process.

[0186] Figure 3 The traditional wavelet transform is used to process seismic data, including the overall and local parts. The overall figure shows that the data processed by wavelet transform has been partially improved, especially the noise in the medium and low frequency bands has been suppressed. However, the local figure shows that the wavelet transform is still insufficient in retaining high-frequency weak signals, and some high-frequency details are suppressed or blurred. At the same time, the hard threshold and soft threshold methods may cause discontinuous transitions between sub-bands, thus affecting the quality of signal restoration.

[0187] Figure 4The denoising effect after traditional curvelet processing is shown, including overall and local images. The overall image shows the decomposition advantages of curvelet transform in multi-scale and multi-directional, especially in the processing of complex seismic signals. The local image further shows the effect of cyclic translation technology in reducing pseudo-Gibbs phenomenon, significantly improving the clarity of signal boundary areas. Compared with wavelet transform, curvelet transform can more effectively preserve the details of seismic data while enhancing the coherence of the signal.

[0188] Figure 5 The overall effect and local details of the method of the present invention are demonstrated. The overall graph intuitively reflects the seismic data after data preprocessing, curvelet transform, multi-scale adaptive block complex domain threshold processing, FNLM denoising and directional smooth diffusion algorithm. The noise is fully suppressed and the signal quality is significantly improved. The local graph shows the details of the denoised data in more detail, especially in the weak signal area, the signal-to-noise ratio is greatly improved, and the pseudo-Gibbs phenomenon that may be introduced by the traditional method is effectively avoided. This further verifies the efficiency and superiority of the method of the present invention.

[0189] Through the comparative analysis of the above diagrams, it can be seen that the present invention is superior to traditional wavelet transform and other methods in terms of denoising performance and signal detail retention, and effectively achieves high-quality seismic data denoising and signal enhancement.

[0190] pass Figures 2 to 5 From the comparative analysis, it can be seen that the method of the present invention is significantly superior to the existing methods in terms of signal-to-noise ratio (SNR) and peak signal-to-noise ratio (PSNR), proving its effectiveness in denoising. In addition, the method of the present invention can effectively remove noise while retaining signal details, providing higher quality seismic data. This provides more stable and accurate input for subsequent processing, inversion analysis and interpretation of seismic data, and has important practical application value.

[0191] In another embodiment of the present invention, a seismic data denoising system is provided. The system can be used to implement the above-mentioned seismic data denoising method, specifically comprising:

[0192] Data preprocessing module: normalize the original seismic data to remove possible systematic deviations and provide stable input for subsequent denoising processing.

[0193] Loop-translation curvelet transform module: Curvelet transform is applied to the normalized seismic data to decompose it into sub-band coefficients of multiple scales and directions, and the cyclic translation technique is used to handle boundary effects and reduce pseudo-Gibbs phenomenon.

[0194] Multi-scale adaptive block complex domain threshold processing module: Adaptive block complex domain threshold processing is performed on the sub-band coefficients after curvelet transformation, and the threshold is adjusted according to the noise level and signal distribution to protect the valid signal and remove noise.

[0195] Fast Non-Local Means (FNLM) Denoising Module: The FNLM algorithm is applied on the inverse curvelet transformed data to suppress residual noise by calculating the similarity between image patches and to optimize the FNLM algorithm to improve computational efficiency.

[0196] Directional smoothing diffusion module: The directional smoothing diffusion algorithm is applied to the denoised data, using the gradient direction information of the data to perform weighted smoothing in the main direction, and the diffusion process is controlled by the regularization parameter to avoid over-smoothing of signal details.

[0197] Result display module: outputs the denoised seismic data and compares it with the original data for quality assessment.

[0198] In another embodiment of the present invention, a terminal device is provided, the terminal device includes a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding functions; the processor described in the embodiment of the present invention can be used for the operation of the seismic data denoising method.

[0199] In another embodiment of the present invention, a storage medium is provided, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device for storing programs and data. It is understandable that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and the extended storage medium supported by the terminal device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory, or a non-volatile memory (non-volatile memory), such as at least one disk storage.

[0200] One or more instructions stored in a computer-readable storage medium can be loaded and executed by a processor to implement the corresponding steps of the seismic data denoising method in the above embodiment; one or more instructions in the computer-readable storage medium are loaded and executed by the processor.

[0201] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0202] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0203] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0204] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0205] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the implementation methods of the present invention, and should be understood that the protection scope of the present invention is not limited to such special statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for denoising seismic data, characterized in that: The following steps are involved: S1. Data preprocessing: normalize the original seismic data to remove possible systematic deviations and provide stable input for subsequent denoising processing; S2. Cyclic shift curvelet transform: Apply curvelet transform to seismic data to decompose it into sub-band coefficients of multiple scales and directions; use cyclic shift technology to reduce the influence of pseudo-Gibbs phenomenon by multiple shifts and averaging the boundary effects and discontinuities in the data; S3, multi-scale adaptive block complex domain threshold processing: adaptive block complex domain threshold processing is performed on each sub-band coefficient after the curvelet transform according to the multi-scale features; The threshold of each scale is adjusted according to the noise level and signal distribution to ensure that the effective signal is protected as much as possible while removing the noise; S4, Fast Non-local Mean (FNLM) denoising: The FNLM algorithm is applied to the data after the inverse curvelet transform to further suppress the residual noise by calculating the similarity between image blocks; in view of the high computational complexity of FNLM, the FNLM algorithm is optimized to reduce the amount of search window calculation, thereby improving the computational efficiency; S5, Directional smooth diffusion: Apply the directional smooth diffusion algorithm to the denoised data, and use the gradient direction information of the data to perform weighted smoothing in the main direction; the diffusion process is controlled by the regularization parameter to ensure that the signal details are not over-smoothed; S6. Result display: Output the denoised seismic data.

2. A seismic data denoising method according to claim 1, characterized in that: The specific sub-steps of S2 are as follows: S21, determining the number of scales and directions of the curvelet transform, and selecting an appropriate curvelet basis function; S22. Perform curvelet transform to decompose the seismic data into sub-band coefficients of different scales and directions. The formula is as follows: in, represents the subband coefficient of the sth scale and the dth direction, where S and D represent the number of scales and directions respectively; S23, apply the cyclic translation technique to each subband coefficient To pan left, right, up or down: Average multiple translation results: Where N is the number of translations, is the average subband coefficient after removing pseudo-Gibbs, Δx represents the unit length of each translation when performing cyclic translation; S24, the subband coefficients after cyclic shift The result of the curvelet transform is synthesized as follows: Among them, IFDCT is the inverse curvelet transform operator, D restored represents the reconstructed seismic data; S25. Save the processed seismic data D restored , for use in subsequent steps.

3. A seismic data denoising method according to claim 2, characterized in that: The specific sub-steps of S3 are as follows: S31, the subband coefficients at each scale Divided into multiple small blocks, the formula is as follows: in, represents the i,jth block in the sth scale; S32. Calculate the complexity of each data block to determine whether it contains significant signal features. The formula is as follows: Among them, μ i,j is the mean of the data block, N is the data block The total number of pixels in Represents the specific value or coefficient of the (x,y)th coordinate in the sth scale and the i,jth block; uses texture features to enhance complexity measurement; S33, threshold setting: according to the set threshold T i,j , process the coefficients of each small block: Among them, α represents the adjustment parameter of the threshold setting, σ represents the standard deviation of noise, and log N represents the logarithm of the number of pixels N in the block, which is used to balance the relationship between the threshold and the size of the data block; For important signal blocks, appropriately lower the threshold T i,j , to protect details; for areas dominated by noise, increase the threshold T i,j , to enhance denoising; S34, according to the set threshold T i,j , apply soft or hard thresholding to the coefficients of each small block; Soft Thresholding: The coefficients below the threshold are reduced, and the coefficients above the threshold are retained and adjusted; Hard Threshold: S35, synthesizing the processed subband coefficients of each block to obtain processed curvelet transform coefficients; S36. Save the seismic data after the adaptive block complex domain threshold processing.

4. A seismic data denoising method according to claim 3, characterized in that: The specific sub-steps of S4 are as follows: S41, dividing the processed data into multiple image blocks, and calculating the similarity between each block; S42. Calculate the weight according to the similarity between blocks. The formula is as follows: Among them, ω i,j It is block B i and B j The weight between them, h is the adjustment parameter used to control the smoothing strength, d(B i ,B j ) 2 Indicates data block B i and B j distance or difference measure between; S43, performing weighted averaging on the pixel values ​​in each image block; using the block with higher similarity as a reference, adjusting the pixel value of the target block; S44, FNLM calculation: Search for similar blocks within a local search window: S45, synthesizing the denoised image blocks back into original data; S46. Save the seismic data after FNLM denoising for subsequent processing.

5. A seismic data denoising method according to claim 4, characterized in that: The specific sub-steps of S5 are as follows: S51, perform gradient calculation on the denoised data to obtain the gradient direction of each pixel, the formula is as follows: Among them, I(x,y) represents the pixel value of the image, and are the horizontal and vertical gradient components respectively; Use the gradient operator to calculate the directional information of the data. The formula is as follows: in, Indicates the brightness change rate of the image in the x direction, represents the brightness change rate of the image in the y direction, and θ represents the direction of the image gradient; S52. Define the diffusion tensor D according to the gradient direction θ(x,y) of each pixel. The formula is as follows: Directional weighted diffusion is performed on each pixel, and the formula is as follows: Where, t is the diffusion time; S53, introduce regularization parameters to control the diffusion intensity and avoid excessive smoothing of the signal; adaptively adjust the parameters in the diffusion process according to the local characteristics of the data; S54. Iteratively apply the directional diffusion equation to the entire data set to gradually reduce small noises. The formula is as follows: Among them, Δt is the time step; Iteration stop condition: the gradient change rate meets the threshold condition; the preset number of iterations is reached; S55. Save the denoised seismic data after directional smooth diffusion.

6. A seismic data denoising system, characterized in that: The system can be used to implement the seismic data denoising method according to any one of claims 1 to 5, specifically comprising: Data preprocessing module: normalizes the original seismic data to remove possible systematic deviations and provide stable input for subsequent denoising processing; Loop-translation curvelet transform module: Curvelet transform is applied to normalized seismic data to decompose it into sub-band coefficients of multiple scales and directions, and the cyclic translation technique is used to process boundary effects and reduce pseudo-Gibbs phenomenon; Multi-scale adaptive block complex domain threshold processing module: Adaptive block complex domain threshold processing is performed on the sub-band coefficients after curvelet transformation, and the threshold is adjusted according to the noise level and signal distribution to protect the valid signal and remove the noise; Fast Non-Local Means (FNLM) Denoising Module: The FNLM algorithm is applied to the inverse curvelet transformed data to suppress residual noise by calculating the similarity between image blocks and optimizing the FNLM algorithm to improve computational efficiency. Directional smoothing diffusion module: applies the directional smoothing diffusion algorithm to the denoised data, uses the gradient direction information of the data to perform weighted smoothing in the main direction, and controls the diffusion process through regularization parameters to avoid over-smoothing signal details; Result display module: outputs the denoised seismic data and compares it with the original data for quality assessment.

7. A computer device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the program, the seismic data denoising method is realized.

8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, the above-mentioned seismic data denoising method is implemented.

Images