A method for suppressing black triangle noise in time-frequency domain

Abstract

The application discloses a kind of black triangle noise time-frequency domain self-adapting suppression method, it is related to geophysical technical field, comprising: obtaining input data;Linear positive normalization of common shot gather is calculated;The amplitude of normalized common shot gather is arranged in ascending order;Quadratic function fitting is carried out to each trace of gather transversely;Along the transverse, the curvature of each trace is calculated to the fitted gather;Maximum value is extracted to curvature profile, and adaptive threshold percentage curve is obtained;Along the horizontal direction, Gaussian function fitting is carried out to the data of calculated curvature, and standard deviation is calculated along the transverse;Adaptive threshold percentage curve is revised, and iteration is carried out until optimal adaptive threshold percentage curve is obtained;Original data is transformed to time-frequency domain, and time-frequency data body is obtained;Adaptive soft threshold is carried out to time-frequency domain data;Inverse transformation is carried out to time domain, and black triangle noise is obtained, finally result is obtained.The application can significantly improve signal-to-noise ratio, and attenuate noise in the maximum extent without damaging effective signal.

Description

Technical Field

[0001] This invention relates to the field of exploration geophysics, and in particular to a time-frequency domain adaptive suppression method for black triangle noise. Background Technology

[0002] Seismic processing is a crucial tool for suppressing various interferences and improving the signal-to-noise ratio (SNR) of raw data, providing strong support for subsequent seismic interpretation. As exploration targets have shifted westward, seismic acquisition has expanded from the simple surface of the eastern regions to deserts, Gobi, and loess plateaus in the west. In these areas with undulating surfaces and thick, low-velocity zones, black triangle noise is generated near the offset, submerging the effective signal and hindering subsequent inversion and interpretation. The constant threshold method, by setting a specific threshold value in the time-frequency domain, attenuates black triangle noise, significantly improving the SNR and amplitude uniformity of the data. However, in practical applications, the biggest problem with the constant threshold method is that noise residue remains at depth after denoising, making precise signal-to-noise separation difficult. Therefore, it is challenging to directly apply it to the processing of large-scale data from western exploration.

[0003] Currently, to address the bottleneck issue of strong black triangle noise in seismic processing, a common approach is to use threshold-based methods to denoise based on amplitude differences. This method has two main problems: (1) Due to the absorption and attenuation of seismic signals, there is a significant amplitude difference between the effective signals from deep and shallow layers, making it difficult for conventional threshold methods to achieve a balance between the two; (2) Selecting the critical amplitude value between black triangle noise and the effective signal is challenging. A small threshold will damage the effective signal, while a large threshold will leave residual noise. Therefore, to maximize the attenuation of black triangle noise and improve the signal-to-noise ratio and the continuity of the phase axis, it is urgent to develop an adaptive threshold-based method for attenuating strong black triangle noise, providing strategic support for the exploration and development of oil and gas in western my country. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention discloses a time-frequency domain adaptive suppression method for black triangle noise. First, a quadratic function is fitted along the horizontal direction of the noise profile to determine the curvature. Then, a specific extraction method is used to obtain a roughly estimated adaptive threshold curve. Next, the curve is corrected based on the standard deviation of the Gaussian function fitted along the horizontal direction of the curvature profile to obtain a precise adaptive threshold curve. Finally, this curve is used for thresholding in the time-frequency domain. This method achieves accurate signal-to-noise separation, providing strong support for subsequent inversion and interpretation, and has good prospects for practical application.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] The first aspect of this invention proposes a time-frequency domain adaptive suppression method for black triangle noise, comprising the following steps:

[0007] S1. Acquire input data, which includes: common shot gather D(t,x) after abnormal trace processing and static correction, where t represents time and x represents the position of the detector point;

[0008] S2. Calculate the linear positive value normalization D_nor(t,x) of the common shot point gather D(t,x) to eliminate the influence of negative values ​​and to consider the relationship between pure amplitudes;

[0009] S3. The trace set D_nor(t,x) after linear positive value normalization is sorted in ascending order of amplitude along the time direction. The trace set D_sort(t,x) after ascending order of amplitude is defined as follows:

[0010] D_sort(t,x)=sort(D_nor(t,x))

[0011] Here, sort represents sorting each row of the matrix in ascending order;

[0012] S4. Based on the fact that the black triangle noise has a much stronger amplitude than the effective signal, perform quadratic function fitting on each channel of the channel set D_sort(t,x) after the amplitude is sorted in ascending order, and obtain D_qua(t,x).

[0013] S5. Calculate the curvature of each trace along the transverse direction for the trace set after quadratic function fitting, obtaining D_cur(t,x). The curvature calculation is defined as:

[0014]

[0015] Where ′ represents the first derivative, ″ represents the second derivative, || represents the absolute value, and D_qua(t) i ,x), D_cur(t i x) represents time t i A data point at which D_cur(t) represents all time intervals. i D_cur(t,x) is composed of t,x.

[0016] S6. Extract the maximum value of the calculated curvature profile D_cur(t,x), extract a trend along the peak point of the profile, and obtain the adaptive threshold percentage curve r(t);

[0017] S7. Apply a Gaussian function to the data D_cur(t,x) used to calculate curvature along the horizontal direction to obtain the Gaussian function-fitted data D_gau(t,x), and calculate the standard deviation σ(t) along the horizontal direction. The standard deviation is:

[0018]

[0019] Where ∑ represents the summation symbol, k represents the number of detector points, μ represents the average value of the data, and D_gau(t) i x) represents time t i A data point, σ(t) for all times. i ) constitutes σ(t);

[0020] S8. Correct the adaptive threshold percentage curve r(t) based on the standard deviation σ(t) of the Gaussian function, iterating until the optimal adaptive threshold percentage curve r_mod(t) is obtained. The correction is as follows:

[0021] r_mod(t) = r(t) + σ(t);

[0022] S9. The original data D(t,x) is transformed to the time-frequency domain using a local time-frequency transform to obtain the time-frequency data volume D_transform(t,x,f), wherein the time-frequency domain data is:

[0023] D_transform(t,x,f)=F ltft (D(t,x))

[0024] Among them, F ltft Represents local time-frequency transformation;

[0025] S10. Based on the modified adaptive threshold percentage curve r_mod(t), apply an adaptive soft threshold to the time-frequency domain data D_transform(t,x,f) to obtain the thresholded time-frequency data volume D_thr(t,x,f).

[0026] S11. Inverse transform D_thr(t,x,f) to the time domain to obtain the black triangle noise D_noise(t,x), and subtract the black triangle noise from the original data D(t,x) to obtain the final noise attenuation result D_denoise(t,x). The final noise attenuation result D_denoise(t,x) is as follows:

[0027] D_denoise(t,x)=D(t,x)-D_noise(t,x)

[0028] in, This represents the local time-frequency inverse transform.

[0029] Optionally, in step S2, the linear positive normalization D_nor(t,x) of the common shot point gather is defined as:

[0030] D_nor(t,x)=nor(D(t,x))

[0031] Here, nor represents the linear positive value normalization of each row of the matrix.

[0032] Optionally, in step S4, D_qua(t,x) is:

[0033]

[0034] Where ∑ represents the summation symbol, and k represents the number of detector points.

[0035] Optionally, in step S10, the time-frequency data volume D_thr(t,x,f) after the threshold is:

[0036]

[0037] Where real(D_transform(t,x,f)) represents the real part of D_transform(t,x,f), img(D_transform(t,x,f)) represents the imaginary part of D_transform(t,x,f), |D_transform(t,x,f)| represents the modulus of D_transform(t,x,f), the Real part term represents the real part of D_thr(t,x,f), and the Imaginary part term represents the imaginary part of D_thr(t,x,f), where thr(t,x,f) is:

[0038]

[0039] Where |D_transform(t,x,f)| represents the modulus of D_transform(t,x,f), r_mod_thr(t,f) is the threshold in the time-frequency domain, and r_mod_thr(t,f) = large(|D_transform(t,x,f)|,r_mod(t)), where the large function represents the r_mod(t) largest value of the time-frequency domain data |D_transform(t,x,f)|.

[0040] A second aspect of the present invention provides a computer device.

[0041] In some embodiments, the computer device includes a processor and a memory for storing processor-executable instructions, wherein the processor executes the instructions to implement the steps of the method described above.

[0042] A third aspect of the present invention provides a computer-readable storage medium.

[0043] In some embodiments, a computer-readable storage medium stores computer instructions that, when executed, implement the steps of the method described above.

[0044] The beneficial effects of this invention are as follows: The adaptive black triangle noise suppression method based on local time-frequency transformation primarily addresses the problem of incomplete attenuation of deep noise in seismic records when using constant thresholding methods to attenuate strong black triangle noise. This invention optimizes the adaptive threshold curve using the standard deviation of a Gaussian function, applies the optimized adaptive threshold curve to the time-frequency domain, and transforms the thresholded data back to the time domain, accurately thresholding the black triangle noise and significantly improving the signal-to-noise ratio (SNR). Statistically, the SNR of the denoising result achieved by this invention is more than 10 times that of the traditional constant thresholding method. To address the imbalance between deep and shallow thresholds in the constant thresholding method, this invention calculates the amplitude value of the inflection point by calculating the curvature, thereby estimating the critical amplitude value between the effective signal and the black triangle noise. Compared to the traditional constant thresholding method, this significantly improves the SNR and attenuates noise to the maximum extent without compromising the effective signal. Simultaneously, the adaptive thresholding also achieves equalization of seismic signals from shallow to deep, which is more conducive to subsequent signal processing. This adaptive black triangle noise suppression method based on local time-frequency transformation can solve the problem that noise remains in the deep region when the noise is attenuated by the constant threshold method, while ensuring that there is no damage to the effective signal. Ultimately, it promotes the practical application of high signal-to-noise ratio pre-stack seismic processing in large-scale 3D oil and gas exploration. Attached Figure Description

[0045] Figure 1 This is a flowchart illustrating the time-frequency domain adaptive suppression method for black triangle noise according to the present invention.

[0046] Figure 2 This is a typical common shot gather containing black triangle noise in land exploration, as shown in an embodiment of the present invention.

[0047] Figure 3 This is an embodiment of the present invention showing the result of curvature data extraction and the result of correction based on standard deviation;

[0048] Figure 4 This is the result of black triangle noise attenuation using a constant threshold method as shown in an embodiment of the present invention;

[0049] Figure 5 This is the result of black triangle noise attenuation in an adaptive thresholding method according to an embodiment of the present invention;

[0050] Figure 6 This is a noise difference profile corresponding to the black triangle noise attenuation in an embodiment of the adaptive thresholding method of the present invention.

[0051] Figure 7This is the time-frequency domain data corresponding to the data before denoising, as shown in an embodiment of the present invention;

[0052] Figure 8 This is the time-frequency domain data corresponding to the denoising result of the constant threshold method shown in an embodiment of the present invention;

[0053] Figure 9 This is the time-frequency domain data corresponding to the denoising result of the adaptive thresholding method shown in an embodiment of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] A time-frequency domain adaptive suppression method for black triangle noise, such as Figure 1 As shown, it includes the following steps:

[0056] S1. Acquire input data, which includes: common shot gather D(t,x) after abnormal trace processing and static correction, where t represents time and x represents the position of the detector point;

[0057] S2. Calculate the linear positive normalization D_nor(t,x) of the common shot gather D(t,x) to eliminate the influence of negative values ​​and to consider the relationship between pure amplitudes; the linear positive normalization D_nor(t,x) of the common shot gather is defined as:

[0058] D_nor(t,x)=nor(D(t,x))

[0059] Where nor represents the linear positive value normalization of each row of the matrix;

[0060] S3. The trace set D_nor(t,x) after linear positive value normalization is sorted in ascending order of amplitude along the time direction. The trace set D_sort(t,x) after amplitude ascending order is defined as:

[0061] D_sort(t,x)=sort(D_nor(t,x))

[0062] Here, sort represents sorting each row of the matrix in ascending order;

[0063] S4. Based on the fact that the black triangle noise has a much stronger amplitude than the effective signal, a quadratic function is fitted to each channel of the channel set D_sort(t,x) arranged in ascending order of amplitude using the least squares method to obtain D_qua(t,x); wherein D_qua(t,x) is:

[0064]

[0065] Where ∑ represents the summation symbol, and k represents the number of detector points;

[0066] S5. Calculate the curvature of each trace along the transverse direction for the trace set after quadratic function fitting, obtaining D_cur(t,x). The curvature calculation is defined as:

[0067]

[0068] Where ′ represents the first derivative, ″ represents the second derivative, || represents the absolute value, and D_qua(t) i ,x), D_cur(t i x) represents time t i A data point at which D_cur(t) represents all time intervals. i D_cur(t,x) is composed of t,x.

[0069] S6. Extract the maximum value of the calculated curvature profile D_cur(t,x), extract a trend along the peak point of the profile, and obtain the adaptive threshold percentage curve r(t);

[0070] S7. Apply a Gaussian function to the data D_cur(t,x) used to calculate curvature along the horizontal direction to obtain the Gaussian function-fitted data D_gau(t,x), and calculate the standard deviation σ(t) along the horizontal direction. The standard deviation is:

[0071]

[0072] Where ∑ represents the summation symbol, k represents the number of detector points, μ represents the average value of the data, and D_gau(t) i x) represents time t i A data point, σ(t) for all times. i ) constitutes σ(t);

[0073] S8. Correct the adaptive threshold percentage curve r(t) based on the standard deviation σ(t) of the Gaussian function, iterating until the optimal adaptive threshold percentage curve r_mod(t) is obtained. The correction is as follows:

[0074] r_mod(t) = r(t) + σ(t);

[0075] S9. The original data D(t,x) is transformed to the time-frequency domain using a local time-frequency transform to obtain the time-frequency data volume D_transform(t,x,f), wherein the time-frequency domain data is:

[0076] D_transform(t,x,f)=F ltft (D(t,x))

[0077] Among them, F ltft Represents local time-frequency transformation;

[0078] S10. Based on the corrected adaptive threshold percentage curve r_mod(t), apply an adaptive soft threshold to the time-frequency domain data D_transform(t,x,f) to obtain the thresholded time-frequency data volume D_thr(t,x,f); the thresholded time-frequency data volume D_thr(t,x,f) is:

[0079]

[0080] Where real(D_transform(t,x,f)) represents the real part of D_transform(t,x,f), img(D_transform(t,x,f)) represents the imaginary part of D_transform(t,x,f), |D_transform(t,x,f)| represents the modulus of D_transform(t,x,f), the Real part term represents the real part of D_thr(t,x,f), and the Imaginary part term represents the imaginary part of D_thr(t,x,f), where thr(t,x,f) is:

[0081]

[0082] Where |D_transform(t,x,f)| represents the modulus of D_transform(t,x,f), r_mod_thr(t,f) is the threshold in the time-frequency domain, and r_mod_thr(t,f) = large(D_transform(t,x,f)|,r_mod(t)), where the large function represents the r_mod(t) largest value of the time-frequency domain data |D_transform(t,x,f)|.

[0083] S11. Inverse transform D_thr(t,x,f) to the time domain to obtain the black triangle noise D_noise(t,x), and subtract the black triangle noise from the original data D(t,x) to obtain the final noise attenuation result D_denoise(t,x). The final noise attenuation result D_denoise(t,x) is as follows:

[0084] D_denoise(t,x)=D(t,x)-D_noise(t,x)

[0085] in, This represents the local time-frequency inverse transform.

[0086] This invention also provides a computer device, including a processor and a memory for storing processor-executable instructions, wherein the processor executes the instructions to implement the steps of the above method.

[0087] This invention also proposes a computer-readable storage medium storing computer instructions that, when executed, implement the steps of the above-described methods. Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0088] The method provided by this invention is applied to common shot gathers containing black triangle noise in typical land exploration data. For this data, such as... Figure 2 As shown, the data is first normalized to linear positive values, and then sorted horizontally in ascending order of amplitude to obtain the trace set. Then, for each trace in the horizontal direction of the ascending trace set, a quadratic function is fitted and its curvature is calculated, as follows: Figure 3 As shown, Figure 4 This is the noise attenuation result of the traditional thresholding method. Although most of the noise has been attenuated, the effective signal at depth is still covered by noise due to the absorption and attenuation of the seismic signal and geometric diffusion, resulting in a relatively low signal-to-noise ratio. Figure 5It is the noise attenuation result of time-frequency domain adaptive thresholding. Based on the extreme points of curvature, a general trend is extracted, and then the standard deviation of the fitted Gaussian function is used to correct the curve and apply it to the time-frequency domain adaptive thresholding. Compared with the traditional thresholding method, it can significantly improve the signal-to-noise ratio of the data and improve the continuity of the in-phase axis. Figure 6 yes Figure 5 The corresponding noise difference profile. Time-frequency domain data corresponding to the original data ( Figure 7 Compared to traditional thresholding methods, which produce time-frequency domain data with residual noise due to incomplete thresholding, such as... Figure 8 As shown. The time-frequency domain adaptive threshold considers the trade-off between deep and shallow amplitude information, generating amplitude-balanced time-frequency domain data, such as... Figure 9 As shown, this method achieves a significant improvement in overall signal-to-noise ratio without losing any useful information. It can be seen that, compared to conventional thresholding methods, setting different thresholds at different depths based on an adaptive threshold curve in the time-frequency domain can substantially reduce noise residue caused by inaccurate thresholds, resulting in high signal-to-noise ratio black triangle noise attenuation. This provides technical support for black triangle noise attenuation in exploration areas with undulating terrain.

[0089] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A time-frequency domain adaptive suppression method for black triangle noise, characterized in that, Includes the following steps: S1. Acquire input data, which includes: common shot gather D(t,x) after abnormal trace processing and static correction, where t represents time and x represents the position of the detector point; S2. Calculate the linear positive value normalization D_nor(t,x) of the common shot point gather D(t,x); S3. The trace set D_nor(t,x) after linear positive value normalization is sorted in ascending order of amplitude along the time direction. The trace set D_sort(t,x) after ascending order of amplitude is defined as follows: D_sort(t,x)=sort(D_nor(t,x)) Here, sort represents sorting each row of the matrix in ascending order; S4. For each trace in the horizontal direction of the trace set D_sort(t,x) after the amplitude is sorted in ascending order, perform quadratic function fitting based on the least squares method to obtain D_qua(t,x); S5. Calculate the curvature of each trace along the transverse direction for the trace set after quadratic function fitting, obtaining D_cur(t,x). The curvature calculation is defined as: Where ′ represents the first derivative, ″ represents the second derivative, || represents the absolute value, and D_qua(t) i ,x), D_cur(t i x) represents time t i A data point at which D_cur(t) represents all time intervals. i D_cur(t,x) is composed of t,x. S6. Extract the maximum value of the calculated curvature profile D_cur(t,x), extract a trend along the peak point of the profile, and obtain the adaptive threshold percentage curve r(t); S7. Apply a Gaussian function to the data D_cur(t,x) used to calculate curvature along the horizontal direction to obtain the Gaussian function-fitted data D_gau(t,x), and calculate the standard deviation σ(t) along the horizontal direction. The standard deviation is: Where ∑ represents the summation symbol, k represents the number of detector points, μ represents the average value of the data, and D_gau(t) i x) represents time t i A data point, σ(t) for all times. i ) constitutes σ(t); S8. Correct the adaptive threshold percentage curve r(t) based on the standard deviation σ(t) of the Gaussian function, iterating until the optimal adaptive threshold percentage curve r_mod(t) is obtained. The correction is as follows: r_mod(t) = r(t) + σ(t); S9. The original data D(t,x) is transformed to the time-frequency domain using a local time-frequency transform to obtain the time-frequency data volume D_transform(t,x,f), wherein the time-frequency domain data is: D_transform(t,x,f)=F ltft (D(t,x)) Among them, F ltft Represents local time-frequency transformation; S10. Based on the modified adaptive threshold percentage curve r_mod(t), apply an adaptive soft threshold to the time-frequency domain data D_transform(t,x,f) to obtain the thresholded time-frequency data volume D_thr(t,x,f). S11. Inverse transform D_thr(t,x,f) to the time domain to obtain the black triangle noise D_noise(t,x), and subtract the black triangle noise from the original data D(t,x) to obtain the final noise attenuation result D_denoise(t,x). The final noise attenuation result D_denoise(t,x) is as follows: D_denoise(t,x)=D(t,x)-D_noise(t,x) in, This represents the local time-frequency inverse transform.

2. The time-frequency domain adaptive suppression method for black triangle noise as described in claim 1, characterized in that, In step S2, the linear positive normalized D_nor(t,x) of the common shot point gather is defined as: D_nor(t,x)=nor(D(t,x)) Here, nor represents the linear positive value normalization of each row of the matrix.

3. The time-frequency domain adaptive suppression method for black triangle noise as described in claim 1, characterized in that, In step S4, D_qua(t,x) is: Where ∑ represents the summation symbol, and k represents the number of detector points.

4. The time-frequency domain adaptive suppression method for black triangle noise as described in claim 1, characterized in that, In step S10, the time-frequency data volume D_thr(t,x,f) after the threshold is: Where real(D_transform(t,x,f)) represents the real part of D_transform(t,x,f), img(D_transform(t,x,f)) represents the imaginary part of D_transform(t,x,f), |D_transform(t,x,f)| represents the modulus of D_transform(t,x,f), the Real part term represents the real part of D_thr(t,x,f), and the Imaginary part term represents the imaginary part of D_thr(t,x,f), where thr(t,x,f) is: Where |D_transform(t,x,f)| represents the modulus of D_transform(t,x,f), r_mod_thr(t,f) is the threshold in the time-frequency domain, and r_mod_thr(t,f) = large(|D_transform(t,x,f)|,r_mod(t)), where the large function represents the r_mod(t) largest value of the time-frequency domain data |D_transform(t,x,f)|.

5. A computer device, characterized in that, It includes a processor and a memory for storing processor-executable instructions, wherein the processor, when executing the instructions, implements the steps of the method according to any one of claims 1 to 4.

6. A computer-readable storage medium storing computer instructions thereon, characterized in that, When the instructions are executed, they implement the steps of the method according to any one of claims 1 to 4.

Images