A method for adaptive denoising of seismic data of sandstone type uranium mine based on sparse transform

Abstract

This invention pertains to data processing technology for seismic exploration of sandstone-type uranium deposits, specifically an adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transform. The method includes: inputting noisy raw two-dimensional seismic data y; initializing the maximum number of iterations I, setting the outer iterative accumulator i=1, and designing a search range J for the denoising coefficient truncation percentage; for each i, calculating the curvelet transform coefficients s of y; for each j within the search range, applying hard thresholding for denoising, and performing an inverse curvelet transform on the denoised coefficients to obtain the denoised seismic data y. j ; Calculate the noise removal ε corresponding to each j j The entropy; compare the calculated noise ε removed for each j. j The entropy is calculated, and the percentage corresponding to the maximum entropy within the search range is taken as the optimal denoising effect, and the corresponding seismic data is output. If i reaches the maximum number of iterations, the loop is exited, and the denoised seismic data corresponding to the current maximum entropy value is used as the output result. This invention can protect the effective signal while suppressing random noise.

Description

Technical Field

[0001] This invention pertains to the processing technology of seismic exploration data for sandstone-type uranium deposits, specifically involving an adaptive denoising method for seismic data of sandstone-type uranium deposits based on sparse transformation. Background Technology

[0002] In seismic exploration of sandstone uranium deposits, the target layer depth is generally less than 1000m, and the acquired seismic data is often subject to severe random noise interference. Furthermore, the low number of shallow stacking layers significantly impacts the signal-to-noise ratio. Random noise attenuation is crucial throughout the data processing workflow; effectively suppressing random noise while preserving the effective signal is a key challenge in sandstone uranium seismic exploration. Traditionally, multiple stacking and median or mean filtering are used to directly eliminate random noise in the spatiotemporal (tx) domain. More complexly, spatiotemporal or frequency-space (fx) domain predictive filters are used to eliminate incoherent noise without compromising the seismic phase axis. To date, currently used random noise reduction methods include time-frequency domain denoising methods, sparse representation-based denoising (including wavelet transform, curvelet transform, and Seislets), overcomplete dictionary learning methods, deep learning-based denoising methods, and combinations of all these methods.

[0003] In summary, the more advanced the method, the higher the computational cost. The kernel function used in sparse transform is particularly suitable for characterizing seismic wavefronts, and denoising methods based on sparse transform can significantly improve the signal-to-noise ratio of seismic data while fully preserving valuable information. However, in the specific implementation of actual data processing, due to the unknown model, there is often a lack of standards for selecting denoising parameters. The final denoising effect is often evaluated using empirical methods such as subjective visual effects, and whether the optimal denoising effect has been achieved remains unknown. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive denoising method for seismic data of sandstone-type uranium deposits based on sparse transformation. This method can protect the effective signal while suppressing random noise in the processing of seismic exploration data of sandstone-type uranium deposits.

[0005] Technical solution to achieve the purpose of this invention:

[0006] An adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transform, the method comprising:

[0007] Step 1: Input the raw 2D seismic data y containing noise;

[0008] Step 2: Initialize the maximum number of iterations I, set the outer iterator i=1, and design the search range J for the denoising coefficient truncation percentage.

[0009] Step 3: For each i, calculate the curvelet positive transform coefficient s of y;

[0010] Step 4: For each j within the search range, apply hard thresholding for denoising, and perform inverse curvelet transform on the denoised coefficients to obtain the denoised seismic data y. j ;

[0011] Step 5: Calculate the noise removal ε for each j. j Entropy;

[0012] Step 6: Compare the noise ε removed for each j calculated. j The entropy is taken as the percentage corresponding to the largest entropy within the search range, which is the best noise reduction effect, and the corresponding seismic data is output.

[0013] Step 7: If i reaches the maximum number of iterations, exit the loop and take the denoised seismic data corresponding to the maximum current entropy as the output result.

[0014] The formula for calculating the curvelet positive transform coefficient s of y in step 3 is as follows:

[0015]

[0016] Where Ψ is the curve wave positive transform matrix. is a curve wave atom, and μ is a curve wave atom parameter.

[0017] The coefficients after hard threshold denoising in step 4 The calculation formula is:

[0018]

[0019] The denoised seismic data y in step 4 j The calculation formula is:

[0020]

[0021] Step 5 specifically involves: setting the mean of the noise data to 0, and calculating the overall variance σ of the noise. j Considering that most of the noise amplitude data are concentrated near the 0 value, the noise data is probabilistically mapped using the probability density function to obtain the mapped data; the mapped data is then mapped to gray levels related to the noise amplitude value to obtain an integer array of noise data; combined with the total length of the vectorized seismic data, the probability of occurrence of amplitude data in each interval is statistically obtained, and the Tsallis entropy of the gray data is calculated.

[0022] The formula for calculating the probability mapping of the noise data is as follows:

[0023]

[0024] Among them, u iFor mapping data, ε j Noise to be removed.

[0025] The formula for remapping the mapped data to a gray level related to the noise amplitude value is as follows:

[0026]

[0027] The probability P of the occurrence of amplitude data in each interval is statistically analyzed. i The calculation formula is:

[0028]

[0029] in, It is an array The count in the i-th interval, array For {z j The summation within each interval yields a new array, N. b To use the Sturges formula to divide the histogram and count the number of intervals.

[0030] The formula for calculating the Tsallis entropy TsEn of the grayscale data is:

[0031]

[0032] The beneficial technical effects of this invention are as follows:

[0033] This invention provides an adaptive denoising method for seismic data of sandstone-type uranium deposits based on sparse transform. Seismic data in the curvelet domain exhibits the best sparsity, and curvelet transform, as a typical sparse transform method, is best suited for seismic wavefront characterization and random noise attenuation in the field of sandstone uranium deposit exploration. The curvelet transform denoising method based on entropy measure parameter optimization can achieve the best denoising effect and effectively avoids subjective evaluation of the denoising effect on actual data. Attached Figure Description

[0034] Figure 1 The flowchart illustrates an adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation, as provided in this invention. Detailed Implementation

[0035] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0036] like Figure 1 As shown, the present invention provides an adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation, which specifically includes the following steps:

[0037] Step 1: Input the raw two-dimensional seismic data y(x) containing noise (which can be single-shot data or post-stack data), and the maximum number of outer layer iterations I.

[0038] Step 2: Initialize the maximum number of iterations I, set the outer iterator i=1, and design the search range J for the denoising coefficient truncation percentage.

[0039] Step 3: For each i, calculate the curvelet positive transform coefficient s of y:

[0040]

[0041] Where Ψ is the curve wave positive transform matrix. is a curve wave atom, and μ is a curve wave atom parameter.

[0042] Step 4: For each j within the search range, sort the curve coefficients in descending order to determine the cutoff threshold s. j (>0), the corresponding percentage is θ j Hard threshold denoising is used:

[0043]

[0044] And based on the denoised coefficients Performing inverse curvelet transform yields the denoised seismic data y. j :

[0045]

[0046] The noise removed is ε j :

[0047] ε j =yy j

[0048] Soft thresholding can be used as an alternative to hard thresholding, but hard thresholding is simpler and different threshold forms do not significantly affect the results.

[0049] Step 5: Calculate the noise removal ε for each j. j Entropy:

[0050] Given that the mean of the noise data is 0, calculate the overall variance σ of the noise. j Considering that most of the noise amplitude data is concentrated around 0, a probability density function is used to perform probability mapping on the noise data:

[0051]

[0052] Among them, u i For mapping data, ε j To remove noise;

[0053] The mapped data is then mapped to gray levels (defined as 1 to C) related to the noise amplitude value:

[0054]

[0055] Where C can be a large integer. Thus, the noise data ε j Transform it into an integer array {z} with a range of [1, C]. j}

[0056] Assuming the total length of the vectorized earthquake data is N, the number of histogram intervals divided using the Sturges formula is: Nb = 1 + log₂N. j The new array is obtained by summing within each interval. Therefore, the probability of occurrence of amplitude data in each interval is calculated:

[0057]

[0058] in, It is an array The count in the i-th interval.

[0059] Calculate the Tsallis entropy of grayscale data:

[0060]

[0061] Generally, q = 2 is chosen. Tsallis entropy is used for two reasons: firstly, it has only one summation term, making it computationally efficient. For small datasets, other entropy measures such as Shannon entropy or Renyi entropy can be used instead; secondly, Tsallis entropy incorporates the statistical properties of extensive systems and has good expressive power for potentially non-Gaussian distributed noisy data volumes.

[0062] Step 6: Compare the TsEn obtained for each j, and take the θ corresponding to the largest TsEn within the search range. j This represents the optimal noise reduction effect; simply output the corresponding seismic data.

[0063]

[0064] The reason for choosing the maximum Tsallis entropy as the optimal solution for denoising is due to the inherent characteristics of noisy data. The more developed the random noise, the lower the signal-to-noise ratio, and the higher the degree of disorder, i.e., the greater the entropy measure. Therefore, the most ordered and cleanest seismic data should be obtained when the entropy of the noisy data reaches its maximum.

[0065] In the process of truncating parameter optimization, in addition to the simple uniform grid search given in this example and gradually narrowing the range and refining the search grid to reduce the step size, nonlinear global optimization algorithms such as Genetic Algorithm, Particle Swarm Optimization, and Differential Evolution can also be directly used to obtain the optimal parameters under the given data accuracy conditions, while the overall logic of the denoising process remains unchanged.

[0066] Step 7: If i reaches the maximum number of iterations, exit the loop and output the denoised data corresponding to the maximum value of the existing Tsallis entropy as the best result.

[0067] The present invention has been described in detail above with reference to the accompanying drawings and embodiments. However, the present invention is not limited to the above embodiments, and various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention. All contents not described in detail in the present invention can be derived from existing technologies.

Claims

1. An adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation, characterized in that, The method includes: Step 1: Input the raw 2D seismic data y containing noise; Step 2: Initialize the maximum number of iterations I, set the outer iterator i=1, and design the search range J for the denoising coefficient truncation percentage. Step 3: For each i, calculate the curvelet positive transform coefficient s of y; Step 4: For each j within the search range, apply hard thresholding for denoising, and perform inverse curvelet transform on the denoised coefficients to obtain the denoised seismic data y. j ; Step 5: Calculate the noise removal ε for each j. j Entropy; Step 6: Compare the noise ε removed for each j calculated. j The entropy is taken as the percentage corresponding to the largest entropy within the search range, which is the best noise reduction effect, and the corresponding seismic data is output. Step 7: If i reaches the maximum number of iterations, exit the loop and take the denoised seismic data corresponding to the maximum current entropy as the output result.

2. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 1, characterized in that, The formula for calculating the curvelet positive transform coefficient s of y in step 3 is as follows: Where Ψ is the curve wave positive transform matrix. is a curve wave atom, and μ is a curve wave atom parameter.

3. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 2, characterized in that, The coefficients after hard threshold denoising in step 4 The calculation formula is:

4. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 3, characterized in that, The denoised seismic data y in step 4 j The calculation formula is:

5. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 4, characterized in that, Step 5 specifically involves: setting the mean of the noise data to 0, and calculating the overall variance σ of the noise. j Considering that most of the noise amplitude data are concentrated near the 0 value, the noise data is probabilistically mapped using the probability density function to obtain the mapped data; the mapped data is then mapped to gray levels related to the noise amplitude value to obtain an integer array of noise data; combined with the total length of the vectorized seismic data, the probability of occurrence of amplitude data in each interval is statistically obtained, and the Tsallis entropy of the gray data is calculated.

6. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 5, characterized in that, The formula for calculating the probability mapping of the noise data is as follows: Among them, u i For mapping data, ε j Noise to be removed.

7. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 6, characterized in that, The formula for remapping the mapped data to a gray level related to the noise amplitude value is as follows:

8. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 7, characterized in that, The probability P of the occurrence of amplitude data in each interval is statistically analyzed. i The calculation formula is: in, It is an array The count in the i-th interval, array For {z j The summation within each interval yields a new array, N. b To use the Sturges formula to divide the histogram and count the number of intervals.

9. The adaptive denoising method for sandstone-type uranium deposit seismic data based on sparse transformation according to claim 8, characterized in that, The formula for calculating the Tsallis entropy TsEn of the grayscale data is:

Images