CN104849757A - System and method for eliminating random noise in seismic signals - Google Patents
System and method for eliminating random noise in seismic signals Download PDFInfo
- Publication number
- CN104849757A CN104849757A CN201510222712.2A CN201510222712A CN104849757A CN 104849757 A CN104849757 A CN 104849757A CN 201510222712 A CN201510222712 A CN 201510222712A CN 104849757 A CN104849757 A CN 104849757A
- Authority
- CN
- China
- Prior art keywords
- frequency subband
- threshold
- random noise
- signal
- seismic signal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Geophysics And Detection Of Objects (AREA)
Abstract
The present invention provides a system and a method for eliminating the random noise in seismic signals, which comprises a curvelet transformation unit, a threshold optimization unit, a threshold de-noising unit, a filter design unit, a spectral whitening processing unit, and a curvelet inverse-transformation unit. The curvelet transformation unit is used for reading original seismic signal data, and obtaining first frequency sub-bands and second frequency sub-bands in various scales and all directions within a curvelet domain through decomposing the data based on the wavelet transformation process. The threshold optimization unit is used for determining the threshold values of curvelet coefficients in various scales and all directions, and obtaining an optimized threshold based on the genetic algorithm, wherein the optimized threshold is set in such a manner that the risk assessment function of the generalized cross validation criteria has a minimum value. The threshold de-noising unit is used for conducting the threshold de-noising treatment on the first and second frequency sub-bands obtained through the decomposing process. The filter design unit is used for designing different self-adaptive spectral whitening filters. The spectral whitening processing unit is used for conducting the spectral whitening treatment on effective signals of the first frequency sub-bands and the second frequency sub-bands in various scales and all directions. The curvelet inverse-transformation unit is used for reconstructing the signals through the curvelet inverse-transformation process to obtain de-noised seismic signal data of higher resolution.
Description
Technical field
The present invention relates to Denoising of Seismic Data technical field, particularly one eliminates random noise system and method in seismic signal.
Background technology
Hydrocarbon resources constantly promotes the progress of the mankind as the main energy sources of society, and the growing oil gas energy demand of people is had higher requirement to oil-gas exploration.But along with deepening continuously of exploration, difficulties in exploration also constantly strengthens, be mainly reflected in complex structure, in the exploration of the impalpable subtle reservoir formation of seismic section feature, therefore denoising carried out to signal, improve the resolution of seismic data but exploration and development of the subtle reservoir formation that can increasing the storage produce little to area and have great significance.
For the raising improving the traditional disposal route many emphasis signal to noise ratio (S/N ratio) of this difficult problem of seismic data resolution, fail by widening of frequency band, resolution to be further improved after effectively improving signal to noise ratio (S/N ratio), at raising signal to noise ratio (S/N ratio) this respect, traditional wavelet transformation has good ability to express to one dimension smooth signal, but and be not suitable for express 2D signal, this makes the seismic signal after Wavelet Denoising Method have lost edge detail information, causes that lineups edge thickens, resolution declines.For this limitation, the warp wavelet that small echo basis grows up has good azimuth characteristic, enables it marginal information and noise information well be separated, while Retain edge information, also serves good effect to the removal of noise.Utilize the feature of the multiple dimensioned multidirectional of warp wavelet, the threshold value under adopting GCV (Generalized Cross Validation criterion) that different scale and direction can be determined, and make GCV risk assessment function have optimal threshold with Genetic algorithm searching; Use corresponding semisoft shrinkage function to obtain subband to warp wavelet decomposition by different threshold rule again to carry out threshold denoising, effectively can improve signal to noise ratio (S/N ratio).Widening in frequency band, spectral whitening process is a kind of effective ways of widen spectrum, and it estimates the frequency content outside this frequency band after carrying out net amplitude filtering to limited frequency band, reaches and opens up wide band object.Spectral whitening process can process in time domain or in frequency field, if process can carry out spectral whitening to the useful signal in the subband of different scale, different directions in bent wave zone, widens its frequency band; Random noise is not then processed.In conjunction with before denoising can improve the resolution of seismic signal further.If application number is the method and system that CN201210483278.X discloses a kind of earthquake data random noise that decays.In the method, geological data is carried out Fourier transform, the geological data of generated frequency-spatial domain; Direction in space carries out complex empirical mode decomposition to the geological data of Frequency-Space Domain, generates multiple modal components; According to geological data and multiple modal components of Frequency-Space Domain, optimization method is utilized to generate self-adapting signal reconstruct operator; According to self-adapting signal reconstruct operator and described multiple modal components, reconstruct generated frequency territory seismic signal; Frequency field seismic signal is carried out inverse fourier transform, generates the time domain seismic signal after random noise attenuation.If application number is that CN201210247721.3 discloses a kind of small scale Threshold Denoising Method based on wavelet transformation, the method small scale is shaken data ground-to-ground and is scanned, correlation coefficient value when obtaining one in window, then a threshold value is set, data in window during this small scale are judged, when seismic trace with seismic signal for time main, adopt conventional wavelet to decompose and conventional hard-threshold or soft-threshold; When seismic trace with noise for time main, after seismic signal carries out WAVELET PACKET DECOMPOSITION, adopt best entropy policy setting below method of floating threshold value.Finally the wavelet scale after denoising is carried out wavelet reconstruction, thus obtain the higher seismic channel set of signal to noise ratio (S/N ratio).
Within 1999, Donoho and Candes proposes warp wavelet theory on the basis that Ridgeletb converts, and it is to the process of multi-dimensional signal having the characteristic being better than wavelet transformation.Warp wavelet has multiple dimensioned, multidirectional, has good resolution characteristic to multi-dimensional signal, and this makes it more effective than wavelet transformation in the expression of multi-dimensional signal feature and extraction.According to the difference of seismic data type, the method of current raising seismic signal resolution have based on VSP (vertieal seismical profile) improve seismic signal resolution method, open up process frequently improves the method for seismic signal resolution, time varying spectrum albefaction improves seismic signal resolution method, deconvolution improves method of seismic signal resolution etc., these methods can reach stress release treatment, improve the effect of seismic signal resolution.
But prior art cannot eliminate the random noise in high-resolution seismic exploration signal.
Summary of the invention
For solving the problem cannot eliminating the random noise in seismic signal in existing raising seismic data resolution technology, be necessary that providing a kind of eliminates random noise system and method in seismic signal.
Random noise system in a kind of elimination seismic signal, comprises as lower unit:
Warp wavelet unit, for reading original seismic signal data, and original seismic signal data march wave conversion is decomposed to the second frequency subband obtaining seismic signal and each yardstick, all directions represent the first frequency subband of the spectral signature of image and the space characteristics of expression image in bent wave zone, wherein the frequency range of first frequency subband is 10-20Hz, and the frequency range of second frequency subband is 50-200Hz;
Threshold optimization unit, for for decomposing the different scale, the bent wave system number under different directions that comprise in the first frequency subband that obtains and second frequency subband, by the threshold value of bent wave system number under Generalized Cross Validation criterion determination different scale, different directions, this threshold value is useful signal and random noise in the cut off value of direction and yardstick; And make Generalized Cross Validation criterion risk assessment function have the optimal threshold of minimum value by genetic algorithm acquisition;
Threshold denoising unit, for by being brought into by each optimal threshold obtained in threshold optimization unit, semisoft shrinkage function obtains first frequency subband to decomposition, second frequency subband carries out threshold denoising;
Design of filter unit, for the difference of the correlativity according to the first frequency subband after denoising in threshold denoising unit, the useful signal in second frequency subband and random noise, designs different adaptive spectral whitening wave filters;
Spectral whitening processing unit, for carrying out spectral whitening to the useful signal of different scale, different directions in first frequency subband, second frequency subband by corresponding adaptive spectral whitening wave filter, widens the frequency band of useful signal; Random noise is not then processed;
Bent ripple inverse transformation unit, for being reconstructed by bent ripple inverse transformation each yardstick after spectral whitening process, the first frequency subband in all directions, second frequency subband, after obtaining denoising and the seismic signal data that improves of resolution.
A kind of random noise method in elimination seismic signal, comprises the steps:
S1, read original seismic signal data, and original seismic signal data march wave conversion is decomposed to the second frequency subband obtaining seismic signal and each yardstick, all directions represent the first frequency subband of the spectral signature of image and the space characteristics of expression image in bent wave zone, wherein the frequency range of first frequency subband is 10-20Hz, and the frequency range of second frequency subband is 50-200Hz;
S2, for decomposing the different scale, the bent wave system number under different directions that comprise in the first frequency subband that obtains and second frequency subband, by the threshold value of bent wave system number under Generalized Cross Validation criterion determination different scale, different directions, this threshold value is useful signal and random noise in the cut off value of direction and yardstick; And make Generalized Cross Validation criterion risk assessment function have the optimal threshold of minimum value by genetic algorithm acquisition;
S3, by be brought into by each optimal threshold obtained in step S2, semisoft shrinkage function obtains first frequency subband to decomposition, second frequency subband carries out threshold denoising;
S4, difference according to the correlativity of the first frequency subband after denoising in step S3, the useful signal in second frequency subband and random noise, design different adaptive spectral whitening wave filters;
S5, by corresponding adaptive spectral whitening wave filter, spectral whitening is carried out to the useful signal of different scale, different directions in first frequency subband, second frequency subband, widen the frequency band of useful signal; Random noise is not then processed;
S6, each yardstick after spectral whitening process, the first frequency subband in all directions, second frequency subband to be reconstructed by bent ripple inverse transformation, after obtaining denoising and the seismic signal data that improves of resolution.
Random noise system and method in elimination seismic signal provided by the invention; to seismic signal march Wave Decomposition; projected in bent wave zone; take full advantage of multiple dimensioned, multidirectional advantage of warp wavelet; and choose optimal threshold by Generalized Cross Validation criterion and genetic algorithm; come thus to carry out denoising to seismic signal, while the effective seismic reflection signals of protection, more accurately can filter noise.Simultaneously according to the difference of the correlativity of useful signal and random noise, design Adaptive spectra prewhitening filter.Only carry out spectral whitening to useful signal, widen its frequency band, outstanding useful signal, reaches the object of the resolution that improve seismic signal.
Accompanying drawing explanation
Fig. 1 is random noise system architecture diagram in the elimination seismic signal that provides of embodiment of the present invention;
Fig. 2 is random noise method flow diagram in the elimination seismic signal that provides of embodiment of the present invention;
Fig. 3 is the schematic diagram of bent ripple basis function in time domain and frequency field;
Fig. 4 is the design sketch of original earthquake data display;
Fig. 5 is the design sketch that after process, signal resolution is improved;
Fig. 6 is the effect contrast figure before and after process;
Fig. 7 is the comparison diagram to signal to noise ratio (S/N ratio) before and after many group seismic data processing.
Embodiment
As shown in Figure 1, random noise system in a kind of elimination seismic signal, comprises as lower unit:
Warp wavelet unit, for reading original seismic signal data, and original seismic signal data march wave conversion is decomposed to the second frequency subband obtaining seismic signal and each yardstick, all directions represent the first frequency subband of the spectral signature of image and the space characteristics of expression image in bent wave zone, wherein the frequency range of first frequency subband is 10-20Hz, different geology composition is represented for seismic section, the frequency range of second frequency subband is 50-200Hz, represents detail section.
Carry out the process such as noise compacting, extending bandwidth there is many limitation owing to analyzing seismic signal in spatial domain, therefore the embodiment of the present invention passes through warp wavelet, seismic signal in spatial domain is projected to bent wave zone, also exist multiple dimensioned and multi-direction in bent wave zone, thus can carry out meticulousr analysis to seismic signal.
Alternatively, geological data to calculator memory, and is shown by the geological data in file reading;
First, the file manipulation function in Matlab can be called, in a binary fashion file reading head, read earthquake number of channels respectively, sampling number, the important informations such as sampling interval according to the storage format of geological data.Seismic signal data in file reading, wherein each seismic trace is made up of the trace header of 240 bytes and seismologic record data, and handles data based on Matlab by matrix, very convenient when therefore processing this 2-D data.
Warp wavelet does inner product by bent ripple basis function and objective function and realizes, in order to build bent ripple basis function, then need introducing two window functions, i.e. radius window W (r) and angle window V (t), wherein r and t is polar coordinate position, and both all meet admissible condition:
Wherein, define
for the basis function of Qu Bo, then other 2
-jthe bent ripple basis function of yardstick can pass through
rotation and translation obtain.The definition anglec of rotation
θ
l=2 π * 2
-[j/2]* l, l=0,1,2 ..., 0≤θ
l≤ 2 π; Translation parameters k=(k
1, k
2) ∈ Z
2.At yardstick 2
-j, direction θ
l, position
the bent ripple basis function at place is defined as:
Wherein R
θit is the rotation to θ radian.Yardstick 2-j all Qu Bo all by
rotation and translation obtain.
Bent wave system number c (j, k, l) is by f ∈ L
2(R
2) and
inner product to define:
Warp wavelet comprises the part compared with large scale and thinner yardstick, comparatively r ∈ (3/4 in large scale respective radius window, 1) part, comparatively r ∈ (1 in small scale respective radius window, 3/2) part, wherein not there is directivity compared with the warp wavelet under large scale, so whole warp wavelet is by compared with the direction element under small scale with compared with little wave component isotropic under large scale.
Frequency window uses concentric square areas to represent bent wave zone in continuous domain, discrete warp wavelet schematic diagram as shown in Figure 2, shadow region in this figure is wedge shape, the yardstick of warp wavelet is divided into the high frequency layer (Fine layer) of outermost from high to low according to frequency, middle intermediate frequency layer (Detail layer), the low frequency layer (Coarse layer) of innermost layer, direction clockwise direction from direction 1 (arrow pointed location) on each yardstick is followed successively by 2,3, N, the subband on each direction is made up of corresponding bent wave system number.
Threshold optimization unit, for for decomposing the different scale, the bent wave system number under different directions that comprise in the first frequency subband that obtains and second frequency subband, by the threshold value of bent wave system number under Generalized Cross Validation criterion determination different scale, different directions, this threshold value is useful signal and random noise in the cut off value of direction and yardstick; And make Generalized Cross Validation criterion risk assessment function have the optimal threshold of minimum value by genetic algorithm acquisition.Add the accuracy of threshold value.
Because noise and useful signal distributing position is in the propagation direction different, so after transforming to bent wave zone, on different scale and direction, the characteristic distributions of noise and useful signal is different, useful signal is mainly distributed in the relatively large position of bent wave system number, and random noise is then mainly distributed on the less position of bent wave system number.Therefore according to the difference that they distribute in bent wave zone, suitable threshold value is set and carries out filtering, effectively can suppress noise like this.
In order to ensure to remove noise as much as possible under the condition that useful signal is injury-free, suitable threshold value is arranged to the bent wave system number under different yardsticks, in embodiments of the present invention, alternatively, apply Generalized Cross Validation criterion GCV determine different scale and direction under threshold value.GCV is a kind of and depends on input and output signal when not needing to estimate noise variance, obtains the asymptotic optimization value of threshold value by minimum error function.According to the following objective function GCV of GCV rule definition (K
j,l), and determine asymptotic optimization threshold value T by objective function:
T＝argmin[GCV(K
j,l)] (6)
Wherein K
j,lfor the threshold value on j yardstick l direction, C
j, l, k, nfor the bent wave system number in j yardstick l direction, C
j, l, k, nfor the bent wave system number after thresholding, N
j,lfor j yardstick l side upsweeps the number of wave system number, N
j, l, 0for by the C of zero setting
j, l, k, nnumber.As GCV (K
j,l) when getting minimum value, just can obtain optimum threshold value T, be used for screening useful signal, thus Attenuating Random Noise.
At GCV (K
j,l) get in excellent process, employing genetic algorithm is determined optimal threshold by this method, and genetic algorithm is a kind of stochastic search methods be widely used, and it is a kind of optimization algorithm set up according to the rule of " survival of the fittest in natural selection; the survival of the fittest ", and the method has multiple spot optimizing and efficiency high.Generalized Cross Validation criterion risk assessment function f it formula is as follows:
When this function f it obtains minimum value, GCV (K
j,l) also will obtain minimum value, now corresponding threshold value is also optimal value.
Threshold denoising unit, for by being brought into by each optimal threshold obtained in threshold optimization unit, semisoft shrinkage function obtains first frequency subband to decomposition, second frequency subband carries out threshold denoising.The characteristic distributions of bent wave system number can be made full use of, the noise more accurately in filtering seismic signal and protect effective seismic reflection signals, improve signal to noise ratio (S/N ratio).
Alternatively, described threshold denoising unit comprises: arrange the first compare threshold and the second compare threshold, the first compare threshold is less than the second compare threshold; The bent wave system number being less than the first compare threshold is set to 0, keeps the bent wave system number being greater than the second compare threshold constant; And the value of other bent wave system numbers is reduced by semisoft shrinkage function.
The optimal threshold obtained in threshold optimization unit is substituted into threshold function table, and here we adopt semisoft shrinkage function.Semisoft shrinkage function combines hard-threshold and soft threshold method, is provided with two threshold values varied in size, and retains the coefficient being greater than threshold value, will be less than the coefficient zero setting of threshold value, suitably reduces the coefficient of other threshold value.So just can carry out threshold denoising to decomposing the subband obtained, improving its signal to noise ratio (S/N ratio).Its threshold function table C'(j, l, k) formula is as follows:
Wherein sgn [C (j, l, the k)] symbol that is bent wave system number.T
1be the first compare threshold, T
2it is the second compare threshold.
Design of filter unit, for the difference of the correlativity according to the first frequency subband after denoising in threshold denoising unit, the useful signal in second frequency subband and random noise, designs different adaptive spectral whitening wave filters.
Alternatively, described design of filter unit comprises:
The related coefficient of the first frequency subband in acquisition threshold denoising unit after denoising, the useful signal in second frequency subband and random noise;
The related coefficient of useful signal and random noise is normalized and obtains normalized related coefficient;
Correlation factor is obtained according to related coefficient; Obtain the envelope of the spectral amplitude of each useful signal in first frequency subband in bent wave zone, second frequency subband;
Envelope according to different correlation factor and spectral amplitude designs different adaptive spectral whitening wave filters.
The seismic signal subband that signal to noise ratio (S/N ratio) is different can be distinguished, subband higher for signal to noise ratio (S/N ratio) under different scale is identified.
Utilize useful signal different with the correlativity of random noise, under Matlab platform, first calculate the correlation coefficient r of useful signal and random noise, its formula is as follows:
r＝C
1(j,l,k)*C
2(j,l,k)
Wherein, C
1(j, l, k) is the bent wave system number of useful signal, C
2(j, l, k) is the bent wave system number of random noise.
Secondly, by related coefficient obtain normalized correlation coefficient r ', its expression formula is as follows:
Step 502: by correlation coefficient r ' obtain correlation factor θ, utilize different θ values to distinguish random noise and useful signal.Be random noise as θ >1, when θ≤1, the less signal to noise ratio (S/N ratio) of θ is larger, arranges different wave filter f (ω), is sef-adapting filter, to carry out spectral whitening process according to the difference of θ.
θ＝|C
2(j,l,k)|/|r'| (11)
Step 601: for the spectral amplitude of every one useful signal in each subband in bent wave zone, asks the envelope e (ω) of its spectral amplitude.Design an Adaptive spectra prewhitening filter f (ω) for strengthening useful signal according to amplitude envelope, its formula is as follows:
In formula, υ is the maximal value of e (ω), ε (θ) is the white noise factor, object is compromise signal to noise ratio (S/N ratio) and resolution, the white noise factor changes along with the change of correlation factor θ, namely larger to the subdata body widening frequency band that signal to noise ratio (S/N ratio) is large, the subdata body widening frequency band little to signal to noise ratio (S/N ratio) is smaller, does not widen frequency band to the data volume being random noise.
Spectral whitening processing unit, for carrying out spectral whitening to the useful signal of different scale, different directions in first frequency subband, second frequency subband by corresponding adaptive spectral whitening wave filter, widens the frequency band of useful signal; Random noise is not then processed.
Alternatively, described spectral whitening processing unit comprises:
The signal to noise ratio (S/N ratio) of the subdata body of different scale, different directions in first frequency subband, second frequency subband is increased according to the white noise factor in spectral whitening wave filter; Thus widen the frequency band of useful signal; Random noise is not then processed.
Like this can effectively opening up wide band while restraint speckle, improve resolution further.
Step 602: adopt ε (θ) to carry out self-adaptation, wherein white noise factor ε (θ) diminishes with the reduction of θ, becomes large with the increase of θ.ε
0and ε
1be two constants.Then use each subdata body C (j, l) obtained by two-dimentional warp wavelet, both improved through Adaptive spectra prewhitening filter the subdata body that resolution also increases signal to noise ratio (S/N ratio)
ε (θ) formula is as follows:
ε(θ)＝ε
0+ε
1θ θ≤1 (13)
Bent ripple inverse transformation unit, for being reconstructed by bent ripple inverse transformation each yardstick after spectral whitening process, the first frequency subband in all directions, second frequency subband, after obtaining denoising and the seismic signal data that improves of resolution.
Alternatively, described bent ripple inverse transformation unit comprises:
To the subdata body adding different scale, different directions in the first frequency subband of signal to noise ratio (S/N ratio), second frequency subband
march ripple inverse transformation, thus after obtaining denoising and the seismic signal data that is enhanced of resolution.
The present invention can test under Matlab R2012b STE.Fig. 4 is original earthquake data, as can be seen from figure we, raw data affects by random noise to such an extent as to a lot of lineups None-identified.Therefore, we will containing noisy geological data march Wave Decomposition, obtains their expression in bent wave zone.The distributing position that noise is concentrated in bent wave zone can be determined according to the bent wave system number of seismic signal, by GCV and genetic algorithm, threshold value be set adaptively again and enter to make an uproar by this threshold value, Fig. 5 is the reconstructed image display effect after denoising, known by after method process of the present invention from the comparison diagram of Fig. 6, horizontal lineups in geological data become more clear, partly the thin strate of None-identified displays after treatment before treatment, and resolution is improved.Test result shows that the inventive method can Attenuating Random Noise protect weak seismic reflection signals effectively, and after the inventive method process, the seismic signal lineups after reconstruct become more obvious, and signal to noise ratio (S/N ratio) and resolution are improved.Known by testing above and analyzing, the inventive method is applied in the processing procedure of weak signal, fully to excavate the characteristic distributions of noise in bent wave zone, compacting noise, and widen signal band by spectral whitening process, effectively improve the resolution of weak seismic reflection signals, signal to noise ratio (S/N ratio) and image quality.As can be seen from Figure 7 the comparison diagram (left side is the histogram after process, and the right is histogram before treatment) organizing signal to noise ratio (S/N ratio) before and after seismic data processing, after process, signal to noise ratio (S/N ratio) promotes obvious more.
As shown in Figure 2, the embodiment of the present invention also provides a kind of and eliminates random noise method in seismic signal, comprises the steps:
S1, read original seismic signal data, and original seismic signal data march wave conversion is decomposed to the second frequency subband obtaining seismic signal and each yardstick, all directions represent the first frequency subband of the spectral signature of image and the space characteristics of expression image in bent wave zone, wherein the frequency range of first frequency subband is 10-20Hz, and the frequency range of second frequency subband is 50-200Hz;
S2, for decomposing the different scale, the bent wave system number under different directions that comprise in the first frequency subband that obtains and second frequency subband, by the threshold value of bent wave system number under Generalized Cross Validation criterion determination different scale, different directions, this threshold value is useful signal and random noise in the cut off value of direction and yardstick; And make Generalized Cross Validation criterion risk assessment function have the optimal threshold of minimum value by genetic algorithm acquisition;
S3, by be brought into by each optimal threshold obtained in step S2, semisoft shrinkage function obtains first frequency subband to decomposition, second frequency subband carries out threshold denoising;
S4, difference according to the correlativity of the first frequency subband after denoising in step S3, the useful signal in second frequency subband and random noise, design different adaptive spectral whitening wave filters;
S5, by corresponding adaptive spectral whitening wave filter, spectral whitening is carried out to the useful signal of different scale, different directions in first frequency subband, second frequency subband, widen the frequency band of useful signal; Random noise is not then processed;
S6, each yardstick after spectral whitening process, the first frequency subband in all directions, second frequency subband to be reconstructed by bent ripple inverse transformation, after obtaining denoising and the seismic signal data that improves of resolution.
Alternatively, described step S3 comprises: arrange the first compare threshold and the second compare threshold, the first compare threshold is less than the second compare threshold; The bent wave system number being less than the first compare threshold is set to 0, keeps the bent wave system number being greater than the second compare threshold constant; And the value of other bent wave system numbers is reduced by semisoft shrinkage function.
Alternatively, described step S4 comprises:
The related coefficient of the first frequency subband in obtaining step S3 after denoising, the useful signal in second frequency subband and random noise;
The related coefficient of useful signal and random noise is normalized and obtains normalized related coefficient;
Correlation factor is obtained according to related coefficient; Obtain the envelope of the spectral amplitude of each useful signal in first frequency subband in bent wave zone, second frequency subband;
Envelope according to different correlation factor and spectral amplitude designs different adaptive spectral whitening wave filters.
Alternatively, described step S5 comprises:
The signal to noise ratio (S/N ratio) of the subdata body of different scale, different directions in first frequency subband, second frequency subband is increased according to the white noise factor in spectral whitening wave filter; Thus widen the frequency band of useful signal; Random noise is not then processed.
Alternatively, to the subdata body march ripple inverse transformation adding different scale, different directions in the first frequency subband of signal to noise ratio (S/N ratio), second frequency subband, thus after obtaining denoising and the seismic signal data that is enhanced of resolution.
The software module that the method described in conjunction with embodiment disclosed herein or the step of algorithm can directly use hardware, processor to perform, or the combination of the two is implemented.Software module can be placed in any other forms of storage medium known in random access memory, internal memory, ROM (read-only memory), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technical field.
Be understandable that, for the person of ordinary skill of the art, other various corresponding change and distortion can be made by technical conceive according to the present invention, and all these change the protection domain that all should belong to the claims in the present invention with distortion.
Claims (10)
1. eliminate a random noise system in seismic signal, it is characterized in that, comprise as lower unit:
Warp wavelet unit, for reading original seismic signal data, and original seismic signal data march wave conversion is decomposed to the second frequency subband obtaining seismic signal and each yardstick, all directions represent the first frequency subband of the spectral signature of image and the space characteristics of expression image in bent wave zone, wherein the frequency range of first frequency subband is 10-20Hz, and the frequency range of second frequency subband is 50-200Hz;
Threshold optimization unit, for for decomposing the different scale, the bent wave system number under different directions that comprise in the first frequency subband that obtains and second frequency subband, by the threshold value of bent wave system number under Generalized Cross Validation criterion determination different scale, different directions, this threshold value is useful signal and random noise in the cut off value of direction and yardstick; And make Generalized Cross Validation criterion risk assessment function have the optimal threshold of minimum value by genetic algorithm acquisition;
Threshold denoising unit, for by being brought into by each optimal threshold obtained in threshold optimization unit, semisoft shrinkage function obtains first frequency subband to decomposition, second frequency subband carries out threshold denoising;
Design of filter unit, for the difference of the correlativity according to the first frequency subband after denoising in threshold denoising unit, the useful signal in second frequency subband and random noise, designs different adaptive spectral whitening wave filters;
Spectral whitening processing unit, for carrying out spectral whitening to the useful signal of different scale, different directions in first frequency subband, second frequency subband by corresponding adaptive spectral whitening wave filter, widens the frequency band of useful signal; Random noise is not then processed;
Bent ripple inverse transformation unit, for being reconstructed by bent ripple inverse transformation each yardstick after spectral whitening process, the first frequency subband in all directions, second frequency subband, after obtaining denoising and the seismic signal data that improves of resolution.
2. random noise system in elimination seismic signal as claimed in claim 1, it is characterized in that, described threshold denoising unit comprises: arrange the first compare threshold and the second compare threshold, the first compare threshold is less than the second compare threshold; The bent wave system number being less than the first compare threshold is set to 0, keeps the bent wave system number being greater than the second compare threshold constant; And the value of other bent wave system numbers is reduced by semisoft shrinkage function.
3. random noise system in elimination seismic signal as claimed in claim 1, it is characterized in that, described design of filter unit comprises:
The related coefficient of the first frequency subband in acquisition threshold denoising unit after denoising, the useful signal in second frequency subband and random noise;
The related coefficient of useful signal and random noise is normalized and obtains normalized related coefficient;
Correlation factor is obtained according to related coefficient; Obtain the envelope of the spectral amplitude of each useful signal in first frequency subband in bent wave zone, second frequency subband;
Envelope according to different correlation factor and spectral amplitude designs different adaptive spectral whitening wave filters.
4. random noise system in elimination seismic signal as claimed in claim 3, it is characterized in that, described spectral whitening processing unit comprises:
The signal to noise ratio (S/N ratio) of the subdata body of different scale, different directions in first frequency subband, second frequency subband is increased according to the white noise factor in spectral whitening wave filter; Thus widen the frequency band of useful signal; Random noise is not then processed.
5. random noise method in elimination seismic signal as claimed in claim 4, it is characterized in that, described bent ripple inverse transformation unit comprises:
To the subdata body march ripple inverse transformation adding different scale, different directions in the first frequency subband of signal to noise ratio (S/N ratio), second frequency subband, thus after obtaining denoising and the seismic signal data that is enhanced of resolution.
6. eliminate a random noise method in seismic signal, it is characterized in that, comprise the steps:
S1, read original seismic signal data, and original seismic signal data march wave conversion is decomposed to the second frequency subband obtaining seismic signal and each yardstick, all directions represent the first frequency subband of the spectral signature of image and the space characteristics of expression image in bent wave zone, wherein the frequency range of first frequency subband is 10-20Hz, and the frequency range of second frequency subband is 50-200Hz;
S2, for decomposing the different scale, the bent wave system number under different directions that comprise in the first frequency subband that obtains and second frequency subband, by the threshold value of bent wave system number under Generalized Cross Validation criterion determination different scale, different directions, this threshold value is useful signal and random noise in the cut off value of direction and yardstick; And make Generalized Cross Validation criterion risk assessment function have the optimal threshold of minimum value by genetic algorithm acquisition;
S3, by be brought into by each optimal threshold obtained in step S2, semisoft shrinkage function obtains first frequency subband to decomposition, second frequency subband carries out threshold denoising;
S4, difference according to the correlativity of the first frequency subband after denoising in step S3, the useful signal in second frequency subband and random noise, design different adaptive spectral whitening wave filters;
S5, by corresponding adaptive spectral whitening wave filter, spectral whitening is carried out to the useful signal of different scale, different directions in first frequency subband, second frequency subband, widen the frequency band of useful signal; Random noise is not then processed;
S6, each yardstick after spectral whitening process, the first frequency subband in all directions, second frequency subband to be reconstructed by bent ripple inverse transformation, after obtaining denoising and the seismic signal data that improves of resolution.
7. random noise method in elimination seismic signal as claimed in claim 6, it is characterized in that, described step S3 comprises: arrange the first compare threshold and the second compare threshold, the first compare threshold is less than the second compare threshold; The bent wave system number being less than the first compare threshold is set to 0, keeps the bent wave system number being greater than the second compare threshold constant; And the value of other bent wave system numbers is reduced by semisoft shrinkage function.
8. random noise method in elimination seismic signal as claimed in claim 6, it is characterized in that, described step S4 comprises:
The related coefficient of the first frequency subband in obtaining step S3 after denoising, the useful signal in second frequency subband and random noise;
The related coefficient of useful signal and random noise is normalized and obtains normalized related coefficient;
Correlation factor is obtained according to related coefficient; Obtain the envelope of the spectral amplitude of each useful signal in first frequency subband in bent wave zone, second frequency subband;
Envelope according to different correlation factor and spectral amplitude designs different adaptive spectral whitening wave filters.
9. random noise method in elimination seismic signal as claimed in claim 8, it is characterized in that, described step S5 comprises:
The signal to noise ratio (S/N ratio) of the subdata body of different scale, different directions in first frequency subband, second frequency subband is increased according to the white noise factor in spectral whitening wave filter; Thus widen the frequency band of useful signal; Random noise is not then processed.
10. random noise method in elimination seismic signal as claimed in claim 9, it is characterized in that, to the subdata body march ripple inverse transformation adding different scale, different directions in the first frequency subband of signal to noise ratio (S/N ratio), second frequency subband, thus after obtaining denoising and the seismic signal data that is enhanced of resolution.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510222712.2A CN104849757B (en) | 2015-04-30 | 2015-04-30 | Eliminate random noise system and method in seismic signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510222712.2A CN104849757B (en) | 2015-04-30 | 2015-04-30 | Eliminate random noise system and method in seismic signal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104849757A true CN104849757A (en) | 2015-08-19 |
| CN104849757B CN104849757B (en) | 2018-02-27 |
Family
ID=53849539
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510222712.2A Expired - Fee Related CN104849757B (en) | 2015-04-30 | 2015-04-30 | Eliminate random noise system and method in seismic signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104849757B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105424329A (en) * | 2015-12-16 | 2016-03-23 | 大豪信息技术（威海）有限公司 | Optical cable identification method and optical cable identification device |
| CN105676292A (en) * | 2016-01-22 | 2016-06-15 | 东华理工大学 | 3D earthquake data de-noising method based on 2D curvelet transform |
| CN109917459A (en) * | 2017-12-12 | 2019-06-21 | 中国石油天然气股份有限公司 | Method, device and system for suppressing earthquake noise |
| CN110244202A (en) * | 2019-06-21 | 2019-09-17 | 三峡大学 | Based on synchronous compression wavelet transformed domain partial discharge of transformer denoising method |
| CN112051612A (en) * | 2019-06-05 | 2020-12-08 | 河北地质大学 | Method and system for eliminating random noise of seismic data |
| CN112066935A (en) * | 2020-09-10 | 2020-12-11 | 上海应用技术大学 | Blast furnace taphole depth intelligent sensing method, device and system |
| CN113049913A (en) * | 2021-03-15 | 2021-06-29 | 广东电网有限责任公司梅州供电局 | Distribution line online monitoring method, device and system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009051899A1 (en) * | 2007-10-17 | 2009-04-23 | Exxonmobil Upstream Research Company | Method to adapt a template dataset to a target dataset by using curvelet representations |
| US20130182533A1 (en) * | 2012-01-12 | 2013-07-18 | Westerngeco L.L.C. | Attentuating noise acquired in an energy measurement |
| CN103645502A (en) * | 2013-12-11 | 2014-03-19 | 中国海洋石油总公司 | Seismic wave attenuation compensation method in curvelet domain |
| CN104007469A (en) * | 2014-05-24 | 2014-08-27 | 长江大学 | Weak seismic signal reconstruction method based on curvelet transform |
-
2015
- 2015-04-30 CN CN201510222712.2A patent/CN104849757B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009051899A1 (en) * | 2007-10-17 | 2009-04-23 | Exxonmobil Upstream Research Company | Method to adapt a template dataset to a target dataset by using curvelet representations |
| US20130182533A1 (en) * | 2012-01-12 | 2013-07-18 | Westerngeco L.L.C. | Attentuating noise acquired in an energy measurement |
| CN103645502A (en) * | 2013-12-11 | 2014-03-19 | 中国海洋石油总公司 | Seismic wave attenuation compensation method in curvelet domain |
| CN104007469A (en) * | 2014-05-24 | 2014-08-27 | 长江大学 | Weak seismic signal reconstruction method based on curvelet transform |
Non-Patent Citations (4)
| Title |
|---|
| 姚恒星: "基于曲波变换的面波去噪方法研究", 《长江大学学报（自科版）》 * |
| 王在民 等: "时频域自动识别与压制高频噪声", 《石油地球物理勘探》 * |
| 蔡炳煌: "基于曲波和遗传算法的图像去噪", 《中山大学研究生学刊（自科科学、医学版）》 * |
| 谷秀平: "基于Curvelet变换的图像去噪和增强", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105424329A (en) * | 2015-12-16 | 2016-03-23 | 大豪信息技术（威海）有限公司 | Optical cable identification method and optical cable identification device |
| CN105424329B (en) * | 2015-12-16 | 2018-05-08 | 一诺仪器（中国）有限公司 | A kind of optical cable recognition methods and optical cable recognition device |
| CN105676292A (en) * | 2016-01-22 | 2016-06-15 | 东华理工大学 | 3D earthquake data de-noising method based on 2D curvelet transform |
| CN105676292B (en) * | 2016-01-22 | 2018-10-19 | 东华理工大学 | A kind of 3D seismic data denoising method based on two-dimentional warp wavelet |
| CN109917459A (en) * | 2017-12-12 | 2019-06-21 | 中国石油天然气股份有限公司 | Method, device and system for suppressing earthquake noise |
| CN112051612A (en) * | 2019-06-05 | 2020-12-08 | 河北地质大学 | Method and system for eliminating random noise of seismic data |
| CN110244202A (en) * | 2019-06-21 | 2019-09-17 | 三峡大学 | Based on synchronous compression wavelet transformed domain partial discharge of transformer denoising method |
| CN110244202B (en) * | 2019-06-21 | 2021-07-20 | 三峡大学 | Transformer partial discharge denoising method based on synchronous compression wavelet transform domain |
| CN112066935A (en) * | 2020-09-10 | 2020-12-11 | 上海应用技术大学 | Blast furnace taphole depth intelligent sensing method, device and system |
| CN113049913A (en) * | 2021-03-15 | 2021-06-29 | 广东电网有限责任公司梅州供电局 | Distribution line online monitoring method, device and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104849757B (en) | 2018-02-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104849757A (en) | System and method for eliminating random noise in seismic signals | |
| CN107817527B (en) | Seismic exploration in desert stochastic noise suppression method based on the sparse compressed sensing of block | |
| CN102998706B (en) | Method and system for attenuating seismic data random noise | |
| CN104020492B (en) | A kind of guarantor limit filtering method of three dimensional seismic data | |
| CN103543469A (en) | Small-scale threshold denoising method based on wavelet transform | |
| CN107102356B (en) | Seismic signal high resolution data processing methods based on CEEMD | |
| CN109031422A (en) | A kind of seismic signal noise suppressing method based on CEEMDAN and Savitzky-Golay filtering | |
| CN101482617A (en) | Synthetic aperture radar image denoising method based on non-down sampling profile wave | |
| CN110208856B (en) | Desert complex noise suppression method based on manifold partition 2D-VMD | |
| CN106680876B (en) | A kind of seismic data joint denoising method | |
| CN106199532B (en) | Based on mixing Fourier-wavelet analysis Gpr Signal noise-reduction method | |
| CN107179550B (en) | A kind of seismic signal zero phase deconvolution method of data-driven | |
| CN102945548A (en) | Directional pyramid filtering-based image processing method and device | |
| CN114779343A (en) | Seismic data denoising method based on curvelet transform-joint bilateral filtering | |
| CN107783191A (en) | The method that hyperspace space-time time-frequency method cuts down seismic prospecting random noise | |
| CN111708087A (en) | Method for suppressing seismic data noise based on DnCNN neural network | |
| CN113655522A (en) | Frequency domain seismic weak signal enhancement method | |
| CN117574062A (en) | Small loop transient electromagnetic signal denoising method based on VMD-DNN model | |
| CN109782346B (en) | Acquisition footprint pressing method based on morphological component analysis | |
| CN111880222A (en) | Shearlet transformation-based seismic image minor fault identification enhancement method | |
| CN115236733A (en) | DAS-VSP data background noise suppression method based on deep learning | |
| CN106291680A (en) | A kind of data low frequency continuation method | |
| CN112764108B (en) | Novel seismic data noise suppression algorithm based on improved empirical wavelet transformation | |
| CN110764147B (en) | Desert exploration weak signal recovery method based on VMD local F-X spectrum decomposition | |
| CN113093282A (en) | Desert data denoising method based on geometric modal characteristic parallel network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Wen Chang Inventor before: Xie Kai Inventor before: Li Jicheng Inventor before: Zhang Long Inventor before: Shen Zhengchun Inventor before: Hu Yu |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180227 Termination date: 20210430 |