CN105974468B - A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting - Google Patents
A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting Download PDFInfo
- Publication number
- CN105974468B CN105974468B CN201610289375.3A CN201610289375A CN105974468B CN 105974468 B CN105974468 B CN 105974468B CN 201610289375 A CN201610289375 A CN 201610289375A CN 105974468 B CN105974468 B CN 105974468B
- Authority
- CN
- China
- Prior art keywords
- data
- reconstruction
- noise
- seismic data
- dimension
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
- G01V1/28—Processing seismic data, e.g. analysis, for interpretation, for correction
- G01V1/282—Application of seismic models, synthetic seismograms
Abstract
The present invention proposes a kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting, first against the deficiency of low-dimensional data reconstruction precision, using Fourier transformation as sparse basis, data reconstruction is carried out by using five dimension seismic data information, convex set projection algorithm is introduced in the process, it is proposed the threshold parameter of index square root attenuation law, each isochronous surface is individually rebuild using hard -threshold operator, then weighted factor is introduced in traditional convex set projection algorithm, so that reconstruction process in reduce influence of the noise to reconstructed results, finally realize a kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting.The method proposed by the present invention for being carried out at the same time five dimension Reconstruction of seismic data and noise compacting, it overcomes conventional two-dimensional or 3D seismic data reconstruction precision is insufficient, and conventional method cannot be carried out at the same time the shortcomings that data reconstruction and noise compacting, reconstruction precision is greatly improved, reduces the requirement to calculator memory.
Description
Technical field
It is specifically a kind of based under Fourier transformation the present invention relates to the method for Reconstruction of seismic data and noise compacting
The method that five dimension Reconstruction of seismic data and noise compacting can be carried out at the same time.
Technical background
In the wild in data acquisition, due to the limitation of collecting device and economic cost, seismic data is along direction in space
Upper usually progress irregular sampling, it is irregular, imperfect so as to cause collected seismic data, there is space aliasing, it is this
The seismic data of sparse distribution is difficult to meet the requirement of subsequent processing, such as velocity analysis, Free Surface Multiple attenuation, offset
Playback etc..In this case it is necessary to develop preferable pre stack data method for reconstructing.
Currently, widely used data re-establishing method is to be based on certain mathematic(al) manipulation, this method does not need underground structure
Prior information can handle the reconstruction of rule missing and irregular missing seismic channel.Simultaneously as the earthquake number that field acquisition is arrived
According to the influence for being frequently subjected to high-frequency random noises interference, the signal-to-noise ratio of earthquake record is reduced, data re-establishing method is influenced
Effect, precision is relatively low after causing missing road fully cannot rebuild or rebuild, and seriously affects the accurate imaging and wave field of signal
Playback, this just needs to carry out proper treatment to random disturbances noise in data reconstruction processes.However it is most of at present effective
Noise drawing method and data re-establishing method are all individually separately to be handled, and earthquake data weight can be carried out at the same time by lacking one kind
Build the method suppressed with noise.
For data reconstruction algorithm, various effective data reconstruction algorithms have been developed at present, wherein convex set projection is calculated
Method is widely used since parameter setting is simple and significant effect, and main thought is exactly to be used in each iterative process
Time-space numeric field data is transformed to frequency wavenumber domain by Fourier transform, and a threshold value is then arranged and retains large magnitude
Useful signal, then inverse fourier transform is used to the useful signal after reservation, finally the original seismic channel that need not be rebuild is set
It changes in the data after inverse fourier transform, by successive ignition, is come out to which seismic trace reconstruction will be lacked.This method is earliest
It is proposed by Bregman (1965), is then widely used in image procossing.POCS algorithms are applied to not advise by Abma (2006) for the first time
The then reconstruction of seismic data achieves preferable application effect, but using linear threshold parameter, convergence rate is slow, operation
Time is long.A kind of index threshold parameter of Gao (2010,2013) propositions and data drive threshold parameter, realize 3-D seismics number
It goes to suppress with noise according to reconstruction, improves the convergence rate of this method, then (2011) Liu, Yang (2012), Zhang
(2013) warp wavelet method is respectively adopted, the two and three dimensions seismic data weight based on POCS algorithms is realized in bent wave zone
It builds.But earthquake data acquisition has evolved to five dimension (xs,ys,xr,yr, t), wherein (xs,ys) and (xr,yr) respectively represent partially
Move away from common point coordinate, time t is TWT, therefore Reconstruction of seismic data and noise drawing method should also develop
To five dimensions, missing road seismic data is accurately reconstructed simultaneously in effectively removal noise.Since time orientation need not weigh
It builds, the quasi- space-time data information by using in five dimension seismic datas of the present invention carries out high precision seismic data reconstruction, together
When introduce weighted factor strategy, finally realize it is a kind of can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting method.
Invention content
The purpose of the invention is to can be to field missing road Reconstruction of seismic data, and at the same time can effectively suppress
Noise jamming improves signal-to-noise ratio, and proposes a kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting.
The present invention proposes a kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting, first against
The deficiency of low-dimensional data reconstruction precision is carried out using Fourier transformation as sparse basis by using five dimension seismic data information
Data reconstruction introduces convex set projection algorithm in the process, the threshold parameter of index square root attenuation law is proposed, using hard threshold
Value operator individually rebuilds each isochronous surface, then introduces weighted factor in traditional convex set projection algorithm so that
Reconstruction process in reduce influence of the noise to reconstructed results, five dimension seismic data weights can be carried out at the same time by finally realizing one kind
Build the method suppressed with noise.
A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting, the Problems of Reconstruction of seismic data are retouched
It states to recover complete data by the effect of linear operator by one group of deficiency of data, it is assumed that such as lower linear forward model:
yobs=Md
Here yobs∈RnRepresent the seismic data of acquisition;d∈RN, and N >=n, indicate no alias data to be reconstructed;M∈
Rn×NIndicate that diagonal matrix, element 1 and 0 indicate known seismic channel and unknown seismic channel respectively.
Assuming that coefficient x is rarefaction representations of the d in Fourier F, then above-mentioned equation is:
yobs=Md=MFHX=Ax
Wherein A=MFH, frequently referred to calculation matrix, above formula can equally be write as
In this expression formula,Represent estimated value, L1Norm is defined asX [i] is i-th in vector x
A element, by solving above-mentioned equation, the data of original no alias, which just reconstruct, to be come.
It is specific to rebuild and noise compacting using convex set projection algorithm in five dimension Reconstruction of seismic data and noise pressing process
Steps are as follows:
Step 1:Input has the five noisy seismic datas of dimension in missing road in the time domain, then uses Fourier transform pairs
Time-domain five ties up seismic data and carries out sparse transformation, obtains Fourier coefficient, and it is suitable to be selected according to the size of Fourier coefficient
Threshold parameter formula;
Step 2:It in Fourier, is handled using hard -threshold operator, namely is more than threshold value λiBent wave system number protect
It stays, and other Fourier coefficient zero setting;
Step 3:Fourier coefficient after thresholding is done into inverse transformation and obtains time-domain seismic data, selects suitable weighting
The factor reduces the influence of noise in iterative process, then will not lack the earthquake number after noisy seismic channel is filled into inverse transformation again
In;
Step 4:Obtained seismic data is finally substituted into step (1), re-starts iteration, is terminated until running n times, i.e.,
Final reconstruction and noise compacting result can be obtained.
Further, the convex set projection algorithm, iteration expression formula are:
dk(t,xs,ys,xr,yr)=yobs(t,xs,ys,xr,yr)+[I-M(xs,ys,xr,yr)]Ft -1Tk(t,xs,ys,xr,
yr)
×Ftdk-1(t,xs,ys,xr,yr), k=1,2 ..., N.
Wherein, dkIndicate (t, x that kth time iteration obtainss,ys,xr,yr) domain reconstruction data, d0Indicate acquired original to number
According to yobs(t,xs,ys,xr,yr), meet d0=yobs, FtAnd Ft -1Indicate the positive inverse-Fourier transform of the four-dimension about time variable t,
Tk(t,xs,ys,xr,yr) indicate hard -threshold operator.
Further, the hard -threshold operator, expression formula are as follows:
Sk-1It indicates the Fourier coefficient that -1 iteration of kth obtains, meets Sk-1=Ftdk-1(t,xs,ys,xr,yr), λ indicates N
Tie up threshold value set, λ={ λ1,λ2,…,λN, and meet | Cd |max=λ1>λ2>…>λN>=ε, N indicate maximum iteration, wherein
ε is the small value close to zero in formula, related with the energy of noise in five dimension seismic datas, different data ε values difference.
Further, described to introduce weighted factor in traditional convex set projection algorithm, expression formula is as follows:
dk(t,xs,ys,xr,yr)=a × yobs(t,xs,ys,xr,yr)+[I-a×M(xs,ys,xr,yr)]Ft -1Tk(t,xs,
ys,xr,yr)×Ftdk-1(t,xs,ys,xr,yr), k=1,2 ..., N.
Wherein a is weighted factor, and range is 0<A≤1, if a=1, equation and conventional convex set projection algorithm one
Sample, at this time raw noise data be brought into rebuild after seismic data in, data reconstruction can only be carried out, noise pressure cannot be carried out
System.The selection of the different data weighting factors is different, related with the intensity of noise energy.
Further, the threshold parameter formula, expression formula are as follows:
Wherein Max is the maximum value of Fourier Transform Coefficients absolute value;N be total iterations, ε be close to zero it is small
Value, it is related with the energy of noise in five dimension seismic datas.
Advantages of the present invention:The present invention, as sparse basis, seismic data information is tieed up by using five using Fourier transformation
Data reconstruction is carried out, introduces convex set projection algorithm in the process, proposes the threshold parameter of index square root attenuation law, is used
Hard -threshold operator individually rebuilds each isochronous surface, then introduces weighted factor in traditional convex set projection algorithm, real
A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting is showed, to overcome conventional two-dimensional or three
The shortcomings that dimension Reconstruction of seismic data precision is insufficient and conventional method cannot be carried out at the same time data reconstruction and noise compacting, reduces
Requirement to calculator memory, is greatly improved reconstruction precision, faint significant wave signal is protected, to make back wave
Lineups are more continuous, clear.
Description of the drawings
Fig. 1 is that data reconstruction suppresses flow chart with noise simultaneously in the embodiment of the present invention.
Fig. 2 is original analog seismic data figure.
Fig. 3 is original plus datagram of making an uproar.
Fig. 4 is 20% random lack sampling figure.
Fig. 5 is five dimension Reconstruction of seismic data result figures.
Fig. 6 is that threshold method five ties up seismic data noise compacting result figure.
Fig. 7 is while five dimension Reconstruction of seismic data suppress result figure with noise.
Fig. 8 is 3D seismic data while rebuilding and noise compacting result figure.
Specific implementation mode
Following case study on implementation is not limited to the scope of the present invention for illustrating the present invention.
Embodiment 1
The step of realizing this method includes mainly the structure of Reconstructed equation, convex set projection algorithm for reconstructing, using weighted factor
Carry out noise compacting, threshold parameter formula etc..It is as follows:
Step 1:The structure of Reconstructed equation.The Problems of Reconstruction of seismic data is described as being passed through by one group of deficiency of data linear
The effect of operator recovers complete data, it is assumed that such as lower linear forward model:
yobs=Md
Here yobs∈RnRepresent the seismic data of acquisition;d∈RN, and N >=n, indicate no alias data to be reconstructed, M ∈
Rn×NIndicate that diagonal matrix, element 1 and 0 indicate known seismic channel and unknown seismic channel respectively.
Assuming that coefficient x is rarefaction representations of the d in Fourier F, then above-mentioned equation is:
yobs=Md=MFHX=Ax
Wherein A=MFH, frequently referred to calculation matrix, above formula can equally be write as
In this expression formula,Represent estimated value, L1Norm is defined asX [i] is in vector x
I-th of element, by solving above-mentioned equation, the data of original no alias, which just reconstruct, to be come.
Step 2:Convex set projection algorithm for reconstructing.It can only be utilized for traditional reconstruction algorithm two-dimentional or 3-D seismics information
Disadvantage proposes five dimension Reconstruction of seismic data methods, and using hard -threshold operator and the new threshold parameter exponentially decayed
Formula, algorithm iteration expression formula are:
dk(t,xs,ys,xr,yr)=yobs(t,xs,ys,xr,yr)+[I-M(xs,ys,xr,yr)]Ft -1Tk(t,xs,ys,xr,
yr)
×Ftdk-1(t,xs,ys,xr,yr), k=1,2 ..., N.
Wherein, dkIndicate (t, x that kth time iteration obtainss,ys,xr,yr) domain reconstruction data, d0Indicate acquired original to number
According to yobs(t,xs,ys,xr,yr), meet d0=yobs, FtAnd Ft -1Indicate the positive inverse-Fourier transform of the four-dimension about time variable t,
Tk(t,xs,ys,xr,yr) indicate that hard -threshold operator, element meet:
Sk-1It indicates the Fourier coefficient that -1 iteration of kth obtains, meets Sk-1=Ftdk-1(t,xs,ys,xr,yr), λ indicates N
Tie up threshold value set, λ={ λ1,λ2,…,λN, and meet | Cd |max=λ1>λ2>…>λN>=ε, N indicate maximum iteration, wherein
ε is the small value close to zero in formula, related with the energy of noise in five dimension seismic datas, different data ε values difference.
Step 3:Noise compacting is carried out using weighted factor.For traditional reconstruction algorithm cannot be carried out at the same time data reconstruction and
A weighted factor is added to the initial data being placed in again in the shortcomings that noise is suppressed, and algorithm iteration expression formula is:
dk(t,xs,ys,xr,yr)=a × yobs(t,xs,ys,xr,yr)+[I-a×M(xs,ys,xr,yr)]Ft -1Tk(t,xs,
ys,xr,yr)×Ftdk-1(t,xs,ys,xr,yr), k=1,2 ..., N.
Wherein a is weighted factor, and range is 0<A≤1, if a=1, equation as conventional POCS algorithms, this
When raw noise data be brought into rebuild after seismic data in, data reconstruction can only be carried out, noise compacting cannot be carried out.No
The selection of the same data weighting factor is different, related with the intensity of noise energy.
Step 4:Threshold parameter formula.Different threshold parameters can obtain different reconstruction effects, and suitable threshold value is joined
Number is in the case where meeting required precision, it is possible to reduce iterations simultaneously save calculating cost, and therefore, the selection work of threshold parameter is particularly
It is important.The present invention on the basis of forefathers, in conjunction with the characteristics of Fourier transformation, press by selectionThe threshold value of rule decaying
Parameter, namely can faster be improved under the premise of ensureing reconstruction precision according to the threshold parameter that index square root rule decays
Convergence rate saves part and calculates the time, which is:
Max is in formula | Cd | maximum value, i.e. the maximum value of Fourier Transform Coefficients absolute value.
Realize that this method concrete operations are:
In order to which the effect of five dimension Reconstruction of seismic data and noise compacting is explained in more detail, the present invention defines Signal to Noise Ratio (SNR)
=20log10||f0||2/||f-f0||2To compare the quality after processing data, unit dB, wherein f0Indicate archetype number
Reconstructed results are indicated according to, f, and signal-to-noise ratio is higher, represents that reconstructed results are closer with model data, and effect is more ideal.
Inventive algorithm is applied to five dimension data theoretical models first, it is assumed that five dimension seismic data (xs,ys,xr,yr,t)
Including three hyperbola lineups, wherein (xs,ys) and (xr,yr) represent offset distance (offsetx,offsety) and concentrically
(CMPx,CMPy) point coordinates, time t is TWT.In order to improve arithmetic speed, space memory is reduced, the theoretical model
Space size is set as 21 × 21 × 21 × 21, and sampled point is 256, sample rate 1ms, and spatial sampling interval is
10m is simulated using 30Hz Ricker wavelets, due to being difficult five dimension data images of display, has therefrom been selected thus in multiple be total to
Point trace gather is shown that Fig. 2 indicates CMPy=10, offsetyWhen=20, (Fig. 2 indicates original analog to three common midpoint gathers
Seismic data figure), it can be seen that three curve lineups are more continuous, and the quality of data is preferable.Fig. 3 is expressed as having added certain energy
Gaussian random noise (Fig. 3 indicates original plus makes an uproar data), simulate field noise jamming data, it is random that 20% then carried out to it
Lack sampling, the results are shown in Figure 4 for lack sampling (Fig. 4 indicates 20% random lack sampling figure), and signal-to-noise ratio is 0.892dB at this time, then
Using the method for the present invention data reconstruction and noise is carried out to suppress, rebuild and the iterative process of noise compacting in, still clock synchronization
Between be sliced and handled, reduce one-dimensional positive inverse-Fourier transform so that reduce the dimension of data reconstruction in calculating process, save
Memory headroom.Individual five dimensions Reconstruction of seismic data is carried out first, and the results are shown in Figure 5, and (Fig. 5 indicates five dimension Reconstruction of seismic data
As a result), it can be seen that reconstruction effect is preferable, and lineups are more continuous, signal-to-noise ratio 6.82dB after reconstruction, the lower original of signal-to-noise ratio
Because being not suppress noise jamming, it is equivalent to weighted factor a=1 at this time.In order to compare while data reconstruction and noise
The effect of compacting carries out the five independent noises of dimension threshold iterative method to the noisy data after reconstruction and suppresses, and noise suppresses result such as Fig. 6
Shown (Fig. 6 indicates that threshold method five ties up seismic data noise and suppresses result), it can be seen that noise pressing result is preferable, random noise
It can almost remove clean, lineups are very continuous, signal-to-noise ratio 15.31dB, but the above processing is all data reconstruction and make an uproar
Acoustic pressure system is all to separate progress, is not unified.It finally carries out data reconstruction simultaneously using the method for the present invention to suppress with noise, at this time
Weighted factor a=0.4, (Fig. 7 indicates that five dimension Reconstruction of seismic data and noise compactings are tied simultaneously to handling result as shown in Figure 7
Fruit), it can be seen that although 80% seismic channel of missing, serious in certain one-dimensional square earthquake trace missing, due to five dimension earthquake numbers
Space-time information can be made full use of to rebuild a certain direction in space according to method for reconstructing, missing road seismic data restore compared with
It is good, and noise jamming is also suppressed, signal-to-noise ratio 16.07dB, and effect that treated is suppressed better than noise after first rebuilding
Effect, to also explanation five dimension seismic datas simultaneously rebuild with noise compacting superiority, in order to further embody its effect,
Also the method for the present invention is used to carry out while rebuilding to suppress with noise to 3D seismic data, the results are shown in Figure 8 (Fig. 8 shows
3D seismic data is rebuild suppress result with noise simultaneously), signal-to-noise ratio 8.84dB, it can be seen that three-dimensional reconstruction algorithm is sampling
It is rebuild simultaneously in the case of rate is low-down and noise pressing result is poor, and five dimension datas are rebuild since two-dimension earthquake is utilized more
The spatial information of data, therefore reconstruction and noise pressing result are more preferable, this also illustrates, by handling isochronous surface, to use
POCS algorithms are carried out at the same time the method significant effect of five dimension Reconstruction of seismic data and noise compacting, can be applied at real data
Reason.
It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments and not
In the case of the spirit or essential attributes of the present invention, the practical work(of the present invention can be realized in other specific forms
Energy.
Claims (1)
1. a kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting, which is characterized in that first against low
The deficiency of dimension data reconstruction precision ties up seismic data information into line number using Fourier transformation as sparse basis by using five
According to reconstruction, convex set projection algorithm is introduced in the process, the threshold parameter of index square root attenuation law is proposed, using hard -threshold
Operator individually rebuilds each isochronous surface, then introduces weighted factor in traditional convex set projection algorithm so that
Influence of the noise to reconstructed results is reduced in reconstruction process, five dimension Reconstruction of seismic data can be carried out at the same time by finally realizing one kind
With the method for noise compacting;It is as follows:
Step 1:The structure of Reconstructed equation;The Problems of Reconstruction of seismic data is described as passing through linear operator by one group of deficiency of data
Effect recover complete data, it is assumed that such as lower linear forward model:
yobs=Md
Here yobs∈RnThe seismic data of acquisition is represented, n indicates the seismic data dimension of acquisition;d∈RZ, and Z >=n, it indicates to wait for
Rebuild without alias seismic data, Z indicates no alias seismic data dimension;M∈Rn×ZIndicate diagonal matrix, element 1 and 0 point
It Biao Shi not known seismic channel and unknown seismic channel;
Assuming that coefficient x is rarefaction representations of the d in Fourier F, then above-mentioned equation is:
yobs=Md=MFHX=Ax
Wherein A=MFH, frequently referred to calculation matrix, subscript hereHIndicate associate matrix;Above formula can equally be write as:
subjectto yobs=Ax
In this expression formula,Represent estimated value, L1Norm is defined asX [i] is i-th yuan in coefficient x
Element, by solving above-mentioned equation, the data of original no alias, which just reconstruct, to be come;
Step 2:Convex set projection algorithm for reconstructing;Two-dimentional or 3-D seismics information lack can only be utilized for traditional reconstruction algorithm
Point proposes five dimension Reconstruction of seismic data methods, and using hard -threshold operator and the threshold parameter exponentially decayed, algorithm
Iteration expression formula is:
dk(t,xs,ys,xr,yr)=yobs(t,xs,ys,xr,yr)+[I-M(xs,ys,xr,yr)]Ft -1Tk(t,xs,ys,xr,yr)×
Ftdk-1(t,xs,ys,xr,yr), k=1,2, N.
Wherein, dkIndicate (t, x that kth time iteration obtainss,ys,xr,yr) domain reconstruction data, d0Indicate acquired original to data yobs
(t,xs,ys,xr,yr), meet d0=yobs, xsIndicate lateral shot point, ysIndicate longitudinal shot point, xrIndicate lateral geophone station, yrTable
Show that lateral geophone station, N indicate that maximum iteration, I indicate unit matrix, FtAnd Ft -1Indicate about time variable t it is four-dimensional just
Inverse-Fourier transform, Tk(t,xs,ys,xr,yr) indicate that hard -threshold operator, element meet:
λk∈λ
Sk-1It indicates the Fourier coefficient that -1 iteration of kth obtains, meets Sk-1=Ftdk-1(t,xs,ys,xr,yr), λ indicates threshold value,
λ={ λ1,λ2,···,λN, and meet | Ftd0|max=λ1> λ2> > λN>=ε, N indicate maximum iteration,
ε is the small value close to zero in Chinese style, related with the energy of noise in five dimension seismic datas, different data ε values difference;
Input has the five noisy seismic datas of dimension in missing road in the time domain, and earthquake is tieed up using Fourier transform pairs time-domain five
Data carry out sparse transformation, obtain Fourier coefficient, are then handled using hard -threshold operator, namely are more than threshold value λkSong
Wave system number retains, and other Fourier coefficient zero setting;
Step 3:Noise compacting is carried out using weighted factor;It cannot be carried out at the same time data reconstruction and noise for traditional reconstruction algorithm
A weighted factor is added to the initial data being placed in again in the shortcomings that compacting, and algorithm iteration expression formula is:
dk(t,xs,ys,xr,yr)=a × yobs(t,xs,ys,xr,yr)+[I-a×M(xs,ys,xr,yr)]Ft -1Tk(t,xs,ys,xr,
yr)×Ftdk-1(t,xs,ys,xr,yr), k=1,2, N
Wherein a be weighted factor, range is in 0 a≤1 <, if a=1, equation with routine convex set projection algorithm as,
Raw noise data are brought into the seismic data after rebuilding at this time, can only be carried out data reconstruction, cannot be carried out noise compacting;
The selection of the different data weighting factors is different, related with the intensity of noise energy;
Fourier coefficient after thresholding is done into inverse transformation and obtains time-domain seismic data, is introduced in traditional convex set projection algorithm
Weighted factor reduces the influence of noise in iterative process, then will not lack the ground after noisy seismic channel is filled into inverse transformation again
It shakes in data, and obtained seismic data is substituted into step 2, re-start iteration, until operation n times iteration terminates, you can
Obtain final reconstruction and noise compacting result;
Step 4:Threshold parameter formula;In conjunction with the characteristics of Fourier transformation, selection is pressedThe threshold parameter of rule decaying, wherein 0
≤ x≤1, i.e., the threshold parameter decayed according to index square root rule faster improve convergence under the premise of ensureing reconstruction precision
Speed saves part and calculates the time, which is:
Max is in formula | Ftd0| maximum value, i.e. the maximum value of Fourier Transform Coefficients absolute value;According to the big of Fourier coefficient
The small suitable threshold parameter formula of selection.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610289375.3A CN105974468B (en) | 2016-05-04 | 2016-05-04 | A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610289375.3A CN105974468B (en) | 2016-05-04 | 2016-05-04 | A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105974468A CN105974468A (en) | 2016-09-28 |
CN105974468B true CN105974468B (en) | 2018-10-19 |
Family
ID=56994453
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610289375.3A Expired - Fee Related CN105974468B (en) | 2016-05-04 | 2016-05-04 | A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105974468B (en) |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107015274A (en) * | 2017-04-12 | 2017-08-04 | 中国石油大学(华东) | One kind missing seismic exploration data recovery and rebuilding method |
CN108387928B (en) * | 2018-02-11 | 2021-06-15 | 中国石油化工股份有限公司 | Data construction method based on seismic feature transformation space |
CN108680953B (en) * | 2018-05-16 | 2019-07-26 | 中国海洋石油集团有限公司 | A kind of seismic data based on inverse proportion model interpolation and denoising method simultaneously |
CN108919357B (en) * | 2018-05-16 | 2019-10-11 | 中国海洋石油集团有限公司 | A kind of ghost reflection drawing method based on frequency spectrum reconfiguration |
CN108983287B (en) * | 2018-10-23 | 2020-08-25 | 东华理工大学 | Curvelet transform anti-aliasing seismic data reconstruction method based on convex set projection algorithm |
CN111337973B (en) * | 2018-12-19 | 2022-08-05 | 中国石油天然气集团有限公司 | Seismic data reconstruction method and system |
CN112445649A (en) * | 2019-08-30 | 2021-03-05 | 中国石油化工股份有限公司 | Seismic missing data recovery method, computer storage medium and computer system |
CN113093273A (en) * | 2019-12-23 | 2021-07-09 | 中国石油天然气集团有限公司 | Three-dimensional seismic data reconstruction method and device |
CN113011394B (en) * | 2021-04-26 | 2021-11-30 | 吉林大学 | Eye movement data preprocessing method and system |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104536044A (en) * | 2015-01-16 | 2015-04-22 | 中国石油大学(北京) | Interpolation and denoising method and system for seismic data |
CN105425301A (en) * | 2016-01-08 | 2016-03-23 | 东华理工大学 | Frequency domain three-dimensional irregular earthquake data reconstruction method |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8103453B2 (en) * | 2008-05-21 | 2012-01-24 | Bp Corporation North America Inc. | Method of seismic data interpolation by projection on convex sets |
-
2016
- 2016-05-04 CN CN201610289375.3A patent/CN105974468B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104536044A (en) * | 2015-01-16 | 2015-04-22 | 中国石油大学(北京) | Interpolation and denoising method and system for seismic data |
CN105425301A (en) * | 2016-01-08 | 2016-03-23 | 东华理工大学 | Frequency domain three-dimensional irregular earthquake data reconstruction method |
Non-Patent Citations (5)
Title |
---|
3D seismic data reconstruction based on complex-valued curvelet transform in frequency domain;Hua Zhang,et al.;《Journal of Applied Geophysics》;20151231;第113卷;摘要,第65页第2栏第24行-第67页第2栏第16行 * |
Convergence improvement and noise attenuation considerations for beyond alias projection onto convex sets reconstruction;Jianjun Gao,et al.;《Geophysical Prospecting》;20131231;第61卷;摘要,第139页第1栏第35行-第145页第2栏第18行及图1 * |
基于Jitter采样和曲波变换的三维地震数据重建;张华 等;《地球物理学报》;20130531;第56卷(第5期);第1638页第2栏第29行-第1639页第2栏第22行 * |
基于一种改进凸集投影方法的地震数据同时插值和去噪;曹静杰等;《地球物理学报》;20150831;第58卷(第8期);第2938页第1栏第26行-第2939页第2栏第11行 * |
基于压缩采样理论的五维地震数据重建;张华 等;《中国地球物理2012》;20121231;第449页第1-33行 * |
Also Published As
Publication number | Publication date |
---|---|
CN105974468A (en) | 2016-09-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105974468B (en) | A kind of method that can be carried out at the same time five dimension Reconstruction of seismic data and noise compacting | |
CN105425301B (en) | A kind of three-dimensional irregular earthquake data re-establishing method of frequency domain | |
Liu et al. | Random noise suppression in seismic data: What can deep learning do? | |
Yang et al. | Deep learning seismic random noise attenuation via improved residual convolutional neural network | |
CN106970419B (en) | A kind of non-homogeneous bent wave 3D seismic data method for reconstructing based on linear Bregman algorithm | |
CN106526677B (en) | A kind of wideband reverse-time migration imaging method of marine adaptive compacting ghost reflection | |
CN108983284B (en) | F-p domain ghost wave compression method suitable for offshore inclined cable data | |
CN107817527A (en) | Seismic exploration in desert stochastic noise suppression method based on the sparse compressed sensing of block | |
CN110320555B (en) | Seismic data reconstruction method | |
CN110018517B (en) | Multi-scale ground micro-seismic reverse-time interference positioning method | |
CN109782339A (en) | A kind of poststack three dimensional seismic data stochastic noise suppression method based on 3D-DnCNN network | |
CN110058305A (en) | A kind of DAS seismic data noise-reduction method based on convolutional neural networks | |
CN105549076B (en) | A kind of seismic data processing technique based on alternating direction method and full Theory of Variational Principles | |
CN105700020A (en) | Random noise suppression method and apparatus for seismic data | |
CN106249291A (en) | A kind of high precision seismic data re-establishing method based on two-dimentional non-homogeneous warp wavelet | |
CN107179550B (en) | A kind of seismic signal zero phase deconvolution method of data-driven | |
CN110490219A (en) | A method of the U-net network based on texture constraint carries out Reconstruction of seismic data | |
CN105182417B (en) | A kind of surface wave separation method and system based on anatomic element analysis | |
CN103364832A (en) | Seismic attenuation qualitative estimation method based on self-adaptive optimal kernel time frequency distribution | |
CN103713324B (en) | Self-adaption curvelet threshold value earthquake denoising method based on local variance analysis | |
CN105869126A (en) | Compression perception denoising method of high-spectral images | |
CN105319593A (en) | Combined denoising method based on curvelet transform and singular value decomposition | |
CN105676292B (en) | A kind of 3D seismic data denoising method based on two-dimentional warp wavelet | |
Zu et al. | 3D deblending of simultaneous source data based on 3D multi-scale shaping operator | |
CN111025385B (en) | Seismic data reconstruction method based on low rank and sparse constraint |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20181019 Termination date: 20190504 |
|
CF01 | Termination of patent right due to non-payment of annual fee |