CN106291696A - A kind of low signal-to-noise ratio seismic signal recognition methods and system - Google Patents

Abstract

The present invention relates to seismic data processing technology field, disclose a kind of low signal-to-noise ratio seismic signal recognition methods and system, described low signal-to-noise ratio seismic signal recognition methods includes: window when seismic signal is decomposed into several, use quadratic polynomial that the seismic signal in window time single is fitted, to realize dip direction estimation；By effective earthquake signal rotation in direction, sampling point lineups place, Shi Chuannei center of being tracked by matching to horizontal level, as the input of Singular Value Decomposition Yu reconstruct；And input of based on Singular Value Decomposition Yu reconstruct, the seismic signal in window time after rotating is carried out Singular Value Decomposition and reconstructs, the seismic signal after output reconstruct.The present invention improves the signal to noise ratio of seismic data, it is achieved that the effective identification to low signal-to-noise ratio seismic signal.

Description

A kind of low signal-to-noise ratio seismic signal recognition methods and system

Technical field

The present invention relates to seismic data processing technology field, particularly relate to a kind of low signal-to-noise ratio seismic signal Recognition methods and system.

Background technology

The oil-gas exploration of the complex areas such as mountain front is one of current important exploration targets, and carries out this kind of oil The common issue faced during gas exploration is how to identify low signal-to-noise ratio earthquake in of a relatively high noise The problem of signal (low signal-to-noise ratio seismic signal).Low signal-to-noise ratio seismic signal refers to the signal that amplitude is the most weak Or the signal flooded by noise, due to low signal-to-noise ratio seismic signal in conventional time-space domain the most easy to identify, Need could it be detected by certain detection means.

Existing identification technology mainly has a following several method:

(1) technology such as multi-fold, combination, superposition.Superposition is the effective ways of Attenuating Random Noise, And good application effect can be obtained, but when subsurface structure is complicated, also deposit in addition to random disturbances Strong linear disturbance when, the effect of multi-fold and superposition will be very restricted.

(2) overall situation singularity value decomposition.Traditional overall singular value decomposition method can preferably be examined Measure the signal of horizontal lineups, but when the horizontal dependency of useful signal is not strong in seismic channel set, the overall situation is strange Different value decomposition method is difficult to reach preferable denoising effect, and can introduce new noise.

Therefore, the shortcoming existed according to existing identification technology, it is necessary to study new for identifying low noise The ratio method of seismic signal, to ensure being normally carried out of oil-gas exploration.

Summary of the invention

It is an object of the invention to provide a kind of low signal-to-noise ratio seismic signal recognition methods and system, be used for solving The problem that existing low signal-to-noise ratio seismic signal recognition methods effect is undesirable.

To achieve these goals, technical scheme provides a kind of low signal-to-noise ratio seismic signal knowledge Other method, including: window when seismic signal is decomposed into several, use quadratic polynomial to window time single Interior seismic signal is fitted, to realize dip direction estimation；By tracked by matching time window in Effective earthquake signal rotation in direction, sampling point lineups place, center is to horizontal level, as local singular value The input of decomposition and reconstruction；And input of based on Singular Value Decomposition Yu reconstruct, to window time after rotating Interior seismic signal carries out Singular Value Decomposition and reconstructs, the seismic signal after output reconstruct.

Preferably, the seismic signal in window time single is fitted, specifically by described employing quadratic polynomial Including: with time window center trace center position as a reference point, set for the secondary that is fitted many The expression formula of item formula；Ask for the Monomial coefficient in the expression formula of quadratic polynomial and quadratic term coefficient；With And first use the Monomial coefficient of quadratic polynomial to carry out the matching of seismic signal, then use this secondary multinomial The quadratic term coefficient of formula carries out the matching of seismic signal.

Preferably, described in ask for the Monomial coefficient in the expression formula of quadratic polynomial, specifically include: adopt By maximum cross-correlation rule, in the initial range of given Monomial coefficient, different coefficients are asked for mutually Correlation, determines the Monomial coefficient in the expression formula that coefficient is quadratic polynomial that cross correlation value is maximum.

Preferably, described input based on Singular Value Decomposition Yu reconstruct, to the ground in window time after rotating Shake signal carries out Singular Value Decomposition, specifically includes: the seismic signal setting input has N road earthquake record, The sampling number of every trace record is M, then these N*M data can be expressed as with matrix form:

In formula, x_NMRepresent the seismic signal of input；

According to singular value decomposition method, matrix A is decomposed into following formula:

$A=\mathrm{U\Λ}{V}^T=\underset{i=1}{\overset{r}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{u}_i{v}_i^T$

In formula, r is rank of matrix, and subscript T represents the transposition of matrix；u_iIt it is matrix A A^TI-th special Levy vector, U=[u₁,u₂,…,u_N], U is the orthogonal matrix on N*M rank；v_iIt it is matrix A^TThe i-th of A is special Levy vector, V=[v₁,v₂,…,v_N], V is the orthogonal matrix on M*M rank；Λ is AA^TOr A^TA eigenvalue Non-negative square root λ is by the diagonal matrix of the order composition that successively decreases, Λ=diag (λ₁,λ₂,…,λ_r) it is N*M rank matrixes； Wherein, the singular value of non-negative square root λ representing matrix A.

Preferably, described to rotate after time window in seismic signal be reconstructed, specifically include: use square Battle array A characterizes front p the singular value restructuring matrix A of useful signal.

Preferably, the matrix A after reconstruct is expressed as:

$A=\underset{i=p}{\overset{q}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{u}_i{v}_i^T$

In formula, 1≤p≤q≤r, q represent the total number of singular value, and r is rank of matrix, and subscript T represents The transposition of matrix；u_iIt it is matrix A A^TIth feature vector, v_iIt it is matrix A^TThe ith feature of A to Amount, λ_iIt it is singular value.

Technical scheme additionally provides a kind of low signal-to-noise ratio seismic signal identification system, including: ground Layer tendency estimation module, window during for seismic signal being decomposed into several, use quadratic polynomial to list Time individual, the seismic signal in window is fitted, to realize dip direction estimation；First local singular value processes Module, effective earthquake in the direction, sampling point lineups place, Shi Chuannei center for being tracked by matching Signal rotation is to horizontal level, as the input of Singular Value Decomposition Yu reconstruct；And second local strange Different value processing module, for input based on Singular Value Decomposition Yu reconstruct, in window time after rotating Seismic signal carries out Singular Value Decomposition and reconstructs, the seismic signal after output reconstruct.

Preferably, described dip direction estimation module includes: quadratic polynomial setting module, for time The center trace center position of window is as a reference point, sets the expression for the quadratic polynomial being fitted Formula；Coefficients calculation block, the Monomial coefficient in the expression formula asking for quadratic polynomial and quadratic term Coefficient；And the Fitting Calculation module, for first using the Monomial coefficient of quadratic polynomial to carry out earthquake letter Number matching, then use the quadratic term coefficient of this quadratic polynomial to carry out the matching of seismic signal.

Preferably, described second local singular value processing module, including: Singular Value Decomposition module, For input based on Singular Value Decomposition Yu reconstruct, the seismic signal in window time after rotating is carried out office Portion's singular value decomposition, it is thus achieved that matrix that seismic signal is corresponding and singular value；And reconstructed module, it is used for adopting Matrix with front p the singular value reconstruct seismic signal characterizing effective seismic signal.

The invention has the beneficial effects as follows: the present invention improves the signal to noise ratio of seismic data, it is achieved that to low letter Make an uproar than effective identification of seismic signal.

Other features and advantages of the present invention will be described in detail in detailed description of the invention part subsequently.

Accompanying drawing explanation

Accompanying drawing is used to provide a further understanding of the present invention, and constitutes the part of description, with Detailed description below is used for explaining the present invention together, but is not intended that limitation of the present invention.? In accompanying drawing:

Fig. 1 is the schematic flow sheet of low signal-to-noise ratio seismic signal recognition methods in embodiments of the present invention；

Fig. 2 is the schematic flow sheet carrying out dip direction estimation in embodiments of the present invention；

Fig. 3 is the dip direction schematic diagram estimated in embodiments of the present invention；

Fig. 4 is the structural representation of low signal-to-noise ratio seismic signal identification system in embodiments of the present invention；

Fig. 5 is original CMP road, the real data sandstone district collection in embodiments of the present invention as example；

Fig. 6 is to use the low signal-to-noise ratio seismic signal recognition methods in embodiments of the present invention to Fig. 5's Original CMP road collection process after result schematic diagram；

Fig. 7 (a) is the normal-moveout spectrum that the original CMP road collection of Fig. 5 is corresponding；

Fig. 7 (b) is the normal-moveout spectrum that the process Hou CMP road collection of Fig. 6 is corresponding；

Fig. 8 is the stacked section schematic diagram that the original CMP road collection of Fig. 5 is corresponding；

Fig. 9 is the stacked section schematic diagram that Tu6 CMP road collection is corresponding；

Figure 10 is the schematic diagram that in Fig. 8, rectangle frame part is amplified；

Figure 11 is the schematic diagram of the rectangle frame part amplification of Fig. 9.

Detailed description of the invention

Below in conjunction with accompanying drawing, the detailed description of the invention of the present invention is described in detail.It should be appreciated that Detailed description of the invention described herein is merely to illustrate and explains the present invention, is not limited to the present invention.

As it is shown in figure 1, present embodiments provide for a kind of low signal-to-noise ratio seismic signal recognition methods, including: Window when seismic signal is decomposed into several, uses quadratic polynomial to enter the seismic signal in window time single Row matching, to realize dip direction estimation；The Shi Chuannei center sampling point lineups that will be tracked by matching Effective earthquake signal rotation in direction, place to horizontal level, defeated as Singular Value Decomposition and reconstruct Enter；And input of based on Singular Value Decomposition Yu reconstruct, the seismic signal in window time after rotating is entered Row Singular Value Decomposition also reconstructs, the seismic signal after output reconstruct.

According to each basic step illustrated in Figure 1, it is known that relative to prior art, present embodiment is main To utilize quadratic polynomial that geological data is fitted estimating dip direction, then by time window center sampling point The seismic signal in the direction at lineups place rotates to horizontal level, is then based on Singular Value Decomposition skill Art processes postrotational geological data to realize the effective identification to low signal-to-noise ratio seismic signal.Therefore, originally Scheme in embodiment relates generally to dip direction and estimates and local singular value two parts of process.

One, dip direction is estimated

Use discrete scanning method that dip direction is estimated, i.e. use orthogonal polynomial method.This The problem that kind way of fitting is primarily present is: when asking for orthogonal polynomial coefficient, due to orthogonal Multinomial contains constant term so that when scanning is relevant, the center of window changes along with the change of scan fraction, Can not accurately determine the position of matching central point, can only suppose that the result of matching is the overall of a regional area Trend.Present embodiment, so that the estimation of tendency is more accurate, uses conventional quadratic polynomial, and It not way of fitting, be fitted geological data estimating that dip direction, the method are to align Hand over the improvement of polynomial fitting method, overcome the deficiency of orthogonal polynomial method.

Estimate to mainly comprise the steps that as in figure 2 it is shown, present embodiment carries out dip direction

Step 12, with time window center trace center position as a reference point, set for being fitted The expression formula of quadratic polynomial.

The expression formula of quadratic polynomial preferably employs following quadratic fit multinomial, so that local lineups energy Meet this quadratic polynomial:

T=t₀+c₁x+c₂x²

Wherein: x is the relative Taoist monastic name of matching center trace, and x has positive negative direction, and its value is X=-M ,-M+1 ... ,-1,0,1..., M-1, M, M are the half of matching number of channels, t₀For matching center sampling point Time.

Step 12, asks for the Monomial coefficient in the expression formula of quadratic polynomial and quadratic term coefficient.

Use maximum cross-correlation rule, in the initial range of given Monomial coefficient, to different coefficients Ask for cross correlation value, determine optimal in the expression formula that coefficient is quadratic polynomial that cross correlation value is maximum Secondary term coefficient.

Here, the object function carrying out cross-correlation is:

$R\left(i\right)=\frac{1}{N}\frac{\underset{t=-l/2}{\overset{l/2}{\mathrm{\Σ}}}{\left(\underset{k=-M}{\overset{M}{\mathrm{\Σ}}}{S}_k(t_0+c_1\left(i\right)\×k+t)\right)}^2}{\underset{t=-l/2}{\overset{l/2}{\mathrm{\Σ}}}\underset{j=-M}{\overset{M}{\mathrm{\Σ}}}{S}_j^2(t_0+c_1\left(i\right)\×j+t)}$

Wherein: R (i) is Shi Chuannei road collection cross correlation value, S_k, S_jFor matching road sample value, l is relevant Time window in number of samples.Coefficient R (i) changes between 0～1, ideally, is homophase at that time in window During direction of principal axis, each road dependency is the strongest, and now R (i) tends to 1, and the least then dependency of R (i) is the poorest.

As it is shown on figure 3, the center trace that matching road integrates is as library track, upper and lower line-to-line is first order matching Coefficient scanning scope, (shown in intermediate line) phase when scanning direction is for calculating lineups direction, sampling point place Pass value is optimal value, after Monomial coefficient determines, then is scanned improving further to quadratic term coefficient Fitting precision.

Step 13, in order to reduce amount of calculation, first uses the Monomial coefficient of quadratic polynomial to carry out earthquake letter Number matching, then use the quadratic term coefficient of this quadratic polynomial to carry out the matching of seismic signal.

The method makes local lineups have linear dependence, it is simple to be rotated into horizontal lineups, in order to more Use singularity value decomposition well.

Two, local singular value processes

(1) Singular Value Decomposition

The effective seismic signal in direction, sampling point lineups place, Shi Chuannei center that will be tracked by matching Rotate to horizontal level, as the input of Singular Value Decomposition Yu reconstruct, if the seismic signal of this input Having N road earthquake record, the sampling number of every trace record is M, then these N*M data can use square Formation formula is expressed as:

In formula, x_NMRepresent the seismic signal of input.

According to singular value decomposition method, matrix A is decomposed into following formula:

$A=\mathrm{U\Λ}{V}^T=\underset{i=1}{\overset{r}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{u}_i{v}_i^T$

In formula, r is rank of matrix, and subscript T represents the transposition of matrix；u_iIt it is matrix A A^TI-th special Levy vector, U=[u₁,u₂,…,u_N], U is the orthogonal matrix on N*M rank；v_iIt it is matrix A^TThe i-th of A is special Levy vector, V=[v₁,v₂,…,v_N], V is the orthogonal matrix on M*M rank；Λ is AA^TOr A^TA eigenvalue Non-negative square root λ is by the diagonal matrix of the order composition that successively decreases, Λ=diag (λ₁,λ₂,…,λ_r) it is N*M rank matrixes, Wherein, the singular value of non-negative square root λ representing matrix A.

(2) matrix reconstruction

If matrix A is the matrix collectively constituted by signal and random noise, then the singular value of matrix A can Situation about concentrating with reflection signal and noise energy.Its front p bigger singular value mainly reflects signal, Remaining singular value then reflects noise.Therefore, only with front p the singular value restructuring matrix of sign useful signal, Can be removed random disturbances.Matrix A after reconstruct is expressed as:

$A=\underset{i=p}{\overset{q}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{u}_i{v}_i^T$

In formula, 1≤p≤q≤r, q represent the total number of singular value, and r is rank of matrix, and subscript T represents The transposition of matrix；u_iIt it is matrix A A^TIth feature vector, v_iIt it is matrix A^TThe ith feature of A to Amount, λ_iIt it is singular value.

The geological data of window time all of is carried out identical process, the process knot of whole road collection can be obtained Really.It should be noted that, when should make during practical operation, window is the most continuously slipping to guarantee Cover whole earthquake record, the illusion on window border during for eliminating, also should there is certain repetition between sliding window.

Corresponding to above-mentioned low signal-to-noise ratio seismic signal recognition methods, as shown in Figure 4, present embodiment also carries Supply a kind of low signal-to-noise ratio seismic signal identification system, including: dip direction estimation module, for by ground Window when shake signal decomposition is several, uses quadratic polynomial to intend the seismic signal in window time single Close, to realize dip direction estimation；First local singular value processing module, for following the trail of by matching Effective earthquake signal rotation in the direction, sampling point lineups place, Shi Chuannei center arrived to horizontal level, as Singular Value Decomposition and the input of reconstruct；And the second local singular value processing module, for based on office Portion's singular value decomposition and the input of reconstruct, carry out local singular value to the seismic signal in window time after rotating and divide Solve and reconstruct, the seismic signal after output reconstruct.

Wherein, described dip direction estimation module includes: quadratic polynomial setting module, for time window Center trace center position as a reference point, set the expression formula of quadratic polynomial for being fitted； Coefficients calculation block, the Monomial coefficient in the expression formula asking for quadratic polynomial and quadratic term coefficient； And the Fitting Calculation module, for first using the Monomial coefficient of quadratic polynomial to carry out the plan of seismic signal Close, then use the quadratic term coefficient of this quadratic polynomial to carry out the matching of seismic signal.

Wherein, described second local singular value processing module, including: Singular Value Decomposition module, use In input based on Singular Value Decomposition Yu reconstruct, the seismic signal in window time after rotating is carried out local Singular value decomposition, it is thus achieved that matrix that seismic signal is corresponding and singular value；And reconstructed module, it is used for using Characterize the matrix of front p the singular value reconstruct seismic signal of effective seismic signal.

This low signal-to-noise ratio seismic signal identification system and above-mentioned low signal-to-noise ratio seismic signal recognition methods Specific implementation process is consistent, is described again here.

By actual seismic data experiments, the scheme that present embodiment uses is to low signal-to-noise ratio seismic signal Treatment effect is obvious, has stronger specific aim.If Fig. 5 is the original CMP in certain real data sandstone district Road collection, vertical coordinate is the time, and unit is the second, and abscissa is collection number of channels, and Fig. 5 is that background noise is strong Seismic signal, wherein useful signal CMP road concentrate present Hyperbolic Feature, owing to signal to noise ratio is low, Effectively lineups are submerged in noise, and vision is difficult to differentiate.Fig. 6 is the low noise using the present embodiment The ratio result obtained after seismic signal recognition methods identification, vertical coordinate is the time, and unit is the second, horizontal Coordinate is collection number of channels, it is known that noise is suppressed very well, lineups signal weak between particularly 3s-5s Revealing, signal to noise ratio is significantly improved.From Fig. 6 it can clearly be seen that at vertical coordinate 2s and 4s Hyperbolic feature useful signal reveals, i.e. weak signal is identified.

Fig. 7 is corresponding normal-moveout spectrum before and after Fig. 5 and Fig. 6 processes, its in length and breadth coordinate be Grid dimension, The focus level of normal-moveout spectrum energy group is one of mode judging useful signal signal to noise ratio, original data processing Front normal-moveout spectrum energy dissipates, and after process, normal-moveout spectrum energy is concentrated, and focusing is more preferable, for improving velocity analysis Precision provides guarantee.From Fig. 7 (a) and Fig. 7 (b) it can be seen that relative Fig. 7 of Fig. 7 (b) after Chu Liing A () Voice segment is more preferable, illustrate that signal to noise ratio is improved

Fig. 8 is initial data stacked section, Fig. 9 for process after superposition plane, its in length and breadth coordinate be net Lattice point number, stacked section is the improvement that effect is described from the angle of achievement, and signal lineups strengthen and become More continuous, understand from Fig. 8 and Fig. 9, higher relative to Limestone pavement signal to noise ratio in sandstone region signal to noise ratio, After process, Limestone pavement lineups have clear improvement, as shown in square frame in figure.Figure 10, Figure 11 correspondence respectively In Fig. 8 and Fig. 9, Blocked portion amplifies result, through the low signal-to-noise ratio seismic signal identification of present embodiment After method processes, lineups seriality improves, and weak energy displays.

In sum, present embodiment improves the signal to noise ratio of seismic data, it is achieved that to low signal-to-noise ratio ground Effective identification of shake signal.

The preferred embodiment of the present invention is described in detail above in association with accompanying drawing, but, the present invention does not limit Detail in above-mentioned embodiment, in the technology concept of the present invention, can be to the present invention Technical scheme carry out multiple simple variant, these simple variant belong to protection scope of the present invention.

It is further to note that each the concrete technical characteristic described in above-mentioned detailed description of the invention, In the case of reconcilable, can be combined by any suitable means, in order to avoid unnecessary Repeating, various possible compound modes are illustrated by the present invention the most separately.

Additionally, combination in any can also be carried out between the various different embodiment of the present invention, as long as its Without prejudice to the thought of the present invention, it should be considered as content disclosed in this invention equally.

Claims (9)

1. a low signal-to-noise ratio seismic signal recognition methods, it is characterised in that including:

Window when seismic signal is decomposed into several, uses quadratic polynomial to believe the earthquake in window time single Number it is fitted, estimates realizing dip direction；

The effective seismic signal in direction, sampling point lineups place, Shi Chuannei center that will be tracked by matching Rotate to horizontal level, as the input of Singular Value Decomposition Yu reconstruct；And

Input based on Singular Value Decomposition Yu reconstruct, carries out office to the seismic signal in window time after rotating Portion's singular value decomposition also reconstructs, the seismic signal after output reconstruct.

Low signal-to-noise ratio seismic signal recognition methods the most according to claim 1, it is characterised in that Seismic signal in window time single is fitted by described employing quadratic polynomial, specifically includes:

With time window center trace center position as a reference point, set for the secondary that is fitted multinomial The expression formula of formula；

Ask for the Monomial coefficient in the expression formula of quadratic polynomial and quadratic term coefficient；And

The Monomial coefficient first using quadratic polynomial carries out the matching of seismic signal, then uses this secondary many The quadratic term coefficient of item formula carries out the matching of seismic signal.

Low SNR signal recognition methods the most according to claim 2, it is characterised in that described Ask for the Monomial coefficient in the expression formula of quadratic polynomial, specifically include: use maximum cross-correlation rule, In the initial range of given Monomial coefficient, different coefficients are asked for cross correlation value, determines cross-correlation Monomial coefficient in the expression formula that coefficient is quadratic polynomial that value is maximum.

Low signal-to-noise ratio seismic signal recognition methods the most according to claim 1, it is characterised in that Described input based on Singular Value Decomposition Yu reconstruct, carries out office to the seismic signal in window time after rotating Portion's singular value decomposition, specifically includes:

If the seismic signal of input has N road earthquake record, the sampling number of every trace record is M, then this N*M data can be expressed as with matrix form:

In formula, x_NMRepresent the seismic signal of input；

According to singular value decomposition method, matrix A is decomposed into following formula:

$A={\mathrm{U\ΛV}}^T=\underset{i=1}{\overset{r}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{u}_i{v}_i^T$

In formula, r is rank of matrix, and subscript T represents the transposition of matrix；u_iIt it is matrix A A^TI-th special Levy vector, U=[u₁,u₂,…,u_N], U is the orthogonal matrix on N*M rank；v_iIt it is matrix A^TThe i-th of A is special Levy vector, V=[v₁,v₂,…,v_N], V is the orthogonal matrix on M*M rank；Λ is AA^TOr A^TA eigenvalue Non-negative square root λ is by the diagonal matrix of the order composition that successively decreases, Λ=diag (λ₁,λ₂,…,λ_r) it is N*M rank matrixes；

Wherein, the singular value of non-negative square root λ representing matrix A.

Low signal-to-noise ratio seismic signal recognition methods the most according to claim 4, it is characterised in that Described seismic signal in window time after rotating is reconstructed, specifically includes: using in matrix A to characterize has Front p the singular value restructuring matrix A of effect signal.

Low signal-to-noise ratio seismic signal recognition methods the most according to claim 5, it is characterised in that Matrix A after reconstruct is expressed as:

$A=\underset{i=p}{\overset{q}{\mathrm{\Σ}}}{\mathrm{\λ}}_i{u}_i{v}_i^T$

In formula, 1≤p≤q≤r, q represent the total number of singular value, and r is rank of matrix, and subscript T represents The transposition of matrix；u_iIt it is matrix A A^TIth feature to

Amount, v_iIt it is matrix A^TThe ith feature vector of A, λ_iIt it is singular value.

7. a low signal-to-noise ratio seismic signal identification system, it is characterised in that including:

Dip direction estimation module, window during for seismic signal being decomposed into several, use secondary multinomial Seismic signal in window time single is fitted by formula, to realize dip direction estimation；

First local singular value processing module, same for the Shi Chuannei center sampling point that will be tracked by matching Effective earthquake signal rotation in direction, phase axle place is to horizontal level, as Singular Value Decomposition and reconstruct Input；And

Second local singular value processing module is for input based on Singular Value Decomposition Yu reconstruct, right Time after rotation, the seismic signal in window carries out Singular Value Decomposition and reconstructs, the earthquake letter after output reconstruct Number.

Low signal-to-noise ratio seismic signal identification system the most according to claim 7, it is characterised in that Described dip direction estimation module includes:

Quadratic polynomial setting module, for time window center trace center position as a reference point, if The expression formula of the fixed quadratic polynomial for being fitted；

Coefficients calculation block, the Monomial coefficient in the expression formula asking for quadratic polynomial and quadratic term Coefficient；And

The Fitting Calculation module, for first using the Monomial coefficient of quadratic polynomial to carry out the plan of seismic signal Close, then use the quadratic term coefficient of this quadratic polynomial to carry out the matching of seismic signal.

Low signal-to-noise ratio seismic signal identification system the most according to claim 7, it is characterised in that Described second local singular value processing module, including:

Singular Value Decomposition module, for input based on Singular Value Decomposition Yu reconstruct, to rotation Time rear, the seismic signal in window carries out Singular Value Decomposition, it is thus achieved that matrix that seismic signal is corresponding and unusual Value；And

Reconstructed module, for using front p the singular value characterizing effective seismic signal to reconstruct seismic signal Matrix.