A kind of electrical network unfavorable geologic body three-dimensional deformation monitoring method
Technical field
The present invention relates to the geodesic survey field of remote sensing image, particularly relate to a kind of electrical network unfavorable geologic body three-dimensional deformation prison
Survey method.
Background technology
At present, the method that under electrical network, unfavorable geologic body deformation obtains is mainly the measurement of the level and GPS measures, but they are all
There is obvious defect: one is the stability problem of bench mark (markstone point);Two be measure result be based on a position, and
Measuring point sparse, this causes providing the deformation of region unfavorable geologic body residing for whole electrical network.It addition, this process needs to expend
Substantial amounts of manpower, financial resources and material resources;Three is that artificial enforcement measurement ratio is relatively hazardous.It is poor that general electrical network lays environment, and physical features is dangerously steep, no
The harm of good geologic body deformation disaster is big, needs anticipated environment residing for electrical network, could lay next step measurement work, measure
In journey, the personal safety to operator ensures that difficulty is big.
Synthetic aperture interferometer radar (InSAR, Interferometric Synthetic Aperture radar) is measured
Have high accuracy, on a large scale, advantage all-time anf all-weather, applied greatly in Ground Deformation is monitored and developed.With
The extreme enrichment of High-resolution SAR Images data, utilize InSAR technology high to the high accuracy monitoring unfavorable geologic body under electrical network
Having a extensive future of the deformation of resolution.
But, it is currently based on the research of the deformation monitoring of unfavorable geologic body under the electrical network of InSAR technology little, is primarily due to
InSAR technology can only obtain the Ground Deformation One Dimensional Projection at radar line of sight direction (LOS), and the shape of unfavorable geologic body under electrical network
Change is often to occur at three-dimensional.InSAR how is utilized still to have very to obtain the three-dimensional deformation of unfavorable geologic body under electrical network
Many algorithms need to break through.
Secondly, in High-resolution SAR Images, by the shadow that radar system running parameter is different with ground target characteristic
Ringing, the backward scattered signal phasor of ground object target is different, and the image information formed also differs, and echo-signal is probably
All backscatter signal vectors in pixel resolution cell, it is also possible to the Scattering of Vector of prevailing scattering object, this
Sample will form the speckle noise in image, causes result of interference to produce error.High score SAR image registration interfere after, shape
The interferometric phase gradient become is little, and phase place easily produces winding, causes solution to twine process difficulty, thus affects the essence of the deformation of acquisition
Degree.
Furthermore, only with single track single pair of ground high resolution SAR data, it is impossible to obtain the time sequence of unfavorable geologic body under electrical network
Row deformation, more can not obtain its three-dimensional deformation.
It addition, the three-dimensional deformation of unfavorable geologic body under acquisition electrical network, the automatic identification of steel tower on unfavorable geologic body in electrical network
It it is requisite step.Relative to optical image, the vision readability of SAR image is poor, in addition by speckle noise and shade,
The folded impact covering, having an X-rayed the geometric distortions such as contraction, its information processing is more difficult.
Summary of the invention
Present invention aim to solve the deficiency of above-mentioned background technology, it is provided that a kind of electrical network unfavorable geologic body three-dimensional shaped
Become monitoring method, use sequential InSAR technology PS-InSAR to process high-resolution interferogram, and by multi-platform and many time domains
Observation data fusion means obtain the three-dimensional deformation of unfavorable geologic body under electrical network, use multiple dimensioned based on TurboPixels
Electric power pylon in partitioning algorithm detection High-resolution SAR Images, can automatically analyze the stability of column foot in conjunction with the above results.
For achieving the above object, the invention discloses a kind of electrical network unfavorable geologic body three-dimensional deformation monitoring method, mainly wrap
Include following steps:
The first step: choose satisfactory multi-platform, the High-resolution SAR Images of many time domains;
Second step: in the employing PS-InSAR technology single described platform High-resolution SAR Images of extraction under electrical network poorly
The deformation of plastid is estimated;
3rd step: estimate based on the deformation of unfavorable geologic body under electrical network in single described platform SAR image, use Kalman
Filtering technique estimates the three-dimensional deformation of electrical network unfavorable geologic body.
Described PS-InSAR technology is used to extract in single described platform High-resolution SAR Images unfavorable geologic body under electrical network
Deformation estimate, comprise the steps:
The first step: extract differential interferometry figure;
Second step: many time domains SAR image is carried out radiant correction.The N width obtained is covered many time domains SAR of survey region
Image carries out radiant correction.Radiation correction method may select relative detector calibration method.The amplitude first calculating every width SAR image is equal
Value:
Wherein, A(x,r)Represent the original amplitude of pixel, x and r represent respectively the orientation of pixel to distance to coordinate, m,
N represent respectively image direction to distance to number, t is the time series of image, then calculate seasonal effect in time series amplitude equal
Value:
Calculate the relative amplitude correction factor of single image:
Utilize amplitude correction factor that every width SAR image is carried out amplitude correction, make N width SAR image amplitude naturalization, for PS
The identification of point provides the foundation;
3rd step: identify the high-quality Permanent scatterers point PS kept in many time domains SAR image;
4th step: the PS point chosen is attached and builds PS network and carries out deformation estimation.Assume that earth's surface is at SAR shadow
As during imaging along radar line of sight direction the most linear and non-linear deformation, then for appointing in the i-th width differential interferometry figure
Anticipate a PS pixel (x, r), its phase value can be with once model representation:
In formula,And TiRespectively represent interfere to time reference line and Space Baseline, Δ h (x, r), v (x, r),Represent respectively vertical error, moment radar line of sight to (LOS) to rate of deformation and permanent set, λ is in radar
Cardiac wave is long, and θ is radar angle of incidence.Permanent set includes non-linear deformation, atmospheric phase and mistake correlated noise;
Use Delaunay triangulation network setpoint distance threshold value to carry out PS net to set up, use adjacent PS point difference method to weaken
The error relevant with space length, using the Linear deformation rate of adjacent two PS difference and elevation correction value as the letter of network adjustment
Digital-to-analogue type, uses the method for regional network least square method and Searching Resolution Space to carry out parametric solution, and carries out residual phase
Spatio-temporal filtering isolates non-linear deformation, the most just estimates total deformation of PS point.
Use described Kalman Filter Technology to estimate the three-dimensional deformation of electrical network unfavorable geologic body, comprise the steps:
The first step: obtain the deformation component of described electrical network unfavorable geologic body based on described differential interferometry figure, it is assumed that different flat
Platform is respectively arranged with N width image, uses PS-InSAR technology to process the SAR image of each platform, forms multiple interferogram, for
I-th width interferogram of certain platform, its interferometric phase is represented by:
T in formulaB、tAFor the major-minor image capturing time, λ is radar center wavelength, x and r is respectively pixel position to, distance
To coordinate, it is assumed that the reference time is t0, this moment radar line of sight is equal to 0 to (LOS) Ground Deformation amount, d (tB,x,r)、d(tA,x,
R) it is that major-minor image is relative to t0The LOS accumulation deformation quantity in moment,Represent landform residual error phase place, incident with radar
Angle θ is inversely proportional to, with interference to vertical parallax B⊥It is directly proportional,For orbit error phase place,For residual phase, bag
Include noise, atmospheric phase and high frequency deformation phase place etc., wherein LOS to deformation d can be write as
D in formulau,de,dnBe respectively earth's surface add up deformation vertically to, East and West direction, the component in north-south, a, b, c are then
LOS at the vertically projection vector in, East and West direction and north-south, α be satellite position to.Use linear model describe LOS to
The low frequency component of deformation, so
D=ν (t-t0)=(a vu+b·ve+c·vn)·(t-t0) (7)
V in formulau,ve,vnBe respectively deformation vertically to, East and West direction, the speed in north-south, utilize the dry of all interferograms
Relate to phase place composition matrix, use method of least square to solve the parameter to be asked in formula (5).
Second step: set up three-dimensional deformation Filtering Model;Three-dimensional deformation Filtering Model based on multi-platform many time domains are set up, false
If any one observation station of the i-th width interferogram, LOS to observation L (i) and the state model of moment i can be expressed as
L (i)=H (i) X (i)+V (i) (8)
X (i)=F (i/i-1) X (i-1)+Γ (i-1) W (i-1) (9)
Wherein X (i)T=[du(i) de(i) dn(i) vu(i) ve(i) vn(i)] it is expressed as state vector;du(i)
de(i) dn(i) vu(i) ve(i) vnI () represents t respectivelyiThe three-dimensional in moment adds up deformation (relative to moment t0) and three-dimensional shaped
Variable Rate, H (i)=[a (i) b (i) c (i) 00 0] is the design matrix of the projection vector composition of the i-th width interferogram;For state-transition matrix;I3It it is the unit matrix of 3 × 3;Δ t is moment ti-1And tiBetween
Time interval;V (i) represents observation noise vector;W(i)T=[wu(i) we(i) wn(i)] it is state-noise vector, Γ (i/i-
1)TFor noise profile matrix;
3rd step: use Kalman Filter Technology to estimate the three-dimensional deformation of electrical network unfavorable geologic body, utilize Kalman filtering
Equation group, can estimate moment tiState vector,
Wherein J (i) is gain matrix;For predictive value and the estimated value of state vector,WithFor corresponding variance-covariance matrix, E is unit battle array, it is assumed that V (i), W (i) are orthogonal white noise,
DW(i)=E (W (i) W (i)T) (12) whereinIt is the variance of the observation noise of the i-th width interferogram, Kalman
Filtering is new to carry out repetitive operation from image to each, but needs to carry to Kalman filtering before performing this operation
Initial state vector, at reference time t0Moment, it will be assumed that the three-dimensional in this moment adds up deformation vector and and corresponding variance
It is 0;Furthermore with short time baseline interferogram, use least square method to a number of time series differential interferometry
Figure carries out adjustment, the averaged deformation speed obtained and variance thereof to replace t0The three-dimensional deformation rate vector in moment and side thereof
Difference, it is sufficiently small that the solution needed for real Data processing, Kalman filtering twines the every error of differential interferometry figure, can be by upper one
Low frequency deformation and landform residual error in the interferogram that epoch is estimated deduct from interferogram, and permanent set carries out the anti-winding of 2 π
Reconciliation twines operation, low frequency deformation is conciliate the residual phase twined and is added, and sight line reforms into observation
WhereinRepresent that the solution of residual phase twines differential interferometry phase place, the SAR image that some moment is obtained, can
With with this moment before multiple moment of acquisition form the interferogram of short baseline, these interferograms are possible not only to filter into Kalman
Ripple provides unnecessary observed quantity, it is also possible to improve the precision that this moment three-dimensional deformation result is estimated.
Beneficial effects of the present invention: combine the feature of electrical network, applies high resolution SAR data, should by traditional PS-InSAR
The deformation monitoring of unfavorable geologic body laid for electrical network, and merge the PS-InSAR sight line of time domain multi-platform, many to observation,
Use kalman filter method to obtain the three-dimensional deformation of unfavorable geologic body under electrical network, overcome unfavorable geologic body three under current electrical network
The shortcoming that dimension deformation acquisition methods is wasted time and energy and cannot be obtained whole regional deformation;Meanwhile, method flow clear in structure, have
Realization is simple, expense is low, monitoring accuracy is high, monitoring range is big, automaticity advantages of higher.
Accompanying drawing explanation
Fig. 1 is PS-InSAR differential SAR Interferometry process chart in one embodiment of the present of invention;
Fig. 2 is that in one embodiment of the present of invention, three-dimensional deformation based on multi-platform SAR image solves relation;
Fig. 3 is that in one embodiment of the present of invention, under PS-InSAR electrical network based on Kalman filtering, unfavorable geologic body is three-dimensional
Deformation monitoring and column foot stability automatically analyze flow chart.
Detailed description of the invention
The present invention is described in further detail with specific embodiment below in conjunction with the accompanying drawings.
A kind of electrical network unfavorable geologic body three-dimensional deformation monitoring method disclosed in the present embodiment, mainly comprises the steps:
The first step: choose satisfactory multi-platform, the High-resolution SAR Images of many time domains.
In the present embodiment, specifically choose covering the TerraSAR-X/TanDEM-X of survey region, ALOS-2,
Sentinel-1 etc. are multi-platform, the high resolution SAR data of many time domains.
Second step: in the employing PS-InSAR technology described single platform High-resolution SAR Images of extraction under electrical network poorly
The deformation of plastid is estimated.
Fig. 1 is PS-InSAR differential SAR Interferometry process chart in one embodiment of the present of invention.Described many obtaining
Platform, many time domains high resolution SAR data after, first have to the SAR data of each single plateau time sequence is carried out PS-
InSAR process, extracts the interference observation of unfavorable geologic body under electrical network.Comprise the following steps:
(1) differential interferometry figure is extracted.First according to the N width SAR image of many time domains of the covering survey region obtained, filling
On the basis of dividing time, Space Baseline and the Doppler frequency center difference on the frequency optimum considered between N width SAR image, according to set up
Comprehensive correlation function chooses image corresponding during correlation coefficient maximum as public main image, using remaining image as from shadow
As registrating with main image respectively.According to the space-time baseline threshold set to the many time domains SAR image after being registered freely
It is combined into the short base line interference pair of multiple how main image, and to all interference to carrying out conventional interference treatment, outside being subsequently introduced
The SRTM data of the 30 meters of resolution in portion carry out the intensive striped that weakening area landform causes, and finally obtain M width differential interferometry figure.
(2) many time domains SAR image is carried out radiant correction.The N width obtained is covered many time domains SAR image of survey region
Carrying out radiant correction, radiation correction method may select relative detector calibration method.First calculate the amplitude average of every width SAR image:
Wherein, A(x,r)Represent the original amplitude of pixel, x and r represent respectively the orientation of pixel to distance to coordinate, m,
N represent respectively image direction to distance to number, t is the time series of image, then calculate seasonal effect in time series amplitude equal
Value:
Calculate the relative amplitude correction factor of single image:
Utilize amplitude correction factor that every width SAR image is carried out amplitude correction, make N width SAR image amplitude naturalization, for PS
The identification of point provides the foundation.
(3) the high-quality Permanent scatterers point PS kept in described many time domains SAR image is identified.Sequential is used to be correlated with phase
Close coefficient, phase place deviation threshold method, the combination of amplitude deviation threshold method, remove those dephasings and do serious impact point, serial identification
The high-quality Permanent scatterers point kept in image.
(4) the PS point chosen is attached and builds PS network and carries out deformation estimation.Assume that earth's surface becomes at SAR image
During picture along radar line of sight direction the most linear and non-linear deformation, then any one in the i-th width differential interferometry figure
Individual PS pixel (x, r), its phase value can be with model representation once:
In formula,And TiRespectively represent interfere to time reference line and Space Baseline, Δ h (x, r), v (x, r),Represent that vertical error, LOS are to rate of deformation and permanent set respectively.Permanent set includes non-linear deformation, air
Phase place and mistake correlated noise.
Use Delaunay triangulation network setpoint distance threshold value to carry out PS net to set up, use adjacent PS point difference method to weaken
The error relevant with space length, using the Linear deformation rate of adjacent two PS difference and elevation correction value as the letter of network adjustment
Digital-to-analogue type, uses the method for regional network least square method and Searching Resolution Space to carry out parametric solution, and carries out residual phase
Spatio-temporal filtering isolates non-linear deformation, the most just estimates total deformation of PS point.
3rd step: estimate based on the deformation of unfavorable geologic body under electrical network in described single platform SAR image, uses Kalman's filter
Wave technology estimates the three-dimensional deformation of electrical network unfavorable geologic body.
Fig. 2 is that in one embodiment of the present of invention, three-dimensional deformation based on multi-platform SAR image solves relation.Use and put down more
Platform, the InSAR observation of many time domains carry out the three-dimensional deformation monitoring of electrical network unfavorable geologic body, its essence is the friendship utilized in measuring
Meeting principle obtains the three dimensional change on earth's surface.And the SAR image of the most multi-platform many time domains is because time and space covers
Cover degree is different, is not specifically designed for them and carries out the data processing method of three-dimensional deformation inverting.
Fig. 3 is that in one embodiment of the present of invention, under PS-InSAR electrical network based on Kalman filtering, unfavorable geologic body is three-dimensional
Deformation monitoring and column foot stability automatically analyze flow chart.Kalman filtering is a kind of dynamic data processing method, and it can turn round and look at
And observation data relatedness in time domain and state, it is also possible to by time series merge that different platform obtains different time
Carve and the most co-orbital InSAR observation, obtain seasonal effect in time series three-dimensional deformation.Mainly comprise the steps:
(1) the deformation component of described electrical network unfavorable geologic body is obtained based on described differential interferometry figure.Assume that different platform is each
There is N width image, use the PS-InSAR technology implemented in step 2 that the SAR image of each platform is processed, formed multiple
Interferogram, for the i-th width interferogram of certain platform, its interferometric phase is represented by:
T in formulaB、tAFor the major-minor image capturing time, λ is radar center wavelength, x and r is respectively pixel position to, distance
To coordinate.Assume that the reference time is t0, this moment radar line of sight is equal to 0 to (LOS) Ground Deformation amount, d (tB,x,r)、d(tA,x,
R) it is that major-minor image is relative to t0The LOS accumulation deformation quantity in moment,Represent landform residual error phase place, incident with radar
Angle θ is inversely proportional to, and is directly proportional vertical parallax B ⊥ to interfering,For orbit error phase place,For residual phase,
Including noise, atmospheric phase and high frequency deformation phase place etc..Wherein LOS to deformation d can be write as
D=[a b c] [du de dn]T
A=cos θ (6)
B=-cos (α-3 pi/2) sin θ
C=-sin (α-3 pi/2) sin θ
D in formulau,de,dnBe respectively earth's surface add up deformation vertically to, East and West direction, the component in north-south, a, b, c are then
LOS at the vertically projection vector in, East and West direction and north-south, α be satellite position to.Use linear model describe LOS to
The low frequency component of deformation, so
D=ν (t-t0)=(a vu+b·ve+c·vn)·(t-t0) (7)
V in formulau,ve,vnBe respectively deformation vertically to, East and West direction, the speed in north-south, utilize the dry of all interferograms
Relate to phase place composition matrix, use method of least square to solve the parameter to be asked in formula (5).
(2) three-dimensional deformation Filtering Model is set up.Three-dimensional deformation Filtering Model based on multi-platform many time domains are set up.Assume
Any one observation station of i width interferogram, LOS to observation L (i) and the state model of moment i can be expressed as
L (i)=H (i) X (i)+V (i) (8)
X (i)=F (i/i-1) X (i-1)+Γ (i-1) W (i-1) (9)
Wherein X (i)T=[du(i) de(i) dn(i) vu(i) ve(i) vn(i)] it is expressed as state vector;
du(i) de(i) dn(i) vu(i) ve(i) vnI () represents t respectivelyiThe three-dimensional in moment add up deformation (relative to
Moment t0) and three-dimensional deformation speed, H (i)=[a (i) b (i) c (i) 00 0] is that the projection vector of the i-th width interferogram is constituted
Design matrix;For state-transition matrix;I3 is the unit matrix of 3 × 3;When Δ t is
Carve ti-1And tiBetween time interval;V (i) represents observation noise vector;W(i)T=[wu(i) we(i) wn(i)] it is state
Noise vector, Γ (i/i-1)TFor noise profile matrix.
(3) Kalman Filter Technology is used to estimate the three-dimensional deformation of electrical network unfavorable geologic body.Utilize Kalman filter equation
Group, can estimate moment tiState vector,
Wherein J (i) is gain matrix;For predictive value and the estimated value of state vector,WithFor corresponding variance-covariance matrix, E is unit battle array.Assume that V (i), W (i) are orthogonal white noise,
DW(i)=E (W (i) W (i)T) (12) whereinIt it is the variance of the observation noise of the i-th width interferogram.Kalman
Filtering is new to carry out repetitive operation from image to each, but needs to carry to Kalman filtering before performing this operation
Initial state vector.At reference time t0Moment, it will be assumed that the three-dimensional in this moment adds up deformation vector and and corresponding variance
It is 0.Furthermore with short time baseline interferogram, use least square method to a number of time series differential interferometry
Figure carries out adjustment, the averaged deformation speed obtained and variance thereof to replace t0The three-dimensional deformation rate vector in moment and side thereof
Difference.It is sufficiently small that solution needed for real Data processing, Kalman filtering twines the every error of differential interferometry figure, can be by upper one
Low frequency deformation and landform residual error in the interferogram that epoch is estimated deduct from interferogram, permanent set is carried out the counter of 2 π and twines
Twining operation around reconciliation, low frequency deformation is conciliate the residual phase twined and is added, the sight line in (8) formula reforms into observation
WhereinRepresent that the solution of residual phase twines differential interferometry phase place.Some is obtained in moment
SAR image, can form the interferogram of short baseline, these interferograms with multiple moment of the acquisition before this moment
It is possible not only to provide unnecessary observed quantity for Kalman filtering, it is also possible to improve the essence that this moment three-dimensional deformation result is estimated
Degree.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention is not limited thereto, and any
Belong to those skilled in the art in the technical scope that the invention discloses, the change that can readily occur in or replacement, all answer
Contain within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with scope of the claims.