CN103968808B - The strict geometric correction method of wide visual field Satellite CCD image - Google Patents

Abstract

The invention discloses the strict geometric correction method of a kind of wide visual field Satellite CCD image, including step: S1, satellite image direction along ng a path is divided multiple nodes；S2, utilizes the preliminary orbit parameter that satellite image provides, the appearance rail initial parameter values of each described node of the Fitting Calculation, camera distortion parameter initial value all 0；S3, acquisition controlling point data, and the coordinate figure at control point is brought into wide visual field satellite image Geometric rectification, and each described control point one equation of the most each row, utilize least square to solve the parameter of Geometric rectification；S4, utilizes the parameter of the Geometric rectification that the model solution in S3 obtains, wide visual field satellite image carries out resampling, and carries out geocoding, completes high accuracy geometric correction.The method of the present invention is corrected precision relative to additive method of the prior art and is improved more than at least 5 times.

Description

The strict geometric correction method of wide visual field Satellite CCD image

Technical field

The present invention relates to photogrammetry and remote-sensing technique field in topography, particularly relate to a kind of wide visual field Satellite CCD The strict geometric correction method of image.

Background technology

The Geometric rectification that satellite image is conventional at present includes rational function model, affine change based on parallel light projection Die change type, multinomial model, co-colouration effect etc., wherein rational function model and affine transformation mould based on parallel light projection Type is suitable for narrow angle of visual field high-resolution satellite image, and multinomial model is suitable for middle low resolution satellite image and correction precision is wanted The image geometry asking the highest is corrected, and above correction model is not the most strict geometric model, can only be by various deformation of image factors Consider and be fitted.

Major part middle high-resolution satellite image at present, the mode using linear array push to sweep obtains image, is currently based on conllinear The linear array satellite image Geometric rectification of equation assume that each base line 6 elements of exterior orientation (3 location parameters, 3 Attitude parameter) along heading linear change, therefore have only to calculate 6 elements of exterior orientation of one of them base line and 6 corresponding variable quantities.It is that approximation is set up that this assumed condition covers the shorter satellite image of track for every scape image , for environment disaster reduction satellite (HJ-1) CCD image etc and be false, the analysis of causes is as follows.

Test and trial production test by the HJ-1 image of whole world zones of different being carried out the prior art of geometric correction, right The orbit parameter of trial zone a large amount of HJ-1 image carries out statistical analysis, is concluded that

A () HJ-1 orbital position data are more stable, with quadratic polynomial regression criterion typically within 15 meters.

B () HJ-1 track profile data stability is poor, in major part scape, attitude can not be with uniform polynomial or triangle letter Number matching, data discrete degree (variance) and data value are originally in an order of magnitude, in accompanying drawing 1,2,3 is HJ-1 image scape respectively The distribution schematic diagram (abscissa is the satellite flight moment, and vertical coordinate is attitude value) of three direction attitudes.

Visible, there is single scape satellite trajectory length, track profile shakiness for wide visual field Satellite CCD image in prior art Fixed, camera distorts relatively big, causes current Geometric rectification to exist and corrects the problem that precision is poor.

Summary of the invention

It is an object of the invention to provide the strict geometric correction method of a kind of wide visual field Satellite CCD image, thus solve Foregoing problems present in prior art.

To achieve these goals, the technical solution used in the present invention is as follows:

The strict geometric correction method of a kind of wide visual field Satellite CCD image, comprises the following steps:

S1, divides multiple nodes by satellite image direction along ng a path；

S2, utilizes the preliminary orbit parameter that satellite image provides, the appearance rail initial parameter values of each described node of the Fitting Calculation, as Machine distortion parameter initial value all 0；

S3, acquisition controlling point data, and the coordinate figure at control point is brought wide visual field satellite image Geometric rectification into, often Individual described control point one equation of the most each row, utilizes least square to solve the parameter of Geometric rectification；

S4, utilizes the parameter of the Geometric rectification that the model solution in S3 obtains, and wide visual field satellite image carries out weight Sampling, and carry out geocoding, complete high accuracy geometric correction.

Preferably, when S1 divides multiple described node, if described control point is evenly distributed in orbital direction, each Described node separation is the most impartial, if described control point skewness in orbital direction, the interval also phase of each described node Should adjust.

Preferably, the quantity at described control point is no less than three times of described number of nodes.

Preferably, each described node includes 6 appearance rail parameters, and described 6 appearance rail parameters are by 3 location parameters and 3 Attitude parameter forms.

Preferably, the often row image appearance rail parameter between each described node uses the mode of interpolation to be calculated.

Preferably, described wide visual field satellite image Geometric rectification is:

$\left[\begin{array}{c}{x}_i\\ 0\\ -f\end{array}\right]=\frac{1}{\mathrm{\λ}}\left[\begin{array}{ccc}{a}_1\left(t\right)& b_1\left(t\right)& c_1\left(t\right)\\ a_2\left(t\right)& b_2\left(t\right)& c_2\left(t\right)\\ a_3\left(t\right)& b_3\left(t\right)& c_3\left(t\right)\end{array}\right]\left[\begin{array}{c}{X}_i-\mathrm{Xs}\left(t\right)\\ Y_i-\mathrm{Ys}\left(t\right)\\ Z_i-\mathrm{Zs}\left(t\right)\end{array}\right]$

Wherein: x_iFor picpointed coordinate, y_iPerseverance is 0, and f is the effective focal length of image；

X_i, Y_i, Z_iTopocentric coordinates corresponding to picture point；

Xs (t), Ys (t), Zs (t) are the projection centre geographical coordinates corresponding to t row；

a_i(t), b_i(t), c_iThe spin matrix that the attitude angle of t projection centre that () (i=1,2,3) is corresponding is constituted；

T () shows that each parameter changes with base line.

The invention has the beneficial effects as follows:

Lagrange appearance rail fitting function is added co-colouration effect by the method for the present invention, and additional one-dimensional high order camera is abnormal Varying model, is integrated in lagrangian fit function parameter, collinearity equation parameter, camera distortion model parameter in a model, Utilize control point and least square optimization to solve all parameters, reach the purpose of high accuracy geometric correction, the side of the present invention Method is corrected precision relative to additive method of the prior art and is improved more than at least 5 times.

Accompanying drawing explanation

Fig. 1 is the angular distribution schematic diagram of φ of three direction attitudes in HJ-1 image scape；

Fig. 2 is the angular distribution schematic diagram of ω of three direction attitudes in HJ-1 image scape；

Fig. 3 is the angular distribution schematic diagram of κ of three direction attitudes in HJ-1 image scape；

Fig. 4 is the wide visual field satellite image geometric model schematic diagram of the present invention；

Fig. 5 is the step schematic diagram of the strict geometric correction method of the wide visual field Satellite CCD image of the present invention.

Detailed description of the invention

In order to make the purpose of the present invention, technical scheme and advantage clearer, below in conjunction with accompanying drawing, the present invention is entered Row further describes.Should be appreciated that detailed description of the invention described herein, only in order to explain the present invention, is not used to Limit the present invention.

There is single scape satellite trajectory length for wide visual field Satellite CCD image, track profile is unstable, camera distortion is relatively big, Causing current Geometric rectification correction precision poor, the present invention proposes Lagrange appearance rail matching additional high order camera distortion Difference correcting method, improves more than 5 times by such satellite image geometric correction precision, within reaching 2 pixels.

The principle of the method for the present invention is: Lagrange appearance rail fitting function adds co-colouration effect, additional one-dimensional High order camera distortion model, is integrated in lagrangian fit function parameter, collinearity equation parameter, camera distortion model parameter In one model, utilize control point and least square optimization to solve all parameters, reach the purpose of high accuracy geometric correction. The details that wherein Lagrange appearance rail fitting function is integrated with collinearity equation is as follows.

By list scape wide visual field CCD satellite image direction along ng a path divide multiple nodes, can decile also can not decile (see figure 4), each track node has 6 appearance rail parameters (3 location parameters, 3 attitude parameters), the often row image between track node Appearance rail parameter uses the mode of interpolation to be calculated.The appearance rail parameter of track node is brought into conllinear side together with interpolating function Journey, is wide visual field satellite image Geometric rectification, sees formula (1).Assume there be n track node, then have 6 × n appearance Rail parameter, at least needs 3 × n control point to resolve above-mentioned parameter by least square adjustment.

The geometric correction method of single base line image is the same with frame width formula image, i.e. the imaging equation of t row is (formula 1):

$\left[\begin{array}{c}{x}_i\\ 0\\ -f\end{array}\right]=\frac{1}{\mathrm{\λ}}\left[\begin{array}{ccc}{a}_1\left(t\right)& b_1\left(t\right)& c_1\left(t\right)\\ a_2\left(t\right)& b_2\left(t\right)& c_2\left(t\right)\\ a_3\left(t\right)& b_3\left(t\right)& c_3\left(t\right)\end{array}\right]\left[\begin{array}{c}{X}_i-\mathrm{Xs}\left(t\right)\\ Y_i-\mathrm{Ys}\left(t\right)\\ Z_i-\mathrm{Zs}\left(t\right)\end{array}\right]$

Wherein: x_iFor picpointed coordinate, y_iPerseverance is 0, and f is the effective focal length of image；

X_i, Y_i, Z_iTopocentric coordinates corresponding to picture point；

Xs (t), Ys (t), Zs (t) are the projection centre geographical coordinates corresponding to t row；

a_i(t), b_i(t), c_iThe spin matrix that the attitude angle of t projection centre that () (i=1,2,3) is corresponding is constituted.

T () shows that each parameter changes with base line.

Instantaneous outer parameter Xs (t) of sensor platform, Ys (t), Zs (t),ω (t) κ (t) is to every a line CCD image Having different numerical value, they are the functions of the appearance rail parameter at track node place, can select conventional Lagrange's interpolation letter Number expresses (formula 2).I.e.

$\mathrm{\ω}\left(t\right)=\underset{k=0}{\overset{m}{\mathrm{\Σ}}}\left(\underset{j\≠k}{\overset{m}{\underset{j=0}{\mathrm{\Π}}}}\frac{t-t_j}{t_k-t_j}\right){\mathrm{\ω}}_l$

$\mathrm{\κ}\left(t\right)=\underset{k=0}{\overset{m}{\mathrm{\Σ}}}\left(\underset{j\≠k}{\overset{m}{\underset{j=0}{\mathrm{\Π}}}}\frac{t-t_j}{t_k-t_j}\right){\mathrm{\κ}}_l$

$\mathrm{Xs}\left(t\right)=\underset{k=0}{\overset{m}{\mathrm{\Σ}}}\left(\underset{j\≠k}{\overset{m}{\underset{j=0}{\mathrm{\Π}}}}\frac{t-t_j}{t_k-t_j}\right){\mathrm{Xs}}_l$

$\mathrm{Ys}\left(t\right)=\underset{k=0}{\overset{m}{\mathrm{\Σ}}}\left(\underset{j\≠k}{\overset{m}{\underset{j=0}{\mathrm{\Π}}}}\frac{t-t_j}{t_k-t_j}\right){\mathrm{Ys}}_l$

$\mathrm{Zs}\left(t\right)=\underset{k=0}{\overset{m}{\mathrm{\Σ}}}\left(\underset{j\≠k}{\overset{m}{\underset{j=0}{\mathrm{\Π}}}}\frac{t-t_j}{t_k-t_j}\right){\mathrm{Zs}}_l$

Whereinω_l, κ_l, Appearance rail parameter for track l node.

The method of the present invention comprises the steps:

(1) satellite image direction along ng a path dividing different node, the general decile of node separation, if control point is at track Being distributed the most uneven on direction, the Area Node interval sparse at control point can suitably increase.

(2) utilizing the preliminary orbit parameter that satellite image provides, the appearance rail initial parameter values of each node of the Fitting Calculation, camera is abnormal Variable element initial value all 0.

(3) utilizing control point to bring wide visual field satellite image Geometric rectification into, each control point is the most each One equation of row, general control point quantity exceedes the half of unknown parameter number, and therefore equation number exceedes unknown parameter Number, utilizes least square to solve the parameter of Geometric rectification.

(4) utilize the above-mentioned model solved, wide visual field satellite image is carried out resampling, and carries out geocoding, complete Geometric correction in high precision.

The method of the application present invention creates following effect:

Gather same control point data, be utilized respectively the method with conventional satellite image geometric correction method to wide visual field Satellite image carries out geometric correction, corrects precision and improves more than at least 5 times.

Accuracy comparison table corrected by table 1

The method that the present invention is described in detail in detail in detail below:

(1) control point is measured: measured Three Dimensional Ground coordinate and the shadow of correspondence thereof of characteristic point on image on the spot by field operation As upper characteristic coordinates, or the method by manual or automatic Image registration, at existing high accuracy orthography or other geography Characteristic coordinates on the Three Dimensional Ground coordinate of characteristic point and the image of correspondence thereof is measured on image on spatial data.Number of control points is not Can be less than three times of the node number arranged in step (3).

(2) track profile and positional parameter are read: from satellite image auxiliary information file, read orbital position and attitude Parameter, typically with the sampling number evidence every specified time interval.

(3) orientation node is arranged: satellite image direction along ng a path divides different nodes, the general decile of node separation, as Really control point is distributed the most uneven in orbital direction, and the Area Node interval sparse at control point can suitably increase.According to Experiment experience, nodal pitch precision between 40-120 kilometer is preferable.Utilize the track initial information matching meter that step (2) provides The initial value of operator node appearance rail parameter, camera distortion parameter initial value all 0.

(4) Model Parameter Optimization solves: brought at control point in wide visual field CCD satellite image rigorous geometric model, utilizes This geometric model parameter of least square method Optimization Solution, makes error global optimum under least square condition at each control point Change.

(5) control point adjusting and optimizing: according to each control point residual error (error) situation during (4th) step model optimization, Adjust or check control point distribution and precise manner, the most suitably accepting or rejecting and increase control point, general control point residual error Middle error is less than 2 pixels, and maximum residul difference is less than error in 2 times.Progressively adjust until global error (middle error) and list Till point tolerance can meet requirement.

(6) image rectification (resampling): utilize the model after optimizing, at image region Law of DEM Data (DEM) under auxiliary, satellite image being carried out geometry resampling and geocoding, complete satellite image just penetrates correction.

By using technique scheme disclosed by the invention, obtain following beneficial effect:

Lagrange appearance rail fitting function is added co-colouration effect by the method for the present invention, and additional one-dimensional high order camera is abnormal Varying model, is integrated in lagrangian fit function parameter, collinearity equation parameter, camera distortion model parameter in a model, Utilize control point and least square optimization to solve all parameters, reach the purpose of high accuracy geometric correction, the side of the present invention Method is corrected precision relative to additive method of the prior art and is improved more than at least 5 times.

The above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art For Yuan, under the premise without departing from the principles of the invention, it is also possible to make some improvements and modifications, these improvements and modifications also should Depending on protection scope of the present invention.

Claims (5)

1. the strict geometric correction method of a wide visual field Satellite CCD image, it is characterised in that comprise the following steps:

S1, divides multiple nodes by satellite image direction along ng a path；

S2, utilizes the preliminary orbit parameter that satellite image provides, and the appearance rail initial parameter values of each described node of the Fitting Calculation, camera is abnormal Variable element initial value all 0；

S3, acquisition controlling point data, and the coordinate figure at control point is brought into wide visual field satellite image Geometric rectification, Mei Gesuo State one equation of the most each row in control point, utilize least square to solve the parameter of Geometric rectification；

S4, utilizes the parameter of the Geometric rectification that the model solution in S3 obtains, wide visual field satellite image is carried out resampling, And carry out geocoding, complete high accuracy geometric correction；Within described high accuracy refers to that geometric correction precision reaches 2 pixels.

The strict geometric correction method of wide visual field Satellite CCD image the most according to claim 1, it is characterised in that described The quantity at control point is no less than three times of described number of nodes.

The strict geometric correction method of wide visual field Satellite CCD image the most according to claim 1, it is characterised in that each Described node includes that 6 appearance rail parameters, described 6 appearance rail parameters are made up of 3 location parameters and 3 attitude parameters.

The strict geometric correction method of wide visual field Satellite CCD image the most according to claim 1, it is characterised in that each institute Stating the often row image appearance rail parameter between node uses the mode of interpolation to be calculated.

The strict geometric correction method of wide visual field Satellite CCD image the most according to claim 4, it is characterised in that described Wide visual field satellite image Geometric rectification is:

Wherein: x_iFor picpointed coordinate, y_iPerseverance is 0, and f is the effective focal length of image；

X_i, Y_i, Z_iTopocentric coordinates corresponding to picture point；

X_S(t),Y_S(t),Z_ST () is the projection centre geographical coordinates corresponding to t row；

a_i(t), b_i(t), c_iThe spin matrix that the attitude angle of t projection centre that () (i=1,2,3) is corresponding is constituted；

T () shows that each parameter changes with base line.