CN105510901A - Optical satellite image time-varying error calibrating method and system based on multiple calibration fields - Google Patents

Abstract

The invention provides an optical satellite image time-varying error calibrating method and system based on multiple calibration fields. In order to accurately calibrate external parameters of an optical remote sensing satellite system, time parameters are considered when a calibration model is built, and a polynomial relationship between an installation angle and normalized time parameters is built and guided into a satellite image strict geometric imaging model to build a timing sequence calibration model; according to a certain time sequence, an image covering the multiple calibration fields is selected as the image to be calibrated for control point measurement; the time of the selected image is normalized; according to the timing sequence calibration model, an error equation is built, and a remote sensing image is calibrated through the built model. The method is feasible and effective, and an adjustment calculating result is stable and reliable.

Description

Error calibrating method and system is become during optical satellite image based on many calibration fields

Technical field

The invention belongs to remote sensing image process field, when relating to the optical satellite image based on many calibration fields, become error calibrating method and system.

Background technology

The calibration of Optical remote satellite image is the basis of optical remote sensing image geometric manipulations, is the prerequisite playing remotely sensing image geometric performance.The model basis of Optical remote satellite image calibration is geometric orientation model, geometric orientation model mainly comprises strict geometry imaging model and two kinds, rational polynominal model (RPC model), RPC model has unity of form, use simple feature, but it can hide the critical parameter information of satellite load simultaneously, be unfavorable for introducing time parameter in a model, and camera established angle parameter is dominant in strict geometry imaging model, can express in a model, strict geometry imaging model or variation on its basis is generally adopted when needing clear and definite camera error of fixed angles.

In view of the imaging characteristics of optical remote sensing image satellite, its imaging system error is mainly camera error of fixed angles, image show as at vertical rail with along the drift error on rail direction, simultaneously due to satellite in orbit time residing environment change, camera error of fixed angles is caused to change in time, and showing as certain timing, therefore will consider time factor when carrying out Optical remote satellite geometric calibration.But, not yet have related art scheme to occur.

Summary of the invention

The present invention becomes the calibrating method of error and the problem of calibration model structure when mainly solving Optical remote satellite image system.

Technical scheme of the present invention provides during a kind of optical satellite image based on many calibration fields and becomes error calibrating method, comprises the following steps:

Step 1, sets up sequential calibration model, and time parameter analogue camera error of fixed angles is introduced on strict geometry imaging model basis, obtains the function model F (X along rail image space residual error, rail image space residual error of hanging down _e), G (X _e);

Described sequential calibration model is as follows,

$\left\{\begin{array}{c}pitch\left(t\right)={\mathrm{ax}}_0+{\mathrm{ax}}_{1}t+{\mathrm{ax}}_2{t}^2+{\mathrm{ax}}_3{t}^3\\ roll\left(t\right)={\mathrm{bx}}_0+{\mathrm{bx}}_{1}t+{\mathrm{bx}}_2{t}^2+{\mathrm{bx}}_3{t}^3\\ yaw\left(t\right)={\mathrm{cx}}_0+{\mathrm{cx}}_{1}t+{\mathrm{cx}}_2{t}^2+{\mathrm{cx}}_3{t}^3\end{array}\right.$

Wherein, independent variable t represents normalized time variable, and (pitch (t), roll (t), yaw (t)) is the camera established angle under time t, outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃) be parameter to be resolved;

Step 2, chooses the image covering multiple calibration field and calibrates image for waiting, carry out reference mark measurement according to certain hour sequence;

Step 3, does normalized to the time of the image chosen;

Step 4, sets up error equation according to sequential calibration model, adjustment resolution error coefficient, and implementation is as follows,

For outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃), set up error equation as follows,

V＝AX-LP

Wherein, A represents the matrix of coefficients of error equation, and X represents outer scaling parameter correction (dax ₀, dax ₁, dax ₂, dax ₃, dbx ₀, dbx ₁, dbx ₂, dbx ₃, dcx ₀, dcx ₁, dcx ₂, dcx ₃), V represents the correction of error equation, and P is the weight matrix of observed reading; L utilizes outer scaling parameter currency and reference mark coordinate errors vector;

$L={\left[\begin{array}{cc}F\left(X_E^o\right)& G\left(X_E^o\right)\end{array}\right]}^T.$

Least square adjustment is utilized to calculate X,

X＝(A ^TPA) ^-1(A ^TPL)

Utilize upgrade outer scaling parameter X _ecurrency, then carry out parameter calculation according to above formula iteration, when outer scaling parameter correction is all less than predetermined threshold value, iteration stopping.

And the reference mark image choosing different time in the whole world carries out precision evaluation to the calibration results.

The present invention also provides during a kind of optical satellite image based on many calibration fields and becomes error scaling system, comprises with lower module:

First module, for setting up sequential calibration model, time parameter analogue camera error of fixed angles is introduced on strict geometry imaging model basis, obtains the function model F (X along rail image space residual error, rail image space residual error of hanging down _e), G (X _e);

Described sequential calibration model is as follows,

$\left\{\begin{array}{c}pitch\left(t\right)={\mathrm{ax}}_0+{\mathrm{ax}}_{1}t+{\mathrm{ax}}_2{t}^2+{\mathrm{ax}}_3{t}^3\\ roll\left(t\right)={\mathrm{bx}}_0+{\mathrm{bx}}_{1}t+{\mathrm{bx}}_2{t}^2+{\mathrm{bx}}_3{t}^3\\ yaw\left(t\right)={\mathrm{cx}}_0+{\mathrm{cx}}_{1}t+{\mathrm{cx}}_2{t}^2+{\mathrm{cx}}_3{t}^3\end{array}\right.$

Wherein, independent variable t represents normalized time variable, and (pitch (t), roll (t), yaw (t)) is the camera established angle under time t, outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃) be parameter to be resolved;

Second module, calibrating image for choosing the image covering multiple calibration field according to certain hour sequence for waiting, carrying out reference mark measurement;

3rd module, for doing normalized to the time of the image chosen;

Four module, for setting up error equation according to sequential calibration model, adjustment resolution error coefficient, implementation is as follows,

For outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃), set up error equation as follows,

V＝AX-LP

Wherein, A represents the matrix of coefficients of error equation, and X represents outer scaling parameter correction (dax ₀, dax ₁, dax ₂, dax ₃, dbx ₀, dbx ₁, dbx ₂, dbx ₃, dcx ₀, dcx ₁, dcx ₂, dcx ₃), V represents the correction of error equation, and P is the weight matrix of observed reading; L utilizes outer scaling parameter currency and reference mark coordinate errors vector;

$L={\left[\begin{array}{cc}F\left(X_E^o\right)& G\left(X_E^o\right)\end{array}\right]}^T.$

Least square adjustment is utilized to calculate X,

X＝(A ^TPA) ^-1(A ^TPL)

Utilize upgrade outer scaling parameter X _ecurrency, then carry out parameter calculation according to above formula iteration, when outer scaling parameter correction is all less than predetermined threshold value, iteration stopping.

And, the 5th module is set, for the reference mark image choosing different time in the whole world, precision evaluation is carried out to the calibration results.

The invention has the advantages that: by the Changing Pattern of analytical error, consider that when calibrating phase factor is on the impact of error, satellite system is in imaging process, error with rail image is basically identical, can be variant between different rail image, by introducing time parameter in calibration model, the function of camera established angle about the time can be obtained in calibration process, then can be obtained by this function for not camera established angle in the same time, thus revise the RPC parameter of this image.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet of the embodiment of the present invention.

Embodiment

The specific embodiment of the invention is described in detail below in conjunction with drawings and Examples.

Optical remote satellite image system external error is mainly derived from camera error of fixed angles, shows for vertical rail with along the drift error on rail direction on image, and almost consistent with the error on rail image, and this error presents certain timing.In order to compensate for optical remote sensing satellite system time become error, the present invention becomes error calibration and modeling method when proposing a kind of Optical remote satellite image system based on many calibration fields, time factor is introduced on strict geometry imaging model basis and sets up sequential calibration model, by carrying out regularly long-term calibration to Optical remote satellite system, thus the time dependent rule of simulation error, become the object of error when reaching bucking-out system.The present invention is for giving top priority to what is the most important, and supposed premise is the known outer calibration carrying out sequential of internal calibration parameter, when internal calibration unknown parameters, the model of internal calibration can be incorporated in model of the present invention equally and carry out inside and outside calibration simultaneously.

See Fig. 1, the flow process of embodiment can be divided into four steps, and concrete grammar, formula and flow process that each step is implemented are as follows:

1. set up sequential calibration model.The present invention considers the image of phase factor to camera established angle, the present invention utilizes a cubic polynomial to carry out matching to the camera established angle under different time sequence, and it can be used as the outside the pale of civilization calibration model of the sequential of Optical remote satellite camera, shown in (1), wherein independent variable t represents normalized time variable, (pitch (t), roll (t), yaw (t)) is the camera established angle under time t.

$\left\{\begin{array}{c}pitch\left(t\right)={\mathrm{ax}}_0+{\mathrm{ax}}_{1}t+{\mathrm{ax}}_2{t}^2+{\mathrm{ax}}_3{t}^3\\ roll\left(t\right)={\mathrm{bx}}_0+{\mathrm{bx}}_{1}t+{\mathrm{bx}}_2{t}^2+{\mathrm{bx}}_3{t}^3\\ yaw\left(t\right)={\mathrm{cx}}_0+{\mathrm{cx}}_{1}t+{\mathrm{cx}}_2{t}^2+{\mathrm{cx}}_3{t}^3\end{array}\right.---\left(1\right)$

Then for outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃), calibration model is set up on the basis of strict geometry imaging model, and strict geometry imaging model formula is as follows:

$\left(\begin{array}{c}x\\ y\\ f+\mathrm{\Δ}f\end{array}\right)={\mathrm{\λR}}_{body}^{cam}\left(pitch\right(t),roll(t),yaw(t\left)\right)(R_{J2000}^{body}{K}_{gs}^{2000}{\left[\begin{array}{c}{X}_g-X_{gps}\left(t_{gps}\right)\\ Y_g-Y_{gps}\left(t_{gps}\right)\\ Z_g-Z_{gps}\left(t_{gps}\right)\end{array}\right]}_{wgs}-{\left[\begin{array}{c}{B}_X\\ B_Y\\ B_Z\end{array}\right]}_{body})---\left(2\right)$

Wherein, (x, y) represents picpointed coordinate, and f represents focal length, and Δ f represents the reduction of focal length, (X _g, Y _g, Z _g) represent the object coordinates of culture point; (X _gps(t _gps), Y _gps(t _gps), Z _gps(t _gps)) and (B _x, B _y, B _z) represent that camera shooting center is at the t time of exposure over the ground respectively _gpsobject coordinates and GPS eccentric error; λ represents scale-up factor; represent respectively and be tied to J2000 coordinate system rotation matrix by WGS84 coordinate, be tied to satellite body coordinate system rotation matrix by J2000 coordinate and be tied to the rotation matrix of camera surving coordinate system by satellite body coordinate.

Make in formula (2)

$\left\{\begin{array}{c}\left[\begin{array}{c}Ux\\ Uy\\ Uz\end{array}\right]=(R_{J2000}^{body}{R}_{wgs}^{J2000}{\left[\begin{array}{c}{X}_g-X_{gps}\left(t_{gps}\right)\\ Y_g-Y_{gps}\left(t_{gps}\right)\\ Z_g-Z_{gps}\left(t_{gps}\right)\end{array}\right]}_{wgs}-{\left[\begin{array}{c}{B}_X\\ B_Y\\ B_Z\end{array}\right]}_{body})\\ R_{body}^{cam}\left(pitch\right(t),roll(t),yaw(t\left)\right)=\left[\begin{array}{c}{a}_1,b_1,c_1\\ a_2,b_2,c_2\\ a_3,b_3,c_3\end{array}\right]\end{array}\right.$

Wherein, (U _x, U _y, U _z) be the coordinate of ground coordinate under body coordinate system, (X _gps(t), Y _gps(t), Z _gps(t)) be time t _gpsunder the object coordinates at camera shooting center over the ground, for the corresponding satellite body coordinate of camera established angle is tied to the rotation matrix of camera surving coordinate system, (a ₁, a ₂, a ₃, b ₁, b ₂, b ₃, c ₁, c ₂, c ₃) be element wherein.

Formula (2) can be converted into formula (3)

$\left\{\begin{array}{c}F\left(X_E\right)=\frac{a_{1}Ux+b_{1}Uy+c_{1}Uz}{a_{3}Ux+b_{3}Uy+c_{3}Uz}-\frac{x}{f}\\ G\left(X_E\right)=\frac{a_{2}Ux+b_{2}Uy+c_{2}Uz}{a_{3}Ux+b_{3}Uy+c_{3}Uz}-\frac{y}{f}\end{array}\right.---\left(3\right)$

F (X _e), G (X _e) represent along rail image space residual error respectively, hang down the function model of rail image space residual error.

2. calibrate choosing and reference mark measurement of image.In certain hour sequence (such as: 365 days), more calibration field image is obtained in order to cross, make the timing of calibration result more reliable, the present invention chooses the remote sensing image covering multiple calibration field in the world, and record the shooting time of these images, on these images, find ground control point and record corresponding picpointed coordinate simultaneously.

3. the time of pair image chosen does normalized process.Suppose that in year, choose n opens the time series of photo as timestamp sequence image sequence undetermined, is arranged as (t from small to large ₁, t ₂... t _n), then to its be normalized into:

$t_j^{\′}=\frac{t_j}{365},(j=1,2...n)---\left(4\right)$

Wherein, t' _jfor the corresponding time normalization result of jth sheet photo, in resolving below, use t' _jreplace original t to carry out adjustment to resolve.

4. row error equation and adjustment are resolved, and obtain the calibration results.According to above-mentioned sequential calibration model row error equation, need to resolve 12 outer scaling parameter and X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃), error equation is as follows:

V＝AX-LP(5)

Wherein $A=\left[\begin{array}{c}\frac{\∂F}{{\∂X}_E}\\ \frac{\∂G}{{\∂X}_E}\end{array}\right]=\left[\begin{array}{cccccccccccc}\frac{\∂F}{{\∂\mathrm{ax}}_0}& \frac{\text{\∂F}}{{\∂\mathrm{ax}}_1}& \frac{\∂F}{{\∂\mathrm{ax}}_2}& \frac{\∂F}{{\∂\mathrm{ax}}_3}& \frac{\∂F}{{\∂\mathrm{ax}}_0}& \frac{\∂F}{{\∂\mathrm{ax}}_1}& \frac{\∂F}{{\∂\mathrm{ax}}_2}& \frac{\∂F}{{\∂\mathrm{ax}}_3}& \frac{\∂F}{{\∂\mathrm{ax}}_0}& \frac{\∂F}{{\∂\mathrm{ax}}_1}& \frac{\∂F}{{\∂\mathrm{ax}}_2}& \frac{\∂F}{{\∂\mathrm{ax}}_3}\\ \frac{\∂G}{{\∂\mathrm{ax}}_0}& \frac{\∂G}{{\∂\mathrm{ax}}_1}& \frac{\∂G}{{\∂\mathrm{ax}}_2}& \frac{\∂G}{{\∂\mathrm{ax}}_3}& \frac{\∂G}{{\∂\mathrm{ax}}_0}& \frac{\∂G}{{\∂\mathrm{ax}}_1}& \frac{\∂G}{{\∂\mathrm{ax}}_2}& \frac{\∂G}{{\∂\mathrm{ax}}_3}& \frac{\∂G}{{\∂\mathrm{ax}}_0}& \frac{\∂G}{{\∂\mathrm{ax}}_1}& \frac{\∂G}{{\∂\mathrm{ax}}_2}& \frac{\∂G}{{\∂\mathrm{ax}}_3}\end{array}\right],$

X＝dX _E＝[dax ₀dax ₁dax ₂dax ₃dbx ₀dbx ₁dbx ₂dbx ₃dcx ₀dcx ₁dcx ₂dcx ₃] ^T，

$L={\left[\begin{array}{cc}F\left(X_E^o\right)& G\left(X_E^o\right)\end{array}\right]}^T$

In formula, F and G is respectively along rail image space residual error F (X _e), hang down rail image space residual error G (X _e) function model, V represents the correction of error equation, be respectively the vector of function model to scaling parameter differentiate gained, for reference mark, L utilizes outer scaling parameter currency with the error vector that reference mark coordinate substitution formula (5) calculates, by object space point coordinate initial value and outer scaling parameter currency substitute into formula (5) and calculate error vector L. i.e. outer scaling parameter currency corresponding F (X _e), G (X _e).

A represents the matrix of coefficients of error equation; X represents outer scaling parameter correction (dax ₀, dax ₁, dax ₂, dax ₃, dbx ₀, dbx ₁, dbx ₂, dbx ₃, dcx ₀, dcx ₁, dcx ₂, dcx ₃); P is the weight matrix of observed reading.

Least square adjustment is utilized to calculate X, such as formula (6):

X＝(A ^TPA) ^-1(A ^TPL)(6)

Formula (7) is utilized to upgrade outer scaling parameter X _ecurrency, then iteration (5) ~ (6) formula carries out parameter calculation.When outer scaling parameter correction is all less than predetermined threshold value (such as 10 ^-12) time, iteration stopping.

$X_E=X_E^o+X---\left(7\right)$

According to above flow process, automatic acquisition the calibration results can be realized.

During concrete enforcement, the reference mark image can also choosing different time in the whole world carries out precision evaluation to the calibration results.The present invention mainly evaluates the absolute geometry positioning precision before and after image calibration, evaluation method is as checkpoint using reference mark, utilize the image space coordinate figure that the RPC parameter calculating inspection point of image is corresponding, difference is done again with the image space coordinate measured, this difference is the absolute geometry positioning precision of image, and the detailed process of precision evaluation is as follows:

1) choose 500 scape reference mark images according to regular hour sequence in the world and carry out accuracy checking, and measure corresponding picpointed coordinate.

2) according to formula (4), time normalization process is carried out to the image chosen.

3) image chosen is calculated to the camera established angle of different time image according to formula (1).

4) according to the RPC parameter of camera established angle correction image.RPC parameter correction method is: on image, divide certain uniform grid, to each graticule mesh central pixel, utilize orientation parameter inside and outside this image, a series of object space point is obtained and this image space central point forms reference mark point pair by forward intersection intersection on object space local a series of elevation datum (from-1000m to 9000m every 1000m elevation datum), according to these points to and RPC recursive model can adjustment resolve RPC parameter, solution process still adopts the method for photogrammetric middle Iterative in prior art, and it will not go into details in the present invention.

5) calculate image space coordinate (x, y) according to image RPC parameter and check point coordinates, its computing formula is as follows:

$\left\{\begin{array}{c}x=\frac{\underset{\mathrm{\α}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\β}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\γ}=0}{\overset{3}{\mathrm{\Σ}}}{\mathrm{Sn}}_{\mathrm{\α}\mathrm{\β}\mathrm{\γ}}{U}^a{V}^{\mathrm{\β}}{W}^{\mathrm{\γ}}}{\underset{\mathrm{\α}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\β}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\γ}=0}{\overset{3}{\mathrm{\Σ}}}{\mathrm{Sd}}_{\mathrm{\α}\mathrm{\β}\mathrm{\γ}}{U}^a{V}^{\mathrm{\β}}{W}^{\mathrm{\γ}}}\\ y=\frac{\underset{\mathrm{\α}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\β}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\γ}=0}{\overset{3}{\mathrm{\Σ}}}{\mathrm{Ln}}_{\mathrm{\α}\mathrm{\β}\mathrm{\γ}}{U}^a{V}^{\mathrm{\β}}{W}^{\mathrm{\γ}}}{\underset{\mathrm{\α}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\β}=0}{\overset{3}{\mathrm{\Σ}}}\underset{\mathrm{\γ}=0}{\overset{3}{\mathrm{\Σ}}}{\mathrm{Ld}}_{\mathrm{\α}\mathrm{\β}\mathrm{\γ}}{U}^a{V}^{\mathrm{\β}}{W}^{\mathrm{\γ}}}\end{array}\right.---\left(8\right)$

In formula, Sn _{α β γ}, Sd _{α β γ}ln _{α β γ}, Ld _{α β γ}represent rational polynominal coefficient (rationalpolynomialcoefficients, RPC), (U ^α, V ^β, W ^γ) simplify be designated as (U, V, W), (U, V, and (x W), y) ground point terrestrial coordinate and the picpointed coordinate of regularization is represented respectively, ground point terrestrial coordinate (Lat, the Lon of they and non-regularization, Height) relation and between picpointed coordinate (s, l) is as formula (9):

$\begin{array}{cc}\begin{array}{c}V=\frac{Lat-LAT\_OFF}{LAT\_SCALE}\\ U=\frac{Lon-LON\_OFF}{LON\_SCALE}\\ W=\frac{Height-H\_OFF}{H\_SCALE}\end{array}& \begin{array}{c}y=\frac{s-S\_OFF}{s\_SCALE}\\ x=\frac{l-L\_OFF}{l\_SCALE}\end{array}\end{array}---\left(9\right)$

Wherein, LAT_OFF, LAT_SCALE, LON_OFF, LON_SCALE, H_OFF and H_SCALE are the regularization parameter of ground coordinate; S_OFF, S_SCALE, L_OFF and L_SCALE are the regularization parameter of image coordinate.

6) picture point residual error (Δ x, Δ y) is calculated.Residual computations formula (10) is as follows:

$\left\{\begin{array}{c}\mathrm{\Δ}x=x-x^{\′}\\ \mathrm{\Δ}y=y-y^{\′}\end{array}\right.---\left(10\right)$

Wherein, the image space that (x', y') is checkpoint measures coordinate.

7) the picture point residual error before and after statistical scaling, calculates its medial error and CE90, and draws corresponding error map.

During concrete enforcement, method provided by the present invention can realize automatic operational scheme based on software engineering, and modular mode also can be adopted to realize corresponding system.The present invention also provides during a kind of optical satellite image based on many calibration fields and becomes error scaling system, comprises with lower module:

First module, for setting up sequential calibration model, time parameter analogue camera error of fixed angles is introduced on strict geometry imaging model basis, obtains the function model F (X along rail image space residual error, rail image space residual error of hanging down _e), G (X _e);

Described sequential calibration model is as follows,

$\left\{\begin{array}{c}pitch\left(t\right)={\mathrm{ax}}_0+{\mathrm{ax}}_{1}t+{\mathrm{ax}}_2{t}^2+{\mathrm{ax}}_3{t}^3\\ roll\left(t\right)={\mathrm{bx}}_0+{\mathrm{bx}}_{1}t+{\mathrm{bx}}_2{t}^2+{\mathrm{bx}}_3{t}^3\\ yaw\left(t\right)={\mathrm{cx}}_0+{\mathrm{cx}}_{1}t+{\mathrm{cx}}_2{t}^2+{\mathrm{cx}}_3{t}^3\end{array}\right.$

Wherein, independent variable t represents normalized time variable, and (pitch (t), roll (t), yaw (t)) is the camera established angle under time t, outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃) be parameter to be resolved;

Second module, calibrating image for choosing the image covering multiple calibration field according to certain hour sequence for waiting, carrying out reference mark measurement;

3rd module, for doing normalized to the time of the image chosen;

Four module, for setting up error equation according to sequential calibration model, adjustment resolution error coefficient, implementation is as follows,

For outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃), set up error equation as follows,

V＝AX-LP

Wherein, A represents the matrix of coefficients of error equation, and X represents outer scaling parameter correction (dax ₀, dax ₁, dax ₂, dax ₃, dbx ₀, dbx ₁, dbx ₂, dbx ₃, dcx ₀, dcx ₁, dcx ₂, dcx ₃), V represents the correction of error equation, and P is the weight matrix of observed reading; L utilizes outer scaling parameter currency and reference mark coordinate errors vector;

$L={\left[\begin{array}{cc}F\left(X_E^o\right)& G\left(X_E^o\right)\end{array}\right]}^T.$

Least square adjustment is utilized to calculate X,

X＝(A ^TPA) ^-1(A ^TPL)

Utilize upgrade outer scaling parameter X _ecurrency, then carry out parameter calculation according to above formula iteration, when outer scaling parameter correction is all less than predetermined threshold value, iteration stopping.

Further, the 5th module is set, for the reference mark image choosing different time in the whole world, precision evaluation is carried out to the calibration results.

Each module specific implementation can see corresponding steps, and it will not go into details in the present invention.

Concrete enforcement described herein is only to the explanation for example of the present invention's spirit.Those skilled in the art can make various amendment or supplement or adopt similar mode to substitute to described concrete enforcement, but can't depart from spirit of the present invention or surmount the scope that appended claims defines.

Claims (4)

1. become an error calibrating method during optical satellite image based on many calibration fields, it is characterized in that, comprise the following steps:

Step 1, sets up sequential calibration model, and time parameter analogue camera error of fixed angles is introduced on strict geometry imaging model basis, obtains the function model F (X along rail image space residual error, rail image space residual error of hanging down _e), G (X _e);

Described sequential calibration model is as follows,

$\left\{\begin{array}{c}pitch\left(t\right)=a{x}_0+a{x}_{1}t+a{x}_2{t}^2+a{x}_3{t}^3\\ roll\left(t\right)={\mathrm{bx}}_0+{\mathrm{bx}}_{1}t+{\mathrm{bx}}_2{t}^2+{\mathrm{bx}}_3{t}^3\\ yaw\left(t\right)={\mathrm{cx}}_0+{\mathrm{cx}}_{1}t+{\mathrm{cx}}_2{t}^2+{\mathrm{cx}}_3{t}^3\end{array}\right.$

Wherein, independent variable t represents normalized time variable, and (pitch (t), roll (t), yaw (t)) is the camera established angle under time t, outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃) be parameter to be resolved;

Step 2, chooses the image covering multiple calibration field and calibrates image for waiting, carry out reference mark measurement according to certain hour sequence;

Step 3, does normalized to the time of the image chosen;

Step 4, sets up error equation according to sequential calibration model, adjustment resolution error coefficient, and implementation is as follows,

For outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃), set up error equation as follows,

V＝AX-LP

Wherein, A represents the matrix of coefficients of error equation, and X represents outer scaling parameter correction (dax ₀, dax ₁, dax ₂, dax ₃, dbx ₀, dbx ₁, dbx ₂, dbx ₃, dcx ₀, dcx ₁, dcx ₂, dcx ₃), V represents the correction of error equation, and P is the weight matrix of observed reading; L utilizes outer scaling parameter currency and reference mark coordinate errors vector;

$L={\left[\begin{array}{cc}F\left(X_E^o\right)& G\left(X_E^o\right)\end{array}\right]}^T.$

Least square adjustment is utilized to calculate X,

X＝(A ^TPA) ^-1(A ^TPL)

Utilize upgrade outer scaling parameter X _ecurrency, then carry out parameter calculation according to above formula iteration, when outer scaling parameter correction is all less than predetermined threshold value, iteration stopping.

2. become error calibrating method during optical satellite image as claimed in claim 1 based on many calibration fields, it is characterized in that: the reference mark image choosing different time in the whole world carries out precision evaluation to the calibration results.

3. become an error scaling system during optical satellite image based on many calibration fields, it is characterized in that, comprise with lower module:

First module, for setting up sequential calibration model, time parameter analogue camera error of fixed angles is introduced on strict geometry imaging model basis, obtains the function model F (X along rail image space residual error, rail image space residual error of hanging down _e), G (X _e);

Described sequential calibration model is as follows,

$\left\{\begin{array}{c}pitch\left(t\right)=a{x}_0+a{x}_{1}t+a{x}_2{t}^2+a{x}_3{t}^3\\ roll\left(t\right)={\mathrm{bx}}_0+{\mathrm{bx}}_{1}t+{\mathrm{bx}}_2{t}^2+{\mathrm{bx}}_3{t}^3\\ yaw\left(t\right)={\mathrm{cx}}_0+{\mathrm{cx}}_{1}t+{\mathrm{cx}}_2{t}^2+{\mathrm{cx}}_3{t}^3\end{array}\right.$

Wherein, independent variable t represents normalized time variable, and (pitch (t), roll (t), yaw (t)) is the camera established angle under time t, outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃) be parameter to be resolved;

Second module, calibrating image for choosing the image covering multiple calibration field according to certain hour sequence for waiting, carrying out reference mark measurement;

3rd module, for doing normalized to the time of the image chosen;

Four module, for setting up error equation according to sequential calibration model, adjustment resolution error coefficient, implementation is as follows,

For outer scaling parameter X _e=(ax ₀, ax ₁, ax ₂, ax ₃, bx ₀, bx ₁, bx ₂, bx ₃, cx ₀, cx ₁, cx ₂, cx ₃), set up error equation as follows,

V＝AX-LP

Wherein, A represents the matrix of coefficients of error equation, and X represents outer scaling parameter correction (dax ₀, dax ₁, dax ₂, dax ₃, dbx ₀, dbx ₁, dbx ₂, dbx ₃, dcx ₀, dcx ₁, dcx ₂, dcx ₃), V represents the correction of error equation, and P is the weight matrix of observed reading; L utilizes outer scaling parameter currency and reference mark coordinate errors vector;

$L={\left[\begin{array}{cc}F\left(X_E^o\right)& G\left(X_E^o\right)\end{array}\right]}^T.$

Least square adjustment is utilized to calculate X,

X＝(A ^TPA) ^-1(A ^TPL)

Utilize upgrade outer scaling parameter X _ecurrency, then carry out parameter calculation according to above formula iteration, when outer scaling parameter correction is all less than predetermined threshold value, iteration stopping.

4. become error scaling system during optical satellite image as claimed in claim 3 based on many calibration fields, it is characterized in that: the 5th module is set, for the reference mark image choosing different time in the whole world, precision evaluation is carried out to the calibration results.