CN104537614A - Orthographic correction method of CCD image of HJ-1 satellite - Google Patents

Abstract

The invention relates to an orthographic correction method of a CCD image of an HJ-1 satellite. A first-stage image of a CCD of an HJ satellite as well as corresponding altitude data and reference data are introduced; ground control points are collectedly uniformly at the first-stage image of the CCD of the HJ satellite; coordinates of the collected control points are introduced into a cubic rational function model and calculation is carried out to obtain a rational polynomial coefficient (RPC), so that a corresponding rational function model is obtained; pixel coordinates of all points of the first-stage image of the CCD of the HJ-1 satellite are inputted into the rational function model to obtain coordinates corresponding to all points of an image after orthographic correction; and with a bilinear interpolation method, interpolation of gray values of all points of the image after correction is carried out on points with pixel coordinates in an entire row or column mode, thereby obtaining an orthographic correction image. According to the invention, compared with the traditional polynomial method, the provided orthographic correction method enables the precision to be improved.

Description

A kind of environment Satellite CCD image ortho-rectification method

Technical field

The present invention relates to a kind of ccd image ortho-rectification method, particularly a kind of environment Satellite CCD sensor image ortho-rectification method, is applicable to the ortho-rectification of an environment Satellite CCD image, belongs to technical field of image processing.

Background technology

" No. one, environment " satellite A star, B star launch on September 6th, 2008 in Taiyuan.A, B two star respectively carried the CCD camera that two resolution are 30 meters, fabric width 360 kilometers.Two stellar associations with can realize two days return to the cycle.Due to characteristics such as temporal resolution are high, macroscopic property is good and multiple dimensioned, this satellite is with a wide range of applications in environment and disaster monitoring and assessment etc.

HJ star CCD image covers about 360 kilometers, and the geometric distortion that its huge fabric width causes HJ ccd image more serious, walk from substar toward edge, geometric distortion is increasing.Eliminate satellite image geometric distortion, improving satellite image positioning precision, is one of basic work of satellite image deep processing process.Geometric distortion is eliminated also referred to as geometry correction, is exactly to adopt geometric manipulations modeling satellite image imaging process, is object coordinates by raw video image space coordinate conversion.The key of geometry correction obtains geometric manipulations model exactly, and geometric manipulations model is generally divided into strict geometric manipulations model and general geometric manipulations model.Strict geometric manipulations model is the imaging mechanism from sensor, according to the feature of line array CCD capable central projection push-broom type imaging, be positioned at for foundation on straight line with imaging moment ground point, sensor projection centre and picture point, the topocentric coordinates utilizing the characteristic parameter of sensor, satellite ephemeris and sensor attitude angle to form and the direct strict geometric relationship formula of picpointed coordinate.When satellite orbit, the various characteristic parameter of sensor are enough accurate, the satellite image geometry location of strict geometric manipulations model tuning can reach very high precision.But the various parameter such as satellite, track all belongs to sensitive information, general user is difficult to obtain, and strict geometric manipulations model is difficult to set up, and HJ satellite image faces same problem, cannot set up strict transaction module.General geometric manipulations model adopts common mathematical function to set up the geometric relationship between object space three dimensional space coordinate image space two dimensional surface corresponding to it coordinate, does not need to obtain satellite orbit, the various characteristic parameter of sensor.So general geometric manipulations model is the utility model of satellite image geometric manipulations always, conventional has multinomial model and rational function model.Multinomial model has that form is simple, parameter is easy to solve, be suitable for the features such as all the sensors adopt by all remote sensing image processing business softwares.But multinomial model is only applicable to the satellite image that landform is smooth, geometry deformation is little, sensor side visual angle is less, lower, the easy generation oscillatory occurences of its correction accuracy and the application requirement of satellite image precise geometrical process cannot be met.HJ Satellite CCD image fabric width 360 kilometers, the edge's deformation of image away from substar is large, adopts multinomial model process HJ Satellite CCD adjustment of image precision lower (being generally 108 meters), does not meet precision and is better than 3 pixels i.e. application demand of 90 meters.Based on above reason, in satellite image process field, multinomial model is replaced by rational function model gradually.Rational function model is the expansion to multinomial model, is more broadly and more perfect a kind of expression-form of general geometric manipulations model.Set up rational function model without the need to obtaining the various sensitive parameter such as satellite, track, and the features such as rational function model correction accuracy is high, the oscillatory occurences that there will not be multinomial model, applicable dissimilar sensor, rational function model has become the de facto standard of satellite image geometry correction, and rational function model coefficient (RPC) file also becomes the standard configuration file that satellite data provides.So rational function model is applicable to the geometry correction of HJ Satellite CCD image.

The key setting up rational function model obtains rational polynominal coefficient (RPC), for various reasons, the HJ satellite image of satellite image supplier distribution does not have rational polynominal coefficient files, so, HJ star CCD image rational polynominal coefficient must be solved.Solve in the industry rational polynominal coefficient common method be the strict model utilizing satellite sensor known set up a group image grid points and correspondence ground grid points as virtual controlling point to resolve rational polynominal coefficient, the subsidiary rational polynominal coefficient files of domestic and international main flow satellite image is all adopt to solve rational polynominal coefficient in this way.But the HJ satellite image of HJ satellite image supplier distribution does not have strict model, so the mode adopting strict model to calculate rational polynominal coefficient is not suitable for HJ satellite image.The present invention is adopted and is a kind ofly solved the new method of rational polynominal coefficient by ground control point without the need to critical alignment model and relevant satellite sensor and orbit parameter.Practice shows, adopts this method process HJ star CCD adjustment of image precision to be 44 meters, is obviously better than the precision of traditional multinomial model 108 meters.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome the deficiencies in the prior art, propose a kind of environment Satellite CCD sensor image ortho-rectification method, the image of environment satellite and reference satellite is utilized to establish rational function model, and utilize this model and interpolating method to calibrate environment satellite image, the present invention has increased substantially the geometric accuracy of HJ star CCD image, meets the demand that environment Satellite CCD image is just penetrating calibration to the full extent.

Technical solution of the present invention is: a kind of environment Satellite CCD image ortho-rectification method, and step is as follows:

(1) the one-level image of an environment Satellite CCD is imported; To import mesh spacing be the SRTM altitude figures of M1 rice and resolution is the panchromatic reference data of LandSat of M2 rice; The scope of described M1 is [30,150], and the scope of M2 is [7.5,30];

(2) evenly gather N number of ground control point in the one-level image imported in step (1), obtain this reference mark at volume coordinate (P, L, H) and the reference mark pixel coordinate (X, Y) in one-level image with reference to image; Described N is more than or equal to 19;

(3) ground control point step (2) gathered is at the volume coordinate (P with reference to image, L, and the pixel coordinate (X of reference mark in one-level image H), Y) rational function model is imported, set up 4N equation, calculate rational polynominal coefficient, i.e. RPC, determine rational function model, specific formula for calculation is as follows:

$\left\{\begin{array}{c}P=\frac{P_1(X,Y,H)}{P_2(X,Y,H)}\\ L=\frac{P_3(X,Y,H)}{P_4(X,Y,H)}\end{array}\right.$

Wherein, P, L and H reference mark of being respectively collection is with reference to the horizontal ordinate of image, ordinate and height value; The pixel horizontal ordinate of the reference mark that X and Y is respectively collection in one-level image and pixel ordinate; P ₁, P ₂, P ₃and P ₄be identic polynomial expression, be respectively after expansion:

P ₁＝a ₁+a ₂X+a ₃Y+a ₄H+a ₅XY+a ₆YH+a ₇XH+a ₈Y ²+a ₉X ²+a ₁₀H ²+a ₁₁XYH+a ₁₂Y ³+a ₁₃YX ²+a ₁₄YH ²+a ₁₅Y ²X+a ₁₆X ³+a ₁₇XH ²+a ₁₈Y ²H+a ₁₉X ²H+a ₂₀H ³

P ₂＝b ₁+b ₂X+b ₃Y+b ₄H+b ₅XY+b ₆YH+b ₇XH+b ₈Y ²+b ₉X ²+b ₁₀H ²+b ₁₁XYH+b ₁₂Y ³+b ₁₃YX ²+b ₁₄YH ²+b ₁₅Y ²X+b ₁₆X ³+b ₁₇XH ²+b ₁₈Y ²H+b ₁₉X ²H+b ₂₀H ³

P ₃＝c ₁+c ₂X+c ₃Y+c ₄H+c ₅XY+c ₆YH+c ₇XH+c ₈Y ²+c ₉X ²+c ₁₀H ²+c ₁₁XYH+c ₁₂Y ³+c ₁₃YX ²+c ₁₄YH ²+c ₁₅Y ²X+c ₁₆X ³+c ₁₇XH ²+c ₁₈Y ²H+c ₁₉X ²H+c ₂₀H ³

P ₄＝d ₁+d ₂X+d ₃Y+d ₄H+d ₅XY+d ₆YH+d ₇XH+d ₈Y ²+d ₉X ²+d ₁₀H ²+d ₁₁XYH+d ₁₂Y ³+d ₁₃YX ²+d ₁₄YH ²+d ₁₅Y ²X+d ₁₆X ³+d ₁₇XH ²+d ₁₈Y ²H+d ₁₉X ²H+d ₂₀H ³

In formula, a _j, b _j, c _j, d _j, j=1,2 ..., 20 is rational polynominal coefficient;

(4) all pixel coordinates of the one-level image (P, L) of an environment Satellite CCD and corresponding height value H are substituted into the rational function model determined in step (3), obtain the coordinate figure (X, Y) correcting rear image;

(5) adopt interpolate value method to process the coordinate figure (X, Y) of image after the correction obtained in step (4), obtain the gray-scale value of integer coordinate values pixel, thus obtain the environment Satellite CCD image after correcting.

N in described step (2) is 81.

Interpolating method in described step (5) is bilinear interpolation method.

The present invention's beneficial effect is compared with prior art:

(1) the present invention adopts reference mark to calculate HJ star ccd image rational function model, and with the rational function model obtained after calculating, ortho-rectification is carried out to HJ star ccd image, compared to traditional polynomial revise method, increase substantially the positioning precision that HJ star CCD corrects rear image.

(2) classic method first carries out geometry preliminary correction to one-level image to generate secondary image, and then adopt general polynomial method to correct secondary image again, many image procossing, image information can be lost more.The inventive method directly carries out single treatment to HJ star CCD one-level image, avoids classic method to carry out secondary treating to HJ star CCD image, loses more image informations.

(3) multinomial model is adopted to carry out timing, correction accuracy near ground control point is higher, but the correction accuracy in other regions is obviously on the low side namely produces oscillatory occurences, and this method relies on its distinctive continuity can allow being more evenly distributed of error of fitting, avoid the oscillatory occurences that error skewness causes.

(4) the inventive method has good confidentiality.This method employing mathematical function sets up the geometric relationship between object space three dimensional space coordinate image space two dimensional surface corresponding to it coordinate, does not need to obtain satellite orbit, sensor sensitive parameter, avoids the leakage of relevant information.

(5) the inventive method has versatility.This method does not rely on concrete satellite sensor model, so this method is not only applicable to process HJ star CCD image, is also applicable to the optical satellite sensor image that other do not provide RPC file.

Accompanying drawing explanation

Fig. 1 is process flow diagram of the present invention.

Fig. 2 is bilinear interpolation method schematic diagram of the present invention.

Embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is further described in detail.

Be illustrated in figure 1 process flow diagram of the present invention, as can be seen from Figure 1, a kind of environment provided by the invention Satellite CCD image ortho-rectification method, is characterized in that step is as follows:

(1) the one-level image of an environment Satellite CCD is imported; To import mesh spacing be the SRTM altitude figures of M1 rice and resolution is the panchromatic reference data of LandSat of M2 rice; The scope of described M1 is [30,150], and the scope of M2 is [7.5,30];

(2) evenly gather N number of ground control point in the one-level image imported in step (1), obtain this reference mark at volume coordinate (P, L, H) and the reference mark pixel coordinate (X, Y) in one-level image with reference to image; Described N is more than or equal to 19;

(3) ground control point step (2) gathered is at the volume coordinate (P with reference to image, L, and the pixel coordinate (X of reference mark in one-level image H), Y) rational function model is imported, set up 4N equation, calculate rational polynominal coefficient, i.e. RPC, determine rational function model, specific formula for calculation is as follows:

$\left\{\begin{array}{c}P=\frac{P_1(X,Y,H)}{P_2(X,Y,H)}\\ L=\frac{P_3(X,Y,H)}{P_4(X,Y,H)}\end{array}\right.$

Wherein, P, L and H reference mark of being respectively collection is with reference to the horizontal ordinate of image, ordinate and height value; The pixel horizontal ordinate of the reference mark that X and Y is respectively collection in one-level image and pixel ordinate; P ₁, P ₂, P ₃and P ₄be identic polynomial expression, be respectively after expansion:

P ₁＝a ₁+a ₂X+a ₃Y+a ₄H+a ₅XY+a ₆YH+a ₇XH+a ₈Y ²+a ₉X ²+a ₁₀H ²+a ₁₁XYH+a ₁₂Y ³+a ₁₃YX ²+a ₁₄YH ²+a ₁₅Y ²X+a ₁₆X ³+a ₁₇XH ²+a ₁₈Y ²H+a ₁₉X ²H+a ₂₀H ³

P ₂＝b ₁+b ₂X+b ₃Y+b ₄H+b ₅XY+b ₆YH+b ₇XH+b ₈Y ²+b ₉X ²+b ₁₀H ²+b ₁₁XYH+b ₁₂Y ³+b ₁₃YX ²+b ₁₄YH ²+b ₁₅Y ²X+b ₁₆X ³+b ₁₇XH ²+b ₁₈Y ²H+b ₁₉X ²H+b ₂₀H ³

P ₃＝c ₁+c ₂X+c ₃Y+c ₄H+c ₅XY+c ₆YH+c ₇XH+c ₈Y ²+c ₉X ²+c ₁₀H ²+c ₁₁XYH+c ₁₂Y ³+c ₁₃YX ²+c ₁₄YH ²+c ₁₅Y ²X+c ₁₆X ³+c ₁₇XH ²+c ₁₈Y ²H+c ₁₉X ²H+c ₂₀H ³

P ₄＝d ₁+d ₂X+d ₃Y+d ₄H+d ₅XY+d ₆YH+d ₇XH+d ₈Y ²+d ₉X ²+d ₁₀H ²+d ₁₁XYH+d ₁₂Y ³+d ₁₃YX ²+d ₁₄YH ²+d ₁₅Y ²X+d ₁₆X ³+d ₁₇XH ²+d ₁₈Y ²H+d ₁₉X ²H+d ₂₀H ³

In formula, a _j, b _j, c _j, d _j, j=1,2 ..., 20 is rational polynominal coefficient;

(4) all pixel coordinates of the one-level image (P, L) of an environment Satellite CCD and corresponding height value H are substituted into the rational function model determined in step (3), obtain the coordinate figure (X, Y) correcting rear image;

(5) adopt interpolate value method to process the coordinate figure (X, Y) of image after the correction obtained in step (4), obtain the gray-scale value of integer coordinate values pixel, thus obtain the environment Satellite CCD image after correcting.

Embodiment

1, the HJ star CCD one-level image that file is called HJ1A-CCD1-8-84-20130414 is imported.

2, altitude figures and the reference data of an environment satellite is imported, the quality of altitude figures and reference data directly affects the precision of just penetrating result, selection principle is five times that altitude figures mesh spacing should not be greater than HJ star CCD image resolution, be not less than HJ star CCD image resolution, reference data resolution relative HJ star CCD image resolution, should not be greater than the latter, should not be less than four times of the latter.The altitude figures that this example adopts is 90 meters of mesh spacing SRTM (Space Shuttle Radar topographic mapping task) that NASA (NASA) issues, and to be that 15 meters of resolution LandSat (Landsat) that NASA (NASA) issues are panchromatic just penetrate data to reference data.

3, on HJ1A-CCD1-8-84-20130414 image and ETM image, evenly 81 ground control points are gathered.

4,81 the reference mark coordinates gathered are imported three rational function models,

Specific formula for calculation is as follows:

$\left\{\begin{array}{c}P=\frac{P_1(X,Y,H)}{P_2(X,Y,H)}\\ L=\frac{P_3(X,Y,H)}{P_4(X,Y,H)}\end{array}\right.$

Wherein, P, L, H are respectively the reference mark of collection in horizontal ordinate, ordinate, the height value with reference to image; X, Y are respectively the pixel horizontal ordinate of reference mark at one-level image of collection, pixel ordinate; P ₁, P ₂, P ₃, P ₄be all identic polynomial expression, after launching respectively be:

P ₁＝a ₁+a ₂X+a ₃Y+a ₄H+a ₅XY+a ₆YH+a ₇XH+a ₈Y ²+a ₉X ²+a ₁₀H ²+a ₁₁XYH+a ₁₂Y ³+a ₁₃YX ²+a ₁₄YH ²+a ₁₅Y ²X+a ₁₆X ³+a ₁₇XH ²+a ₁₈Y ²H+a ₁₉X ²H+a ₂₀H ³

P ₂＝b ₁+b ₂X+b ₃Y+b ₄H+b ₅XY+b ₆YH+b ₇XH+b ₈Y ²+b ₉X ²+b ₁₀H ²+b ₁₁XYH+b ₁₂Y ³+b ₁₃YX ²+b ₁₄YH ²+b ₁₅Y ²X+b ₁₆X ³+b ₁₇XH ²+b ₁₈Y ²H+b ₁₉X ²H+b ₂₀H ³

P ₃＝c ₁+c ₂X+c ₃Y+c ₄H+c ₅XY+c ₆YH+c ₇XH+c ₈Y ²+c ₉X ²+c ₁₀H ²+c ₁₁XYH+c ₁₂Y ³+c ₁₃YX ²+c ₁₄YH ²+c ₁₅Y ²X+c ₁₆X ³+c ₁₇XH ²+c ₁₈Y ²H+c ₁₉X ²H+c ₂₀H ³

P ₄＝d ₁+d ₂X+d ₃Y+d ₄H+d ₅XY+d ₆YH+d ₇XH+d ₈Y ²+d ₉X ²+d ₁₀H ²+d ₁₁XYH+d ₁₂Y ³+d ₁₃YX ²+d ₁₄YH ²+d ₁₅Y ²X+d ₁₆X ³+d ₁₇XH ²+d ₁₈Y ²H+d ₁₉X ²H+d ₂₀H ³

In formula, a _j, b _j, c _j, d _j(j=1,2, _l, 20) and be rational polynominal coefficient (RPC); Calculate 80 rational polynominal coefficients, concrete coefficient is as shown in table 1, namely obtains the rational function model of HJ1A-CCD1-8-84-20130414 image.

Table 1

	a	b	c	d
					1	-5.00E-02	1.00E+00	1.02E-01	1.00E+00
2	-2.22E-01	6.91E-04	1.13E+00	9.53E-03
					3	-1.09E+00	1.22E-03	-2.24E-01	1.69E-03
4	-1.60E-04	4.94E-04	1.75E-03	-2.68E-04
					5	2.99E-03	-4.34E-04	2.85E-02	-2.65E-02
6	-1.21E-04	-7.98E-06	2.85E-03	6.97E-03
					7	-3.49E-04	-7.28E-04	-1.40E-03	-1.12E-02
8	-1.27E-02	1.74E-02	-1.20E-01	2.46E-04
					9	3.89E-04	-3.06E-04	-4.83E-03	6.60E-03
10	-2.20E-04	4.28E-04	6.84E-05	-1.01E-02
					11	2.33E-04	2.17E-04	-1.50E-02	4.69E-04
12	-3.82E-03	-2.30E-04	-3.65E-02	-1.16E-02
					13	9.34E-04	-1.46E-03	7.41E-03	-2.03E-04

[0058]

14	-7.08E-05	2.29E-04	-1.23E-02	-1.56E-03
					15	-1.91E-02	-6.11E-05	-5.32E-03	3.40E-03
16	-2.79E-04	-1.90E-03	-8.02E-04	3.81E-05
					17	-8.81E-05	-9.87E-05	3.22E-03	2.26E-03
18	2.59E-04	1.84E-04	6.20E-03	1.76E-03
					19	1.17E-03	-4.86E-04	2.79E-03	1.36E-03
20	-3.08E-04	5.64E-04	9.88E-04	8.17E-04

5, all pixel coordinates of the one-level image (P, L) of an environment Satellite CCD and corresponding height value H are substituted into the coordinate figure (X, Y) that the rational function model determined in step (4) can obtain correcting rear image.

6, after correcting in theory, image coordinate value is (X, Y) gray-scale value put is exactly the one-level image pixel coordinate (P substituting into rational function model, L) gray-scale value, in fact coordinate figure (X, Y) may not be integer, not necessarily lucky on location of pixels, interpolating method must be utilized could to determine the gray-scale value of target location according to the gray-scale value of surrounding pixel, conventional interpolating method has the most contiguous interpolation method, bilinear interpolation method, cubic convolution interpolation method, comprehensive interpolation precision and operation efficiency, recommend bilinear interpolation method, the gray-scale value of each pixel of image after correcting is obtained after interpolation, the formula of bilinear interpolation method is as follows:

$G_p=[1-\mathrm{dxdy}]\left[\begin{array}{cc}{G}_{00}& G_{01}\\ G_{10}& G_{11}\end{array}\right]\left[\begin{array}{c}1-\mathrm{dx}\\ \mathrm{dx}\end{array}\right]$

In formula, G _pfor p point gray-scale value after interpolation, G ₀₀, G ₀₁, G ₁₀, G ₁₁be respectively the gray-scale value of square four corner pixels, dx, dy are that p point departs from the distance of [0,0] point in x, y direction, and Fig. 2 is shown in by bilinear interpolation method schematic diagram.

Above step (1)-(6) are that a complete employing the inventive method is to HJ star CCD ortho-rectification process.In order to compare the quality of the inventive method and Polynomial Method two kinds of bearing calibrations, traditional Polynomial Method is adopted to correct to HJ1A-CCD1-8-84-20130414 level image below, then the image after two kinds of bearing calibration process gathers checkpoint and carry out error analysis, obtain mean square error of coordinate value and the positioning precision of two kinds of bearing calibrations, through calculating, the precision 139.48 meters of HJ1A-CCD1-8-84-20130414 image polynomial revise method, the inventive method is 46.74 meters.Can find out, for process HJ1A-CCD1-8-84-20130414 image, the precision of the inventive method is better than polynomial revise method.In order to ensure the objectivity of test result, we expand experiment sample, two kinds of bearing calibration accuracy tests are carried out to other 30 scape HJ star ccd images, test result is in table 2, subordinate list result shows, HJ star ccd image adopts the mean accuracy of the inventive method to be 44.1 meters and is far superior to traditional polynomial revise method 108 meters of mean accuracies.

Table 2

The content be not described in detail in instructions of the present invention belongs to the known technology of professional and technical personnel in the field.

Claims (3)

1. environment Satellite CCD image ortho-rectification method, is characterized in that step is as follows:

(1) the one-level image of an environment Satellite CCD is imported; To import mesh spacing be the SRTM altitude figures of M1 rice and resolution is the panchromatic reference data of LandSat of M2 rice; The scope of described M1 is [30,150], and the scope of M2 is [7.5,30];

(2) evenly gather N number of ground control point in the one-level image imported in step (1), obtain this reference mark at volume coordinate (P, L, H) and the reference mark pixel coordinate (X, Y) in one-level image with reference to image; Described N is more than or equal to 19;

(3) ground control point step (2) gathered is at the volume coordinate (P with reference to image, L, and the pixel coordinate (X of reference mark in one-level image H), Y) rational function model is imported, set up 4N equation, calculate rational polynominal coefficient, i.e. RPC, determine rational function model, specific formula for calculation is as follows:

$\left\{\begin{array}{c}P=\frac{P_1(X,Y,H)}{P_2(X,Y,H)}\\ L=\frac{R_3(X,Y,H)}{R_4(X,Y,H)}\end{array}\right.$

Wherein, P, L and H reference mark of being respectively collection is with reference to the horizontal ordinate of image, ordinate and height value; The pixel horizontal ordinate of the reference mark that X and Y is respectively collection in one-level image and pixel ordinate; P ₁, P ₂, P ₃and P ₄be identic polynomial expression, be respectively after expansion:

P ₁＝a ₁+a ₂X+a ₃Y+a ₄H+a ₅XY+a ₆YH+a ₇XH+a ₈Y ²+a ₉X ²+a ₁₀H ²+a ₁₁XYH+

a ₁₂Y ³+a ₁₃YX ²+a ₁₄YH ²+a ₁₅Y ²X+a ₁₆X ³+a ₁₇XH ²+a ₁₈Y ²H+a ₁₉X ²H+a ₂₀H ³

P ₂＝b ₁+b ₂X+b ₃Y+b ₄H+b ₅XY+b ₆YH+b ₇XH+b ₈Y ²+b ₉X ²+b ₁₀H ²+b ₁₁XYH+

b ₁₂Y ³+b ₁₃YX ²+b ₁₄YH ²+b ₁₅Y ²X+b ₁₆X ³+b ₁₇XH ²+b ₁₈Y ²H+b ₁₉X ²H+b ₂₀H ³

P ₃＝c ₁+c ₂X+c ₃Y+c ₄H+c ₅XY+c ₆YH+c ₇XH+c ₈Y ²+c ₉X ²+c ₁₀H ²+c ₁₁XYH+

c ₁₂Y ³+c ₁₃YX ²+c ₁₄YH ²+c ₁₅Y ²X+c ₁₆X ³+c ₁₇XH ²+c ₁₈Y ²H+c ₁₉X ²H+c ₂₀H ³

P ₄＝d ₁+d ₂X+d ₃Y+d ₄H+d ₅XY+d ₆YH+d ₇XH+d ₈Y ²+d ₉X ²+d ₁₀H ²+d ₁₁XYH+

d ₁₂Y ³+d ₁₃YX ²+d ₁₄YH ²+d ₁₅Y ²X+d ₁₆X ³+d ₁₇XH ²+d ₁₈Y ²H+d ₁₉X ²H+d ₂₀H ³

In formula, a _j, b _j, c _j, d _j, j=1,2 ..., 20 is rational polynominal coefficient;

(4) all pixel coordinates of the one-level image (P, L) of an environment Satellite CCD and corresponding height value H are substituted into the rational function model determined in step (3), obtain the coordinate figure (X, Y) correcting rear image;

(5) adopt interpolate value method to process the coordinate figure (X, Y) of image after the correction obtained in step (4), obtain the gray-scale value of integer coordinate values pixel, thus obtain the environment Satellite CCD image after correcting.

2. a kind of environment according to claim 1 Satellite CCD image ortho-rectification method, is characterized in that: the N in described step (2) is 81.

3. a kind of environment according to claim 1 Satellite CCD image ortho-rectification method, is characterized in that: the interpolating method in described step (5) is bilinear interpolation method.