CN114562982B - A weighting method and device for joint adjustment of optical and SAR heterosource satellite images - Google Patents

Abstract

The invention discloses a weight determining method and device for optical and SAR heterologous satellite image joint adjustment, wherein the method comprises the following steps: constructing a SAR image and optical image heterogeneous regional network adjustment model; solving a weight matrix of an error equation set of the image side compensation parameters according to the priori positioning accuracy of the SAR image and the optical image; updating the weight of the compensation parameter of the image side; the device comprises: the system comprises a SAR image and optical image heterogeneous area network adjustment model construction module, a weight matrix solving module of an error equation set of image side compensation parameters and a weight updating module of the image side compensation parameters. According to the method, the contribution degree of the optical and SAR image sources with different geometric qualities to the positioning calculation is controlled, so that a more stable adjustment model is constructed, and the positioning precision of the combined adjustment of the heterogeneous images is improved.

Description

A weight determination method and device for joint adjustment of optical and SAR heterogeneous satellite images

Technical Field

The invention relates to the technical field of SAR image processing, and in particular to a weight determination method and device for joint adjustment of optical and SAR heterogeneous satellite images.

Background Art

At present, optical satellite images and SAR satellite images are the two major image sources for satellite stereo mapping. Joint processing of homologous data is still the main means in the field of remote sensing stereo imaging. Using optical satellite images and SAR satellite images for stereo positioning has its own advantages and disadvantages: optical satellite images have a high signal-to-noise ratio and intuitive interpretation, but when the intersection angle of multiple optical satellite images involved in stereo positioning is small, the positioning results of direct regional network adjustment cannot meet the requirements of stereo positioning accuracy. SAR satellite images have the unique advantage of all-day and all-weather, and their unique side-view characteristics are highly sensitive. With the substantial improvement of the resolution of spaceborne SAR, they can well supplement optical remote sensing and achieve the purpose of auxiliary positioning. When using SAR satellite images for positioning, since its positioning process does not require attitude information, it has a high uncontrolled geometric positioning accuracy after system calibration. Combining SAR images with optical images for stereo positioning can not only give play to the control role of SAR images with high geometric accuracy in the entire regional network, but also because the distance imaging method of SAR images is completely different from the central projection of optical images, the combination of the two can achieve complementary advantages and effectively improve the stereo observation structure.

In the process of joint adjustment of optical and SAR heterogeneous satellite images, due to the different geometric qualities of the optical images and SAR images involved in the adjustment, their contributions to the adjustment accuracy are very different. In addition, the imaging geometric models of optical and SAR images are different, and the corresponding single-view geometric positioning errors have different effects on the adjustment. However, the existing adjustment methods do not take the above different factors into consideration, and all of them are adjusted based on the assumption of indifferent image input, resulting in low accuracy of the existing joint adjustment methods for optical and SAR heterogeneous satellite images.

Summary of the invention

The technical problem to be solved by the present invention is that the existing joint adjustment method of optical and SAR heterogeneous satellite images does not take into account the geometric quality, imaging geometric model and single-view geometric positioning error of the optical images and SAR images involved in the adjustment, resulting in low accuracy of the joint adjustment.

In order to solve the above technical problems, considering that the autonomous positioning accuracy of each type of satellite is relatively stable over a long period of time, its prior value can largely reflect the positioning accuracy of the images involved in the adjustment. In order to improve the joint positioning accuracy, the present invention considers the influence of the positioning accuracy of optical and SAR images in the adjustment model, constructs a robust adjustment model, and proposes a method for setting the weights of observation values in the joint adjustment model based on the prior positioning accuracy of optical and SAR images. By controlling the contribution of optical and SAR image sources of different geometric qualities to the positioning solution, a more robust adjustment model is constructed to improve the positioning accuracy of the joint adjustment of heterogeneous images.

The first aspect of the embodiment of the present invention discloses a weight determination method for joint adjustment of optical and SAR heterogeneous satellite images, the method comprising:

S1, constructing a heterogeneous regional block adjustment model for SAR images and optical images, which includes:

S11, solve the RPC (Rational Polynomial Coefficients) model parameters of SAR images (Synthetic Aperture Radar) and optical images. Generate a set of virtual control point pairs of corresponding image coordinates and corresponding ground coordinates based on the rigorous imaging geometry models of SAR images and optical images respectively. Use the point pair set and the fitting method to solve the coefficients of the RPC model function to obtain the RPC model functions s(B,L,H) and l(B,L,H) of the corresponding images. (s,l) are the two-dimensional image coordinates of the image point in the satellite image, and (B,L,H) are the ground coordinates of the ground point corresponding to the image point in the image in the longitude and latitude coordinate system. The RPC model function is used to convert the ground coordinates in the longitude and latitude coordinate system into two-dimensional image coordinates.

S12, is the measured two-dimensional image coordinate of the i-th image point on the j-th image. The ground coordinates of the image point corresponding to the ground point are (B _k , L _k , H _k ), k is the serial number of the ground point. The image compensation function model of optical image and SAR satellite image is constructed using the affine transformation model, and the measurement error equation of the i-th image point is constructed as follows:

Where s ^(j) ( _Bk , _Lk , _Hk ) and l ^(j) ( _Bk , _Lk , _Hk ) are the results of substituting the ground point coordinates ( _Bk , _Lk , _Hk ) into the RPC model function of the jth image. is the measured coordinate error of the i-th image point on the j-th image, is the image compensation parameter of the jth image.

The error equation of the measurement error of the i-th image point is derived and linearized to obtain the error equation:

In the formula, is the partial derivative of the RPC model function of the jth image with respect to the ground point coordinates (B _k , L _k , H _k ), is the correction value for the unknown number, is the initial value of the ground coordinates of the kth ground point and the initial value of the image compensation parameter of the jth image The initial value of the two-dimensional image coordinates of the i-th image point on the j-th image is calculated. The calculation process is expressed as:

in, is the initial value of the image compensation parameter of the jth image.

S13, constructing a set of image point coordinate measurement error equations for the optical image and the SAR image in the overlapping area.

For n image points in the image, the corresponding image point coordinate measurement error equation group is constructed, and its expression is:

Among them, _AG and _AA are the coefficient matrices of the error equation group, _ΔxG is the ground point coordinate correction, _ΔxA is the image compensation parameter correction, for the image point coordinate measurement error equation group, the corresponding weight matrix is P _I , L _I represents the image point coordinate measurement value, Represents the initial value of the image point coordinate measurement value.

S14, the image-space compensation parameters in the measurement error equation are regarded as virtual observation values of the image-space affine transformation parameters, and the error equation group of the image-space compensation parameters is constructed, and its expression is:

Among them, I is the unit matrix, _VA is the observation error of the virtual observation value of the image-side affine transformation parameter, and its expression is:

_LA is the virtual observation value of the image-side affine transformation parameter, and its expression is:

is the initial value of the virtual observation value, and its value is constantly revised during iteration. Its expression is:

The weight matrix of the error equation group of the image compensation parameters is _PA ;

S15, based on the image point coordinate measurement error equation group and the image square compensation parameter error equation group, a weighted regional block adjustment equation group of heterogeneous images without control points is constructed, and its expression is:

The weight matrix of the weighted regional network adjustment equations of the heterogeneous image _PI and _PA are the weight matrices of the image point coordinate measurement error equation group and the image square compensation parameter error equation group, respectively. The _ΔxG and _ΔxA in the weighted regional network adjustment equation group of heterogeneous images are solved by an iterative method.

S2, the weight matrix of the error equation group of image compensation parameters is solved according to the prior positioning accuracy of SAR images and optical images.

The image point coordinate measurement weight of the image is the inverse of the mean error of the corresponding image point coordinate measurement value. The elements on the diagonal of the weight matrix P _I are the weights of the corresponding image point coordinate measurement values, and the other elements of the weight matrix P _I are 0. For the weight matrix _PA , the mean error of each image compensation parameter is determined according to the prior accuracy information of the image, the weight of each image compensation parameter in the weight matrix _PA is the inverse of the error, the elements on the diagonal of the weight matrix _PA are the weights of the corresponding image compensation parameters, and the other elements of the weight matrix _PA are 0.

As an optional implementation, in the first aspect of the embodiment of the present invention, the method of determining the mean square error of each image compensation parameter according to the priori accuracy information of the image includes:

The constant terms _a0 and _b0 of the image compensation parameters represent the translation information of the satellite in the row and column directions, respectively, which are determined by the a priori absolute positioning accuracy of the image row and column directions. For optical images, the mean error of the constant term of the image compensation parameters is calculated based on the a priori positioning accuracy of the image row and column directions and the known resolution of the image row and column directions;

For SAR images, the first constant term of the image compensation parameter represents the translation information of the satellite along the row direction. The mean error of the first constant term of the image compensation parameter is calculated according to the prior positioning accuracy in the image row direction and the known resolution in the image row direction. The second constant term of the image compensation parameter reflects the measurement error of the range slant range R. Therefore, the mean error of the second constant term of the image compensation parameter is calculated by multiplying the prior positioning accuracy σ _Y in the image column direction by the sine value of the image incident angle α _inc .

The linear term of the image compensation parameter represents the scaling and rotation errors of the image. The prior relative positioning accuracy of the image is determined by the nominal error value of the satellite platform. The mean error of the linear term of the image compensation parameter is determined based on the maximum influence of this accuracy on the image and the image size.

As an optional implementation, in the first aspect of the embodiments of the present invention, after solving the weight matrix of the error equation group of the image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image, the pixel value vector of the row where the image point related to the ground point target is located is used to calculate its cross-correlation matrix, and according to the number of weights of the image compensation parameters, the feature matrix of the corresponding dimension is extracted, and the feature matrix is used as the weighting matrix to perform weighted update on the weights of the image compensation parameters in turn to obtain the updated weights of the image compensation parameters.

As an optional implementation, in the first aspect of the embodiment of the present invention, for the pixel value vectors a _i of the corresponding rows of the image points related to the ground point target in the image, i=1, 2, ..., N, N is the number of the image points related to the ground point target, all pixel value vectors are represented as the acquisition data matrix A:

A＝[a ₁ ,a ₂ ,…,a _N ]，

Calculate the cross-correlation matrix R of the collected data matrix A, and you will get:

R＝ ^ATA ，

Among them, the element r _ij = a _i a _j ^T in the i-th row and j-th column of the cross-correlation matrix R, and the column vector a _j represents the pixel value vector of the j-th image point related to the ground point target in the corresponding row in the image. The principal component analysis method is used to reduce the dimension of the collected data matrix and extract its feature matrix. The specific process is:

Perform eigendecomposition on the cross-correlation matrix R to obtain N eigenvectors and eigenvalues. Select the eigenvectors according to the eigenvalues, and select the eigenmatrix E composed of M eigenvectors greater than a certain threshold, which is recorded as:

E＝[v ₁ ,v ₂ ,…,v _M ]，

Wherein, _vk represents the kth eigenvector, k = 1, 2, ..., M, the eigenvectors are all column vectors, and M is the number of weights of the image compensation parameters; the eigenmatrix E is used as the weighting matrix, and the vector a composed of the weights of the image compensation parameters is weighted updated, that is:

b＝ ^ETa ，

The updated weight vector b of the image-side compensation parameters is obtained.

The second aspect of the embodiment of the present invention discloses a weight determination device for joint adjustment of optical and SAR heterogeneous satellite images, the device comprising: a SAR image and optical image heterogeneous regional network adjustment model construction module, a weight matrix solving module for the error equation group of image compensation parameters, and a weight value updating module for the image compensation parameters. The weight matrix solving module for the error equation group of image compensation parameters is used to solve the weight matrix of the error equation group of image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image.

As an optional implementation, in the second aspect of the embodiment of the present invention, the SAR image and optical image heterogeneous regional block adjustment model construction module is used to construct the SAR image and optical image heterogeneous regional block adjustment model, and its implementation process includes:

Solve the RPC (Rational Polynomial Coefficients) model parameters of SAR images (Synthetic Aperture Radar) and optical images. Generate the corresponding image coordinates and the corresponding ground coordinates of the virtual control point pair set based on the rigorous imaging geometry model of SAR images and optical images respectively. Use the point pair set and the fitting method to solve the coefficients of the RPC model function to obtain the RPC model functions s(B,L,H) and l(B,L,H) of the corresponding image. (s,l) is the two-dimensional image coordinates of the image point in the satellite image, and (B,L,H) is the ground coordinates of the ground point corresponding to the image point in the image in the longitude and latitude coordinate system. The RPC model function is used to convert the ground coordinates in the longitude and latitude coordinate system into two-dimensional image coordinates.

is the measured two-dimensional image coordinate of the i-th image point on the j-th image. The ground coordinates of the image point corresponding to the ground point are (B _k , L _k , H _k ), k is the serial number of the ground point. The image compensation function model of optical image and SAR satellite image is constructed using the affine transformation model, and the measurement error equation of the i-th image point is constructed as follows:

Where s ^(j) ( _Bk , _Lk , _Hk ) and l ^(j) ( _Bk , _Lk , _Hk ) are the results of substituting the ground point coordinates ( _Bk , _Lk , _Hk ) into the RPC model function of the jth image. is the measured coordinate error of the i-th image point on the j-th image, is the image compensation parameter of the jth image.

The error equation of the measurement error of the i-th image point is derived and linearized to obtain the error equation:

In the formula, is the partial derivative of the RPC model function of the jth image with respect to the ground point coordinates (B _k , L _k , H _k ), is the correction value for the unknown number, is the initial value of the ground coordinates of the kth ground point and the initial value of the image compensation parameter of the jth image The initial value of the two-dimensional image coordinates of the i-th image point on the j-th image is calculated. The calculation process is expressed as:

in, is the initial value of the image compensation parameter of the jth image.

The error equations for measuring the coordinates of image points in overlapping areas of optical and SAR images are constructed.

For n image points in the image, the corresponding image point coordinate measurement error equation group is constructed, and its expression is:

Among them, _AG and _AA are the coefficient matrices of the error equation group, _ΔxG is the ground point coordinate correction, _ΔxA is the image compensation parameter correction, for the image point coordinate measurement error equation group, the corresponding weight matrix is P _I , L _I represents the image point coordinate measurement value, Represents the initial value of the image point coordinate measurement value.

The image-space compensation parameters in the measurement error equation are regarded as virtual observation values of the image-space affine transformation parameters, and the error equation group of the image-space compensation parameters is constructed, and its expression is:

Among them, I is the unit matrix, _VA is the observation error of the virtual observation value of the image-side affine transformation parameter, and its expression is:

_LA is the virtual observation value of the image-side affine transformation parameter, and its expression is:

is the initial value of the virtual observation value, and its value is constantly revised during iteration. Its expression is:

The weight matrix of the error equation group of the image compensation parameters is _PA ;

According to the error equations of image point coordinate measurement and image square compensation parameters, the weighted regional block adjustment equations of heterogeneous images without control points are constructed, and the expression is:

The weight matrix of the weighted regional network adjustment equations of the heterogeneous image _PI and _PA are the weight matrices of the image point coordinate measurement error equation group and the image square compensation parameter error equation group, respectively. The _ΔxG and _ΔxA in the weighted regional network adjustment equation group of heterogeneous images are solved by an iterative method.

As an optional implementation manner, in the second aspect of the embodiment of the present invention, the weight matrix solving module of the error equation group of the image compensation parameters is used to solve the weight matrix of the error equation group of the image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image, and specifically includes:

The image point coordinate measurement weight of the image is the inverse of the mean error of the corresponding image point coordinate measurement value. The elements on the diagonal of the weight matrix P _I are the weights of the corresponding image point coordinate measurement values, and the other elements of the weight matrix P _I are 0. For the weight matrix _PA , the mean error of each image compensation parameter is determined according to the prior accuracy information of the image, the weight of each image compensation parameter in the weight matrix _PA is the inverse of the error, the elements on the diagonal of the weight matrix _PA are the weights of the corresponding image compensation parameters, and the other elements of the weight matrix _PA are 0.

As an optional implementation, in the second aspect of the embodiment of the present invention, the method of determining the mean square error of each image compensation parameter according to the priori accuracy information of the image includes:

The constant terms _a0 and _b0 of the image compensation parameters represent the translation information of the satellite in the row and column directions, respectively, which are determined by the a priori absolute positioning accuracy of the image row and column directions. For optical images, the mean error of the constant term of the image compensation parameters is calculated based on the a priori positioning accuracy of the image row and column directions and the known resolution of the image row and column directions;

For SAR images, the first constant term of the image compensation parameter represents the translation information of the satellite along the row direction. The mean error of the first constant term of the image compensation parameter is calculated according to the prior positioning accuracy in the image row direction and the known resolution in the image row direction. The second constant term of the image compensation parameter reflects the measurement error of the range slant range R. Therefore, the mean error of the second constant term of the image compensation parameter is calculated by multiplying the prior positioning accuracy σ _Y in the image column direction by the sine value of the image incident angle α _inc .

The linear term of the image compensation parameter represents the scaling and rotation errors of the image. The prior relative positioning accuracy of the image is determined by the nominal error value of the satellite platform. The mean error of the linear term of the image compensation parameter is determined based on the maximum influence of this accuracy on the image and the image size.

As an optional implementation, in the second aspect of the embodiment of the present invention, the weight update module of the image compensation parameters, after solving the weight matrix of the error equation group of the image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image, uses the pixel value vector of the row where the image point related to the ground point target is located to calculate its cross-correlation matrix, extracts the feature matrix of the corresponding dimension according to the number of weights of the image compensation parameters, uses the feature matrix as the weighting matrix, and performs weighted update on the weights of the image compensation parameters in turn to obtain the updated weights of the image compensation parameters.

As an optional implementation, in the second aspect of the embodiment of the present invention, the weight update module of the image space compensation parameter, for the pixel value vectors a _i , i=1, 2, ..., N, where N is the number of the image points related to the ground point target in the corresponding row in the image, represents all pixel value vectors as the acquisition data matrix A:

A＝[a ₁ ,a ₂ ,…,a _N ]，

Calculate the cross-correlation matrix R of the collected data matrix A, and you will get:

R＝ ^ATA ，

Among them, the element r _ij = a _i a _j ^T in the i-th row and j-th column of the cross-correlation matrix R, and the column vector a _j represents the pixel value vector of the j-th image point related to the ground point target in the corresponding row in the image. The principal component analysis method is used to reduce the dimension of the collected data matrix and extract its feature matrix. The specific process is:

Perform eigendecomposition on the cross-correlation matrix R to obtain N eigenvectors and eigenvalues. Select the eigenvectors according to the eigenvalues, and select the eigenmatrix E composed of M eigenvectors greater than a certain threshold, which is recorded as:

E＝[v ₁ ,v ₂ ,…,v _M ]，

Wherein, _vk represents the kth eigenvector, k = 1, 2, ..., M, the eigenvectors are all column vectors, and M is the number of weights of the image compensation parameters; the eigenmatrix E is used as the weighting matrix, and the vector a composed of the weights of the image compensation parameters is weighted updated, that is:

b＝ ^ETa ，

The updated weight vector b of the image-side compensation parameters is obtained.

The third aspect of the present invention discloses another weight determination device for joint adjustment of optical and SAR heterogeneous satellite images, the device comprising:

A memory storing executable program code;

a processor coupled to the memory;

The processor calls the executable program code stored in the memory to execute part or all of the steps in the weighting method for joint adjustment of optical and SAR heterogeneous satellite images disclosed in the first aspect of the embodiment of the present invention.

The fourth aspect of the present invention discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the weighting method for joint adjustment of optical and SAR heterogeneous satellite images disclosed in the first aspect of an embodiment of the present invention.

Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

The embodiment of the present invention provides support for the construction of an adjustment model based on optical and SAR heterogeneous multi-view images. The embodiment of the present invention controls the contribution of optical and SAR image sources of different geometric qualities to the positioning solution by setting the weights of optical and SAR image observation values in the adjustment model, thereby building a more robust adjustment model and improving the positioning accuracy of the joint adjustment of heterogeneous images.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

1 is a schematic flow diagram of a weighting method for joint adjustment of optical and SAR heterogeneous satellite images disclosed in an embodiment of the present invention;

FIG. 2 is a schematic diagram of the composition of a weight determination device for joint adjustment of optical and SAR heterogeneous satellite images disclosed in an embodiment of the present invention.

DETAILED DESCRIPTION

In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or equipment.

Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present invention. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments. Each is described in detail below.

FIG1 is a flow chart of a method for determining weights for joint adjustment of optical and SAR heterogeneous satellite images disclosed in an embodiment of the present invention; FIG2 is a composition chart of a device for determining weights for joint adjustment of optical and SAR heterogeneous satellite images disclosed in an embodiment of the present invention.

Embodiment 1

This embodiment discloses a weight determination method for joint adjustment of optical and SAR heterogeneous satellite images, the method comprising:

S1, constructing a heterogeneous regional block adjustment model for SAR images and optical images, which includes:

S11, solve the RPC (Rational Polynomial Coefficients) model parameters of SAR images (Synthetic Aperture Radar) and optical images. Generate a set of virtual control point pairs of corresponding image coordinates and corresponding ground coordinates based on the rigorous imaging geometry models of SAR images and optical images respectively. Use the point pair set and the fitting method to solve the coefficients of the RPC model function to obtain the RPC model functions s(B,L,H) and l(B,L,H) of the corresponding images. (s,l) are the two-dimensional image coordinates of the image point in the satellite image, and (B,L,H) are the ground coordinates of the ground point corresponding to the image point in the image in the longitude and latitude coordinate system. The RPC model function is used to convert the ground coordinates in the longitude and latitude coordinate system into two-dimensional image coordinates. The RPC model associates the target image point (s,l) and its ground coordinates (B,L,H) in the form of the ratio of the cubic polynomial of the ground coordinates.

S12, is the measured two-dimensional image coordinate of the i-th image point on the j-th image. The ground coordinates of the image point corresponding to the ground point are (B _k , L _k , H _k ), k is the serial number of the ground point. The image compensation function model of optical image and SAR satellite image is constructed using the affine transformation model, and the measurement error equation of the i-th image point is constructed as follows:

Where s ^(j) ( _Bk , _Lk , _Hk ) and l ^(j) ( _Bk , _Lk , _Hk ) are the results of substituting the ground point coordinates ( _Bk , _Lk , _Hk ) into the RPC model function of the jth image. is the measured coordinate error of the i-th image point on the j-th image. This error is minimized during the block adjustment process. is the image square compensation parameter of the jth image, and the ground point coordinates (B _k , L _k , H _k ) corresponding to the ith image point are obtained by using the subsequent regional block adjustment method.

Δs ^j (s, l) and Δl ^j (s, l) are the image-space compensation functions of the jth image, which are obtained using the affine transformation model and are expressed as follows:

The error equation of the measurement error of the i-th image point is derived and linearized to obtain the error equation:

In the formula, is the partial derivative of the RPC model function of the jth image with respect to the ground point coordinates (B _k , L _k , H _k ), is the correction value for the unknown number, is the initial value of the ground coordinates of the kth ground point and the initial value of the image compensation parameter of the jth image The initial value of the two-dimensional image coordinates of the i-th image point on the j-th image is calculated. The calculation process is expressed as:

in, is the initial value of the image compensation parameter of the jth image.

S13, constructing a set of error equations for the coordinate measurement of image points of the optical image and the SAR image in the overlapping area. The optical image and the SAR image are related to each other through the connection points measured on the image. The connection points are the matching ground object points with the same name in the overlapping area. For m images, all n image points on the image correspond to s ground points in total, and Δx _G is the correction amount of the coordinates of the s ground points, and its expression is:

Δx _G =[ΔB ₁ ,ΔL ₁ ,ΔH ₁ ,…,ΔB _s ,ΔL _s ,ΔH _s ] ^T ,

Δx _A is the image compensation parameter correction of the m images, and its expression is:

set up and

is the coefficient matrix of the error equation for the i-th pixel.

is the measurement error of the image point coordinates of the i-th image point,

Then the error equation of the i-th pixel is expressed as The above error equations are constructed for n image points respectively and integrated to obtain the image point coordinate measurement error equation group corresponding to the n image points.

For n image points in the image, the corresponding image point coordinate measurement error equation group is constructed, and its expression is:

Among them, _AG and _AA are the coefficient matrices of the error equation group, _ΔxG is the ground point coordinate correction, _ΔxA is the image compensation parameter correction, for the image point coordinate measurement error equation group, the corresponding weight matrix is P _I , L _I represents the image point coordinate measurement value, Represents the initial value of the image point coordinate measurement value.

S14, the image-space compensation parameters in the measurement error equation are regarded as virtual observation values of the image-space affine transformation parameters, and the error equation group of the image-space compensation parameters is constructed, and its expression is:

Among them, I is the unit matrix, _VA is the observation error of the virtual observation value of the image-side affine transformation parameter, and its expression is:

_LA is the virtual observation value of the image-side affine transformation parameter, and its expression is:

is the initial value of the virtual observation value, which is set to 0 at the beginning. Its value is constantly modified during iteration. Its expression is:

The weight matrix of the error equation group of the image compensation parameters is _PA ;

S15, based on the image point coordinate measurement error equation group and the image square compensation parameter error equation group, a weighted regional block adjustment equation group of heterogeneous images without control points is constructed, and its expression is:

The weight matrix of the weighted regional network adjustment equations of the heterogeneous image _PI and _PA are weight matrices of the image point coordinate measurement error equation group and the image square compensation parameter error equation group, respectively, and the _ΔxG and _ΔxA in the weighted regional network adjustment equation group of heterogeneous images are solved by an iterative method. The iterative method can adopt the least square method.

S2, the weight matrix of the error equation group of image compensation parameters is solved according to the prior positioning accuracy of SAR images and optical images.

The mean error of the image point coordinate measurement value is related to the accuracy of manual measurement or automatic matching, usually less than 1 pixel, and is basically determined by the optical/SAR image source and resolution. Therefore, for the same type of image source, the mean error of the image point measurement can be determined to be a constant. The image point coordinate measurement weight of the image is the reciprocal of the mean error of its corresponding image point coordinate measurement value. The elements on the diagonal of the weight matrix P _I are the weights of the corresponding image point coordinate measurement values, and the other elements of the weight matrix P _I are 0. For the weight matrix _PA , the mean error of each image compensation parameter is determined according to the prior accuracy information of the image. The weight of each image compensation parameter in the weight matrix _PA is the reciprocal of the error, the elements on the diagonal of the weight matrix _PA are the weights of the corresponding image compensation parameters, and the other elements of the weight matrix _PA are 0.

As an optional implementation manner, in this embodiment, the method of determining the mean square error of each image compensation parameter according to the priori accuracy information of the image includes:

The constant terms _a0 and _b0 of the image compensation parameters represent the translation information of the satellite in the row and column directions, respectively, which are determined by the a priori absolute positioning accuracy of the image row and column directions. For optical images, the mean error of _a0 and _b0 is calculated based on the a priori positioning accuracy _σX , _σY of the image row and column directions and the known resolutions _RX , RY of the image row and _column directions. The calculation process is:

in, represents the mean error of a ₀ , represents the mean error of b ₀ ;

For SAR images, the constant term _a0 of the image compensation parameter represents the information of the satellite translation along the row direction. The mean error of _a0 is calculated based on the prior positioning accuracy _σX in the image row direction and the known resolution _RX in the image row direction. The constant term _b0 of the image compensation parameter reflects the measurement error of the range slant range R. Therefore, the mean error is calculated by multiplying the prior positioning accuracy _σY in the image column direction by the sine value of the image incident angle _αinc . The calculation formula is:

in, represents the mean error of a ₀ , represents the mean error of _b0 ; the slant range refers to the distance from the ground point to the satellite.

The linear terms _a1 , _a2 , _b1 and _b2 of the image compensation parameters represent the scaling and rotation errors of the image. For optical satellite images, this error depends on focal length error, lens distortion, satellite radial position error and gyro drift. For SAR images, this error depends on pulse repetition frequency error, range sampling frequency error, satellite radial position error and satellite along-track velocity error. The a priori relative positioning accuracy of the image is determined by the nominal error value of the satellite platform. The mean square error of the linear terms of the image compensation parameters is determined based on the maximum impact of this accuracy on the image and the image size. Assuming that the maximum impact of this accuracy on the image is M pixels, the mean square errors of _a1 , _a2 , _b1 and _b2 are determined based on M and the image size, as follows:

Where Height and Width are the height and width of the image.

As an optional implementation, in this embodiment, in order to improve the positioning accuracy of the joint adjustment of heterogeneous images, the pixel value vectors of the rows of image points related to the ground point targets are used to calculate their cross-correlation matrix, and the characteristic matrix of the corresponding dimension is extracted according to the number of weights of the image compensation parameters. The characteristic matrix is used as a weighted matrix, and the weights of the image compensation parameters are weighted updated in turn to obtain the updated weights of the image compensation parameters.

As an optional implementation, in this embodiment, for the pixel value vectors a _i of the corresponding rows of the image points related to the ground point targets in the image, i=1, 2, ..., N, N is the number of the image points related to the ground point targets, all pixel value vectors are represented as the acquisition data matrix a:

a＝[a ₁ ,a ₂ ,…,a _N ]，

Calculate the cross-correlation matrix R of the collected data matrix a, and you will get:

R＝a ^T a，

Among them, the element r _ij = a _i a _j ^T in the i-th row and j-th column of the cross-correlation matrix R, and the column vector a _j represents the pixel value vector of the j-th image point related to the ground point target in the corresponding row in the image. The principal component analysis method is used to reduce the dimension of the collected data matrix and extract its feature matrix. The specific process is:

Perform eigendecomposition on the cross-correlation matrix R to obtain N eigenvectors and eigenvalues. Select the eigenvectors according to the eigenvalues, and select the eigenmatrix E composed of M eigenvectors greater than a certain threshold, which is recorded as:

E＝[v ₁ ,v ₂ ,…,v _M ]，

Wherein, _vk represents the kth eigenvector, k = 1, 2, ..., M, the eigenvectors are all column vectors, and M is the number of weights of the image compensation parameters; the eigenmatrix E is used as the weighting matrix, and the vector a composed of the weights of the image compensation parameters is weighted updated, that is:

b＝ ^ETa ，

The updated weight vector b of the image-side compensation parameters is obtained.

The experimental results of the joint adjustment and positioning of optical and SAR images using this method are given below. After the optical/SAR images participating in the prior information joint positioning experiment were geometrically calibrated on-orbit, the geometric information is shown in the following table, including one SAR image and one optical image of a certain area. The constant term weight of the virtual observation value of the image compensation parameter is determined according to its prior plane positioning accuracy, and the first-order term weight is determined according to the maximum pixel offset of the image, that is, the relative mean error. The SAR-262 image is combined with the CCD-513 image, and the weight of the virtual observation value of the SAR/optical image compensation parameter in the combination is adjusted to observe the influence of the image prior accuracy information on the joint positioning accuracy. The maximum pixel offset of the first-order term is fixed to 2 pixels, so that the prior plane positioning accuracy of the SAR/optical image changes from 0.5 meters to 200 meters respectively. The weight is determined according to this method and the regional network adjustment is performed. The mean error of plane and elevation positioning under different constant term weight configurations is counted. The optimal weight range set when the highest accuracy is obtained is basically consistent with the image prior positioning accuracy. Table 1 shows the positioning accuracy (meter) under different constant term accuracy configurations of SAR-262/CCD-513.

Table 1 Positioning accuracy of SAR-262/CCD-513 under different constant item accuracy configurations (meters)

The constant term weight is fixed to the prior plane positioning accuracy of CCD and SAR images, and the maximum pixel offset caused by the first-order term of the image compensation parameter is changed from 0.5 pixels to 200 pixels. The weight is determined according to this method and regional network adjustment is performed to count the plane/elevation errors under different first-order term weight configurations. It can be found that when the maximum offset caused by the first-order term is set to 0.5-2 pixels, the joint positioning obtains better and more stable results, with the best plane accuracy of 2 meters and the best elevation accuracy of 1.7 meters. This result also shows that the image used in this experiment has small distortion in the scene, and the image deformation is basically a translation. Table 2 shows the positioning accuracy (meters) under different first-order term accuracy configurations of SAR-262/CCD-513.

Table 2 Positioning accuracy of SAR-262/CCD-513 under different first-order accuracy configurations (meters)

It can be seen that the embodiment of the present invention provides support for the construction of an adjustment model based on optical and SAR heterogeneous multi-view images. The embodiment of the present invention controls the contribution of optical and SAR image sources of different geometric qualities to the positioning solution by setting the weights of optical and SAR image observation values in the adjustment model, thereby building a more robust adjustment model and improving the positioning accuracy of the joint adjustment of heterogeneous images.

Embodiment 2

The second aspect of the embodiment of the present invention discloses a weight determination device for joint adjustment of optical and SAR heterogeneous satellite images, the device comprising: a SAR image and optical image heterogeneous regional network adjustment model construction module, a weight matrix solving module for the error equation group of image compensation parameters, and a weight value updating module for the image compensation parameters. The weight matrix solving module for the error equation group of image compensation parameters is used to solve the weight matrix of the error equation group of image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image.

As an optional implementation, in this embodiment, the SAR image and optical image heterogeneous regional block adjustment model construction module is used to construct the SAR image and optical image heterogeneous regional block adjustment model, and its implementation process includes:

Solve the RPC (Rational Polynomial Coefficients) model parameters of SAR images (Synthetic Aperture Radar) and optical images. Generate the corresponding image coordinates and the corresponding ground coordinates of the virtual control point pair set based on the rigorous imaging geometry model of SAR images and optical images respectively. Use the point pair set and the fitting method to solve the coefficients of the RPC model function to obtain the RPC model functions s(B,L,H) and l(B,L,H) of the corresponding image. (s,l) is the two-dimensional image coordinates of the image point in the satellite image, and (B,L,H) is the ground coordinates of the ground point corresponding to the image point in the image in the longitude and latitude coordinate system. The RPC model function is used to convert the ground coordinates in the longitude and latitude coordinate system into two-dimensional image coordinates.

is the measured two-dimensional image coordinate of the i-th image point on the j-th image. The ground coordinates of the image point corresponding to the ground point are (B _k , L _k , H _k ), k is the serial number of the ground point. The image compensation function model of optical image and SAR satellite image is constructed using the affine transformation model, and the measurement error equation of the i-th image point is constructed as follows:

Where s ^(j) ( _Bk , _Lk , _Hk ) and l ^(j) ( _Bk , _Lk , _Hk ) are the results of substituting the ground point coordinates ( _Bk , _Lk , _Hk ) into the RPC model function of the jth image. is the measured coordinate error of the i-th image point on the j-th image, is the image compensation parameter of the jth image.

The error equation of the measurement error of the i-th image point is derived and linearized to obtain the error equation:

In the formula, is the partial derivative of the RPC model function of the jth image with respect to the ground point coordinates (B _k , L _k , H _k ), is the correction value for the unknown number, is the initial value of the ground coordinates of the kth ground point and the initial value of the image compensation parameter of the jth image The initial value of the two-dimensional image coordinates of the i-th image point on the j-th image is calculated. The calculation process is expressed as:

in, is the initial value of the image compensation parameter of the jth image.

The error equations for measuring the coordinates of image points in overlapping areas of optical and SAR images are constructed.

For n image points in the image, the corresponding image point coordinate measurement error equation group is constructed, and its expression is:

Among them, _AG and _AA are the coefficient matrices of the error equation group, _ΔxG is the ground point coordinate correction, _ΔxA is the image compensation parameter correction, for the image point coordinate measurement error equation group, the corresponding weight matrix is P _I , L _I represents the image point coordinate measurement value, Represents the initial value of the image point coordinate measurement value.

The image-space compensation parameters in the measurement error equation are regarded as virtual observation values of the image-space affine transformation parameters, and the error equation group of the image-space compensation parameters is constructed, and its expression is:

Among them, I is the unit matrix, _VA is the observation error of the virtual observation value of the image-side affine transformation parameter, and its expression is:

_LA is the virtual observation value of the image-side affine transformation parameter, and its expression is:

is the initial value of the virtual observation value, and its value is constantly revised during iteration. Its expression is:

The weight matrix of the error equation group of the image compensation parameters is _PA ;

According to the error equations of image point coordinate measurement and image square compensation parameters, the weighted regional block adjustment equations of heterogeneous images without control points are constructed, and the expression is:

The weight matrix of the weighted regional network adjustment equations of the heterogeneous image _PI and _PA are the weight matrices of the image point coordinate measurement error equation group and the image square compensation parameter error equation group, respectively. The _ΔxG and _ΔxA in the weighted regional network adjustment equation group of heterogeneous images are solved by an iterative method.

As an optional implementation manner, in this embodiment, the weight matrix solving module of the error equation group of the image compensation parameters is used to solve the weight matrix of the error equation group of the image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image, and specifically includes:

The image point coordinate measurement weight of the image is the inverse of the mean error of the corresponding image point coordinate measurement value. The elements on the diagonal of the weight matrix P _I are the weights of the corresponding image point coordinate measurement values, and the other elements of the weight matrix P _I are 0. For the weight matrix _PA , the mean error of each image compensation parameter is determined according to the prior accuracy information of the image, the weight of each image compensation parameter in the weight matrix _PA is the inverse of the error, the elements on the diagonal of the weight matrix _PA are the weights of the corresponding image compensation parameters, and the other elements of the weight matrix _PA are 0.

As an optional implementation manner, in this embodiment, the method of determining the mean square error of each image compensation parameter according to the priori accuracy information of the image includes:

The constant terms _a0 and _b0 of the image compensation parameters represent the translation information of the satellite in the row and column directions, respectively, which are determined by the a priori absolute positioning accuracy of the image row and column directions. For optical images, the mean error of the constant term of the image compensation parameters is calculated based on the a priori positioning accuracy of the image row and column directions and the known resolution of the image row and column directions;

For SAR images, the first constant term of the image compensation parameter represents the translation information of the satellite along the row direction. The mean error of the first constant term of the image compensation parameter is calculated according to the prior positioning accuracy in the image row direction and the known resolution in the image row direction. The second constant term of the image compensation parameter reflects the measurement error of the range slant range R. Therefore, the mean error of the second constant term of the image compensation parameter is calculated by multiplying the prior positioning accuracy σ _Y in the image column direction by the sine value of the image incident angle α _inc .

The linear term of the image compensation parameter represents the scaling and rotation errors of the image. The prior relative positioning accuracy of the image is determined by the nominal error value of the satellite platform. The mean error of the linear term of the image compensation parameter is determined based on the maximum influence of this accuracy on the image and the image size.

As an optional implementation, in this embodiment, the weight update module of the image compensation parameters, after solving the weight matrix of the error equation group of the image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image, uses the pixel value vector of the row where the image point related to the ground point target is located to calculate its cross-correlation matrix, extracts the feature matrix of the corresponding dimension according to the number of weights of the image compensation parameters, uses the feature matrix as the weighting matrix, and performs weighted update on the weights of the image compensation parameters in turn to obtain the updated weights of the image compensation parameters.

As an optional implementation, in this embodiment, the weight update module of the image space compensation parameter, for the pixel value vectors a _i of the corresponding rows of the image points related to the ground point target in the image, i=1, 2, ..., N, N is the number of the image points related to the ground point target, all pixel value vectors are expressed as the acquisition data matrix A:

A＝[a ₁ ,a ₂ ,…,a _N ]，

Calculate the cross-correlation matrix R of the collected data matrix A, and you will get:

R＝ ^ATA ，

Among them, the element r _ij = a _i a _j ^T in the i-th row and j-th column of the cross-correlation matrix R, and the column vector a _j represents the pixel value vector of the j-th image point related to the ground point target in the corresponding row in the image. The principal component analysis method is used to reduce the dimension of the collected data matrix and extract its feature matrix. The specific process is:

Perform eigendecomposition on the cross-correlation matrix R to obtain N eigenvectors and eigenvalues. Select the eigenvectors according to the eigenvalues, and select the eigenmatrix E composed of M eigenvectors greater than a certain threshold, which is recorded as:

E＝[v ₁ ,v ₂ ,…,v _M ]，

Wherein, _vk represents the kth eigenvector, k = 1, 2, ..., M, the eigenvectors are all column vectors, and M is the number of weights of the image compensation parameters; the eigenmatrix E is used as the weighting matrix, and the vector a composed of the weights of the image compensation parameters is weighted updated, that is:

b＝ ^ETa ，

The updated weight vector b of the image-side compensation parameters is obtained.

It can be seen that the embodiment of the present invention provides support for the construction of an adjustment model based on optical and SAR heterogeneous multi-view images. The embodiment of the present invention controls the contribution of optical and SAR image sources of different geometric qualities to the positioning solution by setting the weights of optical and SAR image observation values in the adjustment model, thereby building a more robust adjustment model and improving the positioning accuracy of the joint adjustment of heterogeneous images.

Embodiment 3

This embodiment discloses another weight determination device for joint adjustment of optical and SAR heterogeneous satellite images, the device comprising:

A memory storing executable program code;

a processor coupled to the memory;

The processor calls the executable program code stored in the memory to execute part or all of the steps in the weighting method for joint adjustment of optical and SAR heterogeneous satellite images disclosed in the first aspect of the embodiment of the present invention.

It can be seen that the embodiment of the present invention provides support for the construction of an adjustment model based on optical and SAR heterogeneous multi-view images. The embodiment of the present invention controls the contribution of optical and SAR image sources of different geometric qualities to the positioning solution by setting the weights of optical and SAR image observation values in the adjustment model, thereby building a more robust adjustment model and improving the positioning accuracy of the joint adjustment of heterogeneous images.

Embodiment 4

The present embodiment discloses a computer storage medium, which stores computer instructions. When the computer instructions are called, they are used to execute some or all of the steps in the weighting method for joint adjustment of optical and SAR heterogeneous satellite images disclosed in the first aspect of the embodiment of the present invention.

The embodiment of the present invention provides support for the construction of an adjustment model based on optical and SAR heterogeneous multi-view images. The embodiment of the present invention controls the contribution of optical and SAR image sources of different geometric qualities to the positioning solution by setting the weights of optical and SAR image observation values in the adjustment model, thereby building a more robust adjustment model and improving the positioning accuracy of the joint adjustment of heterogeneous images.

The device embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those of ordinary skill in the art may understand and implement it without paying creative labor.

Through the specific description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution can be essentially or partly contributed to the prior art in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, and the storage medium includes a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically erasable rewritable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

Finally, it should be noted that the method and device for determining the weights of the optical and SAR heterogeneous satellite images jointly adjusted disclosed in the embodiment of the present invention disclose only the preferred embodiments of the present invention, which are only used to illustrate the technical scheme of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, it should be understood by those skilled in the art that the technical schemes described in the aforementioned embodiments can still be modified, or some of the technical features therein can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical schemes from the spirit and scope of the technical schemes of the various embodiments of the present invention.

Claims (5)

1. A weighting method for optical and SAR heterogeneous satellite image joint adjustment, characterized in that the method comprises: S1, constructing heterogeneous regional network adjustment model of SAR images and optical images; S2, the weight matrix of the error equation group for solving the image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image; The step S1 comprises: S11, solve the rational polynomial model of SAR image and optical image, i.e. RPC, model parameters; generate corresponding image coordinates and corresponding ground coordinates of virtual control point pairs based on the rigorous imaging geometry model of SAR image and optical image respectively, use the point pair set and adopt the fitting method to solve the coefficients of RPC model function, and obtain the RPC model function s(B, L, H) and l(B, L, H) of the corresponding image; (s, l) is the two-dimensional image coordinate of the image point in the satellite image, (B, L, H) is the ground coordinate of the ground point corresponding to the image point in the image in the longitude and latitude coordinate system; RPC model function is used to convert the ground coordinates in the longitude and latitude coordinate system into two-dimensional image coordinates; S12, is the measured two-dimensional image coordinate of the i-th image point on the j-th image. The ground coordinates of the image point corresponding to the ground point are (B _k , L _k , H _k ), k is the serial number of the ground point. The image compensation function model of optical image and SAR satellite image is constructed using the affine transformation model, and the measurement error equation of the i-th image point is constructed as follows: Where s ^(j) ( _Bk , _Lk , _Hk ) and l ^(j) ( _Bk , _Lk , _Hk ) are the results of substituting the ground point coordinates ( _Bk , _Lk , _Hk ) into the RPC model function of the jth image. is the measured coordinate error of the i-th image point on the j-th image, is the image compensation parameter of the jth image; The error equation of the measurement error of the i-th image point is derived and linearized to obtain the error equation: In the formula, is the partial derivative of the RPC model function of the jth image with respect to the ground point coordinates (B _k , L _k , H _k ), is the correction value for the unknown number, is the initial value of the ground coordinates of the kth ground point and the initial value of the image compensation parameter of the jth image The initial value of the two-dimensional image coordinates of the i-th image point on the j-th image is calculated. The calculation process is expressed as: in, is the initial value of the image compensation parameter of the jth image; S13, constructing a set of image point coordinate measurement error equations for the optical image and the SAR image in the overlapping area; For n image points in the image, the corresponding image point coordinate measurement error equation group is constructed, and its expression is: Among them, _AG and _AA are the coefficient matrices of the error equation group, _ΔxG is the ground point coordinate correction, _ΔxA is the image compensation parameter correction, for the image point coordinate measurement error equation group, the corresponding weight matrix is P _I , L _I represents the image point coordinate measurement value, Represents the initial value of the image point coordinate measurement value; S14, the image-space compensation parameters in the measurement error equation are regarded as virtual observation values of the image-space affine transformation parameters, and the error equation group of the image-space compensation parameters is constructed, and its expression is: Among them, I is the unit matrix, _VA is the observation error of the virtual observation value of the image-side affine transformation parameter, and its expression is: _LA is the virtual observation value of the image-side affine transformation parameter, and its expression is: is the initial value of the virtual observation value, and its expression is: The weight matrix of the error equation group of the image compensation parameters is _PA ; S15, based on the image point coordinate measurement error equation group and the image square compensation parameter error equation group, a weighted regional block adjustment equation group of heterogeneous images without control points is constructed, and its expression is: The weight matrix of the weighted regional network adjustment equations of the heterogeneous image _PI and _PA are the weight matrices of the image point coordinate measurement error equation group and the image square compensation parameter error equation group, respectively. The _ΔxG and _ΔxA in the weighted regional network adjustment equation group of heterogeneous images are solved by an iterative method. The step S2 comprises: The image point coordinate measurement weight of the image is the inverse of the mean error of its corresponding image point coordinate measurement value; the elements on the diagonal of the weight matrix P _I are the weights of the corresponding image point coordinate measurement values, and the other elements of the weight matrix P _I are 0; for the weight matrix _PA , the mean error of each image compensation parameter is determined according to the prior accuracy information of the image, the weight of each image compensation parameter in the weight matrix _PA is the inverse of the error, the elements on the diagonal of the weight matrix _PA are the weights of the corresponding image compensation parameters, and the other elements of the weight matrix _PA are 0; After solving the weight matrix of the error equation group of the image compensation parameters according to the prior positioning accuracy of the SAR image and the optical image, the pixel value vector of the row where the image point related to the ground point target is located is used to calculate its cross-correlation matrix. According to the number of weights of the image compensation parameters, the feature matrix of the corresponding dimension is extracted. The feature matrix is used as the weighting matrix, and the weights of the image compensation parameters are weighted updated in turn to obtain the updated weights of the image compensation parameters. 2. The weighting method of the optical and SAR heterogeneous satellite image joint adjustment as claimed in claim 1 is characterized in that the described priori accuracy information according to the image respectively determines the mean error of each image compensation parameter, comprising: The constant terms _a0 and _b0 of the image compensation parameters represent the translation information of the satellite in the row and column directions, respectively, which are determined by the a priori absolute positioning accuracy of the image row and column directions. For optical images, the mean error of the constant term of the image compensation parameters is calculated based on the a priori positioning accuracy of the image row and column directions and the known resolution of the image row and column directions. The calculation process is: in, represents the mean error of a ₀ , represents the mean error of b ₀ ; For SAR images, the first constant term of the image compensation parameter represents the translation information of the satellite along the row direction. The mean error of the first constant term of the image compensation parameter is calculated according to the prior positioning accuracy in the image row direction and the known resolution in the image row direction. The second constant term of the image compensation parameter reflects the measurement error of the range slant range R. Therefore, the mean error of the second constant term of the image compensation parameter is calculated by multiplying the prior positioning accuracy σ _Y in the image column direction by the sine value of the image incident angle α _inc . The linear term of the image compensation parameter represents the scaling and rotation errors of the image. The prior relative positioning accuracy of the image is determined by the nominal error value of the satellite platform. The mean error of the linear term of the image compensation parameter is determined based on the maximum influence of this accuracy on the image and the image size. 3. The weighting method of the combined adjustment of optical and SAR heterogeneous satellite images as claimed in claim 2 is characterized in that, for the pixel value vector a _i of the corresponding row in the image of the image point relevant to the ground point target, i=1,2, ..., N, where N is the number of the image points relevant to the ground point target, all pixel value vectors are represented as a collection data matrix A: A＝[a ₁ ,a ₂ ,…,a _N ]， Calculate the cross-correlation matrix R of the collected data matrix A, and you will get: R＝ ^ATA ， Among them, the element r _ij = a _i a _j ^T in the i-th row and j-th column of the cross-correlation matrix R, and the column vector a _j represents the pixel value vector of the j-th image point related to the ground point target in the corresponding row in the image. The principal component analysis method is used to reduce the dimension of the collected data matrix and extract its feature matrix. The specific process is: Perform eigendecomposition on the cross-correlation matrix R to obtain N eigenvectors and eigenvalues. Select the eigenvectors according to the eigenvalues, and select the eigenmatrix E composed of M eigenvectors greater than a certain threshold, which is recorded as: E＝[v ₁ ,v ₂ ,…,v _M ]， Wherein, _vk represents the kth eigenvector, k = 1, 2, ..., M, the eigenvectors are all column vectors, and M is the number of weights of the image compensation parameters; the eigenmatrix E is used as the weighting matrix, and the vector a composed of the weights of the image compensation parameters is weighted updated, that is: b＝ ^ETa ， The updated weight vector b of the image-side compensation parameters is obtained. 4. A weighting device for joint adjustment of optical and SAR heterogeneous satellite images, the device comprising: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute part or all of the steps in the weighting method for joint adjustment of optical and SAR heterogeneous satellite images disclosed in any one of claims 1 to 3. 5. A computer storage medium storing computer instructions, which, when called, are used to execute some or all of the steps in the weighting method for joint adjustment of optical and SAR heterogeneous satellite images disclosed in any one of claims 1 to 3.