CN105093222A - Automatic extraction method for block adjustment connection points of SAR image - Google Patents

Abstract

The invention provides a method for automatically extracting connection points of SAR image block adjustment, which includes: obtaining the corrected SAR image; generating ground grid points according to the geographical coordinates of the corrected SAR image and external auxiliary terrain information, and calculating the ground grid points in the corrected SAR The corresponding image points on the image; intercept a group of image blocks with the same name from the corrected SAR image, select one of them as a reference image block, and extract the point with the largest value of interest in the reference image block as a matching reference point; on the image block with the same name Search for every image point with the same name that matches the reference point, and get the same-name connection point between the corrected SAR images; convert the same-name connection point between the corrected SAR images to the original SAR image, and get the SAR image block adjustment connection point. This method improves the automation of SAR image block adjustment processing and improves the production efficiency of SAR topographic mapping.

Description

An Automatic Extraction Method of SAR Image Block Adjustment Connection Points

technical field

The invention belongs to the field of digital photogrammetry of remote sensing images, in particular to the field of synthetic aperture radar block adjustment processing.

Background technique

At present, Synthetic Aperture Radar (SAR) terrain mapping technology has been gradually popularized and applied. In cloudy and foggy areas, the application of large-area SAR mapping is an inevitable trend in the future. High-precision positioning of SAR images is the basis of SAR topographic mapping, but in large-area mapping applications, high-precision positioning of SAR images needs to be realized through block adjustment. The block adjustment is to make full use of the redundant observation information on the connection points of the overlapping areas of the images, solve the model orientation parameters, and achieve high-precision positioning, thereby reducing the demand for the number of control points. Therefore, the acquisition of connection points has become the first problem to be solved for high-precision positioning of large-area SAR images. It is very difficult for large-area mapping applications to extract connection points only by manual measurement, especially for airborne SAR mapping, because of the large amount of image data, automatic extraction methods must be developed to meet the requirements of practical applications. need.

At present, the SAR image matching method is used to automatically extract the connection points, but because SAR adopts the distance projection imaging method, the geometric distortion is serious, and there is speckle noise, which makes SAR image matching much more difficult than the optical image method. SAR image matching is a hot and difficult point in the field of SAR remote sensing. For the regional network, each connection point involves multiple images, the corresponding illumination angles are very different, and the resolution is not uniform, which leads to large differences in shape and grayscale of the same target on different images. In particular, the difference between SAR images with different side views is greater, which brings greater difficulties to the area network SAR image matching.

Contents of the invention

In order to solve the above-mentioned technical problems existing in the prior art, the present invention proposes a SAR image automatic matching method that integrates multivariate information. Image matching overcomes the problems of scale inconsistency, complex relative deformation, uncertain initial relative positional relationship of image points, difficulty in matching non-feature points, and large amount of matching calculations in SAR block network image matching, and realizes SAR image block adjustment Efficient automatic extraction of connection points.

The method for automatically extracting connection points of SAR image block adjustment of the present invention comprises the following steps:

Step 1, using the SAR geometric imaging information and external auxiliary terrain information to perform geometric correction on the SAR image to obtain the corrected SAR image;

Step 2, according to the geographical coordinates of the corrected SAR image and the external auxiliary terrain information, generate ground grid points at preset intervals, and calculate the corresponding image points of the ground grid points on the corrected SAR image, and the corresponding image points are used as approximate image point with the same name;

Step 3, for each group of similarly named image points, intercept a group of image blocks with the same name from the corrected SAR image, select one of them as a reference image block from the group of image blocks with the same name, use the feature information of the SAR image, Extract the point with the largest interest value from the reference image block as the matching reference point;

Step 4, for each matching reference point, search for the image point with the same name on the image block with the same name, and obtain the connection point with the same name between the corrected SAR images;

Step 5, according to the geometric positioning information, convert the connection points with the same name between the corrected SAR images obtained in step 4 to the original SAR image, and obtain the SAR image block adjustment connection points.

Preferably, the step 1 is specifically: using the geometric imaging parameters and sensor platform state vector parameters of the SAR image to establish a geometric positioning model, using the geometric positioning model combined with external auxiliary terrain information to perform geometric correction on the SAR image, Get the corrected SAR image.

Preferably, the external auxiliary terrain information is digital elevation model data.

Preferably, said step 2 specifically includes:

Step 2.1, within the scope of the regional network, take the ground grid points with the preset grid spacing, and combine the external auxiliary terrain information to obtain elevation information, and obtain the three-dimensional coordinates of the ground grid points;

In step 2.2, according to the geographical coordinate information of the corrected SAR image, the three-dimensional coordinates obtained in step 2.1 are back-calculated onto the geometrically corrected SAR image;

Step 2.3, for each ground grid point, traverse all the images in the area network, extract all the images imaged on the ground grid point, and obtain a set of image points on the corrected SAR image, which are approximate image points with the same name.

Preferably, said step 3 specifically includes:

Step 3.1, for each group of image points with the same name, take the corresponding image point as the center, and intercept the image block with preset width and height from the corrected SAR image. A group of image blocks approximately corresponds to the same ground area, and the group of image blocks is a group of image blocks with the same name.

Step 3.2, for each group of image blocks with the same name intercepted in step 3.1, select one of the image blocks as a reference image block, perform feature point extraction on the reference image block, and extract the point with the largest interest value as the final matching reference point.

Preferably, the width and height preset in step 3.1 are determined according to a geometric positioning model and an elevation error.

Preferably, in the step 4, for each matching reference point, an image point with the same name is searched on the image block with the same name using the region matching method to obtain the connection point with the same name between the corrected SAR images.

Preferably, the step 5 specifically includes: for correcting the same-named connection points between SAR images, according to the geographic coordinate information of the corrected SAR images, obtaining the plane coordinates (X _Rec , Y _Rec ) of the same-named connection points, using the plane coordinates in the DEM data Interpolate the elevation H _Rec to obtain the three-dimensional geographic coordinates (X _Rec , Y _Rec , H _Rec ) of the connection point with the same name. According to the SAR image geometric positioning model, calculate the image coordinates (x _SAR , y _SAR ).

The method for automatically extracting connection points of SAR regional network adjustments according to the present invention comprehensively utilizes geometric information, feature information, grayscale information and external auxiliary terrain information to overcome the inconsistency of scales in the multi-view image matching in the SAR regional network. For problems such as complex relative deformation, uncertain initial relative positional relationship of image points, difficulty in matching non-feature points, and large amount of matching calculations, fast area network SAR image matching is performed to realize automatic extraction of high-precision connection points with the same name.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present invention more clearly, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. Those of ordinary skill in the art can also obtain other drawings based on these drawings without any creative effort.

Fig. 1 is a flow chart of the method for automatically extracting connection points of SAR image block adjustment according to the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

As shown in Figure 1, the SAR image block adjustment connection point automatic extraction method of the present invention adopts the following steps to carry out:

Step 1, use the SAR geometric imaging information and external auxiliary terrain information to perform geometric correction on the SAR image to obtain the corrected SAR image.

Among them, step 1 is specifically: using the geometric imaging parameters of the SAR image and the sensor platform state vector parameters to establish a geometric positioning model, using the geometric positioning model combined with external auxiliary terrain information, that is, the existing DEM (Digital Elevation Model, digital elevation Model, which is a discrete digital expression of topography, is a digital method for spatially describing ground features) data, geometrically corrects the SAR image, and obtains the corrected SAR image.

Step 2, according to the geographical coordinates of the corrected SAR image and the external auxiliary terrain information, generate ground grid points at preset intervals, and calculate the corresponding image points of the ground grid points on the corrected SAR image, and the corresponding image points are used as approximate Image point of the same name.

Among them, step 2 specifically includes:

Step 2.1, within the scope of the regional network, the ground grid points are taken at a preset grid interval, and combined with external auxiliary terrain information, such as DEM, to obtain elevation information and obtain the three-dimensional coordinates of the ground grid points.

Assuming that the Gaussian plane coordinates of the upper left corner of the entire area network are (X ₀ , Y ₀ ), and the grid spacing is (△X, △Y), then the ground grid point P _{r, c} of the rth row and the cth column The coordinates (X _P , Y _P ) are: X _P =X ₀ +r*△X,Y _P =Y ₀ +c*△Y, where r and c are integers greater than 0. From the ground coordinates (X _P , Y _P ), the elevation value H _P can be interpolated from the DEM to obtain the three-dimensional coordinates (X _P , Y _P , H _P ) of the ground grid point P _r,c .

In step 2.2, according to the geographic coordinate information of the corrected SAR image, the three-dimensional coordinates obtained in step 2.1 are back-calculated onto the geometrically corrected SAR image.

Step 2.3, for each ground grid point, traverse all the images in the regional network, extract all the images imaged on the ground grid point, and obtain a set of image points on the corrected SAR image, this set of image points is approximately the same name image point .

For example, for the ground grid point P _r,c , the image related to m scenes can be obtained from the three-dimensional coordinates (X _P , Y _P , H _P ), m is the number of scenes, and the image of this point on the image of each scene can be obtained Point I _P ¹ , I _P ² ,....,I _P ^m , the image coordinates are (r _P ¹ ,c _P ¹ ),(r _P ² ,c _P ² ),....,(r _P ^m ,c _P ^m ), this group of image points is approximately the same name image point.

Step 3, for each group of similarly named image points, intercept a group of image blocks with the same name from the corrected SAR image, select one of them as a reference image block from the group of image blocks with the same name, use the feature information of the SAR image, The point with the largest interest value is extracted from the reference image block as the matching reference point.

Among them, step 3 specifically includes:

Step 3.1, for each group of approximate image points with the same name, take the corresponding image point as the center, intercept the preset width and height from the corrected SAR image (the width and height are determined according to geometric positioning and elevation error, to ensure accurate image points with the same name are all within the image block area), for image points with multiple degrees of overlap, a group of image blocks where the image point is located approximately corresponds to the same ground area, and this group of image blocks is a group of image blocks with the same name.

For example, for the approximate image points I _P ¹ , I _P ² ,....,I _P ^m corresponding to the grid point P _r,c , the image blocks B _P ¹ , B _P ² ,....,B with the same name can be obtained _P ^m .

Step 3.2, for each group of image blocks with the same name intercepted in step 3.1, select one of the image blocks as the main image block (reference image block), perform feature point extraction on the main image block, and extract the point with the largest value of interest as the final Match main point (match reference point).

This step uses the Harris corner detection operator to extract matching reference points. The basic principle of the Harris algorithm is to take a small window centered on the target pixel, calculate the gray level change after the window moves in any direction, and express it in an analytical form. Assuming that the small window centered on the pixel point (x, y) moves u in the x direction and v in the y direction, Harris gives the analytical expression of the gray scale change measure:

$\begin{array}{c}{\mathrm{<span\; class="notranslate">\; G</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;w</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; u</span>}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; v</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;w</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; u</span>}\frac{<span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; I</span>}}{<span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; v</span>}\frac{<span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; I</span>}}{<span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; o</span>}<span\; class="notranslate">\; (</span>\sqrt{{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; v</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\\ <span\; class="notranslate">\; \&ap;</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;w</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; \&lsqb;</span>{\mathrm{<span\; class="notranslate">\; u</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; v</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}{<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; uvI</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}<span\; class="notranslate">\; \&rsqb;</span><span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; Au</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; C</span>}\mathrm{<span\; class="notranslate">\; u</span>}\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; Bv</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\end{array}$

Among them, G _{x, y} is the grayscale change measure in the window; I is the image grayscale function. Transform G _{x, y} into quadratic form: $G_{x,\mathrm{the\; y}}=\left[\begin{array}{cc}u& v\end{array}\right]m\left[\begin{array}{c}u\\ v\end{array}\right],$ in $m={\mathrm{\&Sigma;w}}_{x,\mathrm{the\; y}}\left[\begin{array}{cc}{I}_x^2& I_x{I}_{\mathrm{the\; y}}\\ I_x{I}_{\mathrm{the\; y}}& I_{\mathrm{the\; y}}^2\end{array}\right].$

Through diagonalization, we get: $G_{x,\mathrm{the\; y}}=R^{-1}\left(\begin{array}{cc}{\mathrm{\&lambda;}}_1& 0\\ 0& {\mathrm{\&lambda;}}_2\end{array}\right)R$

Among them, R is the twiddle factor, and its eigenvalues λ ₁ and λ ₂ reflect the curvature of the image surface in the direction of the two principal axes. In order to avoid finding the eigenvalues of the matrix M, Tr(M) and Det(M) can be used instead of finding λ ₁ and λ ₂ , assuming:

$\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; \&Sigma;w</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}}\left[\begin{array}{cc}{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}& {\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}\\ {\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}& {\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}\end{array}\right]<span\; class="notranslate">\; =</span>\left[\begin{array}{cc}\mathrm{<span\; class="notranslate">\; A</span>}& \mathrm{<span\; class="notranslate">\; B</span>}\\ \mathrm{<span\; class="notranslate">\; C</span>}& \mathrm{<span\; class="notranslate">\; D.</span>}\end{array}\right]$

Then the determinant and trace of the matrix M _{(x, y)} are:

Tr(M)=λ ₁ +λ ₂ =A+B, Det(M)=λ ₁ λ ₂ =AB-C ²

The Harris corner response function (R) expression is thus obtained:

R(x,y)=Det(M)-k(Tr(M)) ² =(AB-C ² )-k(A+B) ²

The response function value is the interest value of feature extraction. In the reference image block, the point with the largest interest value is extracted as the matching reference point.

For example, for image blocks B _P ¹ , B _P ² ,...,B _P ^m with the same name, B _P ^k is selected as the main image block, and the matching reference point I _P ^k _ref is extracted in the image block B _P ^k .

Step 4, for each matching reference point, search for the image point with the same name on the image block with the same name, and obtain the connection point with the same name between the corrected SAR images.

Among them, step 4 is specifically: for each matching reference point, adopt the regional matching method, that is, carry out the gray level correlation matching of the region, and use the pyramid correlation matching strategy to search, in the image blocks B _P ¹ , B _P ² , .. .,B _P ^k-1 ,B _P ^k+1 ,....,B _P ^m search for image points with the same name I _P ¹ _mat ,I _P ² _mat ,....,I _P ^m _mat . The above-mentioned exact matching process is performed on all extracted image points with approximately the same name, and the connection points with the same name between the corrected SAR images in the area network are obtained. In the process of pyramid correlation matching, the parallax extreme value of the image point with the same name on different images can be obtained by analyzing the disparity between the images in this area combined with DEM, and the relevant matching parameters can be set accordingly to speed up the matching speed and accuracy.

Step 5, according to the geometric positioning information, convert the connection points with the same name between the corrected SAR images obtained in step 4 to the original SAR images, and obtain the connection points of the SAR image block adjustment.

Among them, step 5 is specifically: for correcting the connection points with the same name between the SAR images, combined with the elevation information of the DEM, back-calculating the connection points with the same name to the original SAR image, so as to obtain the required SAR image block adjustment processing. A join point of the same name. In the specific processing, for the connection point on the corrected SAR image, according to the geographic coordinate information of the corrected SAR image, the plane coordinates (X _Rec , Y _Rec ) of the connection point are obtained, and then the height H _Rec is interpolated on the DEM data by using the plane coordinates, Get the three-dimensional geographic coordinates (X _Rec , Y _Rec , H _Rec ) of the connection point, and then calculate the image coordinates (x _SAR , y _SAR ) of the connection point on the original SAR image according to the geometric positioning model of the SAR image. In the correction process, the indirect sampling method is used, so the geographic coordinates are calculated from the image points on the corrected SAR image, and then converted to the image coordinates of the original SAR image, the image points are strictly corresponding, and there is no loss of accuracy.

This method greatly enhances the automation of SAR image block adjustment processing, improves the production efficiency of SAR topographic mapping, and is of great significance to the promotion of SAR mapping technology applications.

It should be understood that the above descriptions are only preferred embodiments of the present invention, and are not sufficient to limit the technical solutions of the present invention. For those of ordinary skill in the art, within the spirit and principles of the present invention, they can Additions, substitutions, transformations or improvements are described, and all technical solutions after such additions, substitutions, transformations or improvements shall belong to the protection scope of the appended claims of the present invention.

Claims (8)

1. a SAR image block adjustment connection point automatic extraction method is characterized in that said method comprises the steps: Step 1, using the SAR geometric imaging information and external auxiliary terrain information to perform geometric correction on the SAR image to obtain the corrected SAR image; Step 2, according to the geographical coordinates of the corrected SAR image and the external auxiliary terrain information, generate ground grid points at preset intervals, and calculate the corresponding image points of the ground grid points on the corrected SAR image, and the corresponding image points are used as approximate image point with the same name; Step 3, for each group of similarly named image points, intercept a group of image blocks with the same name from the corrected SAR image, select one of them as a reference image block from the group of image blocks with the same name, use the feature information of the SAR image, Extract the point with the largest interest value from the reference image block as the matching reference point; Step 4, for each matching reference point, search for the image point with the same name on the image block with the same name, and obtain the connection point with the same name between the corrected SAR images; Step 5, according to the geometric positioning information, convert the connection points with the same name between the corrected SAR images obtained in step 4 to the original SAR image, and obtain the SAR image block adjustment connection points. 2. the SAR image block adjustment connection point automatic extraction method according to claim 1, is characterized in that said step 1 is specifically: utilize the geometric imaging parameter that SAR image carries and sensor platform state vector parameter, establish geometric positioning The model uses the geometric positioning model combined with external auxiliary terrain information to perform geometric correction on the SAR image to obtain the corrected SAR image. 3. The method for automatically extracting connection points of SAR image block adjustment according to claim 2, characterized in that: the external auxiliary terrain information is digital elevation model data. 4. the SAR image block adjustment connection point automatic extraction method according to claim 1, is characterized in that described step 2 specifically comprises: Step 2.1, within the scope of the regional network, take the ground grid points with the preset grid spacing, and combine the external auxiliary terrain information to obtain elevation information, and obtain the three-dimensional coordinates of the ground grid points; In step 2.2, according to the geographical coordinate information of the corrected SAR image, the three-dimensional coordinates obtained in step 2.1 are back-calculated onto the geometrically corrected SAR image; Step 2.3, for each ground grid point, traverse all the images in the area network, extract all the images imaged on the ground grid point, and obtain a set of image points on the corrected SAR image, which are approximate image points with the same name. 5. the SAR image block adjustment connection point automatic extraction method according to claim 1, is characterized in that described step 3 specifically comprises: Step 3.1, for each group of image points with the same name, take the corresponding image point as the center, and intercept the image block with preset width and height from the corrected SAR image. A group of image blocks approximately corresponds to the same ground area, and the group of image blocks is a group of image blocks with the same name. Step 3.2, for each group of image blocks with the same name intercepted in step 3.1, select one of the image blocks as a reference image block, perform feature point extraction on the reference image block, and extract the point with the largest interest value as the final matching reference point. 6. The SAR image block adjustment connection point automatic extraction method according to claim 5, characterized in that the preset width and height in the step 3.1 are determined according to a geometric positioning model and an elevation error. 7. the SAR image block adjustment connection point automatic extraction method according to claim 1, is characterized in that in the described step 4, for each matching reference point, adopts the area matching method to search for the image point of the same name on the image block of the same name , get the connection points with the same name between corrected SAR images. 8. The SAR image block adjustment connection point automatic extraction method according to claim 1, wherein said step 5 specifically comprises: for correcting the same-named connection points between SAR images, according to the geographical coordinate information of the correction SAR images, Obtain the plane coordinates (X _Rec , Y _Rec ) of the connection point with the same name, use the plane coordinates to interpolate the elevation H _Rec on the DEM data, and obtain the three-dimensional geographic coordinates (X _Rec , Y _Rec , H _Rec ) of the connection point with the same name, according to the SAR image geometry Locate the model and calculate the image coordinates (x _SAR , y _SAR ) of the connection point with the same name on the original SAR image.