An image registration method for ground-based SAR intermittent measurement

Abstract

The application provides an image registration method for ground-based SAR discontinuous measurement, a first image is acquired by using a ground-based SAR system, a reference coordinate system O-xyz is constructed, then the height of the ground-based SAR is increased along the z axis to acquire a second image; based on interferometric measurement, three-dimensional coordinates of reference points in the first image are acquired, and the first image is taken as a main image of discontinuous measurement; an image acquired at other time is taken as an auxiliary image of discontinuous measurement, then based on a cross-correlation coefficient method, homonymic points of the reference points in the main image in the auxiliary image are acquired, and the distance and azimuth offset between each reference point and the homonymic point are calculated; the distance and azimuth offset are respectively subjected to parameter modeling to construct a linear equation group, then a model coefficient is estimated by using a least square method to calculate accurate distance and azimuth offset; complex information of the homonymic points in the auxiliary image is acquired to realize accurate registration of the discontinuous measurement images.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ground-based SAR, in particular to an image registration method for ground-based SAR discontinuous measurement. BACKGROUND

[0002] Landslide disasters account for more than 70% of the total number of geological disasters in China, and monitoring and early warning of landslides are important links in disaster prevention and reduction. Deformation monitoring is an important basis for landslide prediction and early warning. Spaceborne / ground-based SAR (Synthetic Aperture Radar) based on microwave remote sensing measurement technology has been widely used in the field of landslide disaster monitoring. Spaceborne SAR has a wide coverage, but the revisit period is long and the observation angle is not flexible. Ground-based SAR is an effective supplement to spaceborne measurement methods, which can achieve high-precision, near-real-time deformation measurement.

[0003] The measurement mode of ground-based SAR can be divided into two types: continuous measurement and discontinuous measurement. For landslides with continuous deformation or instability, ground-based SAR can be used to conduct continuous monitoring at fixed observation points for several months or even years. For creep-type landslides with unstable factors but slow deformation (a few centimeters per year), continuous measurement by ground-based SAR will result in a great waste of manpower and material resources. Therefore, the observation period can be flexibly adjusted, and deformation measurement can be carried out periodically, i.e. discontinuous measurement.

[0004] When ground-based SAR is used for discontinuous monitoring, the radar needs to be installed every time, and after a certain number of radar images are obtained, the radar needs to be disassembled. Repeated installation and disassembly of the radar will cause changes in the position and attitude of the radar during multiple measurements, which will inevitably result in re-tracking errors. When two discontinuous measurement images are processed by interference, the re-tracking error will cause the pixels in the interference image to be unable to be directly matched, which will seriously reduce the coherence of the interference image and make it impossible to realize differential interference processing.

[0005] In order to ensure the feasibility of deformation measurement based on ground-based SAR discontinuous measurement technology, high-precision registration of discontinuous measurement ground-based SAR images must be carried out to accurately extract deformation information. Therefore, how to accurately register ground-based SAR discontinuous measurement images is a problem that needs to be solved urgently. SUMMARY

[0006] Therefore, the present application provides an image registration method for ground-based SAR discontinuous measurement, which can accurately model the distance and azimuth offset caused by re-tracking error and accurately register discontinuous measurement images based on interpolation.

[0007] To achieve the above purpose, the technical scheme of the present application is as follows.

[0008] An image registration method for ground-based SAR discontinuous measurement, comprising:

[0009] A first image is acquired by using a ground-based SAR system, a reference coordinate system O-xyz is constructed, then the height of the ground-based SAR is increased along the z-axis to acquire a second image; based on interferometric measurement, the three-dimensional coordinates of reference points in the first image are acquired, and the first image is taken as a main image of discontinuous measurement;

[0010] An image acquired at other time is taken as a secondary image of discontinuous measurement, then based on a cross-correlation coefficient method, homonymic points of the reference points in the main image in the secondary image are acquired, and the distance and azimuth offset between each reference point and the homonymic point thereof are calculated;

[0011] The distance and azimuth offset are respectively subjected to parameter modeling to construct a linear equation group, then a model coefficient is estimated by using a least square method to calculate accurate distance and azimuth offset;

[0012] Complex information of the homonymic points in the secondary image is acquired to realize accurate registration of the discontinuous measurement image.

[0013] The present application has the following beneficial effects:

[0014] The present application can accurately model the distance and azimuth offset caused by heavy rail error, and realizes accurate registration of the discontinuous measurement image based on interpolation, thereby ensuring the feasibility of realizing deformation measurement based on differential interference technology for the radar image acquired in the discontinuous measurement mode. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The present application has the following beneficial effects:

[0016] Figure 2 The present application has the following beneficial effects:

[0017] Figure 3 The present application has the following beneficial effects: DETAILED DESCRIPTION

[0018] The present application will be described in detail below in combination with the drawings and embodiments.

[0019] As shown in the drawings, the present application has the following beneficial effects: Figure 1 A kind of image registration method for ground-based SAR discontinuous measurement of the present embodiment, specifically includes:

[0020] Step one, obtaining a first image by using a ground-based SAR system, constructing a reference coordinate system O-xyz, then improving the height of the ground-based SAR along the z-axis to obtain a second image; obtaining the three-dimensional coordinates of the reference points in the first image based on interferometric measurement, and taking the first image as the main image of the discontinuous measurement;

[0021] In this embodiment, it is assumed that the synthetic aperture center of the ground-based SAR is located at the origin O, the aperture direction is parallel to the x-axis, the y-axis is perpendicular to the x-axis in the horizontal plane, and the z-axis forms a right-handed rectangular coordinate system with the x-axis and the y-axis. Then the radar height is improved in the positive direction of the z-axis, and the radar image 2 is obtained at the height B. The typical value range of B is 10 cm to 30 cm. The pixel points with a signal-to-noise ratio higher than 20 dB are selected from the image 1 as the reference points for interference processing. The complex conjugate multiplication is performed on the two images to obtain an interference phase image.

[0022] For any reference point in the monitoring area, it is assumed that the distances between the reference point and the radar in the two measurements are r and r(B) respectively, as shown in the following formula (1). Figure 2 The interference phase of the reference point and the distance change are related as shown in the following formula (2).

[0023]

[0024] Therefore, the height z of the reference point can be expressed as shown in the following formula (3).

[0025]

[0026] It is assumed that the slant range of a reference point P in the radar image 1 is r p , the azimuth angle is θ p , and the height coordinate z p is obtained based on formula (2). According to the ground-based SAR imaging geometry, the x p and y p coordinates of P are respectively

[0027]

[0028] Based on formulas (1)-(3), the three-dimensional coordinates of all reference points in the radar image 1 are obtained, and the radar image 1 is taken as the main image of the discontinuous measurement.

[0029] Step two, taking an image obtained at other time as the auxiliary image of the discontinuous measurement, then based on the cross-correlation coefficient method, obtaining the homonymic points of the reference points in the main image in the auxiliary image, and calculating the distance and azimuth offset between each reference point and its homonymic point;

[0030] In the specific implementation, after the first interferometric measurement acquires two images, the second measurement is carried out several weeks or months later, and only one image is acquired as an auxiliary image of the intermittent measurement. Since the position and attitude of the ground-based SAR cannot be completely unchanged during the two measurements, a re-tracking error is inevitably caused. For a reference point P1 (r P1 , θ P1 ) in the main image of the intermittent measurement, it is assumed that the homonym of P1 in the auxiliary image of the intermittent measurement is P2 (r P2 , θ P2 ). Due to the re-tracking error, the slant range and azimuth angle of P1 and P2 are different, and their row and column coordinates in the main image and the auxiliary image are different, and thus they cannot be directly matched.

[0031] Therefore, in order to realize the matching of P1 and P2, a cross-correlation coefficient method is adopted. Taking P1 as the center, a smaller rectangular window L1 x L2 in the main image is first constructed, denoted as W M , and then a larger rectangular window L3 x L4 in the auxiliary image is constructed, denoted as W S . Figure 3 A reference point matching schematic diagram is shown in FIG. 1. A rectangular window L1 x L2 in W S is selected, denoted as W , wherein Δθ and Δr represent the offset along the azimuth direction and the range direction respectively, and then the cross-correlation coefficient γ M of W (Δθ,Δr) and W is calculated based on the following formula.

[0032]

[0033] wherein ∑[] represents summation, and * represents complex conjugate.

[0034] The offset Δθ and Δr are adjusted along the azimuth direction and the range direction in the window W S , and γ (Δθ,Δr) is calculated once for each offset pair. When the value of γ (Δθ,Δr) reaches the maximum, the center point of W is taken as the homonym of the reference point P1 in the auxiliary image, such as P2 in the figure, and the offset Δθ m1 and Δr m1 when the cross-correlation coefficient reaches the maximum are recorded. In actual processing, in order to improve the estimation accuracy of Δθ m1 and Δr m1 , W S may be 8 times or 16 times up-sampled. However, due to the existence of matching errors, noise phases and the like, there will be a small error between the measured value and the true value of Δθ m1 and Δr m1 .

[0035] Step three, the distance and azimuth offset are respectively modeled, linear equations are constructed, and then the model coefficients are estimated by least square method to calculate the accurate distance and azimuth offset.

[0036] In the specific implementation, when the ground-based SAR acquires the main image, the three-dimensional coordinates of the aperture center are located at the origin O, and the aperture direction is parallel to the x-axis. It is assumed that when the auxiliary image is acquired, the position and attitude of the radar are changed, the aperture center moves to O'(ε x ,ε y ,ε z ), the angle between the aperture direction and the positive direction of the x-axis is ε α , and the angle between the aperture direction and the x-y plane is ε β . Taking the same named points P1 and P2 in step two as an example to analyze the re-rail error, the slant range difference Δr and the azimuth angle difference Δθ between the two points in the theoretical case are respectively

[0037]

[0038]

[0039] In this embodiment, it is assumed that N reference points q1(r1, θ1), q2(r2, θ2), …, q N (r N ,θ N ) are acquired in the main image. In step 1, the three-dimensional coordinates (x1, y1, z1), (x2, y2, z2), …, (x N ,y N ,z N ) of all reference points are acquired. In step two, the offset (Δr m1 ,Δθ m1 ), (Δr m2 ,Δθ m2 ), …, (Δr mN ,Δθ mN ) between all reference points and the same named points is preliminarily acquired. Based on equation (5), the equation group is constructed,

[0040] Y r =H xyz X xyz +ε r

[0041]

[0042] wherein Y r is a column vector composed of the distance offset between the reference points and the same named points, H xyz is a coefficient matrix, and X xyzε is the column vector of aperture center error to be estimated. r Let X represent the error vector. The least squares method is used to estimate X. xyz

[0043]

[0044] in, T This represents the matrix transpose. Substituting into equation (7), we re-estimate the range offset Y. r

[0045]

[0046] in It is a column vector composed of distance offsets.

[0047] Referring to equations (7)-(9), the azimuth offset is re-estimated in the same way. First, a system of linear equations is constructed.

[0048] Y θ =H αβ X αβ +ε θ

[0049]

[0050] Using the least squares method, re-estimate Y θ

[0051]

[0052]

[0053] Step 4: Use spatial interpolation to obtain the complex information of corresponding points in the auxiliary image, so as to achieve accurate registration of discontinuous measurement images.

[0054] In step three, the distance offset between all reference points in the main image and their corresponding points in the auxiliary image is estimated. and azimuth offset Based on the nth reference point q n Coordinates (r) in the main image n ,θ n This allows us to obtain the coordinates of its corresponding point in the auxiliary image. Assuming in the secondary image (r) n ,θ n The complex information of the pixel at position () is I n Now, interpolation is used to obtain... Reply information Two-dimensional linear interpolation, sinc interpolation, kriging interpolation, etc., can be used. This is the complex information of the nth point with the same name.

[0055] After the above processing, for all the reference points in the main image, the complex information of their homonymic points in the auxiliary image is obtained, i.e. the registration of the discontinuous measurement images is realized. Then the interference processing is performed on each reference point and its homonymic point, i.e. the deformation measurement of the discontinuous measurement images can be realized.

[0056] To sum up, the above is only the preferred embodiment of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An image registration method for ground-based SAR discontinuous measurements, characterized in that, include: A first image is acquired using a ground-based SAR system, a reference coordinate system O-xyz is constructed, and then the height of the ground-based SAR is increased along the z-axis to acquire a second image. Based on interferometry, the three-dimensional coordinates of the reference point in the first image are obtained, and the first image is used as the master image for discontinuous measurements. An image acquired at another time is used as an auxiliary image for discontinuous measurement. Then, based on the cross-correlation coefficient method, the corresponding points of the reference points in the main image in the auxiliary image are obtained, and the distance and azimuth offsets between each reference point and its corresponding points are calculated. The specific method for obtaining the corresponding points of the reference points in the main image in the auxiliary image is as follows: For any reference point in the main image, a small window and a large window are constructed on the main and auxiliary images of discontinuous measurement, respectively, with that reference point as the center. Then, the small window is moved within the large window, and the cross-correlation coefficient is calculated. The center point of the small window with the largest cross-correlation coefficient is selected as the corresponding point. The range and azimuth offsets are parametrically modeled to construct a system of linear equations. Then, the least squares method is used to estimate the model coefficients and calculate the accurate range and azimuth offsets. The complex information of corresponding points in the auxiliary image is obtained to achieve accurate registration of discontinuous measurement images.

2. The image registration method for ground-based SAR discontinuous measurement as described in claim 1, characterized in that, The reference coordinate system O-xyz is constructed by taking the center of the synthetic aperture of the ground-based SAR as the origin O, with the aperture direction parallel to the x-axis, the y-axis perpendicular to the x-axis in the horizontal plane, and the z-axis forming a right-handed rectangular coordinate system with the x-axis and y-axis.

3. The image registration method for ground-based SAR discontinuous measurement as described in claim 2, characterized in that, The range of height values ​​for improving ground-based SAR is 10cm to 30cm.

4. An image registration method for ground-based SAR discontinuous measurement as described in claim 1, 2, or 3, characterized in that, The reference point in the main image is selected from pixels with a signal-to-noise ratio higher than 20dB.

5. An image registration method for ground-based SAR discontinuous measurement as described in claim 1, 2, or 3, characterized in that, The complex information of corresponding points in the auxiliary image is obtained using spatial interpolation, two-dimensional linear interpolation, sinc interpolation, and Kriging interpolation.

Images