Improved point source array on-orbit MTF measurement method, system, and apparatus

Abstract

This invention discloses an improved method, system, and device for on-orbit MTF measurement of point source arrays. The method includes: first, measuring the coordinates of the ground point source array; then, calculating the initial coordinates of the peak values ​​of each point source image based on the initial RPC file generated from satellite imagery; subsequently, establishing a homography transformation model from the object side to the image side of the payload to correct the peak coordinates based on the object-side constraint relationship between the point sources in the point source array; then, aligning the brightness distribution of all point sources with the peak coordinates to obtain a refined point spread function; finally, fitting a two-dimensional PSF and calculating the MTF. This invention is applicable to on-orbit MTF measurement of high-resolution optical remote sensing payloads and has the advantages of noise resistance and high accuracy in image point peak detection.

Description

Technical Field

[0001] This invention belongs to the field of aerospace optical remote sensing technology and relates to an improved method for measuring the on-orbit modulation transfer function (MTF) of point source arrays, particularly for measuring the on-orbit MTF of optical remote sensing payloads for ground-based point source arrays. Background Technology

[0002] Remote sensing, with its comprehensive and objective advantages, has become an important means of Earth observation. With economic development and technological advancements, the demand for high-quality, high-resolution remote sensing data—available 24 / 7 and in all weather conditions—is increasing to support natural resource surveys and assessments, ecological environment monitoring and protection, and emergency management decision-making and response. Remote sensing data is widely used in scientific research, and optical remote sensing payloads face higher requirements for data product quality and long-term accuracy maintenance. In remote sensing images, the Medium Formatting Rendering (MTF) describes the size of the smallest identifiable detail in the image. A higher MTF value means the system can transmit higher spatial frequency information, i.e., it can capture finer details and provide higher clarity and spatial resolution, expanding the application range of remote sensing imagery.

[0003] Optical payloads undergo static MTF measurements in the laboratory before satellite launch. However, due to vibrations during launch and the influence of the on-orbit environment, the electronic components of the remote sensor age, leading to a decline in image quality. Furthermore, dynamically changing imaging conditions also reduce image MTF. To ensure the quality of remote sensing data, periodic dynamic MTF measurements and compensation are necessary during satellite operation. On-orbit MTF measurements are primarily achieved using artificial targets or specially shaped natural features in the image. Depending on the target type, methods include edge-based, pulse-based, and point-source methods. On-orbit MTF detection based on tilted edge targets is currently the most commonly used method. However, since the edge itself does not contain multiple frequency components, it is necessary to recover each frequency from the edge spread function extracted from the image. This process is inevitably affected by noise, introducing additional errors and reducing the accuracy of on-orbit MTF detection. The disadvantage of pulse-based on-orbit MTF detection is that different satellites have different resolutions, making it impossible to set ideal ground targets. Additionally, finding a pulsed natural feature in a fixed direction takes considerable time.

[0004] The point source method is based on the physical definition of MTF and is theoretically the most rigorous method for on-orbit MTF detection of optical remote sensing payloads. It can directly obtain the two-dimensional point spread function (PSF) based on the payload's response to a point source. The process is as follows: Figure 1Point sources are primarily reflective point sources. Remote sensing imaging systems exhibit different response characteristics at different spatial locations and orientations; therefore, a single point source target has insufficient sampling points and is easily affected by radiation from surrounding ground features and atmospheric conditions, leading to inaccurate measurement results. To improve the accuracy and stability of MTF measurement results, multiple point sources are typically combined to reconstruct the PSF, such as... Figure 2 As shown.

[0005] When a remote sensing payload images multiple point sources, the peak radiative response of each point source may not be located exactly at the center of the payload's power scattering (PSF). In most cases, the imaging occurs at sub-pixel positions deviating from the PSF center. Therefore, an accurate and effective method is needed to solve for the peak position to achieve high-precision PSF reconstruction from multiple point sources and ensure the accuracy of MTF evaluation. However, solving for the MTF solely from the local image space is susceptible to noise interference, resulting in poor noise robustness and significant fluctuations in measurement accuracy. Ground point source arrays have coplanar and equidistant object-space geometric constraints. Appropriately using these constraints helps to accurately extract the peak response positions of the point sources, improving the accuracy of MTF detection. Summary of the Invention

[0006] This invention provides a novel on-orbit MTF measurement method for ground-based reflective point source arrays. By fixing the object-space geometric constraints of the point source array, the method uses a homography transformation model to align and refine the brightness response, and then fits the PSF to calculate the MTF.

[0007] The purpose of this invention is to provide an improved on-orbit MTF measurement method for point source arrays, specifically including the following steps:

[0008] Step 1: Design and deploy a ground point source array for the high-resolution optical remote sensing payload, and obtain the geodetic coordinates Q of each point source device. 1_ground and planar coordinates Q 1_plane ;

[0009] Step 2: Based on the initial RPC file generated from satellite imagery, calculate the initial coordinates Q of the peak values ​​of each point source image. 2_initial ;

[0010] Step 3, establish Q 1_plane and Q 2_initial The homography matrix H between objects is used to solve the object-to-image transformation relationship H_m using RANSAC and least-squares overdetermined equations, and the optimized peak coordinates Q are calculated based on the mapping relationship. 2_repair ;

[0011] Step 4: Extract a pixel window of a certain size from the point light source image, based on the peak coordinate Q. 2_repair Align the point source brightness distribution;

[0012] Step 5: Fit the aligned scatter points to a two-dimensional point spread function (PSF), and then calculate the on-orbit modulation transfer function (MTF) using Fourier transform.

[0013] Furthermore, in step 1, the point source array is arranged collinearly with equal intervals, and the interval between adjacent sampling points is determined by the following formula:

[0014]

[0015] Where m is the number of integer pixels, n is the number of sampling points along the track or perpendicular to the track, and the geodetic coordinates Q of each point source device are measured dynamically in real time. 1_ground and planar coordinates Q 1_plane .

[0016] Furthermore, the specific implementation method of step 2 is as follows:

[0017] 2.1 Obtain the RPC file for image production;

[0018] 2.2 Using the RPC model, the geodetic coordinates Q of the point source device are determined. 1_ground The initial peak coordinates Q of the point source image are solved by performing the following formula. 2_initial ;

[0019]

[0020] In the formula, (U,V,W) and (x,y) represent the regularized geodetic coordinates of the ground point and the regularized image point coordinates, respectively, and Num L Den L Num S Den S It is a cubic polynomial in U, V, W; p 1ijk p 2ijk p 3ijk p 4ijk RPC represents a rational polynomial model, where i = 1, 2, 3; j = 1, 2, 3; k = 1, 2, 3, and n is the number of terms in the polynomial;

[0021] 2.3 Repeat step 2.2 until all point source coordinates have been processed, and obtain the initial peak coordinates Q of the point source image. 2_initial .

[0022] Furthermore, the specific implementation method of step 3 is as follows:

[0023] 3.1 Homogeneous pixel coordinate system q a To pixel coordinates q b The formula for calculating the homography matrix H is as follows:

[0024] q a ∝Hq b(4)

[0025] Expand to q a =(u a ,v a ,1) T ,q b =(u b ,v b ,1) T When there are more than 4 pairs of matching points, the least squares method is used to solve the problem. Since the transformation is to homogeneous coordinates, the homography matrix is ​​independent of the scale, so the degree of freedom is 8. H9=1 is used for normalization.

[0026]

[0027] The proportionality to the symbol ∝ is interpreted as the homography matrix constraining q. a and Hq b The directions are the same, without scale constraints. H1, H2, H4, and H5 are the image scale and rotation, H3 is the horizontal displacement, H6 is the vertical displacement, and H7 and H8 are the horizontal and vertical deformations. Homogeneous scale factors are eliminated through cross product calculation; therefore, the above constraints are expressed in the following form:

[0028] q a ×Hq b =0 (6)

[0029] Because q a and Hq b Since they are in the same direction, their cross product is a zero vector;

[0030] When the coordinates of each point on the object plane are Q 1_plane Image coordinates Q 2_initial When the one-to-one correspondence is perfect, there exists a homography matrix H, which can be obtained from equation (4):

[0031] Q 2_initial =H×Q 1_plane (7)

[0032] 3.2 The image-side coordinates obtained by RPC orthographic projection have errors. First, RANSAC is used to identify and eliminate some unreasonable matching points. Second, the least squares overdetermined equations are used to solve H to obtain the transformation relationship Hm from the object side to the image side. The optimized peak coordinates Q can be obtained from equation (4). 2_repair :

[0033] Q 2_repair =H _m ×Q 1_plane (8)

[0034] By mapping and projecting the object coordinates to the same image coordinate, the initial peak coordinates can be corrected.

[0035] Furthermore, the specific implementation method of step 4 is as follows:

[0036] 4.1 Traverse the images of the study area and extract windows within the brightness range of each point light source in the image;

[0037] 4.2, the point source brightness region P i With its corresponding peak coordinate Q 2_repair Alignment, after alignment, the coordinates are K, then we have

[0038] K = P i -Q 2_repair (9)

[0039] 4.3 Repeat steps 4.1-4.2 to obtain the point source brightness scatter plot for each image.

[0040] Furthermore, the specific implementation method of step 5 is as follows:

[0041] 5.1 The aligned point source brightness scatter plots are fitted to the PSF using the nonlinear least squares method;

[0042] 5.2 For a two-dimensional PSF, apply a two-dimensional Fourier transform to convert it from the spatial domain to the frequency domain;

[0043] 5.3 After performing the Fourier transform, a centering shift operation is performed to move the zero frequency of the spectrum to the center position;

[0044] 5.4 After applying the Fourier transform, take the absolute value of the spectrum;

[0045] 5.5, Normalize the spectrum;

[0046] 5.6 Finally, the obtained normalized spectrum is plotted as an MTF curve.

[0047] Furthermore, the formula for calculating PSF in 5.1 is as follows;

[0048]

[0049] In the formula, g(x,y) is the point diffusion response at coordinate position (x,y), k is the coefficient factor; (x0,y0) is the center position, i.e., the image point coordinates; σ and ζ are the standard errors, and b is the response value of background radiation and dark current.

[0050] This invention also provides an improved on-orbit MTF measurement system for point source arrays, comprising the following modules:

[0051] The geodetic coordinate and planar coordinate acquisition module is used to design and deploy ground point source arrays for high-resolution optical remote sensing payloads, and obtain the geodetic coordinates Q of each point source device. 1_ground and planar coordinates Q 1_plane ;

[0052] The peak initial coordinate acquisition module calculates the initial coordinates Q of the peak values ​​for each point source image based on the initial RPC file generated from satellite imagery. 2_initial ;

[0053] The peak coordinate acquisition module is used to establish Q. 1_plane and q 2_initial The homography matrix H between objects is used to solve the object-to-image transformation relationship H_m using RANSAC and least-squares overdetermined equations, and the optimized peak coordinates Q are calculated based on the mapping relationship. 2_repair ;

[0054] The brightness distribution alignment module is used to extract a pixel window of a certain size from the point light source image, based on the peak coordinate Q. 2_repair Align the point source brightness distribution;

[0055] The MTF calculation module is used to fit the aligned scatter points to a two-dimensional point spread function (PSF), and then calculate the on-orbit modulation transfer function (MTF) using Fourier transform.

[0056] The present invention also provides an improved on-orbit MTF measurement device for point source arrays, comprising:

[0057] One or more processors;

[0058] A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to perform the method described in any of the above technical solutions.

[0059] Compared with the prior art, the present invention has the following characteristics and beneficial effects:

[0060] (1) A noise-resistant on-orbit MTF detection method based on object-space constraints between point source arrays is proposed.

[0061] (2) The simultaneous object constraint relationship results in high accuracy of peak image point extraction.

[0062] (3) It can also be used when the image is noisy.

[0063] (4) The PSF is solved directly from the image, eliminating the influence of ground object radiation and noise, and improving the accuracy of the on-orbit MTF detection algorithm.

[0064] (5) It can be widely used for on-orbit MTF detection of high-resolution optical remote sensing payloads. Attached Figure Description

[0065] Figure 1 This is a schematic diagram of the basic detection principle of the point source method in the prior art, (a) input signal, (b) output signal, (c) MTF curve;

[0066] Figure 2 Scatter plots of point source images registered with different number of images in the prior art: (a) single point source, (b) 4 point sources, (c) 16 point sources;

[0067] Figure 3 This is a technical roadmap for the present invention.

[0068] Figure 4 This is a diagram showing the effect of homography transformation correction points in an embodiment of the present invention.

[0069] Figure 5 This is a schematic diagram of the point light source target array layout in an embodiment of the present invention.

[0070] Figure 6 This is a system MTF curve diagram according to an embodiment of the present invention. Detailed Implementation

[0071] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings.

[0072] This invention provides a method for on-orbit MTF measurement of optical remote sensing payloads using ground-deployed point source arrays with object-space constraints. The technical roadmap is shown in Figure 3, and specifically includes the following steps:

[0073] S1, Design a ground point source array, the specific implementation is as follows;

[0074] According to the principle of remote sensing payload PSF measurement, in order to accurately measure the PSF, point source sampling of the payload PSF must meet the following requirements: all sampling points of the payload imaging should uniformly cover a PSF; the sampling points along the orbital and perpendicular directions should be collinearly arranged at equal intervals; and there should be a sufficient number of sampling points to fully sample the PSF. Therefore, the interval between adjacent sampling points can be determined by the following formula:

[0075]

[0076] Where m is the number of integer pixels, and n is the number of sampling points along the track or perpendicular to the track.

[0077] To ensure that solar radiation is similar when remote sensing satellites observe the same area multiple times, most remote sensing satellites have orbits with a certain inclination. Sun-synchronous orbit, optical remote sensing satellite The angle is approximately 98°. At this point, the satellite's Earth observation along its orbit has an angle with true north on the ground. Because the orbital inclinations of various remote sensing satellites differ, to ensure that the array direction is aligned with the satellite's orbital direction during actual deployment, the angle between the array's row direction and true north must be θ = 8°. A layout diagram is shown below. Figure 5 As shown.

[0078] At this time, the interval Δx between adjacent point light sources is:

[0079] Δx=m+0.25 (2)

[0080] In this embodiment, the point source array is arranged in a 4×4 equally spaced collinear configuration, with non-integer pixel intervals between adjacent sampling points. The geodetic coordinates Q of each point source device are measured using real-time kinematic (RTK) measurement. 1_ground and planar coordinates Q 1_plane .

[0081] S2, based on the initial RPC model produced from satellite imagery, calculates the initial coordinates Q of the peak values ​​of each point source image. 2_initial .

[0082] This step further includes:

[0083] 2.1 Obtain the RPC file generated from the image. The RPC file is a file format used to describe the geometric relationships of remote sensing images and is used for geolocation. The RPC model can adapt well to image data of different resolutions and angles.

[0084] 2.2 The initial point source peak coordinates calculated using the RPC model are unaffected by image noise, ensuring relatively accurate absolute spatial positions of the initial peak coordinates. The point source device's object-space geodetic coordinates Q are then used. 1_ground Using the initial RPC model for image production, the initial peak coordinates Q of the point source image are solved by executing the following formula. 2_initial

[0085]

[0086] In the formula, (U,V,W) and (x,y) represent the regularized geodetic coordinates of the ground point and the regularized image point coordinates, respectively, and Num L Den L Num S Den S It is a cubic polynomial in U, V, W; p 1ijk p 2ijk p 3ijk p 4ijk RPC represents a rational polynomial model, where i = 1, 2, 3; j = 1, 2, 3; k = 1, 2, 3, and n is the number of terms in the polynomial;

[0087] 2.3 Repeat step 2.2 until all point source coordinates have been processed, and obtain the initial peak coordinates Q of the 16 sets of point source images. 2_initial As shown in the table below;

[0088]

[0089] S3, build Q 1_plane and Q 2_initial The homography matrix H between objects is used to solve the object-to-image transformation relationship H_m using RANSAC and least-squares overdetermined equations; and the optimized peak coordinates Q are calculated based on the mapping relationship. 2_repair This step further includes:

[0090] 3.1 Homography is a plane-to-plane projection transformation that maps one plane to another while preserving the linearity and parallelism of lines within the same plane. The residual errors after geometric correction in RPC models include imaging distortions such as perspective changes between the object plane and the image plane, and resolution inconsistencies. These slightly complex transformations can be simplified into a relatively simple transformation—homography. When a ground point source array is projected onto the image plane via remote sensing imaging, an 8-DOF homography matrix can be established between the two planes to describe their mapping relationship. The collinearity and coplanarity constraints inherent in the homography mapping relationship correct the projection of ground points onto the image plane.

[0091] Where the homogeneous pixel coordinates q a To pixel coordinates q b The formula for calculating the homography matrix H is as follows:

[0092] q a ∝Hq b (4)

[0093] Expanded to:

[0094]

[0095] The proportionality to the symbol ∝ can be understood as the homography matrix constraining q. a and Hq b The directions are the same, without constraining the scale. H1, H2, H4, and H5 are the image scale and rotation, H3 is the horizontal displacement, H6 is the vertical displacement, H7 and H8 are the horizontal and vertical deformations, and H9 is usually 1. Homogeneous scale factors can be eliminated by cross product calculation, so the above constraints can also be expressed in the following form:

[0096] q a ×Hq b =0 (6)

[0097] Because q a and Hqb Since they are in the same direction, their cross product is a zero vector.

[0098] When the object plane coordinates Q 1_plane Image coordinates Q 2_initial In the case of perfect correspondence, there exists a homography matrix H, which can be obtained from equation (4):

[0099]

[0100] 3.2 The image-side coordinates obtained using RPC orthographic projection have errors. First, RANSAC is used to identify and eliminate some unreasonable matching points, thereby improving the accuracy and reliability of the homography matrix. Second, the least squares overdetermined equations are used to solve H, obtaining the transformation relationship Hm from the object side to the image side:

[0101]

[0102] The optimized peak coordinate Q can be obtained from equation (4). 2_repair :

[0103]

[0104] Thus, by mapping and projecting the object-space coordinates to the same image-space coordinates, the initial peak coordinates can be corrected, and the correction effect is as follows: Figure 4 As shown, the peak coordinates are corrected and adjusted to a stable geometric relationship of collinearity and parallelism.

[0105] S4. Extract a pixel window of a certain size from the point light source image. The window should ideally encompass the entire brightness range and then expand outwards by 1-2 pixels. In this implementation, the selected pixel window size is 7×7. Then, based on the peak coordinate Q... 2_repair Align the point source brightness distribution according to the following rules:

[0106] 4.1 Traverse the images of the study area and extract windows within the brightness range of each point light source in the image;

[0107] 4.2, the point source brightness region P i With its corresponding peak coordinate Q 2_repair Alignment, after alignment, the coordinates are K, then we have

[0108]

[0109] 4.3 Repeat steps 4.1-4.2 to obtain a scatter plot of the brightness of 16 point sources, as shown below. Figure 2 As shown in (c);

[0110] S5, fits the aligned scatter points to the PSF, then calculates the MTF using Fourier transform, further including:

[0111] 5.1 The PSF is fitted using the nonlinear least squares method to the aligned point source brightness scatter plot. The point diffusion response g(x,y) is as follows;

[0112]

[0113] In the formula, k is the coefficient factor; (x0, y0) is the center position, i.e., the image point coordinates; σ and ζ are the standard errors; and b is the system's response to background radiation and dark current.

[0114] In this embodiment, the PSF is solved using the nonlinear least squares method to obtain:

[0115]

[0116] 5.2 For a two-dimensional PSF, a two-dimensional Fourier transform is applied to convert it from the spatial domain to the frequency domain. The Fourier transform converts the function into a frequency spectrum representation, where the frequency spectrum describes the existence of different frequency components.

[0117] 5.3 After performing the Fourier transform, a centering shift operation is required to move the zero frequency of the spectrum to the center position. This is because during the Fourier transform, the zero frequency of the frequency spectrum is usually located in the upper left corner of the spectrum.

[0118] 5.4 After applying the Fourier transform, we usually only focus on the amplitude component of the spectrum, and not the phase information. Therefore, we can take the absolute value of the spectrum.

[0119] 5.5 To compare the MTF of different systems or images, the spectrum needs to be normalized. This is typically achieved by setting the maximum value of the spectrum to 1 so that comparisons can be made at the same scale.

[0120] 5.6 Finally, the obtained normalized spectrum is plotted as an MTF curve. This curve shows the transmission characteristics of the optical system for details at different frequencies, such as... Figure 6 The figure shows the MTF curves of the system along the rail and perpendicular to the rail, calculated by this invention.

[0121] Another embodiment of the present invention provides an improved on-orbit MTF measurement system for point source arrays, comprising the following modules:

[0122] The geodetic coordinate and planar coordinate acquisition module is used to design and deploy ground point source arrays for high-resolution optical remote sensing payloads, and obtain the geodetic coordinates Q of each point source device. 1_ground and planar coordinates Q 1_plane ;

[0123] The peak initial coordinate acquisition module calculates the initial coordinates Q of the peak values ​​for each point source image based on the initial RPC file generated from satellite imagery. 2_initial ;

[0124] The peak coordinate acquisition module is used to establish Q. 1_plane and Q 2_initial The homography matrix H between objects is used to solve the object-to-image transformation relationship H_m using RANSAC and least-squares overdetermined equations, and the optimized peak coordinates Q are calculated based on the mapping relationship. 2_repair ;

[0125] The brightness distribution alignment module is used to extract a pixel window of a certain size from the point light source image, based on the peak coordinate Q. 2_repair Align the point source brightness distribution;

[0126] The MTF calculation module is used to fit the aligned scatter points to a two-dimensional point spread function (PSF), and then calculate the on-orbit modulation transfer function (MTF) using Fourier transform.

[0127] The specific implementation methods of each module are the same as those of each step, and will not be described in this invention.

[0128] Another embodiment of the present invention provides an improved on-orbit MTF measurement device for point source arrays, comprising:

[0129] One or more processors;

[0130] A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to perform the method described in any of the above technical solutions.

[0131] The specific embodiments described herein are merely illustrative of the spirit of the invention. Those skilled in the art to which this invention pertains may make various modifications or additions to the described specific embodiments or use similar methods to substitute them, without departing from the spirit of the invention or exceeding the scope defined by the appended claims.

Claims

1. An improved point source array on-orbit MTF measurement method, characterized in that, Includes the following steps: Step 1, design and layout ground point source array for high resolution optical remote sensing load, obtain geodetic coordinates of each point source device and plane coordinates ; Step 2, based on the initial RPC file produced from satellite imagery, calculate the initial peak coordinates of each point source image ; The specific implementation method of step 2 is as follows: 2.1 Obtain the RPC file generated from satellite imagery; 2.2, using the RPC model, the point source device geodetic coordinates , the following formula to solve the point source image initial peak coordinates ; In the formula, and These represent the regularized geodetic coordinates of ground points and the regularized image point coordinates, respectively. It is about A cubic polynomial; , , , RPC representing a rational polynomial model, where i= 1, 2, 3; ,j= 1, 2, 3; k= 1, 2, 3, n The number of terms in the polynomial; 2.3 Repeat step 2.2 until all point source coordinates have been processed, and obtain the initial peak coordinates of the point source image. ; Step 3, Establish and homography matrix between H The transformation relationship from object to image is solved using RANSAC and the least squares overdetermined equations. H_m And based on the transformation relationship H_m The optimized peak coordinates were calculated. ; The specific implementation method of step 3 is as follows: 3.1 Homogeneous form of object-space image pixel coordinates To the pixel coordinate system of the image homography matrix H The calculation formula is as follows: Expand as , When there are more than 4 matching pairs, the least squares method is used to solve the problem. The homography matrix has 8 degrees of freedom. =1 for normalization; Proportional to the symbol It is understood as a homography matrix constraint. and The direction is the same, but the scale is not constrained. The scale and rotation of the point source image. This is the horizontal displacement. This is the vertical displacement. Let be the deformation amounts in the horizontal and vertical directions; the homogeneous scale factor is eliminated by cross product calculation, therefore the above constraint is expressed in the following form: because and Since they are in the same direction, their cross product is a zero vector; When the plane coordinates of each point source device Initial peak coordinates of the point source image In a one-to-one correspondence, there exists a homography matrix. H From equation (4), we get: 3.2 The image-side coordinates obtained using RPC orthographic projection have errors. First, RANSAC is used to identify and eliminate some unreasonable matching points. Second, the homography matrix is ​​solved using a least-squares overdetermined system of equations. H The transformation relationship from object to image is obtained. H_m The optimized peak coordinates are obtained from equation (7). : At this point, the object coordinates are mapped and projected to the same image, thus correcting the initial peak coordinates of the point source image. Step 4: Extract the pixel window of the point source image, based on the optimized peak coordinates. Align the point source brightness distribution; Step 5: Fit the aligned scatter points to a two-dimensional point spread function (PSF), and then calculate the on-orbit modulation transfer function (MTF) using Fourier transform.

2. An improved on-orbit MTF measurement method for point source arrays as described in claim 1, characterized in that: in step 1, the point source array is arranged collinearly with equal intervals, and the interval between adjacent sampling points is determined by the following formula: in m The integer number of pixels is n, where n is the number of sampling points along the track or perpendicular to the track. The geodetic coordinates of each point source device are measured dynamically in real time. and planar coordinates .

3. The improved on-orbit MTF measurement method for point source arrays as described in claim 1, characterized in that: step 4 is specifically implemented as follows: 4.1 Traverse the images of the study area and extract the windows within the brightness range of each point light source in the point source image; 4.2, [Improve the point source brightness area] With its corresponding optimized peak coordinates Alignment, the coordinates after alignment are K Then there is 4.3 Repeat steps 4.1-4.2 to obtain the point source brightness scatter plot for each point source image.

4. An improved on-orbit MTF measurement method for point source arrays as described in claim 1, characterized in that: The specific implementation method of step 5 is as follows: 5.1 The aligned point source brightness scatter plots are fitted to the PSF using the nonlinear least squares method; 5.2 For a two-dimensional PSF, apply a two-dimensional Fourier transform to convert it from the spatial domain to the frequency domain; 5.3 After performing the Fourier transform, a centering shift operation is performed to move the zero frequency of the spectrum to the center position; 5.4 After applying the Fourier transform, take the absolute value of the spectrum; 5.5, Normalize the spectrum; 5.6 Finally, the obtained normalized spectrum is plotted as an MTF curve.

5. An improved on-orbit MTF measurement method for point source arrays as described in claim 4, characterized in that: The formula for calculating PSF in section 5.1 is as follows; In the formula, coordinate position Point spread response at that location, It is a coefficient factor; It represents the center position, i.e., the coordinates of the image point; σ and ζ are the standard errors. It is the response value of background radiation and dark current.

6. An improved on-orbit MTF measurement system for point source arrays, used to implement the method as described in any one of claims 1-5, characterized in that, Includes the following modules: The geodetic coordinate and planar coordinate acquisition module is used to design and deploy ground point source arrays for high-resolution optical remote sensing payloads, and to obtain the geodetic coordinates of each point source device. and planar coordinates ; The initial peak coordinate acquisition module calculates the initial peak coordinates of each point source image based on the initial RPC file generated from satellite imagery. ; The peak coordinate acquisition module is used to establish... and homography matrix between H The transformation relationship from object to image is solved using RANSAC and the least squares overdetermined equations. H_m And based on the transformation relationship H_m The optimized peak coordinates were calculated. ; The brightness distribution alignment module is used to extract pixel windows from the point source image and align them based on the optimized peak coordinates. Align the point source brightness distribution; The MTF calculation module is used to fit the aligned scatter points to a two-dimensional point spread function (PSF), and then calculate the on-orbit modulation transfer function (MTF) using Fourier transform.

7. An improved point source array on-orbit MTF measurement device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 5.

Images