Space target elongation image astronomic positioning method of large field of view space-based monitoring platform

Abstract

The application discloses a space target elongated star image astronomical positioning method of a large field of view space-based monitoring platform and relates to the technical field of astronomical observation. The method comprises the following steps: dividing an observation image into 16 sub-fields, wherein each sub-field is an independent field and is calculated by an independent processor; adopting an adjacent frame image difference method to remove stars in the observation image and extract space target star images; then adopting a minimum circumscribed rectangle to describe the length and angle of the elongated star image, constructing a degenerate model, performing motion blurred image restoration on the elongated star image, adopting a mutual comparison method to remove error points with large errors and completing space target track correlation; and finally completing accurate astronomical positioning of the elongated star image. The method greatly improves the processing efficiency, and the previously discarded elongated star images are fully utilized again.

Description

Technical Field

[0001] This invention relates to the field of astronomical observation technology, specifically to a method for astronomical positioning of space targets using elongated celestial phenomena on a large field-of-view space-based monitoring platform. Background Technology

[0002] Currently, my country has over 900 artificial satellites operating simultaneously in orbit, undertaking numerous important tasks such as navigation and positioning, and space science research. These application satellites are closely related to the national economy and defense construction; damage to them would have a significant social and economic impact, even jeopardizing national security. Therefore, effective means are needed to achieve precise detection and high-precision positioning of space targets, enabling on-orbit satellites to effectively avoid collisions with space debris or other spacecraft. Compared to traditional ground-based observation platforms, space-based surveillance platforms offer a wider field of view and all-weather observation capabilities, meeting the needs for continuous all-weather monitoring and precise feature measurement of artificial satellites and space debris. However, when space-based surveillance platforms are in orbit, their monitoring of space targets is in a dynamic capture scenario. Especially when the space-based surveillance platform and the space target are at their closest intersection along their respective orbits, the apparent motion velocity of the observed space target suddenly increases, causing the target's image to appear elongated after camera exposure, which is highly detrimental to high-precision positioning of the space target.

[0003] Traditional observation platforms typically employ centroid reconstruction when dealing with elongated celestial phenomena. This involves defining a boundary at the elongated celestial location, locating the geometric centroid of that boundary, and then reconstructing the centroid of the elongated celestial object. However, this method is not highly accurate, especially when processing sequential images, as the reconstruction results of elongated celestial phenomena from frame to frame are unrelated, resulting in significant random errors. It is more suitable for applications where ground-based observation platforms perform observations first and then process the data.

[0004] In summary, the rapid movement of the carrier satellite along with the space-based space target monitoring platform causes the images of space targets to be stretched and blurred, especially when the carrier satellite is close to the space target, this stretching and blurring phenomenon is more severe. This results in a large error in the extraction of the centroid of the target stars, seriously affecting the accuracy of astronomical positioning and precise orbit determination.

[0005] For space-based surveillance platforms, the problem of elongated celestial images in sequence images can be addressed by establishing a fuzzy kernel function and iteratively training it. Based on the point spread function of the celestial image, the degree and direction of the elongation can be accurately predicted, thus more precisely reconstructing the centroid of the elongated celestial image. This provides more accurate and reliable data support for the subsequent astronomical positioning and precise orbit determination of space targets. Therefore, to improve the space situational awareness capability of space-based surveillance platforms, this invention provides an astronomical positioning method for elongated celestial images of space targets on large-field-of-view space-based surveillance platforms. This method can eliminate or mitigate the interference factors of this elongation phenomenon, which is crucial for high-precision orbit determination, tracking, and cataloging of space targets. Summary of the Invention

[0006] To address the problems of inaccurate centering of extended star patterns and low positioning accuracy in existing monitoring platforms, which prevents target association and precise orbit determination, thus abandoning the use of extended star patterns, this invention provides a space target extended star pattern astronomical positioning method for a large field-of-view space-based monitoring platform.

[0007] A method for astronomical positioning of space targets using elongated celestial phenomena on a large field-of-view space-based surveillance platform, comprising the following steps:

[0008] Step 1: Acquire the observed images;

[0009] Step 2: Perform image preprocessing on the observed images described in Step 1;

[0010] Step 3: Extract the star catalog and perform star map matching to establish the mapping relationship between the celestial coordinates of the calibration stars and the image coordinates, and solve the film constant model;

[0011] Step 4: Use the adjacent frame image difference method to remove stars from the preprocessed observation image in Step 2 and extract the space target star image;

[0012] Step 5: Reconstruct the elongated star image of the extracted space target to obtain the centroid coordinates of the reconstructed image; the specific process is as follows:

[0013] Step 51: Use the minimum bounding rectangle to describe the length L and the elongation angle θ of the elongated spatial target star image, and construct a degradation model for motion blur that causes astronomical image quality degradation.

[0014] Step 52: Based on the degradation model, perform motion blur image restoration on the elongated space target constellation to obtain the image coordinates of the restored space target;

[0015] Step 6: Use the cross-comparison method to remove erroneous points and realize the trajectory association of space target stars in the restored image;

[0016] Step 7: Perform precise astronomical positioning of the space target constellations that have completed the track association.

[0017] The beneficial effects of this invention are:

[0018] 1. In the method of this invention, a degradation model of astronomical image deterioration caused by motion blur is used to restore the elongated star image of space target through iterative calculation of the fuzzy kernel function. This method is particularly suitable for application scenarios of space-based monitoring platforms processing sequential images. By repeatedly training the fuzzy kernel function, the direction and length of the point spread function of the elongated star image are estimated by maximum likelihood, thereby realizing the restoration of the elongated and blurred star image and extracting accurate centroid coordinate data, thus achieving high-precision astronomical positioning and precise orbit determination of space target.

[0019] 2. The space-based monitoring platform in the method of this invention has advantages such as being unaffected by weather, continuous 24-hour observation, and high mobility of observation position. However, the observation platform operates in a near-Earth orbit with a high speed, and its relative speed with the observed space target is large, resulting in a large number of elongated space targets in the space-based monitoring platform. This invention aims to solve the problem of the inability to perform target association and precise orbit determination due to the low astronomical positioning accuracy of the space-based monitoring platform for elongated space targets, so that a large number of previously discarded elongated star images can be fully utilized again.

[0020] 3. The method of the present invention adopts a multi-core parallel computing scheme with a field of view to monitor space targets, which greatly improves the processing efficiency and is particularly suitable for space-based observation platforms for monitoring space targets with a large field of view.

[0021] 4. The root mean square error (RMS value) of the astronomical positioning of the reconstructed celestial constellation of the extended space target using the method of the present invention is much smaller than that of the traditional astronomical positioning method of reconstructing the centroid of the space target using the centroid when dealing with the problem of extended celestial constellations. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the 16-segment observation image in the space target elongated astronomical positioning method of a large field-of-view space-based monitoring platform according to the present invention.

[0023] Figure 2 This is a schematic diagram of the differential image method for adjacent frames; where the stars numbered 1 to 9 are stars distributed sequentially in the sky, and the stars numbered A and B are space targets.

[0024] Figure 3 An elongated celestial diagram describing a spatial target using the minimum bounding rectangle;

[0025] Figure 4 A schematic diagram showing the length L of a space target constellation and the elongation angle θ.

[0026] Figure 5Degradation Models of Motion Blur Degrading Astronomical Images

[0027] Figure 6 A schematic diagram of removing stars from an image and extracting space targets using the image difference method between adjacent frames; where (a) is a schematic diagram of the previous frame image and (b) is a schematic diagram of the next frame image.

[0028] Figure 7 Schematic diagram of the principle of space target trajectory association;

[0029] Figure 8 The star image restoration effect of the stretched space target sequence image; where (a) is the stretched space target star image sequence (frames 1-5), and (b) is the restored space target star image sequence (frames 1-5). Detailed Implementation

[0030] Specific Implementation Method 1: Combination Figures 1 to 8 This embodiment describes a method for astronomical positioning of space targets using elongated celestial phenomena on a large field-of-view space-based surveillance platform. This method first divides the captured large field-of-view image into 16 sub-fields of view, such as... Figure 1 As shown, with the increase in the aperture and field of view of the observation telescope, the number of stars in the field of view also increases dramatically. Therefore, the observation image is divided into 16 sub-fields of view. The space-based monitoring platform carries an embedded multi-core parallel processor. Each sub-field of view is an independent field of view and is calculated by an independent processor, improving processing efficiency and meeting the real-time processing requirements of astronomical positioning. Secondly, the adjacent frame image difference method is used to remove stars from the observation image and extract the space target stars. Then, the length and angle of the elongated star image are described by the minimum bounding rectangle, a fuzzy kernel function is defined, and the Lucy-Richardon algorithm is used to restore the motion blur image of the elongated star image. Then, the cross-comparison method is used to remove error points with large errors and complete the track association of space targets. Finally, accurate astronomical positioning of the elongated star image is achieved. The specific process of this implementation is as follows:

[0031] Step 1. Acquire the observed image; the specific process is as follows:

[0032] Step 1-1. Turn on the telescope of the space-based monitoring platform and record the instantaneous root of orbit, attitude parameters, and guidance data (A, H) fed back from the telescope motor encoder, where A is the azimuth angle of the space-based monitoring platform and H is the elevation angle of the space-based monitoring platform. Calculate the telescope's pointing direction in celestial coordinates, i.e., the center pointing direction of the telescope (C_Ra, C_Dec). Set the camera's exposure time T and shooting time t to obtain a series of images as the observation images.

[0033] Steps 1-2. Divide the observed image into 16 sub-fields of view, and calculate and record the center direction (C) of each sub-field of view based on the center direction of the telescope. sub _Ra,C sub _Dec), where sub is the sub-field number and also the core number of the multi-core parallel processor, sub = 0 ~ 15.

[0034] Step 2. Image preprocessing;

[0035] The observed images are preprocessed, including thresholding, dilation, and contour extraction. The centroids of the extracted stars are calculated to determine their image coordinates (x, y). i ,y i (i = 1 to N, where N is the total number of stars). According to the field-of-view division rules, the image coordinates of the stars in each sub-field of view are (x... subi ,y subi ),sub=0~15.

[0036] Step 3. Extract the star catalog; perform star chart matching to establish a film constant model;

[0037] Based on the telescope's center direction (C_Ra, C_Dec), the celestial coordinates of the photographed sky region are calculated using methods of spherical astronomy. Then, combined with the telescope's field of view, a suitable magnitude (e.g., magnitude 8–12) calibration star group (α) within the photographed sky region is extracted from a standard star catalog (e.g., Tycho 2 catalog). l ,δ l (l=1~N) st N st (where α is the total number of stars extracted from the standard star catalog), then the calibration star group in the standard star catalog corresponding to the sub-field of view is (α). subl ,δ subl ).

[0038] In this embodiment, the specific process of establishing the film constant model through star map matching is as follows:

[0039] Step 3-1. Set the image coordinates (x, y) of the stars in the first frame. i ,y i ) and calibration star group (α l ,δ l The triangle matching subgraph isomorphism algorithm is used to perform star map matching, and the star map matching yields the center position coordinates (α0, δ0) of the first frame image.

[0040] Step 3-2. Solve for the extracted calibration star group (α l ,δ l The ideal coordinates of () are calculated as follows:

[0041]

[0042] Step 3-3. Image coordinates of the stars (x i ,y i ) and ideal coordinates (ζ) i ,η i There is also a mapping relationship, which can be described using a six-constant film model:

[0043]

[0044] Where A, B, C, D, E, and F are film constants, and Ω is the ratio of the camera's pixel size to the telescope's focal length; by combining equations (1) and (2), the film constants can be solved by fitting the least squares method, thereby establishing the mapping relationship between the calibrated celestial sphere coordinates and the image coordinates.

[0045] Steps 3-4. Set the image center coordinates (x, y) of each sub-field of view. C ,y C ) sub Substituting equations (1) and (2), we can obtain the center position coordinates (α) of each sub-field in the current frame. sub0 ,δ sub0 Thus, each sub-field of view can then use an independent processor (core) to perform star map matching and film constant calculation within its respective sub-field of view.

[0046] Steps 3-5. As the sequence of images is captured, the center position coordinates (α) of each sub-field of view in the current frame are determined. sub0 ,δ sub0 The coordinates of the celestial sphere (α) at ​​the center position of the previous frame can be used to determine the coordinates of the celestial sphere. sub0 ,δ sub0 The film constant is continuously updated and corrected by combining the shooting time t and the telescope guidance data (A,H).

[0047] Step 4. Use the adjacent frame image difference method to separate stars and space target constellations from the preprocessed observation images.

[0048] In this embodiment, the adjacent frame image difference method is used to remove stars from the observed image and extract the space target constellations; the observation mode of the space-based monitoring platform is set to stellar rate mode, so the stellar motion in the telescope is diurnal apparent motion, that is, the stars in the later frame will move a certain distance to the right relative to the previous frame, and this distance can be calculated by exposure time and movement speed. Figure 6As shown, (a) is a schematic diagram of the previous frame image, and (b) is a schematic diagram of the next frame image. All stars in the previous frame are shifted to the right to obtain new coordinates. The coordinates of the stars in the two frames are compared. If the difference in coordinate distance is less than the matching threshold σ, they are identified as stars and their image coordinates are recorded; otherwise, they are identified as suspected space targets. The specific process is as follows:

[0049] Step 4-1. Settings Let the coordinates of the centroid of the star in the k-th frame of the observed image be set. Let be the coordinates of the centroid of the star in the (k-1)th frame of the observed image. Then the distance between the star in the k-th frame and the (k-1)th frame of the observed image is:

[0050]

[0051] Step 4-2. Based on the characteristic of stars moving at a constant speed on the camera target surface, determine the distance of star movement between two consecutive frames of images. They are equidistant, while the movement distance of the space target stars... Distance from stars There is a significant difference in comparison. The matching coefficient D can be calculated and the following judgment can be made:

[0052]

[0053] Where σ is the matching threshold, which is related to the observation accuracy, and is generally taken as 1 to 2 pixels. When D = 0, it indicates that... and It is a star. When D=1, it means... and This is suspected to be a space object. The orbital distance of space object celestial bodies differs significantly from that of stars, such as... Figure 2 As shown, after translation, the positions of the space target stars in the (k-1)th and kth frame images will be offset. By operating the kth and k+1th frame images in the same way, the suspected space targets can be selected.

[0054] Step 5. Reconstruct the elongated constellation of the space target to obtain the image coordinates (x, y) of the reconstructed space target. ta ,y ta ), ta=1~N ta N ta The total number of space targets; the specific process is as follows:

[0055] Step 5-1. Use the minimum bounding rectangle to describe the elongated constellation of the spatial target. Set the coordinates of the four vertices of the bounding rectangle as A(x) and A(x). a ,y a ),B(x b ,y b ),C(xc ,y c ),D(x d ,y d ),like Figure 3 As shown. The central axis EF of the rectangle intersects AB and CD at E(x). e ,y e ),F(x f ,y f If the coordinates of the endpoint of the major axis EF of the rectangle are:

[0056]

[0057] like Figure 4 As shown, the length L of the elongated spatial target constellation and the elongation angle θ are calculated as follows:

[0058]

[0059]

[0060] In the formula, n is the number of pixels occupied by the elongated spatial target star image in the x-axis direction, and m is the number of pixels occupied by the elongated star image in the y-axis direction.

[0061] For astronomical images that suffer from elongated and blurred star formations due to moving objects, image deconvolution can be used to restore the source image. A degradation model for astronomical image degradation caused by motion blur can be constructed, such as... Figure 5 As shown, the ideal, clear image h(x,y) is degraded into a motion-stretched blurred image b(x,y) under the influence of the blur kernel function k(x,y) and noise γ(x,y). The mathematical expression of the degradation model is as follows:

[0062] b(x,y)=h(x,y)*k(x,y)+γ(x,y)(8)

[0063] Where b(x,y) is the elongated motion-blurred image of the spatial target captured by the camera, h(x,y) is the ideal sharp image, k(x,y) is the blur kernel function, γ(x,y) is additive noise, and (x,y) are the image coordinates. During the exposure time T, the spatial target relative to the space-based monitoring platform can be considered to be moving at a constant speed. Its motion in the x and y directions of the camera is represented by time functions x0(t) and y0(t), respectively. Therefore, during the exposure time T, the motion-blurred image b(x,y) can be considered as the average gray level of the spatial target along the blur direction, which can be specifically expressed as:

[0064]

[0065] By performing a Fourier transform on equation (9) and changing the order of integration, we can obtain:

[0066]

[0067] The degenerate model, after undergoing a Fourier transform, transforms from a spatial domain convolution to a frequency domain convolution according to the convolution theorem:

[0068] B(u,v)=H(u,v)*K(u,v)+Γ(u,v)(11)

[0069] Where B(u,v) and H(u,v) represent the Fourier transforms of the blurred source image and the clear image, respectively, Γ is the Fourier transform of the noise, and K(u,v) is the blur kernel function in the frequency domain. Comparing equations (10) and (11), the blur kernel function is specifically expressed as:

[0070]

[0071] Transforming the K(u,v) fuzzy kernel function to the spatial domain yields the linear uniform velocity fuzzy kernel function, defined as:

[0072]

[0073] As shown in equation (13), the blur kernel function is only related to the stretching angle and the stretching length. Therefore, by obtaining these two parameters, the blur kernel function can be obtained, and thus the clear restored image can be derived.

[0074] Step 5-2. The Lucy-Richardon algorithm is one of the most widely used image restoration techniques. It uses an iterative method and an algorithm to accelerate convergence to restore the image. The Lucy-Richardon algorithm assumes that the pixels in the image follow a Poisson distribution. Based on a Bayesian conditional probability model, it uses maximum likelihood estimation to iteratively calculate the original clear image from the degraded image. Its iterative equation is:

[0075]

[0076] Among them, f w (x, y) represents the w-th iteration of the process of restoring the initial image to a clear image, f w+1 (x,y) is the restoration result of the (w+1)th iteration.

[0077] f can be made 0 (x,y) = b(x,y) is used as the initial condition for iteration. As w increases, f w+1 (x,y) will gradually converge to the ideal clear image h(x,y) with probability, and the centroid coordinates (x,y) of the spatial target reconstruction image will be extracted. ta ,y ta ).

[0078] Step 6. Track association;

[0079] The reconstructed images are correlated with the celestial targets in space by tracking their flight paths, in order to distinguish different celestial targets and eliminate redundant and erroneous data. For example... Figure 7 As shown, Figure 7 This diagram illustrates the principle of multi-space target trajectory association. The coordinates of each suspected target in two consecutive frames are correlated, and data such as x-displacement, y-displacement, x-velocity, and y-velocity are calculated to form a list of suspected targets. The data in the list is filtered, and erroneous data with excessively high movement velocities are removed by using x-displacement and y-displacement thresholds; for example, g-3 in the diagram might be eliminated. The data retained in the list will be iteratively incorporated into the processing of the third frame and subsequent frames.

[0080] The data in the suspicious target list is associated with suspicious targets in the third frame and subsequent frames. Since the space target moves in a circular orbit for a short period of time, and the change in the target's apparent velocity is very small, the space target can be considered to be moving in a uniform linear motion within the exposure time. By comparing the xy displacement and xy velocity vectors, redundant targets in the suspicious target list are removed again, completing the target track association.

[0081] In this embodiment, during the process of associating the tracks of space targets in the reconstructed elongated star image, it is inevitable that there will be missed detections and false detections. Therefore, a cross-comparison method is used to eliminate erroneous points. The specific process is as follows:

[0082] Step 6-1. Let the measurement arc duration of the space target be τ. First, determine three verification points, which are more than τ / 4 apart. Set the three verification points as S1, S2, and S3.

[0083] The apparent trajectory of a spatial target in an image is generally obtained using polynomial fitting. Since the measured arc segment of a spatial target is relatively short, a quadratic polynomial is used to describe the target's trajectory.

[0084]

[0085] Step 6-2. Substitute the three authentication points into the above formula to solve for a0, a1, a2 and b0, b1, b2. Using these six coefficients, calculate the coordinates (x, y, t) of the spatial target across all frames based on the shooting time t. cal ,y cal ), and compared these values ​​with the centroid coordinates (x, y) of the reconstructed image of the spatial target. ta ,y ta For comparison, the difference must satisfy:

[0086]

[0087] Where ε is the rejection threshold, which is related to the observation accuracy, and is generally taken as 1 to 2 pixels. If most of the restored spatial targets meet the conditions of equation (16), it means that the three verification points are correct, and the centroids of the restored spatial targets that do not meet the conditions of equation (16) are rejected. If the centroid coordinates of most targets do not meet the conditions of equation (16), it means that there is a problem with these three points, and the verification points need to be replaced until the conditions are met.

[0088] Step 7. Astronomical positioning;

[0089] In this embodiment, the astronomical positioning of the space target is a positioning method that calculates the celestial coordinates of the space target based on the relative mapping relationship between the position of the space target in the image and the calibration star. It is achieved by solving the film constant model by establishing the relative mapping relationship between the celestial coordinates of the calibration star and the image coordinates of the space target.

[0090] Step 7-1. Using the established calibration star groups (α) of each sub-field of view subl ,δ subl ) and the image coordinates (x) of each sub-field of view star image subi ,y subi The mapping relationship between the two and the film constant model:

[0091]

[0092] Where (α) sub0 ,δ sub0 ) represents the celestial coordinates of the center of each sub-field of view, sub = 0 ~ 15, sub is the sub-field of view number, and also the core number of the multi-core parallel processor.

[0093]

[0094] By combining equations (17) and (18), the film constant (A) of each sub-field of view is solved using the least squares method. sub ~F sub ).

[0095]

[0096] Step 7-2. Calculate the centroid coordinates (x, y) of the restored spatial target image. ta ,y ta Substituting into formula (19), we can obtain the ideal coordinates (ζ) of the space target constellation. ta ,η ta ), and substitute it into the following formula:

[0097]

[0098] After rearranging equation (20), the celestial coordinates (α) of the target star in space can be obtained. ta ,δta ):

[0099]

[0100] This completes the astronomical positioning of the elongated constellations of space targets on the wide-field-of-view space-based surveillance platform.

[0101] Specific Implementation Method Two: Combination Figure 8 This embodiment describes a verification example of the astronomical positioning method for elongated celestial phenomena of a space target on a large field-of-view space-based surveillance platform described in Specific Embodiment 1. To verify the restoration effect and astronomical positioning accuracy of the elongated celestial phenomena of a space target using the astronomical positioning method for a large field-of-view space-based surveillance platform described in Specific Embodiment 1, the method described in this embodiment is used to restore a sequence of images of an elongated laser ranging star (space target) and to conduct an astronomical positioning accuracy verification experiment. The two-row roots (TLE) of the laser ranging star are calculated to obtain its celestial position. Because the orbit determination data of the laser ranging star has high accuracy, it can be used as the true value for the experiment; this value is denoted as the C value. The target's centroid is restored using both the traditional centroid algorithm and the fuzzy kernel function method described in this embodiment, and the target is astronomically positioned. The observed value is denoted as the O value. The positioning error OC value (i.e., observed value - true value, where (OC)) is recorded for each frame of the sequence of images. α The direction of right ascension is (OC). δ (in the direction of declination), and calculate the root mean square error of the astronomical positioning of the target:

[0102]

[0103] In the formula, N f This refers to the number of frames in the image sequence. The star reconstruction effect of the stretched spatial target image sequence is as follows: Figure 8 As shown, the root mean square error of astronomical positioning is calculated as shown in Table 1. Table 1 shows the astronomical positioning error of the method of the present invention for elongated target constellations.

[0104] Table 1

[0105]

[0106] Table 2 shows the astronomical positioning error of the elongated target constellation using the centroid method.

[0107] Table 2

[0108]

[0109] As can be seen from the comparison of Tables 1 and 2 above, the root mean square error (RMS value) of the astronomical positioning of the elongated space target constellation after reconstruction by the method described in this embodiment is much smaller than that of the traditional astronomical positioning method of reconstructing the centroid of the space target using the centroid.

[0110] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0111] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A space-based surveillance platform's space object elongated star image celestial positioning method of large field of view, characterized by: The method is realized by the following steps: Step one, obtaining observation images; Step two, image preprocessing on the observation images in step one; Step three, extracting a star catalog and performing star map matching to establish a mapping relationship between the celestial coordinates of the calibration stars and the image coordinates, and to solve a film constant model; Step four, using adjacent frame image difference method to remove the constant stars in the observation images preprocessed in step two and extract spatial target star images; Step five, restoring the elongated spatial target star images in the extracted spatial target star images to obtain the centroid coordinates of the spatial target restored images; the specific process is as follows: Step five one, using the minimum circumscribed rectangle to describe the length L and the elongation angle θ of the elongated spatial target star image, and constructing a degradation model of astronomical image degradation caused by motion blur; the specific process is as follows: Set the four vertex coordinates of the circumscribed rectangle as A(x a ,y a ), B(x b ,y b ), C(x c ,y c ), D(x d ,y d ), the long axis center axis of the rectangle intersects AB and CD at E(x e ,y e ), F(x f ,y f ), then the endpoint coordinates of the center axis EF are: The formulas of the length L and the elongation angle θ of the elongated spatial target star image are as follows: In the formula, n is the number of pixels of the elongated spatial target star image in the x-axis direction, and m is the number of pixels of the elongated spatial target star image in the y-axis direction; The degradation model of astronomical image degradation caused by motion blur is expressed by the following formula: b(x, y) = h(x, y) * k(x, y) + γ(x, y) In the formula, b(x, y) is a spatial target motion elongation blurred image shot by a camera, h(x, y) is an ideal clear image, k(x, y) is a blur kernel function, and γ(x, y) is an additive noise; Step five two, according to the degradation model, performing motion blurred image restoration on the elongated spatial target star image to obtain the image coordinates of the restored spatial target; Step six, using mutual comparison method to remove error points to realize track association of the spatial target star images in the restored images; Step seven, performing accurate astronomical positioning on the spatial target star images with completed track association.

2. The method of claim 1, wherein the method is a space-based surveillance platform's space object elongated image celestial positioning method for a large field of view. The specific process of step one is as follows: Step one one, calculating the pointing of the telescope in the celestial coordinates, i.e., the center pointing (C_Ra, C_Dec) of the telescope, by recording the orbital instantaneous root, attitude parameters of the space-based monitoring platform, and guide data fed back by the motor encoder of the telescope; Step one two, setting the exposure time T and the shooting time t of the camera to obtain a group of sequence images as observation images; Step one three, divide the observation image into 16 sub-fields averagely, calculate the center pointing of each sub-field according to the center pointing of the telescope (C sub _Ra,C sub _Dec), wherein sub is the sub-field number; The sub-field number sub simultaneously serves as the number of the corresponding core of the embedded multi-core parallel processor carried by the space-based monitoring platform, and sub = 0 ~ 15.

3. The method of claim 1, wherein the method is a space-based surveillance platform's space object elongated star image celestial positioning method for a large field of view. In step three, extracting the star table: according to the center pointing (C_Ra, C_Dec) of the telescope, the celestial coordinates of the photographed sky area are calculated by using the calculation method of spherical astronomy, and the calibrated star group (α l ,δ l ) of the photographed sky area is extracted from the standard star table according to the field of view of the telescope, l = 1 ~ N st , N st is the total number of stars extracted from the standard star table, and the calibrated star group corresponding to the sub-field of view in the standard star table is (α subl ,δ subl ).

4. The method of claim 3, wherein the method is a space-based surveillance platform's space object elongated star image celestial positioning method for a large field of view. In step three, the specific process of establishing the film constant model according to the extracted star catalog is as follows: Step three one, the image coordinates of star image of first frame image and the scaling star group (α l ,δ l ) are matched by triangle matching subgraph isomorphism algorithm, and the center position coordinates (α0,δ0) of first frame image are obtained by star map matching; Step three two, according to the central position coordinates (a0, d0) of the first frame image, solve the ideal coordinates (ζ i ,η i ) of the calibration star group (a l , d l ), and adopt the six-constant film model to describe the mapping relationship between the image coordinates (x i , y i ) and the ideal coordinates (ζ i ,η i ) of the star image, and then establish the mapping relationship between the calibration star celestial coordinates and the image coordinates; Step 33: Based on the mapping relationship obtained in Step 32, obtain the center position coordinates (α) of each sub-field of view in the current frame. sub0 ,δ sub0 ); the center position (α) of each sub-field of view in the current frame. sub0 ,δ sub0 ) from the center position of the previous frame (α) sub0 ,δ sub0 The film constant is updated and corrected based on the shooting time t and the telescope's guidance data.

5. The method of claim 4, wherein the method is a space-based surveillance platform's space object elongated image celestial positioning method for a large field of view. In step 3.2, the calibration star group (α) is solved. l ,δ l Ideal coordinates (ζ) i ,η i The formula is as follows: The mapping relationship between the image coordinates (x i ,y i ) of the star image and the ideal coordinates (ζ i ,η i ) is described by a six-constant film model as follows: Where A, B, C, D, E, F are the plate constants, Ω is the ratio of the pixel size of the camera to the focal length of the telescope, and the ideal coordinates (ζ i ,η i ) are calculated by the mapping relationship formula and the ideal coordinate formula. The plate constants are solved by the least square method, and the mapping relationship between the calibrated celestial coordinates and the image coordinates is established.

6. The method of claim 1, wherein: the space-based monitoring platform is a large field-of-view space-based monitoring platform; and the space object is a space object that is elongated in space. In step four, using adjacent frame image difference method to remove the constant stars in the observation images and extract spatial target star images, setting a configuration coefficient D, and expressing it by the following formula: In the formula, σ is a matching threshold value, is the star image motion distance of the front and rear two frame sequence images, is the motion distance of the space target star image; When D = 0, it indicates that the star images of the kth observation image and the star images of the (k-1)th observation image are both constant star images; when D = 1, it indicates that the star images of the kth observation image and the star images of the (k-1)th observation image are suspected to be spatial targets.

7. The method of claim 1, wherein: The specific process of step five two is as follows: ​ The degradation model is subjected to Fourier transform and convolution theorem to convert the convolution in the spatial domain into the convolution in the frequency domain: B(u, v) = H(u, v) * K(u, v) + Γ(u, v) In the formula, B(u, v) is the Fourier transform of the blurred source image in the frequency domain, H(u, v) is the Fourier transform of the clear image in the frequency domain, Γ is the Fourier transform of the noise, and K(u, v) is the blur kernel function in the frequency domain; The blur kernel function K(u, v) is converted to the spatial domain to obtain a linear uniform blur kernel function as follows: The Lucy-Richardon algorithm is used to restore the motion blurred image and extract the centroid coordinates of the restored space target image.

8. The method of claim 1, wherein the method is a space-based surveillance platform's space object elongated star image celestial positioning method for large field of view. In step six, the mutual comparison method is used to remove error points, and the specific process is as follows: In step six, the measurement arc length of the space target star image is set as τ, three identification points are determined, and the distance between the three identification points is set as τ / 4. A quadratic polynomial is used to describe the trajectory of the space target star image, and the following formula is used to represent the trajectory: Step six two, the three authentication point above track formula, solve the coefficient a0, a1, a2 and b0, b1, b2, according to the shooting time t value calculation space target star image in all frame coordinate (x cal ,y cal ), with the calculated all frame coordinate and space target restoration image centroid coordinate (x ta ,y ta ) comparison, use the following formula is expressed as: In the formula, ε is a removal threshold, if the difference satisfies the formula, the three identification points are correct, and the restored space target centroid that does not satisfy the condition is removed; if the restored space target centroid coordinates do not satisfy the formula, the three identification points are incorrect, and the identification points need to be replaced until the condition is satisfied.

9. The method of claim 1, wherein: The specific process of step seven is as follows: ​ Step seven, according to the mapping relationship between the established calibration star celestial coordinates and the image coordinates of the space target, the film constant model is solved, the mapping relationship between the calibration star group (α subl ,δ subl ) of each sub field and the image coordinates (x subi ,y subi ) of each sub field star image is established, and the film constant model is established: Step seven two, the image coordinates (x ta ,y ta ) of the space target are obtained through the ideal coordinates (ζ ta ,η ta ) of the space target star image by the film constant model, and finally the celestial coordinates (α ta ,δ ta ) of the space target star image are obtained, so as to realize the astronomical positioning of the space target elongated star image of the large field of view space-based monitoring platform.

Images