Space-based wide field of view optical observation space multi-target identification and identification method

Abstract

This invention relates to a space-based wide-field-of-view optical observation method for identifying and verifying multiple targets in space, belonging to the field of aerospace engineering and space target optical monitoring technology. It calculates the target observation direction through TLE cataloging prediction and optical time-of-arrival correction; integrates the film parameter model of the space-based measured image to map the target observation direction to a theoretical position sequence on the image plane; interpolates and extracts the theoretical apparent motion velocity and direction of the target to construct a priori motion database; preprocesses the space-based observation image and performs star detection, and designs kinematic filtering and track association to perform track association and complete trajectory extraction on the extracted moving targets, eliminating false motion points; finally, it performs multi-dimensional matching and comparison of the actual motion velocity and direction characteristics of the stars on the image plane with the preset target prediction data to complete the accurate identification and verification of space targets. This achieves efficient and high-precision identification and verification of multiple targets within a wide field of view, effectively improving the anti-interference capability and correlation of target identification in complex scenes.

Description

Technical Field

[0001] This invention relates to the field of aerospace engineering and space target optical monitoring technology, specifically to a space-based large field-of-view optical observation method for identifying and verifying multiple targets in space. Background Technology

[0002] With the surge in the number of near-Earth space targets, space-based wide-field optical observation systems have become core equipment for space target monitoring, orbit cataloging, and collision warning due to their advantages such as wide coverage, high observation timeliness, and lack of ground environment limitations. Space multi-target identification and verification, as a crucial step in space-based observation data processing, directly determines the reliability of subsequent target cataloging and orbit analysis, and is an important foundation for ensuring the safety of space missions and maintaining order in the space environment.

[0003] Existing multi-target identification and authentication methods in space suffer from numerous technical bottlenecks in space-based large field-of-view scenarios, making it difficult to meet the application requirements of high precision and high efficiency. Firstly, insufficient coordinate system consistency and astronomical effect correction mean that traditional methods often suffer from inconsistencies between the predicted target position and the satellite platform coordinate system. Furthermore, simplified error corrections, such as optical travel time, lead to low accuracy in calculating the observed target pointing and image plane position, creating potential errors for subsequent matching. Secondly, poor correlation between image plane position and apparent motion parameter solutions means that most methods fail to construct accurate film parameter models based on measured reference star authentication results, and the interpolation calculations for apparent motion velocity and direction are inaccurate. First, the accuracy of the matching is insufficient, making it impossible to accurately characterize the motion features of the target. Second, the screening of moving targets and the correlation of tracks are not good. Existing methods have insufficient distinguishability between stars and space targets, the elimination logic based on kinematic features is imperfect, and the correlation of short arc tracks is easily affected by noise, making it difficult to form stable target tracks. Third, the accuracy of matching and verification is limited. Traditional matching often relies on a single position parameter and lacks collaborative verification with multi-vector assumptions and sliding window measured data. Angle error and position error are not jointly evaluated, which can easily lead to misjudgment or omission of matching. Moreover, the output parameters are incomplete and cannot provide comprehensive support for subsequent data processing.

[0004] Furthermore, space-based wide-field-of-view observation scenarios involve high target density, complex background interference, and large amounts of observation data. Existing methods struggle to balance accuracy and efficiency in identification and verification, and cannot meet the real-time processing requirements of batch targets, thus limiting the overall operational efficiency of space-based optical observation systems. Therefore, developing a space-based wide-field-of-view optical observation method for multi-target identification and verification that balances coordinate system uniformity, motion characteristic accuracy, and matching reliability has become an urgent need in the field of space monitoring technology. Summary of the Invention

[0005] The purpose of this invention is to provide a space-based large field-of-view optical observation method for identifying and verifying multiple targets in space, in order to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: a space-based large field-of-view optical observation method for identifying and verifying multiple targets in space, comprising the following steps:

[0007] Step 1: Calculate the target observation pointing direction using TLE cataloging prediction and optical time correction; integrate the film parameter model of the space-based measured image to map the target observation pointing direction into the theoretical position sequence of the image plane; and interpolate to extract the theoretical apparent motion velocity and direction of the target to construct a priori motion library.

[0008] Step 2: Preprocess the space-based observation images and perform star detection to form a set of measured star points; and design kinematic filtering and trajectory association to extract the measured target motion parameters;

[0009] Step 3: Generate multiple predicted trajectory hypotheses based on the theoretical motion features in the prior motion library, and combine them with the extracted local measured target motion parameters. Then, directly use pixel velocity and direction to perform nearest neighbor scoring on the image plane, and take the hypothesis corresponding to the minimum score as the best matching trajectory.

[0010] Step 4: Combine matching scores and confidence tests to output the identification and authentication information of spatial targets.

[0011] Preferably, in step one, based on the identification results of reference stars in the space-based measured image, the film parameter model is calculated, and after geometric distortion correction and poor astronomical effect correction, the film model parameters are described by 4 or 6 parameters; the image plane coordinates of the space target are calculated based on the celestial coordinate information obtained from the space target prediction and the film parameter model.

[0012] Preferably, in step one, cubic spline interpolation is performed on the target prediction position sequence based on the start and end times of the observed arc segment to obtain the image plane position, motion velocity, and acceleration parameters of the spatial target at the midpoint of the prediction arc segment; at the same time, a pixel velocity direction angle is generated for each visible spatial target.

[0013] Preferably, in step two, preprocessing and star detection include:

[0014] An adaptive median filter is used to preserve the core features of the stars; and a grayscale stretching algorithm is used to adjust the dynamic range of the image and enhance the grayscale contrast between the stars and the background.

[0015] The background mean and sigma value of the image are calculated using the 3sigma iterative method. The image is then binarized using "mean + 3σ" as the threshold. Pixels below the threshold are set to zero, while those above the threshold are retained, thus generating a binarized image.

[0016] Based on the generated binary image, the eight-neighbor connected component analysis algorithm is used to traverse the image pixels, mark the star pixel regions that meet the connectivity conditions one by one, assign each independent star region a unique identifier ID, and at the same time remove connected regions with an area smaller than a preset threshold.

[0017] For each valid connected component, the pixel group in the image is transformed into a quantifiable feature vector based on the centroid method.

[0018] Preferably, in step two, the kinematic filtering includes:

[0019] For two adjacent frames of candidate star points, establish nearest neighbor matching to determine the threshold circle radius;

[0020] For successfully matched points, calculate the image plane velocity vector to obtain the average motion characteristics of the stars. and speed standard deviation ;

[0021] For all candidate points in the current frame, if:

[0022] ;

[0023] and ;

[0024] If it is identified as a stellar constellation, it is removed from the moving target candidate pool; the remaining moving target candidate points are retained; among them, It is the velocity vector of the j-th candidate star point detected in the image.

[0025] Preferably, in step two, the track association includes:

[0026] For the retained moving target candidate points, a uniform linear model is adopted in the three-dimensional space of the image plane, and the RANSAC clustering method is used to find collinear points with a number ≥ 3 and The short arc segment, where N is the number of stars matched;

[0027] Each short arc is assigned a unique track ID, and its average pixel velocity and orientation angle are recorded.

[0028] Starting from the end of the short arc above, extrapolate the predicted position of the next frame along the average velocity; and establish a pixel threshold circle around the predicted position:

[0029] If a detection point exists within the threshold, that point is added to the current track and the average speed is updated; otherwise, a "loss counter" is started, and the track is terminated when the number of consecutive lost frames exceeds K.

[0030] Output the parameters of each successfully associated measured target, including track ID, start-end pixel coordinates, average pixel velocity, heading angle, total number of frames, and pixel residual RMS.

[0031] Preferably, in step three, the predicted velocity at the midpoint of the arc segment is used. Centered on With uncertainty as the boundary, radial-azimuth discretization generates N≥8 motion hypotheses:

[0032] ;

[0033] in , Given by the standard deviation of the predicted arc segment velocity and direction, That is, the i-th velocity is assumed to have a scalar value. and direction Decide, and The value is determined based on the quantity N within the corresponding interval;

[0034] Locally measured target motion parameters include the spatial target image plane position based on the midpoint of the predicted arc segment. Speed ​​of movement , velocity direction and acceleration Forming an "instantaneous prediction motion vector package":

[0035] .

[0036] Preferably, in step three, for each hypothesis i, along Extrapolate ΔT to the next frame:

[0037] ;

[0038] And by comparing with the measured trajectory endpoint, a scoring function is constructed:

[0039] ;

[0040] Among them, weight , , ΔT represents the normalized value of the image face diagonal pixels and the maximum acceleration, and ΔT is the inter-frame time difference.

[0041] The hypothesis corresponding to the minimum score is taken as the best matching trajectory:

[0042] ;

[0043] The subsequent extrapolated point sequence is used as the output of the image plane prediction route after matching.

[0044] Preferably, if the minimum score is still greater than the threshold τ, then linear extrapolation is used to complete the missing points of the measured track, or the image plane hypothesis set is re-initialized to trigger "new track generation".

[0045] Preferably, the identification and authentication information of the space target includes the target's unique ID, image plane coordinate velocity (X,Y,VX,VY) and celestial coordinates (α,δ), matching residual, matching score, and UTC time.

[0046] Beneficial Effects: This invention achieves efficient and high-precision identification and authentication of multiple targets within a large field of view by accurately calculating the apparent position and apparent motion parameters of space targets using astrometric film model solutions and velocity vectors as the core matching basis. Specifically, it eliminates stellar interference based on kinematic characteristics and establishes stable track associations based on the assumption of uniform linear motion in short arc segments, significantly improving the accuracy of moving target selection and track reliability. By combining multi-vector assumptions with sliding window measured data, it achieves optimal predicted trajectory matching through a joint "position + velocity" score, greatly reducing the probability of misjudgment and missed judgment. Simultaneously, it outputs complete results including target number, motion parameters, residual data, and multi-dimensional coordinates, providing comprehensive support for subsequent data processing. Ultimately, it balances the accuracy and efficiency of batch target identification and authentication in space-based large field-of-view scenarios, significantly improving the overall operational efficiency of space-based optical observation systems. It provides reliable technical support for space target cataloging updates, orbit analysis, and collision warning, possessing strong engineering application value. Attached Figure Description

[0047] Figure 1 This is a flowchart of the space-based large field-of-view optical observation multi-target identification and authentication method of the present invention;

[0048] Figure 2 This is a schematic diagram of the result of converting the target prediction result into the image plane in an embodiment of the present invention;

[0049] Figure 3 This is a schematic diagram of the star detection results in an embodiment of the present invention;

[0050] Figure 4 This is a schematic diagram of the target identification and verification results in an embodiment of the present invention. Detailed Implementation

[0051] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe a space-based large field-of-view optical observation method for identifying and verifying multiple targets in space, or several specific implementations of this invention, and does not strictly limit the scope of protection specifically claimed by this invention.

[0052] Example: Figure 1 As shown, a space-based large field-of-view optical observation method for identifying and verifying multiple targets in space includes the following steps:

[0053] S1. Calculate the target observation direction through TLE cataloging prediction and optical time correction;

[0054] S11. Using the two-line root number (TLE) catalog library as the data source, the instantaneous position vector of the target in the geocentric inertial frame (TEME) is predicted by the SGP4 / SDP4 model.

[0055] S12. Convert the TME position to the ICRF inertial frame and unify the time system with the satellite platform ephemeris;

[0056] S13. Solve for the emission time using the optical travel time iterative algorithm to obtain the accurate target observation vector:

[0057] ;

[0058] in It is the position vector at the moment the target signal is emitted. It is the position vector at the time of the platform's observation. It is the time when the target signal is transmitted. It is the moment when the target signal arrives at the platform, i.e., the observation moment.

[0059] S14. Outputs the ICRF coordinate system observation unit vector after optical travel time correction; effectively solves the problems of inconsistent coordinate systems and low accuracy in calculating observation pointing and image plane position caused by the simplification of optical travel time correction in traditional methods.

[0060] S2. Based on the film parameter model of the target observation pointing fusion space-based measured image, the target observation pointing is mapped to a sequence of theoretical image plane positions; specifically:

[0061] S21. Establish a rectangular coordinate system on the film with point C as the origin, and the vertical axis... This is the projection of the declination circle, with the direction of increased declination taken as the positive direction; the horizontal axis... Perpendicular to The axis, taking the direction of increasing right ascension as the positive direction, is called the ideal coordinate system;

[0062] Among them, the coordinates of the stars in the ideal coordinate system Its equatorial coordinates It is a one-to-one correspondence, and the correspondence relationship is as follows:

[0063] ;

[0064] The equatorial coordinates of the point of tangency between the telescope's field of view plane and the celestial sphere;

[0065] The formula for calculating equatorial coordinates from ideal coordinates is:

[0066] ;

[0067] S22. Based on the results of star detection and reference star matching, calculate the film model parameters, and after correction for geometric distortion and poor astronomical effects, describe the film model parameters using 4 or 6 parameters; the calculation formula is as follows:

[0068] .

[0069] in The model parameters to be estimated are... x and y are the celestial coordinates, and i and j represent powers; a mapping relationship from celestial coordinates to image plane coordinates is established through stars.

[0070] S23. Based on the celestial coordinate information obtained from the space target prediction, and using the film parameter model, calculate the image plane coordinates of the space target for subsequent calculations of apparent motion velocity, etc.

[0071] S3. Calculate the position sequence of visible arc segments of spatial targets on the image plane, interpolate to calculate the apparent motion velocity and direction of the spatial targets, and construct a prior motion library. This achieves a precise and coordinated characterization of image plane position and motion feature parameters, referencing... Figure 2 As shown; specifically:

[0072] S31. Based on the start and end times of the observed arc segment, the cubic spline interpolation method is used to interpolate the target prediction position sequence, thereby accurately solving for the image plane position, motion velocity and acceleration parameters of the spatial target at the midpoint of the prediction arc segment.

[0073] S32. Calculate the pixel velocity direction angle of each target in the visible space. This provides a reliable basis for motion characteristics in subsequent target matching and verification.

[0074] S4. Preprocess and detect star phenomena in the space-based observation images to form a set of measured star points, for reference. Figure 3 As shown; specifically:

[0075] S41. Image preprocessing: To address the noise interference, uneven grayscale, and background stray light issues in the observed image, an adaptive median filtering algorithm is used to denoise the image, suppressing random noise and salt-and-pepper noise while preserving the core features of the stars; subsequently, a grayscale stretching algorithm is used to adjust the dynamic range of the image and enhance the grayscale contrast between the stars and the background.

[0076] S42. Calculate the background mean and sigma value of the image using the 3sigma iterative method, and perform binarization processing on the image using the mean + 3sigma (assign 0 to pixels below the mean + 3sigma) to accurately separate the star region from the background region and generate a binarized image.

[0077] S43. Perform connected component labeling. Based on the preprocessed binarized image, the eight-neighbor connected component analysis algorithm is used to traverse the image pixels and mark the star pixel regions that meet the connectivity conditions one by one, assigning each independent star region a unique identifier ID. At the same time, tiny connected regions with an area smaller than a preset threshold are removed to eliminate noise residue and interference from isolated false star points, and to accurately extract the effective star regions.

[0078] S44. For each marked valid connected component, calculate the core feature parameters of the star, including using the robust centroid method to solve for the coordinates (X, Y) of the star center and accurately locate the position of the star on the image plane; count the gray values ​​of all pixels in the connected component, calculate the average gray value, maximum gray value and gray integral of the star, characterize the brightness characteristics of the star, and support subsequent track association and astronomical positioning photometry and other processing.

[0079] S5. Based on kinematic characteristics, stellar images in the observed images are removed. Based on the assumption of uniform linear motion in short arc segments, space target trajectories are correlated, and measured target motion parameters, including the apparent velocity and direction of the trajectory, are extracted. This significantly improves the accuracy of moving target selection and the reliability of the trajectories. Specifically:

[0080] S51. Star image removal: For candidate star point sets between two adjacent frames:

[0081] ;

[0082] Establish nearest neighbor matching, threshold circle radius:

[0083] ;

[0084] For points that are successfully matched, calculate the image plane velocity vector:

[0085] ;

[0086] Determine the average motion characteristics of stars:

[0087] ;

[0088] Where N is the number of stars matched.

[0089] Speed ​​standard deviation:

[0090] ;

[0091] For all candidate points in the current frame, if:

[0092] ;

[0093] and ;

[0094] in, It is the velocity vector of the j-th candidate star point detected in the image.

[0095] If the point is identified as a stellar object, it will be removed from the moving target candidate pool; the remaining points will be retained as space target candidates.

[0096] S52. Track Establishment: For the retained candidate moving targets, a uniform linear model is adopted in the three-dimensional space of the image plane (x,y,t).

[0097] ;

[0098] The RANSAC (RANdom SAmple Consensus) clustering method is used to find short arc segments with ≥3 collinear points and χ² / N<2.0 (normalized chi-square less than 2);

[0099] Each short arc is assigned a unique track ID, and its average pixel velocity and heading angle are recorded:

[0100] ;

[0101] S53, Track Extrapolation: End point of the above short arc Starting from the average velocity, extrapolate the predicted position for the next frame:

[0102] ;

[0103] Where ΔT is the time difference between frames.

[0104] Establish a pixel-threshold circle around the predicted location:

[0105] ;

[0106] If a detection point exists within the threshold, that point is added to the current track and the average speed is updated; otherwise, a "loss counter" is started, and the track is terminated when the number of consecutive lost frames exceeds K.

[0107] S54. Track Output: Outputs the ID of each successfully associated track, start-end pixel coordinates, average pixel velocity, heading angle, total number of frames, and pixel residual RMS for subsequent actual-predicted track matching.

[0108] S6. Multiple predicted trajectory hypotheses are generated based on theoretical motion features from the prior motion database. Combined with extracted local measured target motion parameters, nearest neighbor scoring is performed directly on the image plane using pixel velocity and direction. The predicted trajectory with the smallest "position + velocity" error is selected to complete the matching between the measured pixel sequence and the predicted trajectory on the image plane, significantly reducing the probability of misjudgment and missed judgment. Simultaneously, a complete result including target number, motion parameters, residual data, and multi-dimensional coordinates is output, providing comprehensive support for subsequent data processing. Specifically:

[0109] S61. Read the predicted arc segment at the midpoint time t0 after cubic spline interpolation: image plane position. ,speed , velocity direction acceleration This forms an "instantaneous prediction motion vector package":

[0110] ;

[0111] S62, Image plane multi-vector hypothesis generation, with Centered on a 3σ uncertainty boundary, radial-azimuth discretization generates N≥8 motion hypotheses:

[0112] ;

[0113] in , Given by the standard deviation of the predicted arc segment velocity and direction, That is, the i-th velocity is assumed to have a scalar value. and direction Decide, and The value is taken according to the quantity N in the corresponding interval, and the value is generally taken at equal intervals;

[0114] S63, Track Extrapolation and Nearest Neighbor Scoring: For each hypothesis i, along... Extrapolate ΔT (measured frame interval) to the next frame:

[0115] ;

[0116] By comparing the measured track endpoints (xmeas, ymeas), a scoring function is constructed:

[0117] ;

[0118] The weights w1 + w2 + w3 = 1. , This represents the normalized value of the image's diagonal pixels and the maximum acceleration.

[0119] S64. Optimal matching selection: The hypothesis corresponding to the minimum score is taken as the optimal matching trajectory.

[0120] ;

[0121] Its subsequent extrapolated point sequence is used as the output of the matched image plane prediction route;

[0122] S65. If the minimum score is still greater than the threshold τ (pixel-level threshold), then linear extrapolation is used to complete the missing points of the measured track, or the image plane hypothesis set is re-initialized to trigger "new track generation".

[0123] S7. Combine matching score and confidence test to output the identification and authentication information of the spatial target, including the target's unique ID, image plane coordinate velocity (X,Y,VX,VY) and celestial coordinates (α,δ), matching residual, matching score, and UTC time.

[0124] This application also provides a specific embodiment that comprehensively analyzes the effectiveness and superiority of the method provided above; it uses a measured image of a certain space payload, with a project size of 1800×1800 pixels and a sky area coverage of 7°×7°; and executes the method of the present invention, wherein, referring to Figure 2 The diagram shows the result of transforming the target prediction result into the image plane. The yellow line represents the result after transforming the target prediction result into image plane coordinates, and the yellow circle indicates the starting point of the prediction arc. (Reference) Figure 3 The image shown is a schematic diagram of the star chart detection results. The green boxes in the diagram indicate the star chart detection results. Figure 4 The diagram shows the target identification and verification results. The green boxes indicate the results of the actual data track association, the yellow lines represent the prediction results (the starting point yellow circle contains the target number), and the targets covered by the green boxes are the targets identified and verified by the results.

[0125] The embodiments of the present invention have been described in detail above with reference to the examples. However, the present invention is not limited to the above embodiments. For those skilled in the art, after learning the contents described in the present invention, several equivalent changes and substitutions can be made without departing from the principle of the present invention. These equivalent changes and substitutions should also be considered to fall within the protection scope of the present invention.

Claims

1. A method for identifying and certifying space multi-targets in space-based wide-field optical observation, characterized in that: Includes the following steps: Step 1: Calculate the target observation direction using TLE cataloging prediction and optical time correction; It also integrates the film parameter model of space-based measured images to map the target observation direction to the theoretical position sequence of the image plane; and interpolates and extracts the theoretical apparent motion velocity and direction of the target to construct a priori motion library; Step 2: Preprocess the space-based observation images and perform star detection to form a set of measured star points; and design kinematic filtering and trajectory association to extract the measured target motion parameters; Step 3: Generate multiple predicted trajectory hypotheses based on the theoretical motion features in the prior motion library, and combine them with the extracted local measured target motion parameters. Then, directly use pixel velocity and direction to perform nearest neighbor scoring on the image plane, and take the hypothesis corresponding to the minimum score as the best matching trajectory. Step 4: Combine matching scores and confidence tests to output the identification and authentication information of spatial targets.

2. The method according to claim 1, characterized in that: In step one, based on the identification results of reference stars in the space-based measured images, the film parameter model is calculated, and after geometric distortion correction and poor astronomical effect correction, the film model parameters are described by 4 or 6 parameters; the image plane coordinates of the space target are calculated by combining the celestial coordinate information obtained from the space target prediction with the film parameter model.

3. The method of claim 1, wherein the method is a method for identifying and authenticating a plurality of targets in space by optical observation from space. In step one, cubic spline interpolation is performed on the target prediction position sequence based on the start and end times of the observed arc segment to obtain the image plane position, motion velocity, and acceleration parameters of the spatial target at the midpoint of the prediction arc segment; at the same time, a pixel velocity direction angle is generated for each visible spatial target.

4. The method of claim 1, wherein the method is a method for identifying and authenticating a plurality of targets in space by optical observation from space. In step two, preprocessing and star detection include: An adaptive median filter is used to preserve the core features of the stars; and a grayscale stretching algorithm is used to adjust the dynamic range of the image and enhance the grayscale contrast between the stars and the background. The background mean and sigma value of the image are calculated using the 3sigma iterative method. The image is then binarized using "mean + 3σ" as the threshold. Pixels below the threshold are set to zero, while those above the threshold are retained, thus generating a binarized image. Based on the generated binary image, the eight-neighbor connected component analysis algorithm is used to traverse the image pixels, mark the star pixel regions that meet the connectivity conditions one by one, assign each independent star region a unique identifier ID, and at the same time remove connected regions with an area smaller than a preset threshold. For each valid connected component, the pixel group in the image is transformed into a quantifiable feature vector based on the centroid method.

5. The method of claim 1, wherein the method is a method for identifying and authenticating a plurality of targets in space by optical observation from space. In step two, the kinematic filtering includes: For two adjacent frames of candidate star points, establish nearest neighbor matching to determine the threshold circle radius; For the matching successful points, the image plane velocity vector is calculated to obtain the average motion characteristics of the stars and the velocity standard deviation ; For all candidate points in the current frame, if: ； and ; If it is identified as a stellar constellation, it is removed from the moving target candidate pool; the remaining moving target candidate points are retained; among them, It is the velocity vector of the j-th candidate star point detected in the image.

6. The space-based large field-of-view optical observation multi-target identification and authentication method according to claim 5, characterized in that: In step two, the track association includes: For the retained moving target candidate points, a uniform linear model is adopted in the three-dimensional space of the image plane, and the RANSAC clustering method is used to find collinear points with a number ≥ 3 and The short arc segment, where N is the number of stars matched; Each short arc is assigned a unique track ID, and its average pixel velocity and orientation angle are recorded. Starting from the end of the short arc above, extrapolate the predicted position of the next frame along the average velocity; and establish a pixel threshold circle around the predicted position: If a detection point exists within the threshold, that point is added to the current track and the average speed is updated; otherwise, a "loss counter" is started, and the track is terminated when the number of consecutive lost frames exceeds K. Output the parameters of each successfully associated measured target, including track ID, start-end pixel coordinates, average pixel velocity, heading angle, total number of frames, and pixel residual RMS.

7. The space-based large field-of-view optical observation multi-target identification and authentication method according to claim 3, characterized in that: In step three, the velocity at the midpoint of the predicted arc segment is used. Centered on With uncertainty as the boundary, radial-azimuth discretization generates N≥8 motion hypotheses: ； in , Given by the standard deviation of the predicted arc segment velocity and direction, That is, the i-th velocity is assumed to have a scalar value. and direction Decide, and The value is determined based on the quantity N within the corresponding interval; Locally measured target motion parameters include the spatial target image plane position based on the midpoint of the predicted arc segment. Speed ​​of movement , velocity direction and acceleration Forming an "instantaneous prediction motion vector packet": 。 8. The space-based large field-of-view optical observation multi-target identification and authentication method according to claim 7, characterized in that: In step three, for each hypothesis i, along Extrapolate ΔT to the next frame: ； And by comparing with the measured trajectory endpoint, a scoring function is constructed: ； Among them, weight , , ΔT represents the normalized value of the image face diagonal pixels and the maximum acceleration, and ΔT is the inter-frame time difference. The hypothesis corresponding to the minimum score is taken as the best matching trajectory: ； The subsequent extrapolated point sequence is used as the output of the image plane prediction route after matching.

9. The space-based large field-of-view optical observation multi-target identification and authentication method according to claim 8, characterized in that: If the minimum score is still greater than the threshold τ, then linear extrapolation is used to complete the missing points of the measured track, or the image plane hypothesis set is reinitialized to trigger "new track generation".

10. The space-based large field-of-view optical observation multi-target identification and authentication method according to claim 1, characterized in that: The identification and authentication information of the space target includes the target's unique ID, image plane coordinate velocity (X,Y,VX,VY) and celestial coordinates (α,δ), matching residual, matching score, and UTC time.

Images