A method and device for identifying near-earth space targets

Abstract

The present invention discloses a method and device for near-earth space target recognition, including: obtaining space-based measurement pictures, acquiring the morphological information of the space target to be recognized and the centroid position of the space target to be recognized based on the space-based measurement pictures, and acquiring image features based on the morphological information and the centroid position; obtaining observation data, constructing the geometric relationship among the space target to be recognized, the observation platform and the earth center based on the observation data, and determining the target orbit of the space target to be recognized based on the geometric relationship and the Laplace method, and acquiring orbit features; splicing based on the image features and the orbit features to obtain fusion features, and recognizing the space target to be recognized based on the fusion features and the historical catalog database to obtain a recognition result. By combining the image features and the orbit features to generate fusion features, the present invention makes full use of the image characteristics and motion characteristics of space targets, comprehensively describes space targets, and improves the accuracy of space-based optical target recognition.

Description

Technical Field

[0001] The present invention relates to the technical field of space target recognition, and particularly to a method and device for near-earth space target recognition. Background Art

[0002] With the rapid development of human space exploration, as of 2024, the number of various space targets in near-earth space has exceeded 50,000 and is still growing explosively. It is of great engineering significance to conduct key detection and tracking on space targets of relatively important value and obtain the orbital elements, shape, size, and category of various space targets. Space target detection and recognition mainly include two means: ground-based observation and space-based observation. The ground-based observation system is usually composed of an optical telescope and a radar detector, which can detect and track space targets passing overhead. The ground-based space target surveillance system has mature technology and low cost, but it is easily affected by the earth's curvature, atmospheric environment, and lighting and meteorological conditions, and the observation range is limited. The space-based optical observation platform flies around the earth at high speed in orbit, which can obtain a wider field of view, weakens the influence of the earth's curvature, and can increase the observation arc of space targets. In addition, due to the certain altitude of the space-based platform itself, the observation geometry is improved, the adverse effects of the atmosphere during observation are reduced, and it is more conducive to optical observation. Combining the respective advantages of the space-based platform and optical observation, space-based optical observation has obvious advantages over ground-based observation in terms of airspace coverage and surveillance timeliness.

[0003] There are mainly two existing approaches for space-based optical target recognition. One is to extract the target pixel coordinates in the observation payload and obtain the target's relative right ascension and declination based on the coordinate transformation principle, and then complete the relative orbit calculation. Depending on dynamic calculations to determine the target type, misjudgments may sometimes occur, and a large amount of other image information obtained in the observation payload is discarded or wasted. The other is to directly make a judgment based on image characteristics such as gray level and size. At this time, due to ignoring the target's dynamic characteristics, it is easily misled and misjudgments may occur. Summary of the Invention

[0004] The present invention provides a method and device for near-earth space target recognition to improve the accuracy of space-based optical target recognition.

[0005] To solve the above technical problems, an embodiment of the present invention provides a method for near-earth space target recognition, including:

[0006] Obtain a space-based measurement picture, obtain the morphological information of the space target to be recognized and the centroid position of the space target to be recognized based on the space-based measurement picture, and obtain image features based on the morphological information and the centroid position;

[0007] Observation data is obtained based on an observation platform, a geometric relationship model is constructed based on the observation data, and the target orbit of the space target to be identified is determined based on the geometric relationship model and the Laplace method, and orbit features are obtained; the geometric relationship model is a geometric relationship model among the space target to be identified, the observation platform, and the earth center.

[0008] Fusion features are obtained based on the image features and the orbit features, and the space target to be identified is identified based on the fusion features and a historical catalog database, and a space target identification result is obtained.

[0009] In the present invention, the morphological information of a space target and the centroid position of the space target are obtained through a space-based measurement picture, the appearance characteristics of the space target are expressed according to the morphological information and the centroid position, and then image features are generated; at the same time, a geometric relationship among the space target, the observation platform, and the earth center is constructed through observation data, the motion characteristics of the space target are expressed according to the geometric relationship, and then orbit features are obtained; fusion features are generated by combining the image features and the orbit features, and the fusion features make full use of the image characteristics and motion characteristics of the space target and can comprehensively describe the space target; finally, the space target to be identified is identified based on the historical catalog database and the fusion features so as to associate the space target to be identified with prior catalog information, and an identification result of a near-earth space target is obtained, improving the accuracy of the identification result of a near-earth space target in space-based optics.

[0010] Further, the obtaining the morphological information of the space target and the centroid position of the space target based on the space-based measurement picture includes:

[0011] A filtering operator matrix is generated based on a preset filter, and the space-based measurement picture is filtered based on the filtering operator matrix to obtain a filtered picture.

[0012] The filtered picture is binarized based on a preset gray-scale threshold to generate a gray-scale image.

[0013] Edge extraction is performed on the gray-scale image based on a preset first edge detection algorithm to obtain the morphological information of the space target.

[0014] The target contour of the gray-scale image is identified based on a preset second edge detection algorithm, and the centroid position of the space target is obtained based on the target contour.

[0015] In the present invention, through filtering and binarization processing, noise is removed and the image is simplified, and the binarized image is monitored through an edge detection algorithm to extract the morphological information of the space target to be identified. At the same time, through target contour identification and pixel threshold calculation, the centroid position of the space target to be identified is obtained, so that image features for describing the image characteristics of the space target to be identified are generated according to the morphological information and the centroid position.

[0016] Further, the observation data includes the spatial target position vector and the observation platform position vector. The spatial target position vector is the position vector of the spatial target relative to the earth's center, and the observation platform position vector is the position vector of the observation platform relative to the earth's center. The construction of the geometric relationship model based on the observation data includes:

[0017] Construct the geometric relationship based on the spatial target position vector and the observation platform position vector;

[0018] Construct the geocentric distance equation of the spatial target based on the Laplace method, and obtain the initial orbit parameters based on the geometric relationship and the geocentric distance equation;

[0019] Construct an observation model and an error model based on the initial orbit parameters, and optimize and determine the target orbit based on the observation model and the error model to obtain orbit characteristics.

[0020] The present invention uses the position vectors of the spatial target and the observation platform to establish the geometric connection between them and the earth's center, thereby constructing the geocentric distance equation by the Laplace method, obtaining the initial orbit parameters based on the geocentric distance equation and the geometric connection, realizing the conversion from observation data to orbit parameters, obtaining the orbit characteristics of the spatial target to be identified, and describing the dynamic characteristics of the spatial target to be identified based on the orbit characteristics.

[0021] Further, obtaining the fusion feature based on the image feature and the orbit feature, and identifying the spatial target to be identified based on the fusion feature and the historical catalog database to obtain the spatial target identification result, including:

[0022] Perform standardization processing on the image feature and the orbit feature to obtain the standard image feature and the standard orbit feature;

[0023] Perform splicing based on the standard image feature and the standard orbit feature to obtain the fusion feature;

[0024] Input the fusion feature into a pre-trained target recognition model to identify the spatial target to be identified and obtain a preliminary recognition result;

[0025] Verify the preliminary recognition result based on the historical catalog data to obtain the spatial target identification result.

[0026] Through standardization processing, the present invention can convert eigenvalue with different dimensions to the same scale, splice the standardized image features and orbit features together to form a comprehensive feature vector, which contains information in multiple aspects, so as to comprehensively describe the image characteristics and kinematic characteristics of the space target to be recognized through the fusion features, and thus recognize the space target to be recognized, improving the recognition accuracy.

[0027] Further, the preliminary recognition result includes the basic information of the space target to be recognized. Verifying the preliminary recognition result based on the historical catalog data to obtain the space target recognition result includes:

[0028] Matching and verifying the target in the historical catalog database based on the basic information;

[0029] Calculating the orbit error based on the basic information and the verification target, verifying the preliminary recognition result based on a preset orbit error, and obtaining the space target recognition result based on the verification result.

[0030] The present invention uses the basic information in the preliminary recognition result to search for matching targets in the historical catalog database, and quantifies the difference between the preliminary recognition result and the orbit parameters of the targets in the historical catalog database based on the orbit error, so as to verify the preliminary recognition result, and based on a preset orbit error threshold, determine whether the preliminary recognition result meets the requirements, thereby determining the space target recognition result, so as to improve the recognition accuracy and avoid misjudgment.

[0031] In a second aspect, the present invention provides a near-earth space target recognition device, including: an image feature extraction module, an orbit feature extraction module, and a fusion recognition module;

[0032] The image feature extraction module is used to obtain space-based measurement pictures, obtain the morphological information of the space target to be recognized and the centroid position of the space target to be recognized based on the space-based measurement pictures, and obtain image features based on the morphological information and the centroid position;

[0033] The orbit feature extraction module is used to obtain observation data based on an observation platform, construct a geometric relationship model based on the observation data, and determine the target orbit of the space target to be recognized and obtain orbit features based on the geometric relationship model and the Laplace method; the geometric relationship model is a geometric relationship model among the space target to be recognized, the observation platform, and the earth center;

[0034] The fusion recognition module is used to obtain fusion features based on the image features and the orbit features, and recognize the space target to be recognized based on the fusion features and the historical catalog database to obtain the space target recognition result.

[0035] Further, the image feature extraction module is used for:

[0036] Generating a filtering operator matrix based on a preset filter, and performing filtering processing on the space-based measurement picture based on the filtering operator matrix to obtain a filtered picture;

[0037] Performing binarization processing on the filtered picture based on a preset gray threshold to generate a gray image;

[0038] Performing edge extraction on the gray image based on a preset first edge detection algorithm to obtain the morphological information of the space target;

[0039] Identifying the target contour of the gray image based on a preset second edge detection algorithm, and obtaining the centroid position of the space target based on the target contour.

[0040] Further, the observation data includes a space target position vector and an observation platform position vector. The space target position vector is the position vector of the space target relative to the earth's center, and the observation platform position vector is the position vector of the observation platform relative to the earth's center; the orbit feature extraction module is used for:

[0041] Constructing the geometric relationship based on the space target position vector and the observation platform position vector;

[0042] Constructing an earth distance equation of the space target based on the Laplace method, and obtaining initial orbit parameters based on the geometric relationship and the earth distance equation;

[0043] Constructing an observation model and an error model based on the initial orbit parameters, and optimizing and determining the target orbit for the initial orbit parameters based on the observation model and the error model to obtain orbit features.

[0044] Further, the fusion recognition module is used for:

[0045] Performing standardization processing on the image features and the orbit features to obtain standard image features and standard orbit features;

[0046] Performing splicing based on the standard image features and the standard orbit features to obtain fusion features;

[0047] Inputting the fusion features into a pre-trained target recognition model to recognize the space target to be recognized and obtain a preliminary recognition result;

[0048] Verifying the preliminary recognition result based on the historical catalog data to obtain a space target recognition result.

[0049] Further, the preliminary recognition result includes the basic information of the space target to be recognized, and the fusion recognition module is configured to:

[0050] Match and verify the target in the historical catalog database based on the basic information;

[0051] Calculate the orbit error based on the basic information and the verified target, verify the preliminary recognition result based on a preset orbit error, and obtain the space target recognition result based on the verification result. Description of the Drawings

[0052] Figure 1 It is a schematic flowchart of a method for recognizing near-earth space targets provided by an embodiment of the present invention;

[0053] Figure 2 It is a schematic diagram of a space target to be recognized provided by an embodiment of the present invention;

[0054] Figure 3 It is a grayscale feature map of a space target to be recognized provided by an embodiment of the present invention;

[0055] Figure 4 It is a schematic diagram of the geometric relationship at a single word observation moment provided by an embodiment of the present invention. Detailed Embodiment

[0056] The following combines the drawings and embodiments to further describe in detail the specific embodiments of the present invention. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0057] The terms "first" and "second" in the specification, claims and drawings of this application are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0058] Referring to "embodiment" herein means that a specific feature, structure or characteristic described in connection with the embodiment can be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0059] Embodiment 1

[0060] SeeFigure 1 , Figure 1 It is a schematic flowchart of a method for identifying near-earth space targets provided by an embodiment of the present invention. An embodiment of the present invention provides a method for identifying near-earth space targets, including steps 101 to 103, specifically as follows:

[0061] Step 101: Obtain a space-based measurement picture, obtain the morphological information of the space target to be identified and the centroid position of the space target to be identified based on the space-based measurement picture, and obtain image features based on the morphological information and the centroid position;

[0062] In this embodiment, a space-based measurement picture is obtained, and the space-based observation pictures are classified and stored according to image data frames, and the space-based observation pictures are preprocessed. Among them, the preprocessing includes slicing processing and data cleaning, and then the large image in the space-based measurement picture is segmented into smaller sub-images for processing, and at the same time, noise and invalid information are removed to ensure data quality.

[0063] In this embodiment, after the space-based measurement picture is preprocessed, the morphological information of the space target to be identified and the centroid position of the space target to be identified are obtained based on the space-based measurement picture.

[0064] In this embodiment, obtaining the morphological information of the space target and the centroid position of the space target based on the space-based measurement picture includes:

[0065] Generate a filtering operator matrix based on a preset filter, and perform filtering processing on the space-based measurement picture based on the filtering operator matrix to obtain a filtered picture;

[0066] Perform binarization processing on the filtered picture based on a preset gray threshold to generate a gray image;

[0067] Perform edge extraction on the gray image based on a preset first edge detection algorithm to obtain the morphological information of the space target;

[0068] Identify the target contour of the gray image based on a preset second edge detection algorithm, and obtain the centroid position of the space target based on the target contour.

[0069] In this embodiment, the preprocessed image data is loaded into a filtering-related compilation environment, and the gray-scale information of all pixels of the image slices is traversed, and the types and magnitudes of the noises present in the image are determined, such as Gaussian noise, background noise, etc. A filter is selected according to the number of isolated noise points and the noise type of the image slices, and thus a specified filtering operator matrix is generated according to the filter. Among them, the filters include a mean filter, a Gaussian filter, and a median filter. Subsequently, the size of the filtering window is determined, the image pixels are traversed, and the filtered image is obtained by calculating by weighting the surrounding pixel values; the edge pixels of the image slices are all replaced with the gray-scale mean value.

[0070] In this embodiment, specific mathematical operations are performed on the pixels in the image slices to change the appearance of the image or enhance specific features of the spatial targets in the image.

[0071] In this embodiment, the filtered image is binarized. Specifically, a gray-scale threshold is preset, and based on this gray-scale threshold, the pixel values in the gray-scale image are divided into two categories. One category is the pixel values greater than the gray-scale threshold, and the other category is the pixel values less than or equal to the gray-scale threshold. For each pixel in the filtered image, its gray-scale value is compared with the gray-scale threshold. If the gray-scale value of the pixel is greater than the threshold, the gray-scale value of the pixel is retained, representing the target object; if the gray-scale value of the pixel is less than or equal to the threshold, the pixel is set to black (gray-scale is 0), representing the background, until all pixels in the image are binarized to generate a gray-scale image.

[0072] In this embodiment, after the binarization process, the image is binary-segmented according to the processing result, and then the edge of the gray-scale image is extracted based on a preset first edge detection algorithm to obtain the morphological information of the spatial target.

[0073] In this embodiment, in the processing of space-based measurement images, the edge extraction of spatial targets is used to identify and analyze the morphology, structure, and boundaries of celestial bodies and various spatial targets. The principle of edge extraction is to detect the change of pixel values of the determined targets in the image, find the boundaries between different regions in the image, and thus depict the contours and shapes of the targets.

[0074] In this embodiment, the gradient or edge response of the pixel values is calculated based on a preset first edge detection algorithm to detect the edges in the image, and after the edge detection, the broken or discontinuous edge segments are connected based on a preset edge connection algorithm to form a complete edge contour.

[0075] In this embodiment, the first edge detection algorithm includes the Sobel edge detection algorithm, Prewitt, Canny, etc.

[0076] In this embodiment, after obtaining the edge contour, it is also necessary to optimize and refine the detected edges, remove noise and unnecessary details, so as to obtain a more accurate target edge contour, and obtain the morphological information of the space target to be recognized based on the extracted target contour. The morphological information includes features such as size, shape, and direction.

[0077] In this embodiment, through edge extraction, the morphological information of the space target can be effectively extracted from the space-based image, providing important data support for subsequent analysis and research.

[0078] In this embodiment, after completing the edge extraction of the space target to be recognized, the target contour of the grayscale image is recognized based on a preset second edge detection algorithm, and the centroid position of the space target is obtained based on the target contour.

[0079] In this embodiment, in the image processing of space-based measurement of space targets, determining the centroid position of celestial bodies and various space targets is the basis for generating the line-of-sight vector of the space target and completing the determination of the space target orbit, and then performing position measurement, motion analysis, etc. The principle of centroid extraction is to calculate the pixel values and position information of the target in the image to find the centroid position of the target image.

[0080] In this embodiment, based on threshold segmentation and the second edge detection algorithm, the target contour of the space target to be recognized is obtained by recognizing the grayscale image, and the weighted average value of its pixel values is calculated based on the target contour to determine the centroid position. Specifically, the pixel values are used as weights to calculate the centroid position, or the morphological method is used to find the geometric center of the target.

[0081] In this embodiment, due to the relatively fast movement speed of the space target, a trailing phenomenon is likely to occur in the image. In view of this phenomenon, when calculating the centroid coordinates of the space target image, the Gaussian surface fitting method based on anisotropy is preferably used to calculate the centroid position of the space target to be recognized. After obtaining the preliminary centroid position, further correction and optimization can be carried out, such as removing noise interference and eliminating errors, so as to obtain the final centroid position.

[0082] In this embodiment, the centroid position is used to measure the position, motion trajectory, morphological features, etc. of the target. Through centroid extraction, the central position of the space target can be accurately determined, providing basic data for subsequent analysis and research.

[0083] Please refer to Figure 2 , Figure 2 which is a schematic diagram of a space target to be recognized provided by an embodiment of the present invention.

[0084] In this embodiment, image features are obtained based on the morphological information and the centroid position, where the image features are the gray-scale features of the spatial target, including the maximum gray value, the diffusion area of the spatial target, the gray-scale gradient, and the average gray value, which are used to describe the features such as the brightness, size, and shape of the spatial target to be recognized.

[0085] Please refer to Figure 3 , Figure 3 which is a gray-scale feature map of a spatial target to be recognized provided by an embodiment of the present invention.

[0086] In this embodiment, the maximum gray value of the spatial target refers to the maximum gray value of the pixels in the target area. By calculating the maximum gray value, the brightest part of the target can be understood, which helps to evaluate the brightness intensity of the target.

[0087] In this embodiment, the diffusion area of the spatial target refers to the total area of all pixels in the target area. The size information of the target can be obtained by calculating the number or area of non-zero pixels in the target area, and then the size of the target can be evaluated.

[0088] In this embodiment, the average gray value of the spatial target refers to the average of the gray values of the pixels in the target area. By calculating the average gray value, the overall brightness level of the target can be understood, which helps to evaluate the overall brightness characteristics of the target.

[0089] In this embodiment, the gray-scale gradient refers to the change rate of the pixel gray values in the image, which reflects the gray-scale change in the target area. The gray-scale change characteristics of the target can be obtained by calculating the gradient or gradient amplitude of the pixel gray values, which helps to analyze the edge information and texture characteristics of the target.

[0090] In this embodiment, in this embodiment, through filtering and binarization processing, noise is removed and the image is simplified, and the binary image is monitored by an edge detection algorithm to extract the morphological information of the spatial target to be recognized. At the same time, through target contour recognition and pixel threshold calculation, the centroid position of the spatial target to be recognized is obtained, so as to generate image features for describing the image characteristics of the target to be recognized based on the morphological information and the centroid position.

[0091] In this embodiment, after obtaining the gray-scale features, image recognition can be performed on the spatial target to be recognized based on the gray-scale features. Specifically: First, the gray-scale features of each spatial target are extracted, including the maximum gray value, the diffusion area, the gray-scale gradient, the average gray value, etc. The gray-scale features of each spatial target are represented as a feature vector, where each feature corresponds to a dimension. For example, a feature vector may include the maximum gray value, the diffusion area, the gray-scale gradient, the average gray value. Then, machine learning algorithms or classification algorithms are used to classify the feature vectors, so as to realize the type recognition of the spatial target.

[0092] In this embodiment, common classification algorithms include Support Vector Machine (SVM), K-Nearest Neighbor (KNN), decision tree, etc.

[0093] In this embodiment, the result of image recognition will be used as auxiliary data to assist in judging whether the final recognition result is correct.

[0094] Step 102: Obtain observation data based on an observation platform, construct a geometric relationship model based on the observation data, and determine the target orbit of the space target to be recognized and obtain orbit features based on the geometric relationship model and Laplace method; the geometric relationship model is the geometric relationship model among the space target to be recognized, the observation platform, and the earth center.

[0095] In this embodiment, the observation data includes the position vector of the space target and the position vector of the observation platform. The position vector of the space target is the position vector of the space target relative to the earth center, and the position vector of the observation platform is the position vector of the observation platform relative to the earth center; constructing the geometric relationship model based on the observation data includes:

[0096] Construct the geometric relationship based on the position vector of the space target and the position vector of the observation platform;

[0097] Construct an earth center distance equation of the space target based on Laplace method, and obtain initial orbit parameters based on the geometric relationship and the earth center distance equation;

[0098] Construct an observation model and an error model based on the initial orbit parameters, and optimize the initial orbit parameters based on the observation model and the error model to determine the target orbit and obtain orbit features.

[0099] In this embodiment, by using the position vectors of the space target and the observation platform, a geometric connection between them and the earth center is established, so as to construct an earth center distance equation through the geometric connection and Laplace method, obtain initial orbit parameters, thereby realizing the conversion from observation data to orbit parameters, obtain the orbit features of the space target to be recognized, and describe the dynamic characteristics of the space target to be recognized based on the orbit features.

[0100] In this embodiment, identifying orbit parameters based on observation data includes observing an un-cataloged space target using a space-based or ground-based optical platform at a time series t i (i = 1, 2, …, n), extracting the direction vector L of the target relative to the platform, i , and combining with the earth center position vector R of the observation platform i to determine the orbit parameters of the target.

[0101] In this embodiment, the observation data includes the spatial target position vector and the observation platform position vector. Among them, the spatial target vector includes the position, velocity, and acceleration of the spatial target relative to the earth's center, and the observation platform vector includes the space-based observation platform vector and the ground-based observation platform vector. The space-based observation platform vector includes the position, velocity, and acceleration of the space-based observation platform relative to the earth's center, and the ground-based observation platform vector includes the position, velocity, and acceleration of the ground-based observation platform relative to the earth's center.

[0102] In this embodiment, at a time series t i (i = 1, 2, …, n), based on the observation data, the geometric relationship among the spatial target, the ground-based or space-based observation platform, and the earth's center at a single observation moment is constructed.

[0103] Please refer to Figure 4 , Figure 4 which is a schematic diagram of the geometric relationship at a single word observation moment provided by the embodiment of the present invention.

[0104] In this embodiment, the geometric configuration of a single observation of the spatial target is:

[0105] (1)

[0106] The first derivative and the second derivative of the geometric configuration of a single observation of the spatial target are respectively:

[0107] (2)

[0108] (3)

[0109] Among them, , , are respectively the slant range, the first-order and second-order derivatives of the slant range; and are respectively the first-order and second-order derivatives of the observation vector; , , and , , are respectively the position, velocity, and acceleration vectors of the spatial target and the observation platform.

[0110] In this embodiment, an octic equation of the target geocentric distance is constructed based on the Laplace method and the geometric relationship The octic equation,

[0111] (4)

[0112] Among them, A, B, and C are coefficients depending on the relative positions of the earth's center, the observation platform, and the target. Specifically:

[0113] (5)

[0114] (6)

[0115] (7)

[0116] In this embodiment, the octic equation contains information such as the positions, velocities, and accelerations of the space target and the observation platform. The coefficients in the equation depend on the relative positions of the geocenter, the observation platform, and the space target. Therefore, the nature of the roots of the octic equation also depends on the relative position relationship among the three.

[0117] In this embodiment, based on the geocentric distance equation, the position of the space target to be identified at the orbit determination moment and the initial value of the velocity vector , thus based on the position at the orbit determination moment and the initial value of the velocity vector and the geometric relationship, the slant range 0 and the slant range rate are calculated. Finally, the orbital elements are calculated based on the slant range 0 and the slant range rate .

[0118] In this embodiment, it can be obtained from the geometric relationship that there is no essential difference between space-based orbit determination and ground-based orbit determination, that is, the basic formulas are the same. When using the quasi-Laplace method for orbit determination, since the short-arc motion equation of the space target basically follows Kepler motion, it is easy to cause the orbit determination result to converge to the orbit of the observation platform. That is:[[]]

[0119] (8)

[0120] where the subscript 0 represents the orbit determination epoch moment. represents the position of the space target relative to the geocenter at the orbit determination epoch moment, represents the acceleration vector of the space target relative to the geocenter at the orbit determination epoch moment.

[0121] In this embodiment, the trivial solutions of the target geocentric distance are eliminated. After the elimination of trivial solutions operation, the original octic equation is reduced to a septic equation, and the Newton iteration method can be used for solution:

[0122] (9)

[0123] (10)

[0124] (11)

[0125] In the formula, is the value of r0 at the k-th iteration. The initial value r0 is obtained through iterative calculation, and thus the estimated values of the slant range and the slant range rate of the space target can be obtained:

[0126] (12)

[0127] (13)

[0128] Therefore, based on the slant range, the slant range rate, and the geometric relationship, the position and velocity of the space target at the orbit determination moment can be obtained, that is, the initial orbit parameters.

[0129] In this embodiment, the orbit characteristics of the space target are obtained based on the position and velocity of the space target. The orbit characteristics specifically include the orbit altitude, the target flight speed, the acceleration, the orbit type, and the orbit eccentricity.

[0130] In this embodiment, an observation model and an error model are constructed based on the initial orbit parameters, that is, the differential equation of the formula is used to determine the initial value from the boundary conditions (observed values) in reverse.

[0131] In this embodiment, an observation model is constructed based on the initial orbit parameters:

[0132] (14)

[0133] Wherein, = , , represents the orbit determination epoch moment, represents the predicted position and direction of the space target at the future moment , represents the direction angle of the space target relative to the observation platform, represents the elevation angle of the space target relative to the observation platform.

[0134] In this embodiment, the observation model is linearized, and a Taylor expansion is performed near the approximate value of the state initial value . Only the linear terms are retained to obtain its error equation. Specifically:

[0135] (15) (16)

[0136] Wherein, is the observation residual, is the correction amount of the state vector, is the approximate value of the state vector.

[0137] In this embodiment, the least squares method is used to solve the error equation, with the goal of minimizing the sum of the squares of the observation residuals h:

[0138] (17)

[0139] According to the correction , update the initial vector of the initial state orbit of the orbit:

[0140] (18)

[0141] The updated state vector will be closer to the true value. If higher precision is required, iterative updates can be performed until the observation residual h meets the preset precision requirements.

[0142] In this embodiment, by using the position vectors of the space target and the observation platform, the geometric connection between them and the earth's center is established, and then the Laplace method is used to construct the geocentric distance equation. Based on the geocentric distance equation and the geometric connection, the initial orbit parameters are obtained, so as to realize the conversion from observation data to orbit parameters, obtain the orbit characteristics of the space target to be identified, and describe the dynamic characteristics of the space target to be identified based on the orbit characteristics.

[0143] Step 103: Obtain the fusion feature based on the image feature and the orbit feature, and identify the space target to be identified based on the fusion feature and the historical catalog database to obtain the space target identification result.

[0144] In this embodiment, the obtaining the fusion feature based on the image feature and the orbit feature, and identifying the space target to be identified based on the fusion feature and the historical catalog database to obtain the space target identification result includes:

[0145] Perform standardization processing on the image feature and the orbit feature to obtain the standard image feature and the standard orbit feature;

[0146] Perform splicing based on the standard image feature and the standard orbit feature to obtain the fusion feature;

[0147] Input the fusion feature into the pre-trained target recognition model to identify the space target to be identified and obtain the preliminary recognition result;

[0148] Verify the preliminary recognition result based on the historical catalog data to obtain the space target identification result.

[0149] In this embodiment, through standardization processing, eigenvalue with different dimensions can be converted to the same scale. The standardized image features and orbit features are spliced together to form a comprehensive feature vector, which contains information in multiple aspects, so as to comprehensively describe the image characteristics and kinematic characteristics of the space target to be recognized through the fused features, and then recognize the space target to be recognized, improving the recognition accuracy.

[0150] In this embodiment, there are many challenges in recognizing space targets relying only on image features, especially under complex observation conditions and target attitude changes. The photosensitive angle in different postures will significantly affect the image features of the target. In addition, when the target is in the reentry or maneuver ignition phase, the strong correlation between its image brightness and flight phases (such as active phase, boost phase, glide phase, and cruise phase) makes it have a high false alarm rate when simply relying on optical image processing methods to judge the target type. Moreover, for targets with ultra-short arc segments, the determination of the initial orbit of dynamics is often restricted by the singularity of the observation equation, which may lead to non-convergence of the orbit or the existence of trivial solutions, further affecting the reliability of orbit determination. At the same time, for the improvement of short arc orbits, due to the small amount of data redundancy and the small change in the constructed space geometry, the orbit improvement accuracy decreases, thus helping little to improve the success rate of space target recognition. Therefore, by fusing image features and orbit features to comprehensively describe the image characteristics and orbit characteristics of space targets, a space target recognition method that fuses image features and orbit features can make full use of the complementary information of the two. Image features provide the appearance and morphological features of the target, while orbit features provide key parameters for the dynamic behavior and motion state of the target. By combining them with each other, the recognition ability of the target can be significantly enhanced, the recognition accuracy can be improved, and the false alarm rate can be effectively reduced.

[0151] In this embodiment, since the dimensions of image features and orbit features are different, all features are standardized using the Min-Max scaling principle, and the standardized image features and orbit features are directly spliced to form a new fused feature.

[0152] In this embodiment, an identification model is pre-trained based on large model technology, and the space target is recognized based on the trained identification model and the fused features to obtain a preliminary identification result.

[0153] In this embodiment, for the space target with cataloged data, the preliminary identification result also needs to be verified based on the historical cataloged data to obtain the space target identification result.

[0154] In this embodiment, the preliminary identification result includes the basic information of the space target to be recognized. Verifying the preliminary identification result based on the historical cataloged data to obtain the space target identification result includes:

[0155] Match and verify the target in the historical catalog database based on the basic information;

[0156] Calculate the orbit error based on the basic information and the verification target, verify the preliminary recognition result based on a preset orbit error, and obtain the space target recognition result based on the verification result.

[0157] In this embodiment, before identifying the space target, for the identification of the space target in the cataloged data, before identifying the space target, the historical catalog database should be updated to the time node closest to the epoch time of the space target to be identified, so as to ensure the timeliness and accuracy of the orbit data used in the subsequent identification;

[0158] In this embodiment, based on the given forecast start time, forecast end time, and forecast step size, orbit forecasts are performed on all carriers such as satellites in the historical catalog database and the space target to be identified, and the position and velocity information in the J2000 coordinate system is obtained.

[0159] In this embodiment, calculate the orbit error between the space target and all satellite carriers in the catalog database respectively; and sort all the orbit errors. The orbit error is specifically:

[0160] (18)

[0161] (19)

[0162] Among them, is the position error between the space target to be identified and the carrier numbered p in the catalog database, is the position error between the space target to be identified and the carrier numbered p in the catalog database.

[0163] In this embodiment, according to the calculated orbit error, compare it with the set target recognition threshold. If the orbit error is within the target recognition threshold, match the space target to be identified with the target name and ID in the catalog database to confirm its identity. After completing the target recognition, further verify the recognition result. The recognition result can be cross-verified through other external measurement data (such as real-time tracking systems, radar monitoring, etc.) to ensure the accuracy and reliability of the recognition.

[0164] In this embodiment, according to the verification feedback of the recognition result, analyze the error sources and potential problems in the recognition process. According to the feedback information, optimize the algorithms and threshold settings for target recognition to improve the accuracy and efficiency of recognition. With the continuous acquisition of new orbit data, regularly update the catalog database to maintain the timeliness and accuracy of the data.

[0165] In this embodiment, by using the basic information in the preliminary recognition result, a matching target is searched for in the historical catalog database, and the difference between the preliminary recognition result and the orbital parameters of the target in the historical catalog database is quantified based on the orbital error, so as to verify the preliminary recognition result, and based on a preset orbital error threshold, it is determined whether the preliminary recognition result meets the requirements, thereby determining the space target recognition result, so as to improve the recognition accuracy and avoid misjudgment.

[0166] In this embodiment, by combining image processing means such as image filtering, binarization, edge extraction, and centroid calculation to analyze the images taken by the observation payload, the recognition of the space target type based on the image characteristics is achieved by analyzing the gray-scale characteristics of the target, such as the maximum gray value, dispersion area, gray-scale gradient, gray-scale average value, etc. and combining them with the dynamic observation data, which can effectively improve the accuracy of discriminating the space target type. At the same time, by integrating the orbital calculation results of the space target and the image recognition results, the space target is associated and matched with the prior catalog information to assist in improving the accuracy of the initial orbital value estimation of the target, and finally improve the efficiency and accuracy of space target recognition.

[0167] The embodiment of the present invention also provides a near-earth space target recognition device, including: an image feature extraction module, an orbital feature extraction module, and a fusion recognition module;

[0168] The image feature extraction module is used to obtain space-based measurement pictures, obtain the morphological information of the space target to be recognized and the centroid position of the space target to be recognized based on the space-based measurement pictures, and obtain image features based on the morphological information and the centroid position;

[0169] The orbital feature extraction module is used to obtain observation data based on the observation platform, construct a geometric relationship model based on the observation data, and determine the target orbit of the space target to be recognized based on the geometric relationship model and the Laplace method, and obtain orbital features; the geometric relationship model is the geometric relationship model among the space target to be recognized, the observation platform, and the earth center;

[0170] The fusion recognition module is used to obtain fusion features based on the image features and the orbital features, and recognize the space target to be recognized based on the fusion features and the historical catalog database, and obtain the space target recognition result.

[0171] In this embodiment, the image feature extraction module is used for:

[0172] Generate a filtering operator matrix based on a preset filter, and perform filtering processing on the space-based measurement pictures based on the filtering operator matrix to obtain filtered pictures;

[0173] Perform binary processing on the filtered image based on a preset grayscale threshold to generate a grayscale image;

[0174] Extract edges from the grayscale image based on a preset first edge detection algorithm to obtain the morphological information of the space target;

[0175] Identify the target contour of the grayscale image based on a preset second edge detection algorithm, and obtain the centroid position of the space target based on the target contour.

[0176] In this embodiment, the observation data includes the space target position vector and the observation platform position vector. The space target position vector is the position vector of the space target relative to the earth's center, and the observation platform position vector is the position vector of the observation platform relative to the earth's center; the orbital feature extraction module is used for:

[0177] Construct the geometric relationship based on the space target position vector and the observation platform position vector;

[0178] Construct the geocentric distance equation of the space target based on the Laplace method, and obtain the initial orbital parameters based on the geometric relationship and the geocentric distance equation;

[0179] Construct an observation model and an error model based on the initial orbital parameters, and optimize the initial orbital parameters based on the observation model and the error model to determine the target orbit and obtain the orbital features.

[0180] In this embodiment, the fusion recognition module is used for:

[0181] Perform standardization processing on the image features and the orbital features to obtain standard image features and standard orbital features;

[0182] Perform splicing based on the standard image features and the standard orbital features to obtain fusion features;

[0183] Input the fusion features into a pre-trained target recognition model to recognize the space target to be recognized and obtain a preliminary recognition result;

[0184] Verify the preliminary recognition result based on the historical catalog data to obtain the space target recognition result.

[0185] In this embodiment, the fusion recognition module is used for:

[0186] Match and verify the target in the historical catalog database based on the basic information;

[0187] Calculate the orbital error based on the basic information and the verification target, verify the preliminary recognition result based on a preset orbital error, and obtain the spatial target recognition result based on the verification result.

[0188] In an embodiment of the present invention, a terminal device is further provided, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the above-mentioned near-earth space target recognition method is implemented.

[0189] In an embodiment of the present invention, a computer-readable storage medium is further provided. The computer-readable storage medium includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the above-mentioned near-earth space target recognition method.

[0190] Exemplarily, the computer program may be divided into one or more modules. One or more modules are stored in the memory and executed by the processor to complete the present invention. One or more modules may be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program in the terminal device.

[0191] The terminal device may be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor, a memory, and a display. Those skilled in the art can understand that the above components are only examples of the terminal device and do not constitute a limitation on the terminal device. It may include more or fewer components than those described, or combine some components, or different components. For example, the terminal device may further include input / output devices, network access devices, a bus, etc.

[0192] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor, or the processor may also be any conventional processor, etc. The processor is the control center of the terminal device and connects various parts of the entire terminal device through various interfaces and circuits.

[0193] The memory can be used to store computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory, and by invoking the data stored in the memory, the processor can implement various functions of the terminal device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store the operating system, application programs required for at least one function (such as the sound playback function, the text conversion function, etc.); the data storage area can store the data created according to the use of the mobile phone (such as audio data, text message data, etc.). In addition, the memory can include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage devices.

[0194] Among them, if the module for near-earth space target recognition is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. Those of ordinary skill in the art can understand and implement it without creative work.

[0195] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above description is only specific embodiments of the present invention and is not used to limit the protection scope of the present invention. It is particularly pointed out that for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A near-Earth space target recognition method, characterized in that: include: Acquire a space-based measurement picture, acquire morphological information of a space target to be identified and a centroid position of the space target to be identified based on the space-based measurement picture, and acquire image features based on the morphological information and the centroid position; Acquire observation data based on the observation platform, construct a geometric relationship model based on the observation data, and determine the target orbit of the space target to be identified based on the geometric relationship model and the Laplace method to obtain orbital characteristics; the geometric relationship model is a geometric relationship model between the space target to be identified, the observation platform and the center of the earth; Acquiring fusion features based on the image features and the orbital features, and identifying the space target to be identified based on the fusion features and the historical catalog database to obtain a space target identification result, including: performing standardization processing on the image features and the orbital features to obtain standard image features and standard orbital features; Based on the standard image features and the standard orbit features, splicing is performed to obtain fusion features; the fusion features are input into a pre-trained target recognition model to identify the space target to be identified and obtain a preliminary recognition result; based on the historical cataloging data, the preliminary recognition result is verified to obtain a space target recognition result.

2. A near-Earth space target recognition method according to claim 1, characterized in that: The acquiring the morphological information of the space target and the center of mass position of the space target based on the space-based measurement picture includes: Generate a filter operator matrix based on a preset filter, and perform filtering processing on the space-based measurement image based on the filter operator matrix to obtain a filtered image; Binarize the filtered image based on a preset grayscale threshold to generate a grayscale image; Performing edge extraction on the grayscale image based on a preset first edge detection algorithm to obtain morphological information of the space target to be identified; The target contour of the grayscale image is identified based on a preset second edge detection algorithm, and the center of mass position of the space target is acquired based on the target contour.

3. A near-Earth space target recognition method as claimed in claim 2, characterized in that: The observation data includes a space target position vector and an observation platform position vector, wherein the space target position vector is a position vector of the space target relative to the center of the earth, and the observation platform position vector is a position vector of the observation platform relative to the center of the earth; The step of constructing a geometric relationship model based on the observation data comprises: Constructing the geometric relationship based on the space target position vector and the observation platform position vector; Constructing a geocentric distance equation of the space target based on the Laplace method, and obtaining initial orbit parameters based on the geometric relationship and the geocentric distance equation; An observation model and an error model are constructed based on the initial orbit parameters, and the initial orbit parameters are optimized based on the observation model and the error model to determine the target orbit and obtain orbit characteristics.

4. A near-Earth space target recognition method as claimed in claim 1, characterized in that: The preliminary recognition result includes basic information of the space target to be recognized, and the verification of the preliminary recognition result based on the historical cataloging data to obtain the space target recognition result includes: matching a verification target in the historical catalog database based on the basic information; The orbit error is calculated based on the basic information and the verification target, the preliminary recognition result is verified based on a preset orbit error, and the space target recognition result is obtained based on the verification result.

5. A near-Earth space target recognition device, characterized in that: include: Image feature extraction module, track feature extraction module and fusion recognition module; The image feature extraction module is used to obtain a space-based measurement picture, obtain morphological information of a space target to be identified and a centroid position of the space target to be identified based on the space-based measurement picture, and obtain image features based on the morphological information and the centroid position; The orbit feature extraction module is used to obtain observation data based on an observation platform, build a geometric relationship model based on the observation data, and determine the target orbit of the space target to be identified based on the geometric relationship model and the Laplace method to obtain orbit features; The geometric relationship model is a geometric relationship model between the space target to be identified, the observation platform and the center of the earth; The fusion recognition module is used to obtain fusion features based on the image features and the orbital features, and to identify the space target to be identified based on the fusion features and the historical catalog database to obtain a space target identification result, including: performing standardization processing on the image features and the orbital features to obtain standard image features and standard orbital features; Based on the standard image features and the standard orbit features, splicing is performed to obtain fusion features; the fusion features are input into a pre-trained target recognition model to identify the space target to be identified and obtain a preliminary recognition result; based on the historical cataloging data, the preliminary recognition result is verified to obtain a space target recognition result.

6. A near-Earth space target identification device as claimed in claim 5, characterized in that: The image feature extraction module is used to: Generate a filter operator matrix based on a preset filter, and perform filtering processing on the space-based measurement image based on the filter operator matrix to obtain a filtered image; Binarize the filtered image based on a preset grayscale threshold to generate a grayscale image; Extracting edges of the grayscale image based on a preset first edge detection algorithm to obtain morphological information of the space target; The target contour of the grayscale image is identified based on a preset second edge detection algorithm, and the center of mass position of the space target is acquired based on the target contour.

7. A near-Earth space target identification device as claimed in claim 6, characterized in that: The observation data includes a space target position vector and an observation platform position vector, wherein the space target position vector is a position vector of the space target relative to the center of the earth, and the observation platform position vector is a position vector of the observation platform relative to the center of the earth; the orbit feature extraction module is used to: Constructing the geometric relationship based on the space target position vector and the observation platform position vector; Constructing a geocentric distance equation of the space target based on the Laplace method, and obtaining initial orbit parameters based on the geometric relationship and the geocentric distance equation; An observation model and an error model are constructed based on the initial orbit parameters, and the initial orbit parameters are optimized based on the observation model and the error model to determine the target orbit and obtain orbit characteristics.

8. A near-Earth space target identification device as claimed in claim 7, characterized in that: The preliminary recognition result includes basic information of the space target to be recognized, and the fusion recognition module is used to: matching a verification target in the historical catalog database based on the basic information; The orbit error is calculated based on the basic information and the verification target, the preliminary recognition result is verified based on a preset orbit error, and the space target recognition result is obtained based on the verification result.

Images