Methods, devices, and media for suppressing stellar false alarms for ground-based telescope observations

Abstract

The embodiment of the application provides a method, device and medium for suppressing star false alarm of a ground-based telescope, analyzes a constructed false alarm data set, determines characteristic coefficient data such as a star speed size approximation coefficient, a star speed direction approximation coefficient, a star eccentricity approximation coefficient, a star phase angle approximation coefficient and a frame breaking penalty coefficient corresponding to a false alarm result, identifies star false alarms in a detection result in combination with a PSO-SVM identification algorithm, solves the problem that in a star dense scene, the same threshold cannot be used to distinguish stars from space targets in different scenes in space target detection, and further leads to the problem of false alarms, improves the detection accuracy of star false alarms, and effectively suppresses star false alarms in false alarm data.

Description

Technical Field

[0001] This application relates to the field of aerospace technology, and in particular to a method and apparatus for suppressing false alarms in stellar observations by ground-based telescopes, an electronic device, and a storage medium. Background Technology

[0002] False alarms are a significant problem in space target detection based on ground-based telescope observations. Inappropriate detection algorithms can lead to a large number of false alarms, causing unnecessary waste of resources for early warning systems, security systems, and related human and material resources, and also greatly affecting the real-time performance of subsequent processing.

[0003] In practical applications, the conventional research approach is to remove images with poor imaging quality (excessive noise, heavy stray light, cloud obstruction, etc.) to avoid false alarms to a certain extent. However, in order to discover more faint targets, the camera exposure time is usually increased and the detection signal-to-noise ratio threshold is lowered during image preprocessing. This results in dense distribution of star images, generating many false temporal motion trajectories that are easily confused with spatial targets, thus affecting the detection effect. Summary of the Invention

[0004] This application provides a method for suppressing false alarms in stellar observations by ground-based telescopes. This method addresses the problem that in dense stellar scenes, the same threshold cannot distinguish between stars and space targets in different scenarios, leading to false alarms.

[0005] Accordingly, embodiments of this application also provide a device for suppressing false alarms in stellar observations by ground-based telescopes, an electronic device, and a storage medium to ensure the implementation and application of the above methods.

[0006] To address the aforementioned problems, this application discloses a method for suppressing false alarms in stellar observations using ground-based telescopes. The method includes:

[0007] Construct a false alarm dataset;

[0008] Detect the shape parameter data, trajectory data, and the fragmented frame data corresponding to the trajectory data in the false alarm dataset;

[0009] Based on the trajectory data, determine the stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset;

[0010] The stellar eccentricity approximation coefficient and stellar phase angle approximation coefficient of the false alarm dataset are determined based on the shape parameter data.

[0011] The frame breakage penalty coefficient of the false alarm dataset is determined based on the frame breakage data;

[0012] Based on the stellar velocity magnitude convergence coefficient, the stellar velocity direction convergence coefficient, the stellar eccentricity convergence coefficient, the stellar phase angle convergence coefficient, and the frame break penalty coefficient, the false alarm dataset is identified using a support vector machine algorithm based on particle swarm optimization to obtain stellar false alarms;

[0013] False alarms from the stars are suppressed.

[0014] This application also discloses an electronic device, including: a processor; and a memory storing executable code thereon, which, when executed, causes the processor to perform one or more of the methods for suppressing false alarms in stellar observations by ground-based telescopes as described in the embodiments of this application.

[0015] This application also discloses a machine-readable medium storing executable code, which, when executed, causes a processor to perform one or more of the methods for suppressing false alarms in stellar observations by ground-based telescopes as described in this application.

[0016] Compared with the prior art, the embodiments of this application have the following advantages:

[0017] In this embodiment, the constructed false alarm dataset is analyzed to determine the characteristic coefficients corresponding to the false alarm results, such as the stellar velocity magnitude convergence coefficient, stellar velocity direction convergence coefficient, stellar eccentricity convergence coefficient, stellar phase angle convergence coefficient, and frame break penalty coefficient. Combined with the PSO-SVM recognition algorithm, stellar false alarms in the detection results are identified. This solves the problem that in dense stellar scenes, the same threshold cannot distinguish between stars and space targets in different scenes, thus causing false alarms. It improves the detection accuracy of stellar false alarms and effectively suppresses stellar false alarms in the false alarm data. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the steps of an embodiment of a method for suppressing false alarms in stellar observations using a ground-based telescope, as described in this application.

[0019] Figure 2a This is a schematic diagram of stellar eccentricity processing, representing an embodiment of a method for suppressing false alarms in stellar observations using a ground-based telescope according to this application.

[0020] Figure 2b This is a schematic diagram of stellar phase angle processing, representing an embodiment of a method for suppressing false alarms in stellar observations using a ground-based telescope according to this application.

[0021] Figure 3 This is a schematic diagram of the iterative process of a support vector machine algorithm based on particle swarm optimization, which is an embodiment of a method for suppressing false alarms in stellar observations by ground-based telescopes according to this application.

[0022] Figure 4 This is a schematic diagram of a classification confusion matrix of an embodiment of a method for suppressing false alarms in stellar observations using ground-based telescopes, as described in this application.

[0023] Figure 5 This is a schematic diagram of the structure of a device provided in an embodiment of this application. Detailed Implementation

[0024] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0025] Reference Figure 1 This is a flowchart illustrating the steps of an embodiment of a method for suppressing false alarms in stellar observations using a ground-based telescope, as described in this application. The steps include:

[0026] Step 101: Construct a false alarm dataset.

[0027] In step 101, a false alarm dataset is constructed, which is the data that needs to be identified as a stellar false alarm.

[0028] Specifically, multiple observation images of the telescope under different observation scenarios in staring mode can be collected. After removing images with excessive stray light and noise, the dataset composed of observation images is detected based on traditional algorithms. The detection results are labeled as false alarms to construct the corresponding false alarm dataset.

[0029] Step 102: Detect the shape parameter data, trajectory data, and the broken frame data corresponding to the trajectory data in the false alarm dataset.

[0030] In step 102, the false alarm dataset is detected to identify the shape parameter data, trajectory data, and corresponding frame breaks in the trajectory data. Based on the detected shape parameter data, trajectory data, and corresponding frame breaks in the false alarm dataset, characteristic coefficients of false alarms, such as stellar velocity magnitude convergence coefficient, stellar velocity direction convergence coefficient, stellar eccentricity convergence coefficient, stellar phase angle convergence coefficient, and frame break penalty coefficient, can be further calculated to achieve the identification of stellar false alarms.

[0031] The shape parameter data consists of the shape parameters of the stars in the observed images, including their phase angles and eccentricities. The trajectory data consists of trajectories composed of trajectory points identified from multiple consecutive frames of observed images, with each trajectory corresponding to a set of trajectory points. The frame-breaking data corresponding to the trajectory data describes the frame-breaking situation of the trajectory.

[0032] Step 103: Determine the stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset based on the trajectory data.

[0033] In this embodiment, stellar proximity refers to the degree of similarity between the trajectory characteristics (velocity magnitude, direction, star image eccentricity, and phase angle) of false alarm data and the inherent characteristics of a known star (due to the extreme distance between stars, they appear to be stationary or moving in uniform straight lines during observation, and their star image shape is stable). That is, the motion characteristics of real space targets (such as debris) are significantly different from those of stars, while the characteristics of stellar false alarms (such as star misidentification) are highly similar to those of stars. In order to quantify the degree of proximity between stellar false alarms and stellar characteristics through multiple dimensions, distinguish stellar false alarms (characteristics highly similar to stars) from non-stellar targets, and provide a quantitative judgment basis for the subsequent accurate elimination of stellar false alarms caused by star misidentification, it is necessary to determine the corresponding stellar velocity magnitude proximity coefficient, stellar velocity direction proximity coefficient, stellar eccentricity proximity coefficient, and stellar phase angle proximity coefficient based on stellar proximity.

[0034] Specifically, in step 103, the stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset can be determined based on the trajectory data corresponding to the false alarm dataset.

[0035] In one embodiment, the trajectory data includes a set of trajectory points, the set of trajectory points including a start point and an end point, and step 103 specifically includes:

[0036] Obtain the stellar velocity and the frame number difference between the start point and the end point;

[0037] Determine the straight line trajectory corresponding to the set of trajectory points;

[0038] Calculate the first foot of the perpendicular from the starting point to the straight trajectory, and calculate the second foot of the perpendicular from the ending point to the straight trajectory;

[0039] Calculate the average trajectory velocity and the magnitude of the trajectory velocity based on the starting point, the ending point, the straight trajectory, the frame number difference, the first perpendicular foot, and the second perpendicular foot;

[0040] The stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset are determined based on the stellar motion velocity, the average trajectory velocity, and the magnitude of the trajectory velocity.

[0041] In this embodiment of the application, the global velocity of the star is first calculated, and the solution process is as follows:

[0042] The direction of stellar motion velocity is calculated based on any two consecutive star images in the observation image queue. According to the results of star image matching, three common stars with the largest brightness and area and the best image preservation in the two images are found as control points to form a control triangle, excluding those triangles with collinear points and sides that are too short. The movement of the star image is caused by the change in telescope pointing, which can be ignored as an affine transformation. The model is constructed as shown in the following set of equations (1):

[0043] (1)

[0044] in, Let be the position of the star in the (k+1)th frame in the pixel coordinate system. Let be the position of the star in the k-th frame in the pixel coordinate system. This represents the translation of the star's coordinates. For the image rotation angle, Let be the affine scaling factor. To facilitate solving the system of equations, equation (1) is transformed based on the following system of equations (2) to obtain system of equations (3):

[0045] (2)

[0046] (3)

[0047] in, Given the conditions, The solution is to be found. Substituting the three sets of control points will solve the problem. Due to the large distortion and star image blurring, there is a certain error in the centroid positioning. Therefore, equation (3) does not have an exact solution. The optimal solution is obtained by the least squares method, and equation (4) can be further obtained:

[0048] (4)

[0049] Depend on It can be seen that the transformation of the stellar background is approximately a rigid body transformation, and This indicates that, within the application context of this embodiment, the movement of the stellar background can be considered as linear motion. Therefore, the stellar velocity in the pixel coordinate system can be calculated as follows: Then the direction of motion of a star can be represented as a vector. .

[0050] Assuming the detected trajectory As a set of trajectory points, where... for The starting trajectory point (i.e., the starting endpoint). for The terminal trajectory points (i.e., the endpoints). After least squares calibration, the straight-line trajectory corresponding to the set of trajectory points is determined as follows: Where k represents The slope in pixel coordinates, b represents The intercept in the pixel coordinate system is used to find the starting endpoint through the system of equations (5). and endpoint exist The first foot of the pendulum Second foot :

[0051] (5)

[0052] At this point, based on the following formulas (6) and (7), according to the starting point... End point Straight line trajectory Frame number difference between the start and end points, first perpendicular foot Second foot The average velocity of the trajectory can be obtained separately. and trajectory velocity magnitude :

[0053] (6)

[0054] (7)

[0055] in, express The frame number difference between the first and last trajectory points (i.e., the frame number difference between the start and end points).

[0056] Therefore, based on the speed of star motion Average velocity of the trajectory and trajectory velocity magnitude Determine the magnitude and direction convergence coefficients of stellar velocities in the false alarm dataset.

[0057] In one embodiment, determining the stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset based on the stellar motion velocity, the trajectory average velocity, and the trajectory velocity magnitude includes:

[0058] Calculate the magnitude approximation coefficient of the star's velocity based on the star's motion velocity and the magnitude of its trajectory velocity;

[0059] Calculate the stellar velocity direction convergence coefficient based on the stellar motion velocity and the average velocity of the trajectory;

[0060] The magnitude approximation coefficient of the stellar velocity is calculated using the following formula:

[0061]

[0062] in, The magnitude approximation coefficient of the star's velocity is... The magnitude of the trajectory velocity. The velocity of the star;

[0063] The stellar velocity direction convergence coefficient is calculated using the following formula:

[0064]

[0065] in, This is the stellar velocity direction approximation coefficient. The average velocity of the trajectory, The velocity of the star's motion, the express and The angle between two vectors.

[0066] In this embodiment, the stellar velocity magnitude convergence coefficient can be further calculated using formula (8) based on the magnitude of the stellar motion velocity and the trajectory velocity, and the stellar velocity direction convergence coefficient can be calculated using formula (9) based on the stellar motion velocity and the average trajectory velocity.

[0067] (8)

[0068] (9)

[0069] in, express and The angle between two vectors is obtained by the following formula (10):

[0070] (10)

[0071] in, This represents the approximation coefficient between the magnitude of the trajectory velocity and the magnitude of the star's velocity (stellar velocity approximation coefficient). This represents the convergence coefficient between the trajectory direction and the star's direction (the star's velocity direction convergence coefficient). For the magnitude of the trajectory velocity, The speed of star motion, This represents the average velocity of the trajectory. Specifically, and The closer the value is to 0, the greater the similarity between the target and the star's motion, and the higher the probability of a false alarm.

[0072] Step 104: Determine the stellar eccentricity approximation coefficient and stellar phase angle approximation coefficient of the false alarm dataset based on the shape parameter data.

[0073] In step 104, the stellar eccentricity approximation coefficient and stellar phase angle approximation coefficient of the false alarm dataset can be determined based on the shape parameter data.

[0074] Specifically, in one embodiment, step 104 includes the following steps:

[0075] The step of determining the stellar eccentricity approximation coefficient of the false alarm dataset based on the shape parameter data includes:

[0076] The stellar eccentricity is determined based on the shape parameters.

[0077] Determine the set of trajectory point eccentricities corresponding to the set of trajectory points;

[0078] The stellar eccentricity approximation coefficient is calculated based on the stellar eccentricity and the set of trajectory point eccentricities.

[0079] The stellar eccentricity approximation coefficient is calculated using the following formula:

[0080]

[0081] in, The eccentricity approximation coefficient of the star is... Let k be the set of trajectory points, and k be the trajectory point currently traversed. The eccentricity of the star is... Let be the eccentricity of the k-th trajectory point in the set of trajectory point eccentricities. ( ) represents the number of trajectory points in the set of trajectory points.

[0082] The step of determining the stellar phase angle approximation coefficient of the false alarm dataset based on the shape parameter data includes:

[0083] Determine the stellar phase angle based on the shape parameters;

[0084] Determine the set of trajectory point phase angles corresponding to the set of trajectory points;

[0085] The stellar phase angle convergence coefficient is calculated based on the stellar phase angle and the set of trajectory point phase angles;

[0086] The stellar phase angle approximation coefficient is calculated using the following formula:

[0087]

[0088] in, The phase angle approximation coefficient of the star is... Let k be the set of trajectory points, and k be the trajectory point currently traversed. The phase angle of the star, Let the phase angle be the phase angle of the k-th trajectory point in the set of trajectory point phase angles. ( ) represents the number of trajectory points in the set of trajectory points.

[0089] According to the camera imaging principle, with a fixed exposure time, the target's speed is directly proportional to the length of the star image. Therefore, this application embodiment further describes stellar false alarms from the perspectives of star image eccentricity and phase angle. To stably acquire stellar image characteristics, this application embodiment adopts a statistical distribution approach.

[0090] In the observation images obtained in this application embodiment, stars occupy the vast majority of the star images. Therefore, by statistically analyzing the probability distribution of the phase angle and eccentricity of the star images, the shape parameters with a larger proportion are selected as the basis for star shape. Thus, the eccentricity and phase angle of the star can be determined, which can effectively ensure its accuracy and robustness.

[0091] Reference Figure 2a This is a schematic diagram of stellar eccentricity processing, representing an embodiment of a method for suppressing false alarms in stellar observations using ground-based telescopes, and, with reference to... Figure 2b This is a schematic diagram of stellar phase angle processing, which is an embodiment of a method for suppressing false alarms in stellar observations using a ground-based telescope, according to this application. Figure 2a and Figure 2b The statistical results of the probability distribution of eccentricity and phase angle of stars in observation images under different scenarios reflect the distribution of star shape parameters well, so that the eccentricity and phase angle of stars can be determined accordingly.

[0092] Assuming the stellar eccentricity obtained from statistical results is... The phase angle of the star is In the set of trajectory points In the equation, the set of eccentricity of trajectory points and the set of phase angle of trajectory points are respectively... and At this point, the stellar eccentricity approximation coefficient can be obtained using the following formula (11), and the stellar phase angle approximation coefficient can be obtained using the following formula (12):

[0093] (11)

[0094] (12)

[0095] in, The eccentricity of the target star image is represented by the approximation coefficient of the star's eccentricity (star eccentricity approximation coefficient). The approximation coefficient between the phase angle of the target star image and the phase angle of the star (stellar phase angle approximation coefficient). Let k be the set of trajectory points, where k is the trajectory point currently visited. For stellar eccentricity, Let be the eccentricity of the k-th trajectory point in the set of trajectory point eccentricities. For stellar phase angle, Let the phase angle be the phase angle of the k-th trajectory point in the set of trajectory point phase angles. ( ) represents the number of trajectory points in the trajectory point set.

[0096] Specifically, and The closer the value is to 0, the greater the similarity between the target and the star's shape, and the higher the probability of a false alarm.

[0097] Step 105: Determine the frame break penalty coefficient of the false alarm dataset based on the frame break data.

[0098] In this embodiment, frame breakage refers to the discontinuous breakage of the trajectory of false alarm data in continuous observation images. That is, due to sampling errors of the observation equipment, the target briefly moving out of the field of view, or noise interference, the trajectory is missing trajectory points in some frames of images, forming an incomplete broken trajectory. In other words, the real target trajectory is usually continuous and complete, while the false alarm trajectory is prone to frequent frame breakage due to interference. In order to quantify the severity of trajectory frame breakage (such as the number of frame breakages and the length of the frame breakage interval), a penalty mechanism is set to distinguish between false alarm trajectories with frequent frame breakages and continuous and stable real trajectories. This provides a quantitative indicator for subsequent identification of false alarms based on frame breakage features. It is necessary to determine the corresponding frame breakage penalty coefficient based on the frame breakage.

[0099] Trajectory correlation for broken frame queues is a key indicator of detection algorithm performance, but it is also one of the main sources of false alarms. Broken frames can occur in three ways:

[0100] 1. The non-uniform light coverage causes the star images of some trajectory points to be obscured;

[0101] 2. The target trajectory point is covered or overlapped by other stars, causing the centroid extraction to deviate from the search path;

[0102] 3. Due to the limitations of the association algorithm, trajectory points arranged in reverse time sequence may occur, which need to be excluded.

[0103] In one embodiment, step 105 specifically includes:

[0104] Obtain the total number of frames in the detection queue of the trajectory data;

[0105] The frame break penalty coefficient is calculated based on the total number of frames in the detection queue of the trajectory data and the number of trajectory points in the trajectory point set.

[0106] The frame breakage penalty coefficient is calculated using the following formula:

[0107]

[0108] in, The frame break penalty coefficient is... This represents the total number of frames in the detection queue of the trajectory data. The number of trajectory points in the set of trajectory points.

[0109] In this embodiment, the total number of frames in the trajectory data detection queue and the number of trajectory points in the trajectory point set can be used as fragmented frame data. Specifically, the more fragmented frames a trajectory has, the higher the probability that the trajectory is a false detection. Therefore, this embodiment assigns a specific value to each detected trajectory. Set the corresponding frame drop penalty coefficient to evaluate the probability of it being a false alarm. Specifically, the frame drop penalty coefficient is calculated by the following formula (13):

[0110] (13)

[0111] in, This represents the total number of frames in the detection queue of the trajectory data. The number of trajectory points in the set of trajectory points. Representing the trajectory The frame drop penalty is the ratio of the number of times a frame is dropped to the total number of frames in the queue. Specifically, The closer the value is to 0, the higher the reliability of the trajectory.

[0112] Step 106: Based on the stellar velocity magnitude convergence coefficient, the stellar velocity direction convergence coefficient, the stellar eccentricity convergence coefficient, the stellar phase angle convergence coefficient, and the frame break penalty coefficient, the false alarm dataset is identified using a support vector machine algorithm based on particle swarm optimization to obtain stellar false alarms.

[0113] In step 106, the extracted feature coefficient data are combined with a support vector machine algorithm based on particle swarm optimization to obtain the final recognition result.

[0114] The particle swarm optimization-support vector machine algorithm is also known as the PSO-SVM algorithm. In other embodiments, other classification methods can be used to identify false alarms in the stellar dataset based on stellar velocity magnitude convergence coefficients, stellar velocity direction convergence coefficients, stellar eccentricity convergence coefficients, stellar phase angle convergence coefficients, and frame break penalty coefficients.

[0115] One embodiment of this application includes the following steps before step 106:

[0116] The support vector machine algorithm to be trained, the particle swarm optimization parameters, and the fitness function are obtained. The kernel function of the support vector machine algorithm to be trained is a Gaussian kernel function. The support vector machine algorithm to be trained has a corresponding penalty factor, and the Gaussian kernel function has a corresponding kernel scale.

[0117] Based on the particle swarm optimization parameters and the fitness function, the particle swarm optimization method is used to optimize the penalty factor and the kernel scale of the support vector machine algorithm to be trained, so as to obtain the target penalty factor and the target kernel scale.

[0118] The support vector machine algorithm corresponding to the target penalty factor and the target kernel scale is used as the support vector machine algorithm based on particle swarm optimization.

[0119] In this embodiment of the application, before step 106, it is necessary to determine the support vector machine algorithm based on particle swarm optimization.

[0120] Specifically, the Support Vector Machine (SVM) algorithm to be trained is obtained. This algorithm has a corresponding kernel function and penalty factor. The penalty factor and kernel function are key factors affecting the performance of the SVM algorithm. The penalty factor controls the trade-off between maximizing the margin and minimizing the classification error. The kernel function is the core of the SVM algorithm for handling nonlinear problems, mapping low-dimensional inseparable data to a high-dimensional space to make it linearly separable without complex high-dimensional calculations.

[0121] The Gaussian kernel function performs well in processing sample data with high feature dimensionality and is not limited by the number of samples. Furthermore, it also achieves good results in classification where prior knowledge is insufficient. Because the various features of the sample data in this embodiment have complex coupling relationships, the Gaussian kernel function is chosen as the kernel function for the support vector machine algorithm to be trained.

[0122] As mentioned above, the penalty factor in the SVM algorithm... and Gaussian kernel bandwidth The kernel size (i.e., the Gaussian kernel function's kernel scale) determines the classification performance. Related schemes typically determine the penalty factor and kernel scale based on experience or trial and error; however, this approach lacks sufficient theoretical basis and has significant uncertainty. Obtaining optimal classification parameters often requires continuous adjustment based on experience, which is a waste of manpower and time. Therefore, this application's embodiment employs a support vector machine algorithm based on particle swarm optimization, combined with the feature coefficient data of the five types of false alarm data to be identified calculated above, to identify stellar false alarms.

[0123] The Particle Swarm Optimization (PSO) algorithm is a population-based parallel global heuristic search algorithm. Individual members in the population cooperate and share information to find the optimal solution. Its advantages include fewer parameters to adjust, fast convergence, less susceptibility to local optima, and applicability to optimization in high-dimensional spaces. Its specific steps can be represented as follows:

[0124] Step 1: Initialize the population. Randomly generate N particles, each represented as a vector with dimensions equal to the number of target parameters D. The initial position and initial velocity of each particle are randomly determined. Then, the i-th... The particle in the first The position and velocity vectors for the next iteration are given by the following formula (14):

[0125] (14)

[0126] in, It is the first The particle in the first The position of the next iteration. It is the first The particle in the first The velocity vector of the next iteration, where N represents the number of particles and D is the number of target parameters to be optimized.

[0127] Step 2: Design the corresponding fitness function based on the optimization objective. .

[0128] Step 3: Based on the fitness function, select the first... The generation The historical optimal position of each particle and the optimal position of the group It is expressed as the following formula (15):

[0129] (15)

[0130] in, For the first The generation The optimal position of each particle in its historical records. For the first The generation The optimal position of the population of individual particles. For N particles in the first... The position of the next iteration.

[0131] Step Four: The particle velocity and position update formulas are as follows: Formula (16):

[0132] (16)

[0133] in, Indicates the first Sub-particles speed; Indicates the first Sub-particles Location; Indicates the inertia factor; For the first The generation The optimal position of each particle in its historical records; For the first The generation The optimal group position of each particle; and Represents the learning factor, with a value range of 1000-12000. ; and For taking values Random numbers between [a certain range]. To avoid the algorithm missing a local optimum, it is usually limited to [a certain value]. The value range is 1 / 5 of the length of the particle position value interval, that is... Need to meet To balance the global and local search capabilities of the PSO algorithm, an inertia factor is set. The linear decreasing trend is shown in the following formula (17):

[0134] (17)

[0135] in, Indicates the initial setting factor. This represents the final factor value. This indicates the maximum number of iterations.

[0136] This application's embodiments utilize PSO to optimize SVM parameters, transforming the SVM parameter selection problem into a global search optimization problem that is stable and accurate. In one embodiment, the specific process of using PSO to optimize SVM parameters to obtain a support vector machine algorithm based on particle swarm optimization includes the following steps:

[0137] Step 201: Obtain the extracted feature data and annotation set, normalize the data, and divide the data into training set and test set using the SVM algorithm.

[0138] Step 202: Select the Gaussian kernel function as the kernel function for the SVM algorithm to be trained, and set the kernel size σ and penalty coefficient c to be optimized using PSO. Subsequently, based on the particle swarm optimization parameters and fitness function of PSO, the penalty factor and kernel size of the support vector machine algorithm to be trained are optimized using the particle swarm optimization method. Specifically, the basic PSO parameters (i.e., particle swarm optimization parameters) are set, and the specific setting method can be found in Table 1 below. The fitness function is also defined. The formula is as follows (18):

[0139] (18)

[0140] in, Indicates the size of the test set. This indicates the number of correctly classified sample data in the test set. The closer the value is to 1, the better the classification effect.

[0141] Table 1 Particle Swarm Optimization Parameter Settings

[0142]

[0143] Step 203: Initialize the population, including particle velocity and particle position, and build a corresponding SVM model based on the information of each particle, and perform training and classification.

[0144] Step 204: Calculate the SVM classification accuracy of each particle based on the training results, calculate the fitness of each particle, and determine the current historical optimal position of each particle and the optimal position of the group.

[0145] Step 205: Determine whether the iteration limit has been reached or the optimal SVM parameters (i.e., the target penalty factor and the target kernel size) have been obtained. If not, update the next generation velocity and position of each particle, redetermine σ and c, and return to step 204 for retraining; if yes, output the optimal SVM model (i.e., the support vector machine algorithm based on particle swarm optimization) corresponding to the target penalty factor and the target kernel size.

[0146] Step 206: Test the model using the test set and evaluate the classification performance based on the test results.

[0147] Based on steps 201 to 206 above, the penalty factor and kernel scale of the support vector machine algorithm to be trained are optimized using the particle swarm optimization method according to the particle swarm optimization parameters and fitness function, so as to obtain the target penalty factor and target kernel scale, and thus the support vector machine algorithm corresponding to the target penalty factor and target kernel scale is used as the support vector machine algorithm based on particle swarm optimization.

[0148] Reference Figure 3This is a schematic diagram of the iterative process of a support vector machine algorithm based on particle swarm optimization, which is an embodiment of a method for suppressing false alarms in stellar observations by ground-based telescopes according to this application.

[0149] Figure 3 This graph illustrates the optimization process of Gaussian SVM based on the PSO algorithm. The horizontal axis represents the number of iterations, and the vertical axis represents the classification error, i.e., the proportion of misclassified samples to the total number of samples. Black dots represent the historical optimal position of the group after each update; dark gray dots (i.e.,...) represent the optimal position of the group after each update. Figure 3 The points marked with penalty factor, kernel size and minimum error represent the optimal parameter results after reaching the maximum number of iterations. That is, after this point, the algorithm reaches final convergence and the classification result cannot be further optimized. In this example, the final optimized result is: target penalty factor is 127.25, target kernel size is 86.08, and minimum error after optimization is 0.033.

[0150] In one embodiment, step 106 specifically includes:

[0151] The stellar velocity magnitude approximation coefficient, stellar velocity direction approximation coefficient, stellar eccentricity approximation coefficient, stellar phase angle approximation coefficient, and frame break penalty coefficient corresponding to the false alarm dataset are used as the feature coefficient data of the false alarm dataset;

[0152] The feature coefficient data is input into the particle swarm optimization-based support vector machine algorithm to obtain the classification result of the particle swarm optimization-based support vector machine algorithm for the false alarm dataset, and the classification result includes the stellar false alarms.

[0153] In this embodiment, after obtaining the support vector machine algorithm based on particle swarm optimization, the stellar velocity magnitude convergence coefficient, stellar velocity direction convergence coefficient, stellar eccentricity convergence coefficient, stellar phase angle convergence coefficient, and frame break penalty coefficient corresponding to the false alarm dataset can be used as feature coefficient data of the false alarm dataset. These are then input into the support vector machine algorithm based on particle swarm optimization for classification, yielding the classification result output by the algorithm. This classification result is used to determine whether the false alarm dataset is a stellar false alarm. The classification result can include stellar false alarms.

[0154] Step 107: Suppress false alarms from the stars.

[0155] In step 107, after identifying stellar false alarms in the false alarm data, stellar false alarms can be suppressed.

[0156] To further verify the method for suppressing false alarms in stellar observations by ground-based telescopes as described in this application, 5000 observation images (images with excessive stray light and noise have been removed) were collected under different observation scenarios in staring mode. Traditional algorithms were used to detect false alarms in each dataset, and the results were labeled to determine whether they were false alarms. A corresponding false alarm data test set was created, as shown in the table below. This table shows the data volume, total number of detected targets, and number of false alarms for different datasets. This false alarm data test set is used to test the method for suppressing false alarms in stellar observations by ground-based telescopes as described in this application.

[0157] Table 2 Application of traditional algorithms on datasets

[0158]

[0159] After constructing a false alarm data test set, the false alarm data in the test set is identified using a method for suppressing false alarms in stellar observations by ground-based telescopes, as shown in the embodiments of this application, and the stellar false alarms are determined to obtain the test results.

[0160] Reference Figure 4 This is a schematic diagram of a classification confusion matrix for an embodiment of a method for suppressing false alarms in stellar observations using ground-based telescopes, as described in this application.

[0161] Test results passed Figure 4 The classification confusion matrix represents the comprehensive test results on sample data from various observation periods. The horizontal axis represents the true category, and the vertical axis represents the predicted category of the false alarm suppression method in this embodiment. Different gray-scale blocks are used to distinguish different data. Specifically, two blocks 1 represent the number and total percentage of correctly identified spatial targets and false alarms, respectively; two blocks 2 represent the number and total percentage of incorrectly identified samples; four blocks 3 represent the percentage of correctly identified targets within that category; and block 4 represents the total number and percentage of correctly identified targets. This shows that the recognition accuracy reaches 96.7%, indicating that the false alarm suppression method in this embodiment has good robustness.

[0162] In this embodiment, the constructed false alarm dataset is analyzed to determine the characteristic coefficients corresponding to the false alarm results, such as the stellar velocity magnitude convergence coefficient, stellar velocity direction convergence coefficient, stellar eccentricity convergence coefficient, stellar phase angle convergence coefficient, and frame break penalty coefficient. Combined with the PSO-SVM recognition algorithm, stellar false alarms in the detection results are identified. This solves the problem that in dense stellar scenes, the same threshold cannot distinguish between stars and space targets in different scenes, thus leading to false alarms. It improves the detection accuracy of stellar false alarms and effectively suppresses stellar false alarms in the false alarm data. The recognition accuracy of stellar false alarms can reach 96.7%, effectively suppressing stellar false alarms in the false alarm data.

[0163] This application also provides a non-volatile readable storage medium storing one or more modules (programs). When these modules are applied to a device, they enable the device to execute the instructions for the method steps in this application.

[0164] This application provides one or more machine-readable media storing instructions that, when executed by one or more processors, cause an electronic device to perform one or more of the methods described in the above embodiments. In this application, the electronic device includes various types of devices such as terminal devices and servers (clusters).

[0165] The embodiments of this disclosure can be implemented as an apparatus configured as desired using any suitable hardware, firmware, software, or any combination thereof, including electronic devices such as terminal devices, servers (clusters), etc. Figure 5 An exemplary apparatus 500 is schematically shown that can be used to implement the various embodiments described in this application.

[0166] In one embodiment, Figure 5 An exemplary device 500 is shown, which includes one or more processors 502, a control module (chipset) 504 coupled to at least one of the processors 502, a memory 506 coupled to the control module 504, a non-volatile memory (NVM) / storage device 508 coupled to the control module 504, one or more input / output devices 510 coupled to the control module 504, and a network interface 512 coupled to the control module 504.

[0167] Processor 502 may include one or more single-core or multi-core processors, and processor 502 may include any combination of general-purpose processors or special-purpose processors (e.g., graphics processors, application processors, baseband processors, etc.). In some embodiments, device 500 can serve as a terminal device, server (cluster), or other device as described in the embodiments of this application.

[0168] In some embodiments, apparatus 500 may include one or more computer-readable media (e.g., memory 506 or NVM / storage device 508) having instructions 514 and one or more processors 502 that are combined with the one or more computer-readable media and configured to execute the instructions 514 to implement the module and thus perform the actions described in this disclosure.

[0169] In one embodiment, the control module 504 may include any suitable interface controller to provide any suitable interface to at least one of the processors 502 and / or any suitable device or component communicating with the control module 504.

[0170] The control module 504 may include a memory controller module to provide an interface to the memory 506. The memory controller module may be a hardware module, a software module, and / or a firmware module.

[0171] Memory 506 may be used, for example, to load and store data and / or instructions 514 for device 500. In one embodiment, memory 506 may include any suitable volatile memory, such as suitable DRAM. In some embodiments, memory 506 may include double data rate type quad synchronous dynamic random access memory (DDR4 SDRAM).

[0172] In one embodiment, the control module 504 may include one or more input / output controllers to provide an interface to the NVM / storage device 508 and (one or more) input / output devices 510.

[0173] For example, NVM / storage device 508 may be used to store data and / or instructions 514. NVM / storage device 508 may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable (one or more) non-volatile storage devices (e.g., one or more hard disk drive (HDD), one or more optical disc (CD) drives, and / or one or more digital universal optical disc (DVD) drives).

[0174] NVM / storage device 508 may include storage resources that are physically part of a device on which device 500 is mounted, or that are accessible to the device but do not necessarily have to be part of the device. For example, NVM / storage device 508 may be accessed via a network via one or more input / output devices 510.

[0175] One or more input / output devices 510 may provide an interface for device 500 to communicate with any other suitable device. Input / output devices 510 may include communication components, audio components, sensor components, etc. A network interface 512 may provide an interface for device 500 to communicate via one or more networks. Device 500 may wirelessly communicate with one or more components of a wireless network according to any of one or more wireless network standards and / or protocols, such as accessing a wireless network based on communication standards, such as WiFi, 2G, 3G, 4G, 5G, etc., or a combination thereof.

[0176] In one embodiment, at least one of the processors 502 may be logically packaged with one or more controllers (e.g., memory controller modules) of the control module 504. In one embodiment, at least one of the processors 502 may be logically packaged with one or more controllers of the control module 504 to form a system-in-package (SiP). In one embodiment, at least one of the processors 502 may be integrated with the logic of one or more controllers of the control module 504 on the same die. In one embodiment, at least one of the processors 502 may be integrated with the logic of one or more controllers of the control module 504 on the same die to form a system-on-a-chip (SoC).

[0177] In various embodiments, device 500 may be, but is not limited to, a server, desktop computing device, or mobile computing device (e.g., laptop computing device, handheld computing device, tablet computer, netbook, etc.). In various embodiments, device 500 may have more or fewer components and / or different architectures. For example, in some embodiments, device 500 includes one or more cameras, a keyboard, a liquid crystal display (LCD) screen (including a touchscreen display), a non-volatile memory port, multiple antennas, a graphics chip, an application-specific integrated circuit (ASIC), and a speaker.

[0178] The detection device can use a main control chip as a processor or control module, and sensor data, position information, etc. can be stored in a memory or NVM / storage device. The sensor group can be used as an input / output device, and the communication interface can include a network interface.

[0179] As the device embodiment is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be found in the description of the method embodiment.

[0180] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0181] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable terminal device for suppressing false alarms in stellar observations using ground-based telescopes to produce a machine such that the instructions, which execute via the processor of the computer or other programmable terminal device for suppressing false alarms in stellar observations using ground-based telescopes, generate instructions for implementing the flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of this application. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0182] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable terminal device for suppressing false alarms in stellar observations by ground-based telescopes to operate in a specific manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction device, which is implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0183] These computer program instructions can also be loaded onto a computer or other programmable terminal device for suppressing false alarms in stellar observations by ground-based telescopes, causing a series of operational steps to be executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0184] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0185] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0186] The foregoing has provided a detailed description of a method and apparatus for suppressing false alarms in stellar observations by ground-based telescopes, an electronic device, and a storage medium. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and its core ideas. At the same time, those skilled in the art will recognize that there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for suppressing false alarms in stellar observations using ground-based telescopes, characterized in that, The method includes: Construct a false alarm dataset; Detect the shape parameter data, trajectory data, and the fragmented frame data corresponding to the trajectory data in the false alarm dataset; Based on the trajectory data, determine the stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset; The stellar eccentricity approximation coefficient and stellar phase angle approximation coefficient of the false alarm dataset are determined based on the shape parameter data. The frame breakage penalty coefficient of the false alarm dataset is determined based on the frame breakage data; Based on the stellar velocity magnitude convergence coefficient, the stellar velocity direction convergence coefficient, the stellar eccentricity convergence coefficient, the stellar phase angle convergence coefficient, and the frame break penalty coefficient, the false alarm dataset is identified using a support vector machine algorithm based on particle swarm optimization to obtain stellar false alarms; False alarms from the stars are suppressed.

2. The method according to claim 1, characterized in that, The trajectory data includes a set of trajectory points, which includes a start point and an end point. Determining the stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset based on the trajectory data includes: Obtain the stellar velocity and the frame number difference between the start point and the end point; Determine the straight line trajectory corresponding to the set of trajectory points; Calculate the first foot of the perpendicular from the starting point to the straight trajectory, and calculate the second foot of the perpendicular from the ending point to the straight trajectory; Calculate the average trajectory velocity and the magnitude of the trajectory velocity based on the starting point, the ending point, the straight trajectory, the frame number difference, the first perpendicular foot, and the second perpendicular foot; The stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset are determined based on the stellar motion velocity, the average trajectory velocity, and the magnitude of the trajectory velocity.

3. The method according to claim 2, characterized in that, The step of determining the stellar velocity magnitude convergence coefficient and stellar velocity direction convergence coefficient of the false alarm dataset based on the stellar motion velocity, the average trajectory velocity, and the magnitude of the trajectory velocity includes: Calculate the magnitude approximation coefficient of the star's velocity based on the star's motion velocity and the magnitude of its trajectory velocity; Calculate the stellar velocity direction convergence coefficient based on the stellar motion velocity and the average velocity of the trajectory; The magnitude approximation coefficient of the stellar velocity is calculated using the following formula: in, The magnitude approximation coefficient of the star's velocity is... The magnitude of the trajectory velocity. The velocity of the star; The stellar velocity direction convergence coefficient is calculated using the following formula: in, This is the stellar velocity direction approximation coefficient. The average velocity of the trajectory, The velocity of the star's motion, the express and The angle between two vectors.

4. The method according to claim 1, characterized in that, The step of determining the stellar eccentricity approximation coefficient of the false alarm dataset based on the shape parameter data includes: The stellar eccentricity is determined based on the shape parameters. Determine the set of trajectory point eccentricities corresponding to the set of trajectory points; The stellar eccentricity approximation coefficient is calculated based on the stellar eccentricity and the set of trajectory point eccentricities. The stellar eccentricity approximation coefficient is calculated using the following formula: in, The eccentricity approximation coefficient of the star is... Let k be the set of trajectory points, and k be the trajectory point currently traversed. The eccentricity of the star is... Let be the eccentricity of the k-th trajectory point in the set of trajectory point eccentricities. ( ) represents the number of trajectory points in the set of trajectory points.

5. The method according to claim 1, characterized in that, The step of determining the stellar phase angle approximation coefficient of the false alarm dataset based on the shape parameter data includes: Determine the stellar phase angle based on the shape parameters; Determine the set of trajectory point phase angles corresponding to the set of trajectory points; The stellar phase angle convergence coefficient is calculated based on the stellar phase angle and the set of trajectory point phase angles; The stellar phase angle approximation coefficient is calculated using the following formula: in, The phase angle approximation coefficient of the star is... Let k be the set of trajectory points, and k be the trajectory point currently traversed. The phase angle of the star, Let the phase angle be the phase angle of the k-th trajectory point in the set of trajectory point phase angles. ( ) represents the number of trajectory points in the set of trajectory points.

6. The method according to claim 1, characterized in that, The step of determining the frame-drop penalty coefficient of the false alarm dataset based on the frame-drop data includes: Obtain the total number of frames in the detection queue of the trajectory data; The frame break penalty coefficient is calculated based on the total number of frames in the detection queue of the trajectory data and the number of trajectory points in the trajectory point set. The frame breakage penalty coefficient is calculated using the following formula: in, The frame break penalty coefficient is... This represents the total number of frames in the detection queue of the trajectory data. The number of trajectory points in the set of trajectory points.

7. The method according to claim 1, characterized in that, Before identifying the false alarm dataset using a support vector machine algorithm based on particle swarm optimization according to the stellar velocity magnitude convergence coefficient, the stellar velocity direction convergence coefficient, the stellar eccentricity convergence coefficient, the stellar phase angle convergence coefficient, and the frame break penalty coefficient, the method includes: The support vector machine algorithm to be trained, the particle swarm optimization parameters, and the fitness function are obtained. The kernel function of the support vector machine algorithm to be trained is a Gaussian kernel function. The support vector machine algorithm to be trained has a corresponding penalty factor, and the Gaussian kernel function has a corresponding kernel scale. Based on the particle swarm optimization parameters and the fitness function, the particle swarm optimization method is used to optimize the penalty factor and the kernel scale of the support vector machine algorithm to be trained, so as to obtain the target penalty factor and the target kernel scale. The support vector machine algorithm corresponding to the target penalty factor and the target kernel scale is used as the support vector machine algorithm based on particle swarm optimization.

8. The method according to claim 7, characterized in that, The step of identifying false alarm datasets using a preset recognition algorithm based on the stellar velocity magnitude convergence coefficient, the stellar velocity direction convergence coefficient, the stellar eccentricity convergence coefficient, the stellar phase angle convergence coefficient, and the frame break penalty coefficient, to obtain stellar false alarms, includes: The stellar velocity magnitude approximation coefficient, stellar velocity direction approximation coefficient, stellar eccentricity approximation coefficient, stellar phase angle approximation coefficient, and frame break penalty coefficient corresponding to the false alarm dataset are used as the feature coefficient data of the false alarm dataset; The feature coefficient data is input into the particle swarm optimization-based support vector machine algorithm to obtain the classification result of the particle swarm optimization-based support vector machine algorithm for the false alarm dataset, and the classification result includes the stellar false alarms.

9. An electronic device, characterized in that, include: processor; and A memory having executable code stored thereon, which, when executed, causes the processor to perform a method for suppressing false alarms in stellar observations by ground-based telescopes as described in any one of claims 1-8.

10. A machine-readable medium having executable code stored thereon, which, when executed, causes a processor to perform a method for suppressing false alarms in stellar observations by ground-based telescopes as described in any one of claims 1-8.

Images