Multi-information fusion ghost point elimination and target tracking method

Abstract

The invention discloses a multi-information fusion ghost point elimination and target tracking method. The target tracking precision is improved by eliminating false associated point locations by using the scattering characteristic of target angle cooperative positioning. Aiming at the problem that the tracking precision is reduced due to a large number of false associated ghost points, the invention provides a two-stage ghost point elimination and target tracking algorithm based on angle measurement and target motion characteristic fusion. According to the algorithm, a cooperative positioning strategy of first correlation and then estimation is adopted, and a vision field grid map and an energy accumulation matrix are constructed by establishing a mapping relation between angle measurement noise and positioning errors; the method comprises the following steps: analyzing spatial geometric distribution characteristics between a real target and false correlation ghost points in a vision field, designing a novel elimination criterion based on the idea of Hough transformation, and realizing primary coarse elimination of the ghost points; by researching the distribution characteristics and motion characteristics of a target positioning fuzzy region, a prediction tracking gate is constructed by using motion parameter identification, and secondary fine elimination of ghost points is realized from the kinematics level.

Description

Technical Field

[0001] The present invention relates to the field of multi-target tracking, and in particular to a multi-information fusion ghost point elimination and target tracking method. Background Art

[0002] In the field of passive detection, aircraft swarms equipped with optical infrared guidance devices extract line-of-sight angle information relative to the target and estimate the target's spatial position using the principle of direction-finding line intersection. During multi-target tracking, when the measurement origin is uncertain, multi-path direction-finding cross-localization inevitably generates a large number of false positioning points, creating difficult-to-eliminate spurious tracks in the time series, seriously impacting multi-target tracking accuracy. As the number of aircraft and targets increases, the number of false points generated by direction-finding cross-localization increases exponentially. Furthermore, complex battlefield electromagnetic environments present measurement noise and significant clutter interference, and in some mission scenarios, even decoy targets are present. Measurement errors can prevent the directional angle rays of aircraft toward the same target from precisely intersecting after positioning, instead dispersing them within a certain range of positioning ambiguity. These ambiguous points, combined with false correlation points, significantly increase the difficulty of identifying false points. When clutter and decoy targets are observable, their intersection points share the same geometric characteristics as the true target, creating stubborn, difficult-to-identify ghost points, placing even higher demands on target tracking accuracy.

[0003] Current research on ghost point removal focuses on two core paradigms: data association optimization and enhanced feature discrimination. In terms of methodological development, researchers have developed a multi-dimensional approach by integrating geometric constraint theory, pattern recognition techniques, and uncertainty reasoning methods. First, based on geometric feature difference analysis, false point identification is achieved by constructing spatial distribution constraints or motion feature discrimination models. Second, by enhancing measurement redundancy, association ambiguity is reduced through multi-view observation data fusion or joint analysis of temporal features. Third, improved pattern recognition algorithms are used to accumulate energy and extract features from raw measurement information, establishing more robust discrimination criteria. During technological evolution, position optimization strategies based on sensor maneuvers have improved observation information completeness, while improved transform domain analysis methods have enhanced feature separability in complex scenarios. However, existing methods still face significant contradictions between accuracy and computational efficiency, and between scenario universality and algorithmic complexity. These problems are particularly evident in insufficient robustness in dynamic environments, limited efficiency in processing high-dimensional data, and inability to discriminate under multiple interference coupling.

[0004] In summary, with the increase in the number of observed targets and the increasing complexity of the battlefield electromagnetic environment, a large number of false targets in direction-finding correlation positioning significantly reduce the multi-target tracking accuracy and system computational efficiency, restricting the development of the field of multi-target tracking in multi-aircraft clusters. Summary of the Invention

[0005] The purpose of the present invention is to provide a multi-information fusion ghost point removal and target tracking method, which can effectively remove associated ghost points and improve the multi-target tracking accuracy from the perspectives of field of view, homology and consistency.

[0006] The technical solution to achieve the purpose of the present invention is: a multi-information fusion ghost point elimination and target tracking method, comprising:

[0007] Step 1: Establish a mapping relationship between angle measurement noise and positioning error, and construct a view grid map and energy accumulation matrix;

[0008] Step 2: Analyze the spatial geometric distribution characteristics between the real target and the false associated ghost points in the field of view, design a elimination criterion based on the idea of ​​Hough transform, and achieve the first-level rough elimination of ghost points;

[0009] Step 3: By identifying the scatter characteristics and motion characteristics of the fuzzy area of ​​the target, the motion parameter is used to construct a prediction tracking gate to achieve secondary precise elimination of ghost points from the kinematic level.

[0010] Furthermore, in step 1, a view grid map and an energy accumulation matrix are constructed, specifically:

[0011] Based on the aircraft position, detection distance, and detection angle, the aircraft fuzzy detection field of view is constructed. In the multi-aircraft multi-target tracking system, the position of the aircraft members at the cluster boundary is taken as the coordinate origin O, with the forward detection direction as the y-axis to construct a relative rectangular coordinate system. The distance between the two aircraft is the positioning baseline length D, the maximum forward detection distance of the aircraft is recorded as r, and the target's measured azimuth relative to the aircraft is recorded as θ. Based on the aircraft position, the primary fuzzy locatable field of view boundary in the two-dimensional XOY plane is determined as:

[0012]

[0013] Among them, S l and S r is the x-axis field of view boundary, S u and S b is the y-axis field of view boundary, and are the minimum values ​​of the aircraft coordinate sets, and are the maximum values ​​of the aircraft coordinate set respectively; the target associated point coordinates are recorded as N is the total number of associated points; the criterion for the associated intersection points to fall into the primary fuzzy locatable visual field is designed as follows

[0014]

[0015] By statistically analyzing the grid distribution of the associated points in the view map, an energy accumulation matrix is ​​constructed. Combining the geometric distribution characteristics of ghost points and real targets, the first-level elimination of false associated ghost points is achieved. The horizontal and vertical view indices in the two-dimensional XOY plane are respectively I row and I col , according to the suspected target coordinates falling into the primary fuzzy locatable field of view The index calculation rules can be designed as follows:

[0016]

[0017] Among them, g is the prior parameter of the grid unit designed based on the optimal topological configuration;

[0018] When the index is calculated according to the above rules, the point position and the grid size are not completely in a multiple relationship, and a rounding operation needs to be performed to obtain a valid established index; considering that when the associated point position falls on the boundary of the view area map, the rounding operation will make the grid map boundary index not necessarily fall within the view area, so filling processing is required to enforce the index number that exceeds the map boundary to make it fall into the map to avoid point loss; in the presence of observation errors, the associated point position of the real target will form a positioning fuzzy area; since the division of grid cells is fixed and orderly, the intersection points in the positioning fuzzy area of ​​the same target may be scattered in different grids, resulting in energy accumulation and dissipation problems; in order to avoid the target fuzzy area being located at the grid boundary, a single grid cell has no The method does not cover all target suspected points, which leads to the problem of incomplete statistics of target associated points under the grid index. By constructing multiple sets of cross-grid maps and increasing the grid density, the loss of block boundary features is prevented, thereby ensuring the target positioning accuracy. The cross design of the grid map is carried out from the x-axis and y-axis of the field of view, and the grid is densely distributed by moving the grid by the cross offset. Although the index boundaries of different grid maps are different, the position coordinates of the target suspected points under the locatable field of view map are unified, ensuring the consistency of multiple sets of grid indexes. The discrete points of direction finding cross positioning are arranged in the field of view plane, and the discrete points accumulated by the grid cells are statistically voted to complete the construction of the energy accumulation matrix under the grid map.

[0019] Furthermore, in step 2, a new elimination criterion is designed based on the idea of ​​Hough transform, specifically:

[0020] The intersection points of the corresponding targets theoretically overlap, so the points will fall in the same accumulation unit. The voting results of each accumulation unit are accumulated to obtain the energy matrix of different accumulation units. The more overlapping points there are, the higher the accumulation value and the maximum energy value. In each observation period, each target can always get intersection points, which are mutually coincident. However, the false positioning points obtained due to the wrong angle correlation intersection will be scattered throughout the detection field of view, and their distribution characteristics are determined by the spatial layout of the aircraft and the position of the target. No matter how the relative positions of the aircraft and the target are distributed and how the number of sensors is selected, the target position containing the correct angle correlation intersection is always the largest. Therefore, to obtain the measurement belonging to the target from many intersection points, it is judged according to the overlap of the intersection points. It is known that the real intersection points of the target overlap and the number of overlaps is Based on the calculated energy accumulation matrix, judgment is made in two cases:

[0021] (1) When the intersection points do not overlap or the number of overlaps is less than When , the intersection point is judged as a ghost point and removed;

[0022] (2) When the number of overlapping intersections is equal to And there are m such coincident points, then these m points are judged as target points; if there are m+m * There are m such coincident points, then there are still m * A false point still needs to be judged.

[0023] Furthermore, in step 3, the prediction tracking gate is constructed using motion parameter identification and a two-level precise ghost point elimination scheme is designed, specifically:

[0024] The secondary point cluster elimination is based on the target motion characteristics identification of the historical trajectory, and the target scattering area is calculated by combining the target positioning fuzzy area and the current motion characteristics, and then the secondary prediction tracking gate is generated; the point clusters falling within the prediction tracking gate are the effective target intersection positioning points under the continuous time series, and the targets falling outside the prediction tracking gate are false detection targets caused by measurement noise, and will not obey the motion constraints at the track information level; by identifying the current motion parameters of the target, assuming that the current speed of the target is v d , the flight sampling time is t, the target maximum speed is v m , then the maximum maneuvering distance of the target is

[0025] L t =v m t (4)

[0026] Take a point on the boundary of the positioning fuzzy area as the center of the circle and L t Draw a circle with radius L, which is the target maneuvering dispersion circle centered on this point. The outer edges of all target maneuvering dispersion circles form an extended closed boundary, which is approximately equivalent to an ellipse. a and L bare the major and minor axes of the elliptical target scattering area, respectively. The minor axis is approximately half of the major axis, and the major axis is approximately the sum of the distance from the current target position to the longitudinal end point of the positioning ambiguity area and the target prediction step length. The approximate calculation formula is as follows:

[0027]

[0028] Δθ max is the direction finding error, θ1 and θ2 are the direction finding angles between the carrier aircraft and the target, and γ is the cross-positioning intersection angle;

[0029] It can be seen that at the target maximum speed v m Under certain conditions, the geometric characteristics of the target elliptical dispersion area and the direction-finding error Δθ max , the direction-finding azimuth angle θ is related to the flight sampling interval t; the current cross-positioning coordinates are selected as the center of the elliptical target scattering area, and the major and minor axis parameters of the elliptical target scattering area obtained by identification calculation are combined to generate the elliptical scattering area equation; the coordinates of the target direction-finding cross-positioning are recorded as (x T ,y T ), then the criterion for a point to fall within the ellipse can be expressed as

[0030]

[0031] Considering the target's motion situation, the scattering area will have directionality, that is, the target scattering ellipse will have a rotation angle α; for an ellipse with a rotation angle α, the general criterion for a point falling within the ellipse can be expressed as

[0032]

[0033] The rotation angle α is determined by the current position of the target (x T ,y T ) and the next moment predicted position (x′ T ,y′ T ) to perform differential calculation:

[0034] α=aractan((y T -y′ T ) / (x T -x′ T )) (8)

[0035] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the program.

[0036] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0037] A computer program product comprises a computer program, which implements the steps of the above method when executed by a processor.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] (1) Aiming at the target positioning uncertainty caused by inaccurate angle measurement in collaborative positioning, the mapping relationship between angle measurement noise and target positioning error was explored, a view grid map and energy accumulation matrix were constructed, and a first-level ghost point elimination criterion was designed based on the ghost point distribution characteristics.

[0040] (2) Taking into account the distribution characteristics of the fuzzy geometric area of ​​target positioning and the target motion characteristics, a prediction tracking gate is constructed in combination with motion parameter identification to achieve effective two-level ghost point elimination;

[0041] (3) A collaborative positioning strategy of "association first, then estimation" is adopted, and a two-stage ghost point elimination and target tracking algorithm based on the fusion of angle measurement and target motion characteristics is proposed. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is the multi-observation platform single target positioning and tracking scenario in the present invention.

[0043] Figure 2 This is the multi-observation platform multi-target false correlation scenario in the present invention.

[0044] Figure 3 It is the angle collaborative target observation model in the two-dimensional plane in the present invention.

[0045] Figure 4 It is a schematic diagram of the ambiguity area of ​​dual-station angle collaborative positioning in the present invention.

[0046] Figure 5 This is a schematic diagram of the design of the dual-station angle collaborative positioning prediction tracking gate in the present invention.

[0047] Figure 6 It is a schematic diagram of multi-maneuverable target task setting in the present invention.

[0048] Figure 7 This is a schematic diagram of the target tracking number elimination result in the present invention.

[0049] Figure 8 It is a schematic diagram of the full OSPA process in the present invention.

[0050] Figure 9 This is a schematic diagram of the target tracking number elimination result in the present invention.

[0051] Figure 10 This is a schematic diagram of the target tracking number elimination result in the present invention.

[0052] Figure 11 It is a schematic diagram of the discrete moment tracking scenario under the simulation experiment in the present invention.

[0053] Figure 12 This is a schematic diagram of a discrete-time secondary prediction gate under a simulation experiment in the present invention.

[0054] Figure 13 It is a schematic diagram (cluster) of target elimination at discrete moments in the simulation experiment of the present invention.

[0055] Figure 14 This is a schematic diagram of target x-direction motion speed identification in the simulation experiment of the present invention.

[0056] Figure 15 This is a schematic diagram of the target y-direction motion speed identification under the simulation experiment of the present invention.

[0057] Figure 16 It is a schematic diagram of the probability interaction results of the maneuver model under the simulation experiment in the present invention.

[0058] Figure 17 This is a comparison chart of point elimination performance under different grid sizes in the simulation experiment of the present invention. DETAILED DESCRIPTION

[0059] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings.

[0060] This paper proposes a two-stage ghost point removal and target tracking algorithm based on a "correlation-first, estimation-later" strategy. By analyzing the mapping relationship between angle measurement noise and target positioning error, a view grid map and energy accumulation matrix are constructed. A first-stage ghost point removal criterion is designed based on the distribution characteristics of ghost points. Furthermore, a predictive tracking gate is constructed by comprehensively considering the distribution characteristics of the target positioning fuzzy geometric area and the target motion characteristics, combined with motion parameter identification, to achieve effective second-stage ghost point removal. The implementation process includes the following five specific steps.

[0061] Step 1: Multi-station angle co-location model

[0062] According to the principle of infrared passive detection, the target position can be calculated by using the position information and target angle information provided by different observation base stations. For the multi-aircraft multi-target tracking system, assuming that the position coordinates of the target in the scene are X T =[x T ,y T ,z T ] T , the coordinates of the i-th aircraft member are S i =[x i ,y i ,z i ], the observation is the azimuth angle β between the i-th aircraft and the target iand pitch angle The specific expressions are:

[0063]

[0064] β i =arctan((y T -y i ) / (x T -x i )) (10)

[0065] Considering the dual-station angle collaborative target positioning scenario, the xoy plane is used as the horizontal plane to establish a three-dimensional Cartesian coordinate system, such as Figure 1 As shown. When the aircraft coordinates are known, formulas (9)-(10) can be sorted out to obtain:

[0066]

[0067] It is worth mentioning that the distribution relationship between the target and the aircraft observation platform does not affect the azimuth and relative positioning relationship, but the azimuth value range β must be ensured. i ∈[0,2π]. Considering that the number of aircraft in the cluster system is n and n≥2, formula (11) can be written as:

[0068] A i X T =b i (12)

[0069] Among them, A i and b i The specific form is as follows:

[0070]

[0071] At this point, the least squares expression for solving the target position matrix can be obtained:

[0072] X T =[x T ,y T ,z T ] T =(A T A) -1 A T b (15)

[0073] Where A and b are i and b i The matrix is ​​formed by sorting the rows from smallest to largest subscripts.

[0074] Step 2: Collaborative multi-target ghost point analysis from multiple combat angles

[0075] When there are multiple targets, each aircraft obtains multiple angle information, and it is unknown whether the measurement information is generated by the target and which target it is generated by. The cross-positioning of different direction-finding information will produce a large number of false points. In order to illustrate the relationship between the aircraft tracking field of view and its parameters, and to intuitively describe the generation and distribution characteristics of ghost points, the aircraft position on the two-dimensional plane is denoted as S i (x i ,y i ), i=1,2,…,n, n is the total number of aircraft members in the cluster system, and the target position is T k (x k ,y k ), k=1,2,…,m, where m is the total number of tracked targets. Take 3 aircraft tracking 3 targets as an example, Figure 2 A schematic diagram of false associations in a typical two-dimensional mission scenario is provided. The black solid dots in the figure represent real targets, while the green solid dots represent falsely associated ghost points. Ghost points are an inevitable problem in multi-target tracking. By analyzing the main factors that cause ghost points and adopting different ghost point removal algorithms based on different application scenarios, their impact on multi-target tracking accuracy can be reduced.

[0076] In general, it is assumed that the number of aircraft in the multi-aircraft cluster cooperative multi-target tracking system is n>2, the number of moving targets is m, the detection probability of the optoelectronic platform carried by each aircraft is 1, and all angle observations are from the target. In each observation cycle, the aircraft are grouped into two groups, and the maximum number of intersections generated is Among them are The number of real intersections (intersections corresponding to the target) is When the measurement error is not considered, these intersection points are coincident. According to the relationship between the number of aircraft and targets, the distribution of intersection points at ghost points is as follows:

[0077] (1) n>m, the maximum number of intersections at the ghost point is

[0078] (2) n = m, the maximum number of intersections at the ghost point is And there can only be one point in this situation;

[0079] (3) n<m, the maximum number of intersections at the ghost point is And there may be multiple points of this situation.

[0080] Step 3: Grid map construction in clutter environment

[0081] Based on the aircraft position, detection distance, detection angle and other parameters, the aircraft fuzzy detection field of view is constructed, which can preliminarily eliminate false correlation intersections that fall outside the detection field of view, reduce computational complexity and reduce the impact of ghost points on target tracking accuracy. In the multi-aircraft multi-target tracking system, the position of the aircraft members at the cluster boundary is used as the coordinate origin O, with the forward detection direction as the y-axis to construct a relative rectangular coordinate system. The distance between the two aircraft is the positioning baseline length D, and the maximum forward detection distance of the aircraft is recorded as r. Based on the aircraft position, the primary fuzzy locatable field of view boundary can be determined as follows:

[0082]

[0083] Among them, S l and S r is the x-axis field of view boundary, S u and S b is the y-axis field of view boundary, and are the minimum values ​​of the aircraft coordinate sets, and are the maximum values ​​of the aircraft coordinate sets. The target associated point coordinates are N is the total number of associated points. The criterion for the associated intersection points to fall into the primary fuzzy locatable visual field can be designed as follows:

[0084]

[0085] Affected by the angle measurement error, the target positioning accuracy will be restricted. The three-dimensional space target observation model can be derived by the two-dimensional approximation method. Figure 3 A two-dimensional plane angle cooperative target observation model is proposed to analyze the coupling relationship between angle error and positioning accuracy. The angle measurement information of the two aircraft is as follows:

[0086]

[0087] Where: (θ1, θ2) is the true azimuth; is the measured azimuth; (v1, v2) is the measurement noise.

[0088] Let (x1, y1) and (x2, y2) be the real positions of the two aircraft respectively. is the target position information with errors, γ is the azimuth angle between the two aircraft relative to the target, and γ = θ2-θ1. Figure 3 As shown, its position relationship can be described as follows:

[0089]

[0090] At the same time, the target coordinates (x T ,y T )

[0091]

[0092] Performing Taylor expansion on the above equation at (θ1, θ2) and ignoring the second-order and higher-order terms, the position error expression can be approximated.

[0093]

[0094] According to the geometric relationship, the distance between the aircraft and the target (r1, r2) is calculated. A right triangle is constructed with r2 as the hypotenuse. The geometric relationship is:

[0095]

[0096] Then we can solve

[0097]

[0098] Calculating the partial differentials, Equation (21) can be rewritten as

[0099]

[0100] The target estimated position error is

[0101]

[0102] If the position error of the aircraft is not considered, and it is assumed that the maximum direction-finding error of the two aircraft (or a single aircraft at different times) is the same and is Δθ max , then the target's true position should be within the quadrilateral ABCD where the aircraft's detection fan-shaped area intersects, such as Figure 4 As shown. Since the direction finding error is within ±Δθ max The target's true position can be any value within the range of 1 / 4, so the target's true position may appear at any point in the quadrilateral ABCD. Since the target's true position within the quadrilateral ABCD cannot be determined, the quadrilateral ABCD is called the positioning ambiguity zone. The law of sines states that the ratio of each side to the sine of its opposite angle is equal. When the baseline distance D is constant, the triangle cosine theorem can be used to calculate the lengths of sides AU1, BU1, and CU1 as follows:

[0103]

[0104] Obviously, quadrilateral ABCD is an axially symmetrical figure, and its area is equal to twice the area of ​​triangle ABC. The area calculation of triangle ABC can be regarded as taking AB as the base, and the height corresponding to the base can be approximated by a corresponding arc, that is,

[0105]

[0106] If we consider Δθ max Since it is a small quantity, ignoring its influence in the calculation of trigonometric functions can simplify the calculation, that is, the area of ​​quadrilateral ABCD can be approximately expressed as

[0107]

[0108] It can be seen that S ABCD In addition to the distance D from the baseline, the direction finding error Δθ max In addition to being related to the direction of the carrier relative to the target, it is also related to the direction θ1 and θ2 of the carrier relative to the target. Generally speaking, Δθ max Depends on the performance of the carrier's infrared direction finding. max Under certain conditions, the smaller the baseline distance D is, the greater the S ABCD If the baseline distance D is constant, then S ABCD The size of θ mainly depends on θ1 and θ2.

[0109] In actual aircraft swarm applications, the baselines between aircraft can be obtained and used for formation adjustment and maintenance, and can be considered as prior information. However, the observation angle of non-cooperative targets changes dynamically and randomly with the target's movement, which is unknown observation information and contains observation errors. Therefore, the area and distribution of the positioning ambiguity zone are more affected by the observation angle. Consider when the baseline distance is constant, let

[0110] W=sinθ1sinθ2 / sin 3 (θ1+θ2) (32)

[0111] To make S ABCD Minimum, should meet Arrangement can be solved

[0112]

[0113] That is, when the baseline distance is constant, the area of ​​the corresponding positioning ambiguity zone is minimum only when the base angle satisfies θ1 = θ2 = π / 6. In this case, the intersection angle γ = 120°. The area of ​​the positioning ambiguity zone can be rewritten as:

[0114]

[0115] Ghost point removal relies on ensuring that overlapping points fall into the same accumulation unit. Therefore, a grid map is first constructed to segment the positioning field of view, generating a number of grid accumulation units. The size of each unit is related to the distribution of the current target's positioning ambiguity zone. Given that the size of the target's positioning ambiguity zone changes under dynamic positioning topology, which is detrimental to the continuity of the target over time, it is possible to design prior parameters for the grid units based on the optimal topological configuration and correction parameters.

[0116] Step 4: Energy accumulation matrix next level ghost point elimination method

[0117] By statistically analyzing the grid distribution of associated points in the view map, constructing an energy accumulation matrix, and combining the geometric distribution characteristics of ghost points and real targets, the first-level elimination of false associated ghost points can be achieved. Specifically, the coordinates of the suspected target falling into the primary fuzzy locatable view are The index calculation rules are designed as follows:

[0118]

[0119] When performing index calculations according to the above rules, the point and grid size are not completely multiples, so a rounding operation is required to obtain a valid established index. Considering that when the associated point falls on the viewport map boundary, the rounding operation may cause the grid map boundary index to not fall within the viewport, so a certain amount of padding is required to enforce constraints on index numbers that exceed the map boundary so that they fall within the map to avoid point loss. Specifically, taking the row index as an example, there are:

[0120]

[0121] Among them, I max This is the theoretical maximum index boundary when building a raster map. Index values ​​exceeding this value will be outside the viewshed map.

[0122] In the presence of observation errors, the associated points of the true target will form a certain range of positioning ambiguity. Because the division of grid cells is fixed and orderly, the intersection points within the positioning ambiguity zone of the same target may be scattered in different grids, leading to energy accumulation and dissipation problems. To avoid the problem that a single grid cell cannot cover all suspected target points when the target ambiguity zone is located at the grid boundary, thereby causing incomplete statistics of target associated points under the grid index, multiple sets of cross-grid maps can be constructed and the grid density increased to prevent the loss of block boundary features, thereby ensuring target positioning accuracy. The cross-grid design can be expanded along the x-axis and y-axis of the field of view, and the grid is densely distributed by shifting the grid by a certain cross-offset. Although the index boundaries of different grid maps may vary, the position coordinates of the suspected target points under the locatable field of view map are unified, ensuring the consistency of multiple sets of raster indexes.

[0123] The discrete points of direction-finding cross-positioning are arranged in the viewing plane, and the discrete points accumulated by the grid units are statistically voted. Among them, the intersection points of the corresponding targets theoretically overlap, so the points will fall in the same accumulation unit. The voting results of each accumulation unit are accumulated to obtain the energy matrix of different accumulation units. Obviously, the more overlapping points there are, the higher the accumulation value is and the maximum energy value is. In each observation cycle, each target can always get intersection points, which are mutually overlapping. However, the false positioning points obtained by the incorrect angle correlation intersection will be scattered throughout the detection field of view, and their distribution characteristics are determined by the spatial layout of the aircraft and the position of the target. Regardless of the relative positions of the aircraft and the target and the number of sensors, the target position containing the correct angle correlation intersection is always the largest. Therefore, to obtain the measurement belonging to the target from many intersection points, it can be judged based on the overlap of the intersection points. It is known that the real intersection points of the target overlap and the number of overlaps is Then according to the accumulated results, we can make judgments in two situations:

[0124] (1) When the intersection points do not overlap or the number of overlaps is less than When , the intersection point is judged as a ghost point and can be removed;

[0125] (2) When the number of overlapping intersections is equal to And there are m such coincident points, then these m points are judged as target points; if there are m+m * There are m such coincident points, then there are still m * A false point still needs to be judged.

[0126] The accumulated unit energy matrix value is p, and the minimum energy value of the suspected target cluster is Where c (c < 1) is the scaling factor. If p < p T , the cluster does not hold; if p≥p T , and determine that the cluster is a suspected target point cluster.

[0127] Step 5: Secondary ghost point removal method under the prediction tracking door

[0128] The core idea of ​​secondary point cluster elimination is to identify the target motion characteristics based on the historical trajectory, and calculate the target scattering area by combining the target positioning fuzzy area and the current motion characteristics, and then generate a secondary prediction tracking gate. The point cluster that falls within the prediction tracking gate is the effective target intersection positioning point under the continuous time series. The target that falls outside the prediction tracking gate is a false detection target due to measurement noise, and will not obey the motion constraint at the track information level. Therefore, under reasonable design conditions, the problem of false detection targets can be further eliminated by the prediction tracking gate of motion characteristics. Specifically, the size and shape of the target scattering area depends not only on the geometric characteristics of the positioning fuzzy area, but also on the target's heading and speed. When the target speed and heading are unknown, the target may start from any point in the positioning fuzzy area and maneuver in any direction within a certain speed range. By identifying the current motion parameters of the target, assuming that the current speed of the target is v d , the flight sampling time is t, the target maximum speed is v m , then the maximum maneuvering distance of the target is

[0129] L t =v m t (37)

[0130] Take a point on the boundary of the positioning fuzzy area as the center of the circle and L t Draw a circle with the radius as the target maneuvering dispersion circle centered at this point. The outer edges of all target maneuvering dispersion circles form an extended closed boundary, which is approximately equivalent to an ellipse, such as Figure 5 As shown in the figure, U1 and U2 are the positions of the carrier aircraft. The position of U1 is taken as the coordinate origin O, and U2 is placed on the positive half axis of the x-axis. The distance D between U1 and U2 is the positioning baseline length, Δθ max is the direction-finding error, θ1 and θ2 are the direction-finding angles between the carrier and the target, γ is the cross-positioning intersection angle, quadrilateral ABCD is the positioning ambiguity area, and T is the target scatter center, which is also the cross-positioning point at the current moment.

[0131] Remember L a and L b are the major and minor axes of the elliptical target scattering area, respectively. The minor axis is approximately half of the major axis, and the major axis is approximately the sum of the distance from the current target position to the longitudinal end point of the positioning ambiguity area and the target prediction step length. The approximate calculation formula is as follows:

[0132]

[0133] It can be seen that at the target maximum speed v m Under certain conditions, the geometric characteristics of the target elliptical dispersion area and the direction-finding error Δθ max, the direction finding azimuth angle θ and the flight sampling interval t. Select the current cross positioning coordinates as the center of the elliptical target scattering area, and combine the major and minor axis parameters of the elliptical target scattering area obtained by identification calculation to generate the elliptical scattering area equation. The coordinates of the target direction finding cross positioning are recorded as (x T ,y T ), then the criterion for a point to fall within the ellipse can be expressed as

[0134]

[0135] Considering the target's motion, the scattering area will have a certain directionality, that is, the target scattering ellipse will have a certain rotation angle α. For an ellipse with a rotation angle α, the general criterion for a point to fall within the ellipse can be expressed as

[0136]

[0137] The rotation angle α can be obtained by performing differential calculation on the target’s current position information and the predicted position information at the next moment:

[0138] α=aractan((y T -y′ T ) / (x T -x′ T )) (41)

[0139] Example

[0140] In order to verify the effectiveness of the proposed algorithm in eliminating ghost points and the accuracy of tracking model parameter identification when tracking maneuvering targets, a simulation case is designed assuming that four friendly aircraft are hovering to track six enemy moving targets. The task setting is as follows: Figure 6 The simulation parameters are as follows: the aircraft hovering positions are: (0m, 500m, 3000m), (900m, 600m, 3000m), (2000m, 850m, 3000m), (3000m, 500m, 3000m), and the enemy target motion pattern and parameters are shown in Table 1.

[0141] Table 1. Initial parameters and task settings for multi-target motion

[0142]

[0143] The forward detection range of the aircraft equipped with a reconnaissance payload is 5000m, the detection azimuth angle is 80°, and the detection pitch angle is 25°. At the same time, it is assumed that the sampling period of the infrared seeker is T = 0.1s, and the direction finding error is δ θ=0.1°. The energy accumulation unit size in the algorithm design is approximately 25m, 30m, and 30m. The clutter of each cycle is randomly distributed in the aircraft's field of view according to a uniform distribution. The given unit clutter number is λ=5, and the unit area clutter parameter is γ=λ×10 -6 , that is, every 1×10 -6 m 2 λ clutter is generated within the tracking area. The total number of clutter in each tracking period can be determined according to the size of the field of view, and the probability of clutter detection is assumed to be 0.5. Considering that the CV model can be regarded as a special form of the CT model that keeps the conversion rate zero, the CA model and the CT model are used as maneuvering target tracking models in this section. For the turning tracking model, the turning rate parameter is selected as [-8o / s 2o / s 5o / s] to generate three turning tracking models under different turning rates. The initial probability of the model set is μ = [0.50.5], and the model transformation probability matrix is ​​selected as

[0144] To verify the effectiveness of the algorithm, we analyzed its performance in terms of its ability to remove falsely associated ghost points and its target tracking accuracy. The OSPA distance is a commonly used evaluation metric in multi-target tracking, used to measure the difference between target estimation and the actual situation. It essentially measures the degree of difference between different sets. The distance between any two vectors is defined as follows:

[0145] d c (x,y)=min(c,d(x,y)) (42)

[0146] Where, parameter c represents the cutoff point, which is a constant greater than 0, reflecting the sensitivity of OSPA distance to potential error. Suppose the true target position set X = {x1, x2, ..., x m}, target estimated position set Y={y1,y2,…y n}, the p-order OSPA distance between the two is defined as

[0147]

[0148] Where A n is the set of all permutations of {1, 2, …, n}, where π(i) represents the i-th element of the π-th combination. p represents the order, reflecting the sensitivity of the OSPA distance to outliers. The sensitivity of target estimation error is adjusted by adjusting the values ​​of c and p. In this patent, c = 100m and p = 1 are selected.

[0149] Figure 7 It shows the full tracking trajectory of the target on the x and y axes. Figure 8 is the OSPA result of multi-target tracking. From the tracking accuracy point of view, Figure 8As shown in the figure, the OSPA distance tracking algorithm is optimized to reduce the distance at each level. After removing the first-level energy accumulation matrix, the OSPA distance is 7.0975 meters, and after the second-level prediction tracking gate processing, it is reduced to 1.5626 meters, and the overall tracking accuracy is improved by 77.98%. Figure 9 This is a schematic diagram of the elimination of associated points under various levels of algorithms. Figure 10 = is the number of multi-target tracking achieved by each algorithm level. From the perspective of tracking effectiveness, each algorithm level achieved varying degrees of falsely associated target rejection. The total number of observable targets within the field of view, including clutter, was approximately 80. This number was reduced to approximately 40 after the first-level energy matrix processing, for an overall rejection rate of 50.384%. Further refinement of this rejection by the second-level prediction gate enabled effective target tracking. Experimental results demonstrate that all six targets were effectively tracked in a complex environment with both falsely associated points and clutter.

[0150] In order to further demonstrate and verify the effectiveness of the algorithm and the logic of the multi-level target elimination algorithm, take the discrete time t = 100 as an example. Figure 11-13 The multi-target elimination and tracking effects of the algorithm proposed in this chapter are given. Figure 11 This is a schematic diagram of a discrete-time target tracking scenario. The five-pointed star represents the friendly aircraft, the blue snowflakes represent all points associated with the observation angle (sources include both targets and clutter), the red hollow circles represent suspected target points that meet the energy accumulation criterion after first-level elimination on the grid map, and the green solid circles represent target points that meet the second-level elimination criteria under the predictive tracking gate. The figure shows that the point elimination capability increases step by step. The statistical results of point and target measurement and elimination are recorded in Table 2.

[0151] Table 2 Statistics of multi-station cooperative multi-maneuvering target elimination and tracking performance

[0152]

[0153] Under the first-level ghost point removal, there may still be suspected target clusters that meet the point cluster generation rules. Under the second-level tracking prediction gate, a prediction tracking gate can be constructed according to the motion characteristics of the target, such as Figure 12-13 Indicated by the yellow oval. Figure 12 This is a schematic diagram of the elimination of suspected target clusters under the secondary prediction tracking gate. Figure 13This is a partial enlargement of the target rejection of six targets under the second-level prediction and tracking gate. The diamond-shaped mark in the figure is the center position of the prediction and tracking gate at the previous moment. The center of the elliptical prediction and tracking gate at the current moment is estimated based on the target's motion state (the yellow hollow circle in the figure). The red points are the real targets that fall into the second-level prediction and tracking gate after the first-level rejection. This can be used to eliminate the suspected target point clusters outside the tracking gate. Specific tracking performance statistics are shown in Table 2. As can be seen from the table, the first-level point rejection percentage under the energy accumulation matrix at the current moment is 59.14%, and the second-level point rejection capability under the prediction and tracking gate is 91.82%. The first-level OSPA distance is 6.6096m, and the second-level OSPA distance is 1.4142m, which improves the accuracy by approximately 78.6%.

[0154] Figure 14 and Figure 15 The x-axis and y-axis velocity identification results of six enemy targets are shown in the figure. The target motion parameters are basically effectively identified, which can be used to eliminate stubborn ghost points and decoy targets with stable tracks according to the prior constraints of target motion characteristics. In view of the characteristics of target maneuverability, the algorithm design includes the design of interactive multi-model tracking algorithm. Figure 16 A schematic diagram of the interactive probability switching results of each target tracking model is given. As can be seen from the figure, when the motion mode of each target changes, the central tracking model switches accurately, which also ensures the accuracy of multi-target tracking and provides accuracy support for the construction of the secondary tracking prediction gate of the proposed algorithm. In addition, in order to explore the impact of the proposed algorithm under different grid sizes in the algorithm design, Figure 17 The algorithm performance comparison results for different parameter variations are presented. The graph shows that when the grid size is too small, the energy accumulated by the accumulation unit is insufficient, which can easily lead to the loss of valid associated points, resulting in missed targets and increased traversal pressure. When the grid size is too large, the accumulation unit may contain clutter or false associated points, resulting in a high false detection rate of target points, ultimately affecting target tracking accuracy and point rejection efficiency. Therefore, in algorithm design, it is important to incorporate positioning error into parameter binding to enhance the robustness and accuracy of the algorithm.

Claims

1. A multi-information fusion ghost point elimination and target tracking method, characterized in that: include: Step 1: Establish a mapping relationship between angle measurement noise and positioning error, and construct a view grid map and energy accumulation matrix; Step 2: Analyze the spatial geometric distribution characteristics between the real target and the false associated ghost points in the field of view, design a elimination criterion based on the idea of ​​Hough transform, and achieve the first-level rough elimination of ghost points; Step 3: By identifying the scatter characteristics and motion characteristics of the fuzzy area of ​​the target, the motion parameter is used to construct a prediction tracking gate to achieve secondary precise elimination of ghost points from the kinematic level.

2. The multi-information fusion ghost point elimination and target tracking method according to claim 1, characterized in that: Step 1: Construct the viewshed grid map and energy accumulation matrix, specifically: Based on the aircraft position, detection distance, and detection angle, the aircraft fuzzy detection field of view is constructed. In the multi-aircraft multi-target tracking system, the position of the aircraft members at the cluster boundary is taken as the coordinate origin O, with the forward detection direction as the y-axis to construct a relative rectangular coordinate system. The distance between the two aircraft is the positioning baseline length D, the maximum forward detection distance of the aircraft is recorded as r, and the target's measured azimuth relative to the aircraft is recorded as θ. Based on the aircraft position, the primary fuzzy locatable field of view boundary in the two-dimensional XOY plane is determined as: Among them, S l and S r is the x-axis field of view boundary, S u and S b is the y-axis field of view boundary, and are the minimum values ​​of the aircraft coordinate sets, and are the maximum values ​​of the aircraft coordinate set respectively; the target associated point coordinates are recorded as N is the total number of associated points; the criterion for the associated intersection points to fall into the primary fuzzy locatable visual field is designed as follows By statistically analyzing the grid distribution of the associated points in the view map, an energy accumulation matrix is ​​constructed. Combining the geometric distribution characteristics of ghost points and real targets, the first-level elimination of false associated ghost points is achieved. The horizontal and vertical view indices in the two-dimensional XOY plane are respectively I row and I col , according to the suspected target coordinates falling into the primary fuzzy locatable field of view The index calculation rules can be designed as follows: Among them, g is the prior parameter of the grid unit designed based on the optimal topological configuration; When the index is calculated according to the above rules, the point position and the grid size are not completely in a multiple relationship, and a rounding operation needs to be performed to obtain a valid established index; considering that when the associated point position falls on the boundary of the view area map, the rounding operation will make the grid map boundary index not necessarily fall within the view area, so filling processing is required to enforce the index number that exceeds the map boundary to make it fall into the map to avoid point loss; in the presence of observation errors, the associated point position of the real target will form a positioning fuzzy area; since the division of grid cells is fixed and orderly, the intersection points in the positioning fuzzy area of ​​the same target may be scattered in different grids, resulting in energy accumulation and dissipation problems; in order to avoid the target fuzzy area being located at the grid boundary, a single grid cell has no The method does not cover all target suspected points, which leads to the problem of incomplete statistics of target associated points under the grid index. By constructing multiple sets of cross-grid maps and increasing the grid density, the loss of block boundary features is prevented, thereby ensuring the target positioning accuracy. The cross design of the grid map is carried out from the x-axis and y-axis of the field of view, and the grid is densely distributed by moving the grid by the cross offset. Although the index boundaries of different grid maps are different, the position coordinates of the target suspected points under the locatable field of view map are unified, ensuring the consistency of multiple sets of grid indexes. The discrete points of direction finding cross positioning are arranged in the field of view plane, and the discrete points accumulated by the grid cells are statistically voted to complete the construction of the energy accumulation matrix under the grid map.

3. The multi-information fusion ghost point elimination and target tracking method according to claim 2, characterized in that: Step 2: Design the elimination criteria based on the idea of ​​Hough transform, specifically: The intersection points of the corresponding targets theoretically overlap, so the points will fall in the same accumulation unit. The voting results of each accumulation unit are accumulated to obtain the energy matrix of different accumulation units. The more overlapping points there are, the higher the accumulation value and the maximum energy value. In each observation period, each target can always get intersection points, which are mutually coincident. However, the false positioning points obtained due to the wrong angle correlation intersection will be scattered throughout the detection field of view, and their distribution characteristics are determined by the spatial layout of the aircraft and the position of the target. No matter how the relative positions of the aircraft and the target are distributed and how the number of sensors is selected, the target position containing the correct angle correlation intersection is always the largest. Therefore, to obtain the measurement belonging to the target from many intersection points, it is judged according to the overlap of the intersection points. It is known that the real intersection points of the target overlap and the number of overlaps is Based on the calculated energy accumulation matrix, judgment is made in two cases: (1) When the intersection points do not overlap or the number of overlaps is less than When , the intersection point is judged as a ghost point and removed; (2) When the number of overlapping intersections is equal to And there are m such coincident points, then these m points are judged as target points; if there are m+m * There are m such coincident points, then there are still m * A false point still needs to be judged.

4. The multi-information fusion ghost point elimination and target tracking method according to claim 3, characterized in that: Step 3: Use motion parameter identification to build a prediction tracking gate and design a two-level precise ghost point elimination scheme, specifically: The secondary point cluster elimination is based on the target motion characteristics identification of the historical trajectory, and the target scattering area is calculated by combining the target positioning fuzzy area and the current motion characteristics, and then the secondary prediction tracking gate is generated; the point clusters falling within the prediction tracking gate are the effective target intersection positioning points under the continuous time series, and the targets falling outside the prediction tracking gate are false detection targets caused by measurement noise, and will not obey the motion constraints at the track information level; by identifying the current motion parameters of the target, assuming that the current speed of the target is v d , the flight sampling time is t, the target maximum speed is v m , then the maximum maneuvering distance of the target is L t =v m t (4) Take a point on the boundary of the positioning fuzzy area as the center of the circle and L t Draw a circle with radius L, which is the target maneuvering dispersion circle centered on this point. The outer edges of all target maneuvering dispersion circles form an extended closed boundary, which is approximately equivalent to an ellipse. a and L b are the major and minor axes of the elliptical target scattering area, respectively. The minor axis is approximately half of the major axis, and the major axis is approximately the sum of the distance from the current target position to the longitudinal end point of the positioning ambiguity area and the target prediction step length. The approximate calculation formula is as follows: Δθ max is the direction finding error, θ1 and θ2 are the direction finding angles between the carrier aircraft and the target, and γ is the cross-positioning intersection angle; It can be seen that at the target maximum speed v m Under certain conditions, the geometric characteristics of the target elliptical dispersion area and the direction-finding error Δθ max , the direction-finding azimuth angle θ is related to the flight sampling interval t; the current cross-positioning coordinates are selected as the center of the elliptical target scattering area, and the major and minor axis parameters of the elliptical target scattering area obtained by identification calculation are combined to generate the elliptical scattering area equation; the coordinates of the target direction-finding cross-positioning are recorded as (x T ,y T ), then the criterion for a point to fall within the ellipse can be expressed as Considering the target's motion situation, the scattering area will have directionality, that is, the target scattering ellipse will have a rotation angle α; for an ellipse with a rotation angle α, the general criterion for a point falling within the ellipse can be expressed as The rotation angle α is determined by the current position of the target (x T ,y T ) and the next moment predicted position (x′ T ,y′ T ) to perform differential calculation: α=aractan((and T -and' T ) / (x T -x′ T )) (8)。 5. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 4 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

7. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images