JP2001153947A - Tracking processor and processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a tracking processor and a processing method in which the track keeping performance can be enhanced under unwanted signal environment while keeping the size of a gate at a track keeping level. SOLUTION: The tracking processor comprises a means 1 for calculating a prediction vector and a prediction error covariance matrix based on the theorem of Kalman filter, an element 2 for delaying the prediction vector and the prediction error covariance matrix by 1 sampling, means 3 for calculating a gate size for maximizing the difference of number between target signals and unwanted signals in a gate, means 4 for making a decision whether an observation vector exists in the gate or not, a smoothing means 5 for integrating a gain matrix, a smoothed vector, and a smoothed error covariance matrix while weighting through the use of observation vector and prediction vector in the gate, an observing means 6 for inputting a target search probability calculated based on the signal/noise ratio, an erroneous alarm probability of unwanted signal, an observation error and an observation vector to a gate decision means, and a means 7 for displaying the smoothed position.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a tracking processing apparatus and method for tracking a target, and more particularly to a tracking processing apparatus and method for enhancing tracking in an unnecessary signal environment.

[0002]

2. Description of the Related Art A conventional tracking processing apparatus will be described with reference to the drawings. FIG. 13 shows, for example, “Samuel S. Bl
ackman, Multiple-Target Tracking with Radar Applica
tions, Artech House, Dedham, 1986 ”p83-p113 (especially p
FIG. 88 is a block diagram illustrating a configuration of a tracking processing device configured using the gate generation method illustrated in (88).

[0003] In FIG. 13, reference numeral 1 denotes a sample after one sampling (time k
The prediction means 2 which outputs the prediction vector and the prediction error covariance matrix of (+1), and the prediction vector and the prediction error covariance matrix after one sampling (time k + 1) obtained from the prediction means 1 The delay element 4 for outputting the prediction vector and the prediction error covariance matrix of k), the prediction vector and the prediction error covariance matrix of the current time (time k) obtained from the delay element 2 and the observation element 6 described later. Observation noise vector, NN (Nearest Neighbor
The gate determination means for inputting the gate size obtained from the gate size calculation means 21 by the r) method and outputting the error covariance matrix of the observation vector at the current time (time k) determined to be present in the gate and the predicted observation value. It is.

Reference numeral 5 denotes an observation vector at the current time (time k) determined to be present in the gate obtained by the gate determination means 4, an error covariance matrix of predicted observation values, and a current time obtained from the prediction means 1. A smoothing means 6 for inputting a prediction vector and a prediction error covariance matrix of the above and outputting a smoothed vector and a smoothed error covariance matrix at the current time; and 6, an observation vector (observation position) expressed by polar coordinates obtained from a built-in sensor. And an observation noise vector represented by polar coordinates, a distance (prediction vector) between the sensor position obtained from the prediction means 1 and the target predicted position (predicted vector), and an observation vector represented by north reference rectangular coordinates (observed position); This is an observation unit that outputs an observation noise vector expressed in polar coordinates, a detection probability, a false alarm probability, and an unnecessary signal occurrence frequency.

Further, reference numeral 7 denotes a display means for inputting a smoothing vector obtained from the smoothing means 5 and outputting a smoothing position and a smoothing speed obtained by cutting out only a position component from the smoothing vector, and 21 denotes a gate judgment. NN (Neares) which inputs the current time prediction observation error covariance matrix obtained from the means 4, the unnecessary signal occurrence frequency obtained from the observation means 6, and the detection probability and outputs the gate size to the gate determination means 4
t Neighbor) method.

The tracking processing apparatus shown in FIG. 13 performs prediction according to a target motion model defined by Equation 1 and a target observation model defined by Equation 2 on the assumption that a Kalman filter is used. I have.

[0007]

(Equation 1)

Here, x _k is a state vector, Φ _k is a state transition matrix, Γ1 _k is a drive noise vector conversion matrix, and w _k is a drive noise vector.

In addition, the lower right subscript k in the code notation of a matrix and a vector represents the case of the sampling time t _k unless otherwise specified. For example, in the case of x _k , “the state of the sampling time t _k Vector "
For x _k-1, a means "state vector of the sampling time t _k-1", when the [Phi _k-1, which means "the state transition matrix of the sampling time t _k-1".

[0010]

(Equation 2)

Note that Z _k is an observation vector, H _k is an observation matrix, Γ2 _k is a transformation matrix of an observation noise vector, and v _k is an observation noise vector.

In the configuration shown in FIG. 13, the prediction means 1 calculates a prediction vector by using Expression 3 and calculates a prediction error covariance matrix by using Expression 4 according to the motion model of Expression 1.

[0013]

(Equation 3)

[0014]

(Equation 4)

In addition,x _k(-) Is a prediction vector,x
_k(+) Is a smooth vector, P_k(-) Is the prediction error covariance row
Column, Q_kIs the driving noise covariance matrix, P_k(+) Indicates smoothing error
It is a covariance matrix.

Hereinafter, the upper right suffix T in the code notation of a matrix and a vector represents the transpose of the matrix and the vector, unless otherwise specified.

In the delay element 2, the prediction vector and the prediction error covariance matrix at the sampling time k + 1 calculated by the prediction means 1 are delayed by one sampling to obtain the prediction vector and the prediction error at the sampling time k. Output the covariance matrix.

Next, the gate size calculating means 21 and the gate judging means 4 based on the NN method in FIG. 13 will be described. Gate size calculation means 21 by NN method
First, the gate determination means 4 will be described. Further, the gate will be described in the description of the gate determination means 4.

A gate is a range in which a target is expected to exist. FIG. 14 illustrates a gate. In FIG. 14, P0 is a predicted observation vector representing the center position of the gate, and D1, D2, and D3 are observation vectors existing in the gate, respectively. Actually, the observation vector D1,
Although more observation vectors than D2 and D3 may be obtained, in FIG. 14, it is assumed that only three observation vectors D1, D2, and D3 are obtained in the gate for the sake of explanation.

In addition to the target signal, the observation vector
Unwanted signals that are signals other than the target signal may also be included. Further, the dotted lines in FIG. 14 indicate the predicted observation vector P0 and the observation vectors D1 and D existing in the gate, respectively.
2, the distance normalized by the probability density with D3.

The distance normalized by the probability density is defined as a correlation distance, and FIG. 15 shows the correlation distance. FIG.
In, a gate is created based on the predicted observation vector from the predicted observation vector and the prediction error covariance matrix. FIG.
In the set of observation vectors obtained by the observation means 6 shown in FIG. 3, for example, the observation vectors included in the gate in FIG. At this time, regarding the probability calculated from the probability density from the error covariance matrix of the predicted observation value, when the correlation distance α1 is closer than the correlation distance α2, the probability from the correlation distance α1 at that time is the probability from the correlation distance α2. With a higher probability.
In the following, unless otherwise specified, performing weighted integration with a correlation distance has the same meaning as weighting with a probability from the correlation distance.

In the conventional tracking processing apparatus, in FIG. 14, only the signal D1 closest to the predicted observation vector among the plurality of observation vectors existing in the gate is used, and the smoothing which is the current estimated value of the target is performed by the Kalman filter. I want a vector. The method of calculating the smooth vector through the processing in the gate is called the Nearest Neighbor method. On the other hand, in FIG. 14, among the plurality of observation vectors existing in the gate, the observation vectors D in all the gates are shown.
1, D2, and D3, and the predicted observation vector, using the respective correlation distances, the observation vector D
A method of calculating a smooth vector as a current estimated value of a target by weighting all of D1, D2, and D3 is described in All Neighbours.
Use the r method. Nearest Neighbor method, All Neighb
Unless otherwise specified, our system is abbreviated as NN system or AN system.

The gate determination in the conventional tracking processing device is performed by the NN method. The gate judgment is shown in FIG.
In the gate determination means 4 in, whether or not the observation vector obtained from the observation means 6 is a target signal,
It is determined whether the signal is an unnecessary signal.

[0024]

(Equation 5)

Here, Z _k (-) is a predicted observation vector, S _k
Is the error covariance matrix of the predicted observations, and d is the gate size. Equation 5 represents a gate that is a region where a target is predicted to be present. Hereinafter, -1 in the upper right suffix of the code notation of a matrix represents an inverse matrix unless otherwise specified. For example, S _k ^-1 represents the inverse of the error covariance matrix of the predicted observation.

As for the necessity of a gate, a radar is considered as a sensor included in the observation means 6, for example. In this case, the target cannot be detected when the direction is 180 degrees opposite to the direction of the radio wave transmitted from the antenna, that is, when the target exists behind the antenna. That is, since the range where the target is detected is not the whole space but a part of the space, a gate is required. In Equation 5, the factors that determine the size of the gate are:
This is the error covariance matrix of the gate size d or the predicted observation value.

FIG. 16 shows an example of a gate when the order of the observation vector is two-dimensional. In this case, the gate is inside the probability ellipse whose mean is determined by the prediction vector and whose variance is determined by the observation error covariance matrix. Further, the error covariance matrix of the predicted observation value, which is one of the factors that determine the size of the gate, represents the error between the predicted observation vector and the state vector that is the true value of the target. Here, when an unnecessary signal is present, a target observation position far from the true target position is obtained, the value of the error covariance matrix of the predicted observation value increases, and the gate size increases. Further, as in the second term on the right-hand side of Equation 4, if the observation accuracy is poor, the covariance matrix of the observation noise vector increases, so that the gate also increases with respect to the parameter d that determines the gate size at a fixed value.

Further, the gate determining means 4 in FIG. 13 will be described based on the Kalman filter theory. FIG.
The gate determination means 4 in 3 determines the observation vector obtained from the observation means 6 using Equation 5. The predicted observation vector in Expression 5 is calculated by Expression 6, and the error covariance matrix of the predicted observation value in Expression 5 is calculated by Expression 7. Here, the calculation formulas of Expressions 5, 6, and 7 follow the observation model of Expression 2.

[0029]

(Equation 6)

[0030]

(Equation 7)

Here, P _k (-) is a prediction error covariance matrix.

The size of the gate corresponds to a right-sided test in mathematical statistics. That is, Equation 5 secures a certain probability of the target reference for the signal from the target to be present in the gate. The predicted observation vector of Equation 6 represents the center of the gate. It is assumed that the observation vector in Expression 5 follows a multivariate normal distribution in which the average is the predicted observation vector of Expression 6, and the variance is the error covariance matrix of the predicted observation values of Expression 7. At this time, the left side of Equation 5 follows a chi-square distribution with N degrees of freedom, where N is the degree of the observation vector.

Next, the gate size calculating means 21 based on the NN method will be described. First, a parameter d for determining the gate size used for the gate determination means 4 in FIG. That is,
G0 calculated by the equation 8 is replaced with a parameter d for determining the gate size.

[0034]

(Equation 8)

Note that n is the order of the observation vector, π is the pi, P _D is the detection probability, β _k is the frequency of occurrence of unnecessary signals, and d 0 is the optimal parameter.

Here, the calculation method of Expression 8 will be described. In the NN method used in the conventional tracking processing device, the target is not detected, and the position of the observation vector detected from the target is located at the center of the gate rather than the position where the unnecessary signal exists in the gate. desirable. That is, FIG.
In, the target signal is preferably the observation vector D1 with respect to the predicted observation vector P0 which is the center of the gate. Therefore, for example, based on the assumption that the unnecessary signal exists in the gate in a uniform distribution, the probability that the unnecessary signal exists at a certain position in the gate without detecting the target is g1 (z).
Is represented by Expression 9.

[0037]

(Equation 9)

Note that z is a random variable having the same dimension as the observation vector. In addition, the probability density function g2 (z) of the observation vector detected from the target is expressed by Equation 10 based on the assumption that the mean follows the predicted observation vector and the variance follows the multivariate normal distribution of the error covariance matrix of the predicted observation value. Become.

[0039]

(Equation 10)

Note that exp is the number of napiers (2.7182)
...). In the NN method used in the conventional tracking processing device, as described above, the target is not detected, and the position of the observation vector detected from the target is smaller than the position where the unnecessary signal exists in the gate. It is desirable to be in the center. That is, it is desirable that the relationship of Expression 11 be established.

[0041]

[Equation 11]

Therefore, the transformation is performed by substituting Equations 9 and 10 into Equation 11, and the gate size satisfying Equation 11 is expressed by Equation 8
Is derived by The smoothing means 5 in FIG. 13 calculates the gain matrix K _k using Equation 12, calculates the smoothing vector x _k (+) using Equation 13, and calculates the smoothing error covariance matrix P _k (+) using Equation 14.

[0043]

(Equation 12)

[0044]

(Equation 13)

[0045] Incidentally, z _{k, j} is the observation vector where the observation vector to be used for updating the smoothing vector, when the number of observation vector obtained when the sampling time t _k and the m _k, the observation vector The set ｚ z _{k, 1} , ..., z
_{k, j, ···, z k} , mk} ( where, j = 1, ···,
_mk ), z _{k, j} is used.

[0046]

[Equation 14]

As the observation vector used for updating the smoothed vector of Expression 13, one selected by the gate determination unit 4 from the set of observation vectors obtained by the observation unit 6 is used.

FIG. 17 is a diagram for explaining a coordinate system related to the observation means 6 in FIG. In FIG. 17, O
Is a sensor, T is a tracking target, R is a distance between the tracking target T and the sensor O, and E is a line segment OT connecting the sensor O and the tracking target T.
Is an elevation angle formed by the XY plane, and B is an azimuth formed by an orthogonal projection vector of the line segment OT connecting the sensor O and the tracking target T onto the XY plane with the X axis. [R, E, B] represents "polar coordinates", and [X, Y, Z] represents "north reference rectangular coordinates". Hereinafter, unless otherwise specified, the coordinates are simply referred to as “coordinates”, and the north reference rectangular coordinates [X, Y, Z]
Is expressed.

The covariance matrix of the observation noise vector in Equation 2 and the observation noise vector in Equation 7 is expressed in polar coordinates. Also, the vectors and matrices in Equations 1, 2, 3, 4, 5, 6, 7, 12, 12, 13, and 14 other than the observation noise vector and the covariance matrix of the observation noise are north. Expressed in reference rectangular coordinates.

The observation means 6 shown in FIG. 13 observes a target position in the polar coordinates of FIG. 17, and converts observation information including an observation vector and an observation noise vector represented by the polar coordinates of the target into the north reference orthogonal in FIG. Convert to coordinates and output. The display means 7 displays the smoothed value calculated by the smoothing means 5.

[0051]

As described above, the conventional tracking processing device does not consider all the observation signals in the gate, and therefore, when calculating the parameter d for determining the gate size, the detection probability is particularly 1 unit. In this case, the value of the parameter d becomes infinite. That is, in this case, since the area inside the gate coincides with the entire space,
There is a problem that it does not make sense as a gate. In this case, in an unnecessary signal environment, it is easy to erroneously track the unnecessary signal.

In the conventional example, since only one observation vector in the gate is used to calculate the smoothed vector, when the signal-to-noise ratio obtained from the sensor is small, if the observation vector varies, one of them is used. If the selected observation vector is an unnecessary signal, there is a problem that tracking cannot be maintained because the smoothed vector varies and is obtained. further,
Each time you have to recalculate the gate size,
There is a problem that the calculation load increases.

The present invention has been made in order to solve the above-described problems of the conventional example.
Keep the gate size large enough to keep track,
An object of the present invention is to provide a tracking processing apparatus and method capable of enhancing tracking maintenance in an unnecessary signal environment.

[0054]

A tracking processing apparatus according to the present invention performs prediction according to a target motion model and an observation model based on Kalman filter theory, and calculates a prediction vector and a prediction error covariance matrix. A prediction element set in advance in consideration of a delay element for delaying the prediction vector and the prediction error covariance matrix calculated by the above-described prediction means by one sampling and all observation vectors in the gate which is a target existence expectation area. By using the variance ratio of the observation error and the error, the constant that determines the ratio of the number of target signals to the number of unnecessary signals, the observation error calculated based on the signal-to-noise ratio, the detection probability of the target, and the false alarm probability of the unnecessary signal, A gate size calculation unit that calculates a gate size that maximizes a difference between the number of target signals and unnecessary signals in the Gate determination means for determining whether or not an observation vector exists within the gate based on the distance between the predicted observation vector and the observation vector, which are the center of the gate set by the calculated gate size, and the gate determination means By using the observation vector determined to be present in the gate and the prediction vector obtained by the prediction means, the gain matrix is calculated by the distance between the prediction observation vector at the center of the gate calculated by the gate determination means and the observation vector, A smoothing means for obtaining a smoothed vector and a smoothed error covariance matrix by weight integration; a sensor; a target detection probability calculated based on a signal-to-noise ratio from the sensor; a false alarm probability of an unnecessary signal; An observation unit for inputting the error and the observation vector to the gate determination unit, and a target calculated by the smoothing unit. It is obtained and a display means for displaying a smooth position from the smoothing vector is an estimate of standing.

The smoothing means sets a limited number of observation vectors in the gate determined by the gate determination means, and uses the distance between the observation vector in the gate and the predicted observation vector which is the center of the gate, The weighted integrated gain matrix, smoothed vector and smoothed error covariance matrix are calculated.

Further, the prediction means, the delay means, the gate size calculation means, the gate determination means and the smoothing means are provided by a first prediction means, a first delay means, a first gate size calculation means, A second prediction unit, a second delay unit, a second gate size calculation unit, a second gate determination unit, and a second prediction unit having the same functions as the first gate determination unit and the first smoothing unit. A second smoothing means is further provided, and a ratio between the number of target signals in the gate and the number of unnecessary signals is provided for each of the plurality of smoothing means, and the first gate size calculating means and the second gate size calculating means have a ratio. Means for varying the ratio between the number of target signals and the number of unnecessary signals input by changing the ratio between the number of target signals and the number of unnecessary signals in the gate; and a plurality of means respectively calculated by the first and second smoothing means. Of smooth vector The smaller the residual of observation vectors of 置成 min at the same time, as a smoothing position of the tracking target goals,
A display determining means for determining whether one of the output results of the first smoothing means and the second smoothing means is to be displayed on the display means.

Further, the prediction means, the delay means, the gate size calculation means, the gate determination means, the smoothing means, the observation means, and the display means are referred to as first prediction means, first delay means, A first gate size calculation unit, a first gate determination unit, a first smoothing unit, a first observation unit, and a first display unit, whereas a second prediction unit having the same function as these, A second delay unit, a second gate size calculation unit, a second gate determination unit, a second smoothing unit, a second observation unit, and a second display unit. And that the difference between the smoothed vector obtained from the first smoothing means and the observed vector obtained from the first observing means is smaller than a preset threshold value based on the output of Sa In this case, the ratio between the number of target signals and the number of unnecessary signals used for calculating the gate size by the first gate size calculation means is converted to the ratio between the number of target signals and the number of unnecessary signals used for the second gate size calculation means. It is characterized by further comprising diversion means.

Also, only in the initial time zone, the control signal is changed in a ratio between the target signal number and the unnecessary signal number so that the ratio of determining the target signal number and the unnecessary signal number in the gate is kept low and the target signal number is emphasized. And an initial time zone gate control means.

When the signal-to-noise ratio obtained from the sensor of the above-mentioned observation means is larger than a certain threshold value, the weighting of the prediction vector and the observation vector at the time of calculating the smoothed vector is performed by the prediction observation vector which is the center of the gate. The smoothing means is further provided with a limiting means for providing the smoothing means with a control signal for limiting only to the weighting of the observation vector and the prediction vector closest to.

Further, the tracking processing method according to the present invention
(A) performing prediction according to a target motion model and an observation model based on the theory of a Kalman filter, calculating and outputting prediction information including a prediction vector and a prediction error covariance matrix, and (b) a signal pair obtained from a sensor. Inputting noise ratio information; and (c) calculating observation information including a target detection probability, an unnecessary signal false alarm probability, an observation error, and an observation vector based on the signal-to-noise ratio information input. (D) inputting a variance ratio between a prediction error and an observation error set in advance, (e) target observation information calculated from the signal-to-noise ratio, and the prediction error and the observation error set in advance. Using the dispersion ratio, the gate size that determines the size of the gate that is the target existence expected area is calculated as the gate size that maximizes the difference between the number of unnecessary signals and the number of target signals. (F) determining an observation vector considered as a target signal from a set of observation vectors including an unnecessary signal and a target signal based on the calculated gate size; and (g) determining the target signal in the gate. Using the obtained observation vector, the prediction vector and the prediction error, a gain matrix integrated by weighting using the distance between the observation vector in the gate and the position component of the prediction vector that is the center of the gate, and a weighted integrated smoothing (H) displaying a smoothed position from a weighted integrated smoothed vector, which is a current estimated value of a target, and (i) displaying a smoothed position including a vector and a weighted integrated smoothed error covariance matrix. When processing from step a) to step (h) is continued,
After the step (h) has passed, returning to the prediction information calculation processing step (a) via a delay element.

[0061]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiment 1 A tracking processing device according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a tracking processing device according to Embodiment 1 of the present invention. In FIG. 1, FIG.
The same parts as those of the conventional example shown in FIG. As a new code, reference numeral 3 denotes a gate size calculation unit according to the first embodiment. The gate size calculation unit 3 considers all observation vectors in the gate in the gate that is the target existence expected area. A dispersion ratio between a prediction error and an observation error set in advance, a constant for determining a ratio between the number of target signals and the number of unnecessary signals, an observation error calculated based on a signal-to-noise ratio obtained by the observation means 6, a target detection probability , By using the false alarm probability of unnecessary signals,
A gate size that maximizes the difference between the number of target signals and unnecessary signals in the gate is calculated.

That is, the parameter d used for the gate determination means 4, that is, the so-called gate size d is calculated by the gate size calculation means 3 described later. Here, a method of determining a parameter d for determining the gate size is as follows.
As shown in FIG. 16, for the target signal, the gate is made as large as possible to take in the target signal as much as possible, and for the unnecessary signal, the gate is made small so that the unnecessary signal is not taken in as much as possible. Want to. As a method of realizing this, a parameter d is determined so as to minimize the function Λ (d) shown in Expression 15.

[0063]

(Equation 15)

Note that P _Gk is a target existence expectation probability in the gate,
V _Gk is the gate volume, C is a constant, Λ (d) is the parameter d
Is a function of Here, the target existence expectation probability in the gate in Expression 15 is determined by Expression 16, and the gate volume is expressed by Expression 17
Is determined by The gate volume represents the size of the gate. For example, if the order of the observation vector is 3rd, the volume is 2
The area is represented if:

[0065]

(Equation 16)

[0066]

[Equation 17]

Note that Γ (·) is a Γ function. Also, d
Unless otherwise specified, et denotes a determinant of a matrix.

In the case of a three-dimensional radar, for example, the frequency of occurrence of unnecessary signals in Equation 15 is 3 as shown in Equation 18.
It is obtained from the distance between the sensor position of the three-dimensional radar and the target predicted position, the distance resolution, the elevation angle resolution, the azimuth angle resolution, and the false alarm probability.

[0069]

(Equation 18)

[0070] The distance between the r _pk sensor position and the target predicted position, [Delta] r is the distance resolution, .DELTA.e elevation angle resolution, [Delta] b
Is the azimuth resolution, and P _FA is the false alarm probability.

The first term on the right side of the parameter d for determining the gate size in Expression 15 represents the number of target signals in the gate, and is a value to be increased. The second term on the right side represents the number of unnecessary signals in the gate, and is a value to be reduced. The positive constant C in Equation 15 is C
As the value of is larger, the number of unnecessary signals in the gate in the second term on the right side of Formula 15 is emphasized, and the function of reducing the number of unnecessary signals detected in the gate is performed.

That is, the parameter d is such a value that maximizes the function Λ (d) of the parameter d that determines the gate size in Expression 15. From now on, this value is defined as a parameter d0. Here, maximizing the function Λ (d) of the parameter d that determines the gate size in Equation 15 is equivalent to maximizing the difference between the target signal and the unnecessary signal.

In order to calculate the parameter d0 that maximizes the function Λ (d) of the parameter d that determines the gate size in Equation 15, the function Λ (d) is differentiated once with respect to the parameter d. Function し た differentiated once with parameter d
(D) is defined as “Λ (d) dash”. ThatΛ
(D) When the dash is set to 0 and the parameter d that gives the extreme value is calculated, the parameter d0 that gives the extreme value is obtained as shown in Expression 19.

[0074]

[Equation 19]

When the target is traveling straight from true sensor to sensor from true north at an elevation angle of 0 degrees, the error covariance matrix of the predicted observation value is calculated using the predicted position error and the observed position error in the north reference Cartesian coordinates. It can be expressed as in Equation 20.

[0076]

(Equation 20)

Σ _{x, p} is the standard deviation of the x component of the predicted position error, σ _{y, p} is the standard deviation of the y component of the predicted position error,
_{z, p} is the standard deviation of the z component of the predicted position error, σ _{x, 0} is the standard deviation of the x component of the observed position error, σ _{y, 0} is the y of the observed position error
The component standard deviation, σ _{z, 0,} is the standard deviation of the z component of the observation position error.

The signal-to-noise ratio obtained from each sensor in the observation means 6, the distance observation noise constant determined by each sensor, the elevation angle observation noise constant determined by each sensor, and the azimuth angle observation noise constant determined by each sensor. The standard deviation of the distance observation noise, the standard deviation of the elevation angle observation noise, and the standard deviation of the azimuth angle observation noise are determined by Expression 21.

[0079]

(Equation 21)

Σ _r is the standard deviation of the distance observation noise, σ _e
The standard deviation of the elevation angle measurement noise, sigma _b is the standard deviation of the azimuth observation noise, k _r is a constant distance measurement noise determined by the sensors, standard deviation, k _e is a constant elevation measurement noise determined by the sensors, k _b Is a constant of azimuth observation noise determined by each sensor, and S / N is a signal-to-noise ratio.

The product of the square of the standard deviation of the X component of the observation noise, the square of the standard deviation of the Y component of the observation noise, and the square of the standard deviation of the Z component of the observation noise is the target predicted position and the sensor position. , The standard deviation of the distance observation noise, the standard deviation of the elevation angle observation noise, and the standard deviation of the azimuth angle observation noise,
It can be represented by Equation 22.

[0082]

(Equation 22)

Note that r _p is the distance between the sensor and the predicted position.

The detection probability of the target can be expressed by Expression 23 using the false alarm probability and the signal-to-noise ratio.

[0085]

(Equation 23)

Now, in Expression 19, Expression 18, Expression 20, Expression 21,
Substituting Equations 22 and 23 and rearranging them results in Equation 24.

[0087]

(Equation 24)

In Expression 24, a constant C, a distance resolution, an elevation resolution, an azimuth resolution, a false alarm probability, a signal-to-noise suppression ratio, a distance observation noise constant determined by each sensor, an elevation observation noise constant determined by each sensor, The azimuth angle observation noise constant determined by each sensor, the X component of the variance ratio between the prediction error and the observation error, obtained by dividing the square of the standard deviation of the X component of the prediction position error by the square of the standard deviation of the X component of the prediction position error. , The variance ratio of the prediction error and the observation error obtained by dividing the square of the standard deviation of the Y component of the prediction position error by the square of the standard deviation of the Y component of the observation noise.
The Z component of the variance ratio between the prediction error and the observation error, which is obtained by dividing the square of the standard deviation of the Z component of the prediction position error by the square of the standard deviation of the Z component of the observation noise, The value to set.

Therefore, the gate size calculating means 3 calculates the gate size d0 using the equation (24). As a result, the parameter d in Equation 5 of the gate determination unit 4 is obtained.
Use this parameter d0.

The smoothing means 5 uses the correlation distance calculated between the predicted observation vector, which is the center of the gate, calculated by the gate determination means 4 and a plurality of observation vectors existing in the gate, as shown in Expressions 12 and 13, the gain matrix, the smooth vector, and the smooth error covariance matrix represented by Expression 14 are weighted and integrated by the correlation distance, and the weighted integrated gain matrix, the weighted integrated smooth vector, and the weighted integrated smooth error covariance matrix are obtained. calculate.

The weighted and integrated smoothed vector and the weighted and integrated smoothed error covariance matrix, which are the output results of the smoothing means 5, are input to the smoothing vector and the smoothed error covariance matrix of the prediction means 1. The weighted and integrated smoothed vector, which is the output result of the smoothing unit 5, is a smoothed vector input to the display unit 7.

Next, FIG. 2 is a flowchart for explaining the operation of the first embodiment. The tracking processing method will be described with reference to this flowchart. First, step ST1
In, prediction information including a prediction vector and a prediction error covariance matrix is calculated.

Next, in step ST2, the signal-to-noise ratio from the sensor is input.

In step ST3, using the signal-to-noise ratio from the sensors and the constants of the distance, elevation, and azimuth observation noise determined for each sensor, the observation information including the observation position and the observation noise by the sensor is calculated in polar coordinates. And then
Observation position and observation noise, which are the observation information input in the polar coordinates, are converted to the north reference rectangular coordinates, and the target detection probability is calculated from the signal-to-noise ratio from the sensor and the false alarm probability. Calculated as observation information.

Next, in step ST4, the variance ratio between the predicted position error and the observed position error set in advance is input.

In step ST5, using the observation information calculated in ST3 and the variance ratio between the predicted position error and the observation position error calculated in advance in step ST4, the unnecessary signal in the gate and the target The gate size d0 is calculated so that the number of signals is maximized. Further, the constant C in Expression 24 includes signals from targets other than the tracking target in the unnecessary signal, and can cope with multiple targets.

Next, in step ST6, using the gate size previously calculated in step ST5, a set of observation vectors including the unnecessary signal and the target signal satisfying Equation 5 is used to determine the observation signal considered to be the target signal. Choose a vector.

Next, in step ST7, the observation vector considered to be the target signal selected in step ST6, and the observation error covariance matrix from the observation vector,
And a prediction vector calculated in the prediction information calculation in step ST1, a weighted integrated gain matrix calculated using the correlation distance calculated by the gate determination process from the prediction error covariance matrix, a weighted integrated smoothed vector, Calculate smoothing information including a weighted integrated smoothing error covariance matrix.

Next, in step ST8, the smoothed position is displayed on a monitor or the like so that the operator can confirm the smoothed position from the weighted and integrated smoothed vector which is the current estimated value of the target.

Next, in step ST9, if the process from step ST1 to ST8 is continued, the process returns to the process of calculating the prediction information in step ST1 via a delay element or the like.

Therefore, in the conventional tracking processing device, the parameter d for determining the gate size is calculated by using Expression 8, but in Expression 8, when the detection probability is 1, the value of the parameter g0 in Expression 8 is changed. It will be infinite. That is, in Equation 8, the area inside the gate coincides with the entire space, and thus does not make sense as a gate. Further, in the conventional example, a smoothed vector is calculated by using only one observation vector in the gate. If the one selected observation vector is an unnecessary signal, tracking cannot be maintained because the smooth vector varies.

On the other hand, in the first embodiment, the parameter d for determining the gate size is set such that the difference between the number of unnecessary signals in the gate and the number of target signals is maximized.
By determining 0 according to Equation 8, the gate can be appropriately sized even in an unnecessary signal environment or in a free space free of unnecessary signals.

Further, in the first embodiment, a conventional tracking processing apparatus is used to calculate a weighted integrated smoothed vector using a correlation vector calculated from a predicted observation vector at the center of a gate and an observation vector in the gate. , The variation of the smooth vector is suppressed, and the tracking is easily maintained.
Also, compared to the conventional method, it is not necessary to calculate the error covariance matrix of the predicted observation value as in Equation 7 at each sampling time,
Calculation load is reduced.

The parameter d for determining the gate size
Is determined by Equation 24 such that the difference between the number of unnecessary signals in the gate and the number of target signals is maximized, so that the gate can be appropriately sized even in an unnecessary signal environment or a free space free of unnecessary signals. Can be.

Further, in order to calculate a smoothed vector which is weighted and integrated using a predicted observation vector at the center of the gate and a correlation distance calculated from the observation vector in the gate,
Compared with the conventional tracking processing device, the variation of the smooth vector is suppressed, and the tracking is easily maintained. Further, the calculation load can be reduced as compared with using Equation 19 as it is.

Embodiment 2 FIG. 3 is a block diagram showing a configuration of a tracking processing device according to Embodiment 2 of the present invention. 3, the same parts as those in the first embodiment shown in FIG. 1 are denoted by the same reference numerals, and the description thereof will be omitted. As a new code, reference numeral 8 denotes a smoothing unit that limits the number in the gate. The smoothing unit 8 sets a limited number to the observation vector in the gate determined by the gate determination unit 4 and performs the observation in the gate. Using the distance between the vector and the predicted observation vector that is the center of the gate, a gain matrix integrated with weights, a smooth vector, and a smooth error covariance matrix are calculated.

FIG. 14 is used to explain the smoothing means 8 in which the number in the gate is limited. In FIG. 14, the observation vector D1 and the observation vector D1
2. Assume that the observation vector is D3.

At this time, in the smoothing means 8 which limits the number in the gate, the observation vector D2 having the longest correlation distance from the center P0 of the gate, for example, among the observation vectors in the plurality of gates is discarded, and the correlation distance Is weighted and integrated by the correlation distance using the observation vectors D1 and D3, which are relatively short, and the observation vectors in the gates whose number is limited, the smoothed vector with the weighted integration, the gain matrix with the weighted integration, Calculate the smoothed error covariance matrix subjected to the weighted integration. Here, the limitation on the number of use of the in-gate observation vectors is determined in advance before operating the tracking processing device.

Next, FIG. 4 is a flowchart for explaining the operation of the second embodiment. In the flowchart shown in FIG. 4, the same portions as those of the steps shown in the flowchart according to the first embodiment shown in FIG. 2 are denoted by the same reference numerals, and description thereof will be omitted. As a new step, there is a step ST7b.
Only T7b will be described.

In step ST7b, step S7
Of all the observation vectors in the gate output at T6, the number of observation vectors to be weighted and integrated is limited, and each observation vector is weighted by the correlation distance from the observation vector with the limited number. , A gain matrix weighted and integrated by correlation distance, a smoothed error covariance matrix weighted, and a smoothed vector weighted integrated.

Therefore, according to the second embodiment, in the first embodiment, the parameter d for determining the gate size is set so that the difference between the number of unnecessary signals in the gate and the number of target signals is maximized. Is determined by the equation (24), so that the gate can be appropriately sized even in an unnecessary signal environment or a free space free of unnecessary signals. In the second embodiment, the weighted integration of the smoothed vector, the smoothed error covariance matrix, and the gain matrix by the plurality of observation vectors in the gate is performed. By providing a restriction, a calculation load is slightly applied compared to the conventional tracking processing device, and a smoothed vector obtained by integrating weights using a correlation distance calculated from a predicted observation vector at the center of the gate and an observation vector in the gate. Is calculated, the variation of the smooth vector is suppressed, and tracking is easily maintained as compared with the conventional tracking processing device. That is, the calculation load is lighter than in the first embodiment.

Embodiment 3 FIG. 5 is a block diagram showing a configuration of a tracking processing device according to Embodiment 3 of the present invention. 5, the same components as those in the first embodiment shown in FIG. 1 are denoted by the same reference numerals, and the description thereof will be omitted. As new codes, 11 to 15 denote the same prediction means 1, delay element 2, gate size calculation means 3, gate determination means 4, and smoothing means 5 as those of the first embodiment, While the delay element, the first gate size calculation means, the first gate determination means and the first smoothing means are used, the second prediction means, the second delay element, and the second Gate size calculation means, second gate determination means, and second gate determination means
Is a smoothing means.

Reference numeral 9 denotes a ratio between the number of target signals in the gate and the number of unnecessary signals for each of the plurality of smoothing means, and the first gate size calculating means 3 and the second gate calculating means 13 control the ratio of the number of unnecessary signals in the gate. The ratio varying means 10 between the number of target signals and the number of unnecessary signals, which is input while changing the ratio between the number of target signals and the number of unnecessary signals, includes a plurality of smoothing circuits calculated by the first smoothing means 5 and the second smoothing means 15, respectively. The output result of any one of the first smoothing means 5 and the second smoothing means 15 is displayed as the smoothed position of the target to be tracked, with the smaller of the residual of the observation vector at the same time as the position component of the vector. 7 is a display judging means for judging whether or not to display.

In the structure shown in FIG. 5, the gate size calculation means 3 and the second gate size calculation means 13
4, the gate size is calculated. At this time, the ratio varying means 9 between the target signal number and the unnecessary signal number has a function of changing the constant C in Expression 24. Further, the constant C in Equation 24 is changed in the gate size calculation means 3 and the second gate size calculation means 13, respectively.

The display judging means 10 comprises the smoothing means 5 and the second
According to the magnitude of the residual between the position components of the plurality of smoothed vectors calculated by the smoothing means 15 and the observation vector at the same time, the smoothed position of the target to be tracked is determined, and the smoothing means 5 and the second smoothing means 15 It is determined whether any one of the outputs is displayed.

FIG. 6 is a flowchart for explaining the operation of the third embodiment. In the flowchart shown in FIG. 6, the same portions as those of the steps shown in the flowchart according to the first embodiment shown in FIG. 2 are denoted by the same reference numerals, and description thereof will be omitted. There are steps ST10 and ST11 as new steps, and the steps ST10 and ST11 will be mainly described below.

In step ST10, the ratio between the number of target signals and the number of unnecessary signals in the gate is substituted, the gate size is calculated in the next step ST5 for each ratio, and the gate determination process is performed in the next step ST6. In the next step ST7, the smoothing information is calculated,
Calculate the smoothed error covariance matrix. Further next step S
At T11, the position components of the smoothed vector obtained as a result of the smoothing information calculation in step ST7 calculated based on the ratio between the number of target signals and the number of unnecessary signals are represented by the magnitude of the residual difference between these smoothed vectors and the observation vector at the same time. Thus, only one smooth position of the tracking target is displayed for each tracking target.

Therefore, according to the third embodiment, a plurality of ratios between the number of target signals and the number of unnecessary signals in a gate are given, a plurality of gate sizes are calculated, gates are set, and a smooth vector of each gate is set. Among them, one of the smaller residuals is displayed as compared with the observation vector at the same time, so that the operator who sees the display can easily confirm that the tracking is maintained.

Embodiment 4 FIG. 7 is a block diagram showing a configuration of a tracking processing device according to Embodiment 4 of the present invention. 7, the same parts as those in the first embodiment shown in FIG. 1 are denoted by the same reference numerals, and the description thereof will be omitted. As new codes, reference numerals 11 to 17 denote prediction means 1, delay element 2, gate size calculation means 3, gate determination means 4, smoothing means 5, observation means 6, and display means 7 similar to those in the first embodiment. 1st prediction means, 1st delay element, 1st gate size calculation means, 1st gate determination means, 1st smoothing means, 1st
, The second predicting means, the second delay element,
A second gate size calculation unit, a second gate determination unit,
A second smoothing unit, a second observation unit, and a second display unit.

Also, 18 indicates that the observation vector exists in the gate based on the output from the first gate determination means 4, and the smoothed vector obtained from the first smoothing means 5 and the first observation When it is determined that the difference between the observation vectors obtained from the means 6 is smaller than a preset threshold,
It is determined that the tracking maintenance is good, and the ratio between the number of target signals and the number of unnecessary signals used in the calculation of the gate size by the first gate size calculation means 3 is determined by the second gate size calculation means 13.
Means for diverting the ratio of the number of target signals to the number of unnecessary signals.

FIG. 8 is a flowchart for explaining the operation of the fourth embodiment. In the flowchart shown in FIG. 8, the same portions as those of the steps shown in the flowchart according to Embodiment 1 shown in FIG. 2 are denoted by the same reference numerals, and description thereof will be omitted. As a new step, there is a step ST12.
T12 will be mainly described.

In step ST12, the ratio between the number of target signals and the number of unnecessary signals in the gate used for calculating the gate size of a certain sensor is determined as the ratio between the number of target signals and the number of unnecessary signals used for calculating the gate size of the target sensor. input.

That is, in this step ST12, the ratio diverting means 18 determines that the observation vector exists in the gate based on the output from the first gate judging means 4 and the smoothed vector obtained from the smoothing means 5 first. When it is determined that the difference between the observation vector obtained from the first observation unit 6 and the observation vector is smaller than a preset threshold, it is determined that tracking is good, and the first gate size calculation unit 3 calculates the gate size. The ratio between the number of target signals and the number of unnecessary signals used for
Is diverted to the gate size calculation means 13 of FIG.

Therefore, according to the fourth embodiment, the ratio between the number of target signals and the number of unnecessary signals used for calculating the gate size of a certain sensor with good tracking maintenance is determined by the number of target signals used for calculating the gate size of the target sensor. And the number of unnecessary signals, the gate size of the target sensor becomes an appropriate size, and tracking can be easily maintained.

Embodiment 5 FIG. FIG. 9 is a block diagram showing a configuration of a tracking processing device according to Embodiment 5 of the present invention. 9, the same components as those of the third embodiment shown in FIG. 5 are denoted by the same reference numerals, and the description thereof will be omitted. As a new code, 19 is a ratio between the number of target signals and the number of unnecessary signals such that the ratio of determining the number of target signals and the number of unnecessary signals in the gate is kept low and the number of target signals is emphasized only in the initial time zone. This is an initial time zone gate control means provided to the variable means 9.

That is, the initial time zone gate control means 19
Is transmitted to the ratio varying means 9 for the number of target signals and the number of unnecessary signals.
From the first sampling of tracking start to N sampling,
The ratio between the target signal number in the gate and the unnecessary signal in each sensor is set so that the target signal number is more important than the unnecessary signal number.
The value of the constant C in Equation 15 is given smaller. here,
N is a positive integer of 2 or more.

FIG. 10 is a flowchart for explaining the operation of the fifth embodiment. In the flowchart shown in FIG. 10, the same portions as those of the steps shown in the flowchart according to the third embodiment shown in FIG. 6 are denoted by the same reference numerals, and description thereof will be omitted. As a new step, step ST13 exists. Hereinafter, step ST13 will be mainly described.

In step ST13, from 1 sampling to N sampling, the constant C for determining the ratio between the number of target signals and the number of unnecessary signals in Equation 15 is controlled so that the number of target signals is more important than the number of unnecessary signals. , And outputs a control signal that gives a smaller value of the constant C, and inputs the signal to the substitution of the ratio between the number of target signals and the number of unnecessary signals in step ST10.

In the initial time zone where only a few samplings have elapsed since the initial sampling, if the ratio between the number of target signals and the number of unnecessary signals is given greater importance to the number of unnecessary signals than the number of target signals and the gate is made smaller, The number of unnecessary signals in the gate can be reduced, but in the unstable initial time zone, the setting location of the gate may vary and it may not be possible to set up near the target signal. The effect of making the gate size smaller and changing the gate setting location and not being able to track is greater than increasing the unnecessary signal by increasing
It is preferable to emphasize the number of target signals over the number of unnecessary signals.

Therefore, according to the fifth embodiment,
The ratio between the number of target signals and the number of unnecessary signals is focused on the number of target signals over the number of unnecessary signals only in the initial time period, so that the gate size is increased, whereby stable tracking can be maintained.

Embodiment 6 FIG. FIG. 11 is a block diagram showing a configuration of a tracking processing device according to Embodiment 6 of the present invention. 11, the same parts as those of the second embodiment shown in FIG. 3 are denoted by the same reference numerals, and the description thereof will be omitted. As a new code, 20 is the center of the gate when the signal-to-noise ratio obtained from the sensor of the observation means 6 is larger than a certain threshold, and weights the prediction vector and the observation vector when performing the smooth vector calculation. NN method limiting means for providing the smoothing means 8 with a control signal for limiting only the weighting of the observation vector closest to the predicted observation vector and the predicted vector.

That is, the NN method limiting means 20 gives a control signal for applying the NN method to the smoothing means 8 having a limited number in the gate when the observation accuracy obtained from the observation means 6 is poor.

FIG. 12 is a flowchart for explaining the operation of the sixth embodiment. In the flowchart shown in FIG. 12, the same portions as those of the steps shown in the flowchart according to the second embodiment shown in FIG. 4 are denoted by the same reference numerals, and description thereof will be omitted. As a new step, there is a step ST14. Hereinafter, the step ST14 will be mainly described.

In step ST14, among the observation values including a plurality of target signals and unnecessary signals obtained in the gate, the observation value closest to the center of the gate based on the NN method is determined by the smoothing process. The target control signal is input to the next step ST7b for calculating the smoothed information by limiting the number in the gate.

Therefore, in the sixth embodiment, when the observation accuracy is good and the observed values do not vary, the gate method in the gate is more effective than the AN method in which a plurality of observation signals in the gate are weighted and integrated to calculate a smoothed value. In the NN method using the observation value closest to the center of, the accuracy of the tracking between the true value of the target and the smoothed value, that is, the so-called tracking accuracy, is improved, and stable tracking maintenance can be obtained.

[0136]

As described above, according to the tracking processing apparatus and method according to the present invention, the size of the gate is maintained at such a level that tracking can be maintained without applying a calculation load, and the gate processing is performed under an unnecessary signal environment. Tracking maintenance can be enhanced.

Also, the smoothing means sets a limited number of observation vectors in the gate determined by the gate determination means, and uses the distance between the observation vector in the gate and the predicted observation vector at the center of the gate to perform weighting integration. Since the calculated gain matrix, smoothed vector and smoothed error covariance matrix are calculated, the gate can be appropriately sized even in an unnecessary signal environment or a free space free of unnecessary signals, and the variation of the smoothed vector is suppressed. Tracking is easily maintained, and the calculation load is reduced.

A plurality of ratios between the number of target signals and the number of unnecessary signals in the gate are given, and a plurality of gate sizes are calculated.
The gate is set, and one of the smoothed vectors at each gate is compared with the observation vector at the same time to display one of the smaller residuals, so that the operator who sees the display can easily confirm that the tracking is maintained. Become.

The ratio between the number of target signals and the number of unnecessary signals used for calculating a gate size with good tracking maintenance by a certain sensor is as follows.
By diverting to the ratio between the number of target signals and the number of unnecessary signals used for calculating the gate size of the target sensor, the gate size of the target sensor becomes an appropriate size, and tracking can be easily maintained.

Further, the ratio between the number of target signals and the number of unnecessary signals is
Only in the initial time period, by focusing on the target signal number over the unnecessary signal number, and by increasing the gate size, stable tracking can be maintained.

Further, when the observation accuracy is good and the observation values are not varied, the observation method closest to the center of the gate in the gate is compared with the AN method in which a plurality of observation signals in the gate are weighted and integrated to calculate a smooth value. The NN method using the values improves the residual of the true value of the target and the smoothed value, that is, the so-called tracking accuracy, and achieves stable tracking maintenance.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a tracking processing device according to Embodiment 1 of the present invention.

FIG. 2 is a flowchart illustrating the operation of the first embodiment of the present invention.

FIG. 3 is a block diagram showing a configuration of a tracking processing device according to Embodiment 2 of the present invention.

FIG. 4 is a flowchart illustrating the operation of the second embodiment of the present invention.

FIG. 5 is a block diagram showing a configuration of a tracking processing device according to Embodiment 3 of the present invention.

FIG. 6 is a flowchart illustrating the operation of the third embodiment of the present invention.

FIG. 7 is a block diagram showing a configuration of a tracking processing device according to Embodiment 4 of the present invention.

FIG. 8 is a flowchart illustrating an operation according to the fourth embodiment of the present invention.

FIG. 9 is a block diagram showing a configuration of a tracking processing device according to Embodiment 5 of the present invention.

FIG. 10 is a flowchart illustrating the operation of the fifth embodiment of the present invention.

FIG. 11 is a block diagram showing a configuration of a tracking processing device according to Embodiment 6 of the present invention.

FIG. 12 is a flowchart illustrating the operation of the sixth embodiment of the present invention.

FIG. 13 is a block diagram illustrating a configuration of a tracking processing device according to a conventional example.

FIG. 14 is an explanatory diagram of a gate.

FIG. 15 is an explanatory diagram illustrating a correlation distance by defining a distance normalized by a probability density as a correlation distance.

FIG. 16 is an explanatory diagram showing an example of a gate when the order of an observation vector is two-dimensional.

17 is an explanatory diagram of a coordinate system according to the observation means 6 in FIG.

[Explanation of symbols]

1 prediction means (first prediction means), 2 delay elements (second
, 3 gate size calculation means (first gate size calculation means), 4 gate determination means (first gate determination means), 5 smoothing means (first smoothing means), 6
Observation means (first observation means), 7 display means (first display means), 8 smoothing means in which the number in the gate is limited,
9 Target signal number and unnecessary signal ratio variable means, 10 display determination means, 11 second prediction means, 12 second delay element, 13 second gate size calculation means, 14 second gate determination means, 15th 2 smoothing means, 16 second observation means, 17 second display means, 18 means for diverting the ratio of the number of target signals to the number of unnecessary signals, 19 initial time zone gate control means, 20 means for limiting the NN method.

 ──────────────────────────────────────────────────続 き Continuing from the front page (72) Inventor Shingo Tsujimichi 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Inside Mitsubishi Electric Corporation (72) Inventor Yoshio Kosuge 2-3-2 Marunouchi, Chiyoda-ku, Tokyo F term in Mitsubishi Electric Corporation (reference) 5J070 AC02 AC12 AC13 AH04 AH12 AH19 AH50 AJ03 AK22 AL01 BB02 BB03 BB04 BB06 BB16 BG06

Claims (7)

[Claims] 1. A prediction means for performing prediction according to a target motion model and an observation model based on the theory of a Kalman filter to calculate a prediction vector and a prediction error covariance matrix, and a prediction vector and a prediction error calculated by the prediction means. Considering the delay element that delays the variance matrix by one sampling and all observation vectors in the gate that is the target existence expected area, the dispersion ratio of the prediction error and the observation error set in advance, the target number of signals and the unnecessary signal Maximize the difference between the number of target and unwanted signals in the gate by using a constant that determines the ratio of numbers, the observation error calculated based on the signal-to-noise ratio, the detection probability of the target, and the false alarm probability of the unwanted signal. A gate size calculating means for calculating a gate size to be set, and a gate set by the gate size calculated by the gate size calculating means. Gate determining means for determining whether or not the observation vector exists in the gate based on the distance between the predicted observation vector and the observation vector, which is the center of the gate, and the observation determined to be present in the gate by the gate determining means. Using a vector and a prediction vector obtained by the prediction means, a gain matrix, a smooth vector, and a smooth error covariance matrix are calculated by the distance between the prediction observation vector and the observation vector which are the center of the gate calculated by the gate determination means. A smoothing means for obtaining by integrating weights, a sensor, a detection probability of a target calculated based on a signal-to-noise ratio from the sensor, a false alarm probability of an unnecessary signal,
A tracking processing device comprising: an observation unit that inputs an observation error and an observation vector to the gate determination unit; and a display unit that displays a smooth position from a smooth vector that is a current estimated value of a target calculated by the smoothing unit. . 2. The tracking processing device according to claim 1, wherein the smoothing means sets a limited number of observation vectors in the gate determined by the gate determination means, and sets the number of observation vectors in the gate and the center of the gate. A tracking processing device, which calculates a weighted integrated gain matrix, smoothed vector, and smoothed error covariance matrix using a distance from a predicted observation vector. 3. The tracking processing device according to claim 1, wherein said prediction means, said delay means, said gate size calculation means, said gate determination means and said smoothing means are first prediction means, first delay Means, a first gate size calculating means, a first gate determining means, and a first smoothing means, whereas a second predicting means, a second delaying means, and a second gate having the same functions as those described above. In addition to a size calculating means, a second gate determining means and a second smoothing means, the ratio of the number of target signals in the gate to the number of unnecessary signals is provided for each of the plurality of smoothing means, and the first gate size calculation is performed. Means for changing the ratio between the number of target signals and the number of unnecessary signals in the gate to the second gate size calculating means and the ratio of the number of target signals to the number of unnecessary signals; the first smoothing means; From smoothing means 2 Either the first smoothing means or the second smoothing means determines the smaller one of the residuals of the observation vectors at the same time as the position components of the plurality of smoothed vectors calculated as the smoothed position of the tracking target. A tracking processing device, further comprising: display determination means for determining whether one output result is to be displayed by the display means. 4. The tracking processing device according to claim 1, wherein said prediction means, said delay means, said gate size calculation means, said gate determination means, said smoothing means, said observation means and said display means are provided with a first Prediction means, first delay means, first gate size calculation means, first gate determination means, first smoothing means, first observation means, and first display means. A second prediction unit, a second delay unit, a second gate size calculation unit, a second gate determination unit, a second smoothing unit, a second observation unit, and a second display unit having the same function. The observation vector is present in the gate based on the output from the first gate determination unit, and the observation vector obtained from the first observation unit and the observation vector obtained from the first observation unit are determined. The difference If the ratio is determined to be smaller than the set threshold value, the ratio between the number of target signals and the number of unnecessary signals used in the calculation of the gate size by the first gate size calculation means is diverted to the second gate size calculation means. A tracking processing device further comprising means for diverting the ratio between the number of target signals and the number of unnecessary signals to be processed. 5. The tracking processing device according to claim 3, wherein the control signal is controlled such that the ratio of determining the number of target signals in the gate and the number of unnecessary signals is kept low only in the initial time zone and the number of target signals is emphasized. A tracking processing apparatus further comprising an initial time zone gate control means for giving to the means for varying the ratio between the number of target signals and the number of unnecessary signals. 6. The tracking processing device according to claim 2, wherein when a signal-to-noise ratio obtained from a sensor of the observation means is larger than a certain threshold value, a prediction vector and an observation vector for performing a smooth vector calculation are calculated. A tracking processing device, further comprising a limiting unit that gives the smoothing unit a control signal that limits the weighting to only the observation vector closest to the predicted observation vector at the center of the gate and the prediction vector weight. (A) performing prediction according to a target motion model and an observation model based on Kalman filter theory, and calculating and outputting prediction information including a prediction vector and a prediction error covariance matrix; and (b) a sensor. Inputting signal-to-noise ratio information obtained from (c), and (c) observation including target detection probability, unnecessary signal false alarm probability, observation error, and observation vector based on the signal-to-noise ratio information input. Calculating information; (d) inputting a variance ratio between a prediction error and an observation error set in advance; and (e) target observation information calculated from the signal-to-noise ratio and a prediction set in advance. The difference between the number of unnecessary signals and the number of target signals is maximized as the gate size that determines the size of the gate, which is the target existence expected area, using the variance ratio between the error and the observation error. (F) determining an observation vector considered as a target signal from a set of observation vectors including an unnecessary signal and a target signal based on the calculated gate size; Using the observation vector determined as the target signal of the target signal, the prediction vector and the prediction error, the gain matrix obtained by weighting and integrating using the distance between the observation vector in the gate and the position component of the prediction vector that is the center of the gate, Calculating smoothing information including a weighted integrated smoothed vector and a weighted integrated smoothing error covariance matrix; and (h) displaying a smoothed position from the weighted integrated smoothed vector that is the current estimated value of the target. (I) When the processing from the above step (a) to the step (h) is continued, the step (h) Tracking processing method having Tsu via a delay element after a lapse of up and returning the prediction information calculation process step of (a).