JP2003149328A - Target correlation device of radar - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the problem that a process cannot be carried out for all targets under multiple targets and an unneeded signal environment as a conventional system has only one kind of a target correlation process. SOLUTION: This device comprises a target track information storage part 6 sequentially storing a current position of a moving target generated based on observational data input from a radar 2, a Nearest Neighbor Data Association correlation processor 4, a method regarding a target candidate nearest to an estimated position of the moving target among target candidates existing in a software gate as the current position, an All Neighbor Data Association correlation processor 5, a method calculating the current position of the moving target using all the target candidates existing in the software gate, and a correlation process selection part 3 selecting either of the processors to be used for the correlation process.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a target correlation device for generating target track information from observation data input from a radar.

[0002]

2. Description of the Related Art Conventionally, a target correlation processing apparatus that estimates the current position of a moving target from radar observation data and sequentially generates a track of the moving target has performed only one type of correlation processing in one system. . For correlation processing, Ne
arest Neighbor Data Association Correlation processing and All Neighbor Data Association
The neighbor data association correlation process is known.

FIG. 10 shows the configuration of a conventional target correlator using the Nearest Neighbor Data Association correlation process.

The radar 2 obtains observation data by transmitting and receiving radio waves to and from the moving target 1. Neares to the detection data with a predetermined signal strength in the observation data
The t Neighbor Data Association correlation processing unit 4 sets a software gate and sets detection data existing in the software gate as a target candidate. Among the plurality of target candidates thus obtained, the target candidate closest to the predicted position of the moving target is set as the current position of the moving target. The target track information storage unit 6 uses the Nearest Neighbor Data Ass
By sequentially registering and updating the current position of the moving target generated by the ociation correlation processing unit 4, the track of the moving target 1 is generated and managed.

FIG. 11 is a diagram showing moving targets and target candidates. The Nearest Neighbor Data Association correlation processing unit 4 calculates the current predicted position 8 based on the past position 7 of the moving target T1 in the past. Also, Nearest Neig
The hbor Data Association correlation processing unit 4 determines the movement target T1.
The software gate 9 centering on the current predicted position 8 of is calculated. The distances from the current predicted position 8 to the target candidates 10 and 11 existing in the software gate 9 are obtained, and the position of the target candidate 10 closest to the current predicted position 8 is estimated as the current position of the moving target T1.

FIG. 12 shows the Nearest Neighbor Data Associ
ation is a diagram showing a specific processing flow in the correlation processing unit 4.

The position of the moving target T1 in the past,
Prediction processing is performed based on the speed and the observation time to obtain the current predicted position 8 (S1), the software gate 9 centering on the current predicted position 8 is calculated (S2), and the detection data is stored in the software gate 9. Determine if it exists (S
3) The target candidate existing in the software gate 9 and the target candidate having the shortest distance between the current predicted predicted position 8 are obtained (correlated) (S4), and the position of the closest target candidate 10 is moved to the current position of the moving target Is updated / registered in the target track information storage unit 6 (S5).

As described above, Nearest Neighbor Data
In the correlation processing by the Association correlation processing unit 4, the distance between the current predicted position 8 of the moving target T1 and the target candidates 10 and 11 in the software gate 9 is calculated, and the respective distances are compared to select the smallest one. Therefore, there is an advantage that the amount of calculation is relatively small and therefore the computer resources are small.

On the other hand, when an unnecessary signal exists in the observation data and a reflected signal such as clutter is included as a target candidate, and the signal exists near the current predicted position, the correlation processing is correctly performed. There was a problem that it was difficult to do. In addition, when there are a plurality of moving targets and the moving targets approach and intersect, it is difficult to correctly perform the correlation processing.

Therefore, conventionally, as a method for solving the above-mentioned problem, a plurality of hypotheses that define which target candidate corresponds to a moving target are generated, and movement is performed based on the reliability obtained for each hypothesis. A method of estimating the current position of the target is used, and this method is based on the All Neighbor Data Association.
It is called correlation processing. In this method, the current position of the moving target is stochastically calculated based on the positions of the target candidates.
The accuracy of tracking the moving target is improved as compared with the est Neighbor Data Association process.

[0011]

However, as the number of moving targets or the number of target candidates in the software gate increases, the number of possible associations, that is, the number of hypotheses increases exponentially, and For example, a large-scale computer resource is required to perform the correlation processing calculation within the sampling time of observation data.

The present invention has been made in order to solve the above problems, and a target of a radar capable of performing an appropriate target correlation processing according to the computer resources of the correlation processing device or the situation of observation data. It is to provide a correlator.

[0013]

In a radar correlation processing apparatus according to the present invention, a current position of a moving target is selected from a plurality of target candidates based on observation data for at least one moving target obtained from the radar. In a radar correlation processing device for estimating, a target candidate closest to the current predicted position is selected from the plurality of target candidates with respect to the current predicted position of the moving target calculated based on the past position of the moving target. A plurality of hypotheses in which the moving target and the target candidate are associated with each other, and the current position of the moving target is calculated based on the reliability calculated for each hypothesis. Second correlation processing means for estimating, target track storage means for sequentially storing the estimated position of the moving target, and first correlation processing means and second correlation processing means depending on the situation. Among shall and a selection means for selecting one of the unit either.

In the radar correlation processing apparatus according to the present invention, the selecting means compares the number of moving targets stored in the target track storing means with a predetermined threshold value, If it is larger, the first correlation processing means is selected, and if it is smaller than the threshold value, the second correlation processing means is selected.

Further, in the radar correlation processing apparatus according to the present invention, the selecting means compares the number of moving targets stored in the target track storing means with a predetermined threshold value, If it is larger, the first correlation processing means is selected, and if it is smaller than the threshold value, the second correlation processing means is selected.

Further, in the radar correlation processing apparatus according to the present invention, the selecting means refers to preset information indicating whether to select the first correlation processing means or the second correlation processing means. Either one of the first correlation processing means and the second correlation processing means is selected based on the pre-article setting information.

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiment 1. FIG. 1 is a diagram showing the overall configuration of a correlation processing device according to the present invention. Hereinafter, in this embodiment, the same components as those in the conventional example are designated by the same reference numerals.

The radar 2 obtains observation data for the moving target 1 by transmitting and receiving radio waves to and from the moving target 1. The correlation processing selection unit 3 (selection means) is a Nearest Neighbor Data Association correlation processing unit 4 depending on the situation.
(First correlation processing means), or AllNeighbor Data Associa
tion correlation processing unit 5 (second correlation processing means) is selected. Ne
The arest Neighbor Data Association correlation processing unit 4 sets the target candidate closest to the predicted position of the moving target among the target candidates existing in the software gate as the current position of the moving target. All Neighbor Data Association correlation processing unit 5
Generates a plurality of hypotheses that define which target candidate corresponds to the moving target 1, and estimates the current position of the moving target from the reliability obtained for each hypothesis. The target track information storage unit 6 (target track storage means) is a Nearest Neighbor.
Data Association Correlation Processing Unit 4 or All Neighbor
By sequentially registering and updating the current position of the moving target estimated by the Data Association correlation processing unit 5, a track of the moving target 1 is generated and managed.

First, the All Neighbor Data Association correlation process will be described.

FIG. 2 is a diagram showing the relationship between the position of a moving target in the All Neighbor Data Association correlation processing and the positions of a plurality of target candidates in the software gate calculated around the predicted current position of the moving target. . Figure 3
It is a flowchart of the correlation processing of the All Neighbor Data Associatio correlation processing unit 5.

Hereinafter, according to the flowchart of FIG. 3, All
Neighbor Data Association correlation processing will be described. The symbols appearing in the description correspond to the symbols in FIG.

Based on the past position 7 of the moving target T1, the current predicted position 8 of the moving target T1 is calculated based on the following equation (S6).

[0023]

[Equation 1]

In the above equation, X_k-1(+) Is the movement target
A position vector representing the past position 7 of T1 and X
_k(-) Is a position vector indicating the current predicted position 8 of the moving target T1.
It's Koutor. φ_k-1Is the movement of the moving target, eg
Moving target determined by assuming
It is a transition matrix at the position of T1, expressed as a matrix with 3 rows and 3 columns.
Be done. The subscript k is the current time t._kIn response to
And k-1 is the past time t _k-1, For example
It corresponds to the previous time only by the interval.

The prediction error covariance matrix P _k (-) at the current time is calculated for the current predicted position 8 obtained by "Equation 1" based on the following equation (S7).

[0026]

[Equation 2]

In the above equation, P _k (-) is the current time t
the prediction error covariance matrix at _k , P
_k−1 (+) is the prediction error covariance matrix at the past time t _k−1 . Further, Ω _k−1 is the past time t
It is a driving noise matrix from _k-1 to the current time t _k , and Γ _k-1 is a conversion matrix for the driving noise matrix Ω _{k-1 at the} past time t _k-1 .

Next, the chi-square test value χ ² is calculated from the prediction error covariance matrix P _k (-) obtained by "Equation 2".

[0029]

[Equation 3]

In the above equation, Z _k (-) is the current time t
It is a vector representing the position of detection data, which is data having a predetermined signal strength in the observation data at _k . When there are multiple detection data, chi-square test is performed for each detection data. For each detection data, it is determined by "Equation 4" whether the chi-square test value χ ² exceeds the threshold value D1 (S8).

[0031]

[Equation 4]

When it is less than the threshold value, it is determined that there is no correlation as a correlation result, and when it is more than the threshold value, it is determined that there is correlation as a correlation result, and the detection data determined to have correlation is set as a target candidate. That is, the range of the software gate 9 is determined by the threshold value judgment represented by "Equation 4".

Next, a hypothesis that associates the obtained target candidate with the moving target is generated.

An example of the generated hypothesis is shown in FIG. The hypothesis reliability for each hypothesis shown in FIG. 4 is calculated by "Equation 5" (S10).

[0035]

[Equation 5]

Here, i is an index representing a hypothesis,
The total number of hypotheses is represented by Ik. β _{k, i} is the reliability of the hypothesis i at time t _{k ,} γ _{k, i} is the hypothesis i at time t _k
Represents the probability that

Next, "Equation 6" which is a weighted average by hypothesis reliability

[Equation 6] The result calculated by is output to the target track information storage unit 6 as the current position of the moving target (S11).

Here, i is an index representing the type of hypothesis
, The total number of hypotheses is m_kIt is represented by. β _{k, i}Is time t_kTo
Reliability of hypothesis i in_,X_{k, i}(+) Indicates time t
_k , Goals associated with moving goals in hypothesis i
It is a complementary position vector. In the example of FIG. 4, in hypothesis 1,
Since there is no target candidate corresponding to the moving target, X
_{k, 1}The value of (+) is 0. In hypothesis 2, move
Since the target candidate corresponding to the target is the target candidate 10, X
_{k, 2}(+) Is Z representing the position vector of the target candidate 10₁
(-). In hypothesis 3, it corresponds to the moving target.
Since the target candidate is the target candidate 11, X_{k, 3}(+) Is
Z representing the position vector of the target candidate 11_Two(-). Well
, Β_{k, 1,} β_{k, 2} , Β_{k, 3,}Each value
Are 0.14, 0.14 and 0.72 respectively according to FIG.
Is.

[0039] In this way, the All Neighbor Data Associatio
The n correlation processing unit 5 can use all target candidates existing in the software gate 9. Further, since it is possible to generate a hypothesis that the target candidate is an unnecessary signal, it is possible to perform more accurate correlation processing when there are relatively many moving targets or in an environment where there are many unnecessary signals. On the other hand, as described above, a large-scale computer resource is required as the number of moving targets or the number of target candidates in the software gate increases.

The current position of the moving target obtained by the above method does not always match the position of the target candidate. The following method may be adopted as a method for matching the current position of the moving target with the position of any of the plurality of target candidates. First,
In the hypothesis having the highest reliability among the plurality of hypotheses, the position of the target candidate associated with the moving target is set as the current position of the moving target. In this way, the current position of the moving target matches the position of a certain target candidate.

Next, Nearest Neighbor Deta Associat
The ion correlation processing will be described. The drawings used for explanation are
11 and 12 used in the description of the conventional example are used.

FIG. 11 shows the Nearest Ne according to the present embodiment.
FIG. 6 is a diagram showing a relationship between a position of a moving target and a position of a plurality of target candidates in a software gate centering on a predicted position of the target in the correlation processing of the corner data association correlation processing unit 4.

The Nearest Neighbor Data Association correlation processing unit 4 calculates the current predicted position 8 based on the position 7 of the moving target T1 in the past. Also, Nearest Ne
The ighbor Data Association correlation processing unit 4 determines the movement target T
Software gate 9 centered on the current predicted position 8 of 1
To calculate. The calculation for obtaining the current predicted position 8 and the software gate 9 is performed by the above All Neighbor Data Associ
ation It is obtained by the same method as the correlation processing (“Equation 1” to “Equation 4”). The distances from the current predicted position 8 to the target candidates 10 and 11 existing in the software gate 9 are obtained, and the position of the target candidate 10 closest to the current predicted position 8 is estimated as the current position of the moving target T1.

FIG. 12 shows the Nearest Ne according to the present embodiment.
FIG. 7 is a diagram showing a specific processing flow in the ighbor Data Association correlation processing unit 4.

The position of the moving target T1 in the past,
Prediction processing is performed based on the speed and the observation time to obtain the current predicted position 8 (S1), the software gate 9 centering on the current predicted position 8 is calculated (S2), and the target candidate is stored in the software gate 9. It is determined whether or not it exists (S3),
A target candidate having the shortest distance between the target candidate existing in the software gate 9 and the current predicted prediction position 8 is obtained (corresponding) (S4), and the position of the closest target candidate 10 is set as the current position of the moving target. It is updated / registered in the track information storage unit 6 (S5).

As described above, Nearest Neighbor Data
The correlation processing by the Association correlation processing unit 4 calculates the distance between the current predicted position 8 of the moving target T1 and the target candidates 10 and 11 in the software gate, compares the respective distances, and selects the smallest one. Since this is a method, there is an advantage that the amount of calculation is relatively small and therefore the computer resources are small. On the other hand, if an unnecessary signal exists in the observation data and a reflected signal such as clutter is included as a target candidate, and the unnecessary signal exists near the predicted position, highly reliable correlation processing is performed. Is difficult.
Further, even when there are a plurality of moving targets and the moving targets approach and intersect, it is difficult to perform correct correlation processing.

Therefore, in the correlation processing apparatus according to this embodiment, the correlation processing selection unit 3 causes the Nearest N according to the situation.
eighbour Deta Association correlation processing unit 4 and All Neighbor
The borData Association correlation processing unit 5 is switched to perform the correlation processing.

The first selection method of the correlation processing performed by the correlation processing selection section 5 will be described with reference to FIG. All Neig
In the case of the hbor Data Association correlation processing unit 5, the calculation amount of the correspondence between the target candidate and the moving target (hypothesis reliability calculation) and the moving target update processing is Nearest Neighbor Data Associ
ation Larger than the correlation processing unit 4. Therefore, as the number of target candidates increases, the number of hypotheses, and eventually the processing load, increases. On the other hand, if the total number of target candidates is not so large, the processing can be performed on all the target candidates in the software gate without running out of computer resources.

As shown in FIG. 5, the total number of target candidates obtained from the observation data input from the radar at a certain time is compared with a predetermined threshold value (S12). If the total number of target candidates is larger than the threshold value, the correlation processing selection unit 3 determines that computer resources sufficient to process all target candidates cannot be secured, and switches the processing to the Nearest Neighbor Data Association correlation processing unit 4. Conversely, if the total number of target candidates is less than the threshold value, the processing is switched to the All Neighbor Data Association correlation processing unit 5.

Next, the second method of selecting the correlation processing performed by the correlation processing selecting section 5 will be described with reference to FIG. Al
l In the case of the Neighbor Data Association correlation processing unit 5, the calculation amount of the process of associating the target candidate with the moving target (hypothesis reliability calculation) and updating the position of the moving target is the Nearest Neighbor
Since the data association correlation processing unit 4 is larger than the Data Association correlation processing unit 4, the number of hypotheses, and thus the processing load, increases as the number of moving targets increases, as in the above case. On the contrary, if the total number of moving targets is not so large, processing can be performed on all moving targets without running out of computer resources.

As shown in FIG. 6, the total number of moving targets input from the radar at a certain time is compared with a predetermined threshold value (S13). If the total number of moving targets is greater than the threshold value, the correlation processing selection unit 3 determines that computer resources sufficient to process the moving targets cannot be secured, and the Nearest Neighbor
or Data Association The processing is switched to the correlation processing unit 4. Conversely, if the total number of moving targets is less than the threshold value, then All Ne
The processing is switched to the ighbor Data Association correlation processing unit 5.

Next, the third selection method of the correlation processing performed by the correlation processing selection section 5 will be described with reference to FIG.
Although the All Neighbor Data Association correlation processing unit 5 can perform highly reliable correlation processing, it requires a large amount of computer resources in an environment where the total number of target candidates in the software gate is large.
Although the data association correlation processing unit 4 is slightly less reliable than the All Neighbor Data Association correlation processing, the data association correlation processing unit 4 can perform the correlation processing for all movement targets without preparing large-scale computer resources.

Therefore, when the reliability of the correlation processing is required, when a large-scale computer resource can be realized, and when the number of moving targets or the total number of expected target candidates is not so large, the All Neighbor Data The Association correlation processing unit 5 is selected, and on the contrary, the Nearest Neighbor Data Association correlation processing unit 4 is selected for a system that does not require reliability but requests processing to be performed on all moving targets. Correlation processing must be performed according to the status of the system used. As shown in FIG. 7, the correlation processing unit 5 refers to the pre-setting information set or for each system in advance using either correlation processing (S14),
For example, when the system is started up, an appropriate correlation process is selected by referring to the information.

Next, the fourth selection method of the correlation processing performed by the correlation processing selection section 5 will be described with reference to FIG.
This method will be described first using the case of FIG. 9 as an example. In FIG. 9, the moving target T3 and the moving target T4 are past positions, for example, the position vectors 13 and 14 at the sampling time immediately before the current time, and the moving velocities of the moving target T3 and the moving target T4 at the past time. The determination as to whether the moving target T3 and the moving target T4 are close to each other is obtained from the vectors 15 and 16 by the inequality of "Equation 7".

[0055]

[Equation 7]

Here, Δt is a minute time from the current time, and "Equation 8" is established.

[0057]

[Equation 8]

"Equation 9" is derived from "Equation 7" and "Equation 8".

[0059]

[Equation 9]

[0060]

[Equation 10]

The case where "Equation 10" is satisfied is a case where the moving target is close. In FIG. 8, first, it is determined whether "Equation 9" is satisfied (S15), and if so, it is determined whether "Equation 10" is satisfied (S16).
When both of the numbers 10 "are satisfied, the moving target is close to each other, and there is a high possibility that the tracks of the moving target will approach and intersect further in the future, and therefore the correlation of the moving targets cannot be correctly determined. Therefore, in such a case, processing is performed by the highly reliable All Neighbor Data Association correlation processing unit 5, and if not, Ne
The arest NeighborData Association correlation processing unit 4 performs processing.

[0062]

As described above, in the target correlator according to the first aspect of the invention, the selecting means selects the first correlation processing means and the second correlation processing means according to the situation to perform the correlation processing. However, there is an effect that highly reliable correlation processing can be performed in a small target environment, and target correlation processing can be performed for a moving target even in a multi-target / unnecessary signal environment.

As described above, the target correlation apparatus according to the second aspect of the present invention prevents insufficient computer resources by causing the first correlation processing means to perform the correlation processing when the number of target candidates is larger than the threshold value. It has the effect of enabling it.

As described above, the target correlation apparatus according to the third aspect of the present invention makes it possible to prevent the shortage of computer resources by causing the first correlation processing means to execute the calculation when the number of moving targets is larger than the threshold value. There is an effect.

[0065] As described above, the target correlation device according to the fourth invention, by setting whether to advance using either correlation process for each system, one device applicable correlation processing in any system There is an effect that can be provided in.

As described above, the target correlation apparatus according to the fifth aspect of the invention selects the second correlation processing means when the distance between moving targets is small and the tracks of the moving targets are approaching or intersecting. Correlation processing is performed by the above, and if not, the first correlation processing means is selected and the correlation processing is performed. Normally, the first correlation processing means performs the correlation processing, and for the target candidates that are difficult to determine among the target candidates, the second correlation processing means performs the correlation processing to reduce the computer load. Has the effect of

[Brief description of drawings]

FIG. 1 is a configuration diagram of a correlation processing device according to an embodiment of the present invention.

FIG. 2 All Neighbor D in the embodiment of the present invention
7 is an example of a moving target and a target track for explaining the processing in the ata Association correlation processing unit 5.

FIG. 3 All Neighbor D in the embodiment of the present invention
It is a flowchart of an ata Association correlation process.

FIG. 4 All Neighbor Data Associat in the present invention
It is an example of a hypothesis generated in the ion correlation processing unit 5.

FIG. 5 is a flowchart of a first method of the correlation processing selection unit 3 in the present invention.

FIG. 6 is a flowchart of a second method of the correlation processing selection unit 3 in the present invention.

FIG. 7 is a flowchart of a third method of the correlation processing selection unit 3 in the present invention.

FIG. 8 is a flowchart of a fourth method of the correlation processing selection unit 3 in the present invention.

FIG. 9 is a diagram for explaining a fourth method of the correlation processing selection unit 3 in the present invention.

FIG. 10 is a configuration diagram of a conventional device.

FIG. 11 shows a conventional device and an embodiment of the present invention.
It is a figure showing the movement target and a target candidate for demonstrating the process in Nearest Neighbor Data Association correlation process part 4.

FIG. 12 shows a conventional device and an embodiment of the present invention.
It is a flowchart of a Nearest Neighbor Data Association correlation process.

[Explanation of symbols]

1 moving target, 2 radar, 3 correlation processing unit switching unit, 4
Nearest Neighbor Data Association Correlation processing unit, 5
All Neighbor Data Association correlation processing unit, 6 target track information storage unit, 7 past position of moving target T1, 8
Current predicted position of moving target T1, 9 software gates, 10, 11 target candidate, 12 current position of moving target generated in association with target candidate 10, 13 position vector of moving target T2, 14 position of moving target T3 Vector, 15 velocity target T2 velocity vector, 16 velocity target T3 velocity vector.

Claims (5)

[Claims] 1. A radar correlation processing apparatus for estimating the current position of a moving target from a plurality of target candidates based on observation data for at least one moving target obtained from a radar, based on the past position of the moving target. A first correlation processing unit that sets a target candidate closest to the current predicted position among the plurality of target candidates to the current predicted position of the moving target calculated as the current position of the moving target; Second correlation processing means for generating a plurality of hypotheses associated with candidates and estimating the current position of the moving target based on the reliability calculated for each hypothesis; Target track storage means for sequentially storing positions, and selection means for selecting one of the first correlation processing means and the second correlation processing means according to the situation Target correlator radar characterized by and. 2. The selecting means compares the number of target candidates with a predetermined threshold value, selects the first correlation processing means if it is larger than the threshold value, and selects the second correlation processing means if it is smaller than the threshold value.
2. The radar target correlation device according to claim 1, wherein the correlation processing means is selected. 3. The selecting means compares the number of moving targets stored in the target track storing means with a predetermined threshold value, and if the number is larger than the threshold value, selects the first correlation processing means. 2. The radar target correlation apparatus according to claim 1, wherein the second correlation processing means is selected if the threshold correlation value is smaller than the threshold value. 4. The selection means refers to preset information for selecting the first correlation processing means or the second correlation processing means, and the first correlation processing means based on the preset information. 2. The radar target correlation device according to claim 1, wherein either one of the second correlation processing means and the second correlation processing means is selected. 5. When there are a plurality of moving targets, the selecting means, based on the distance between the moving targets and the speed of the moving targets,
The second correlation processing means is selected when the distance between the moving targets is small and the tracks are approaching or intersecting, and the first correlation processing means is selected otherwise. 1. The target correlation device of the radar according to 1.