JP2001228246A - Apparatus and method for tracking target - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve the possibility that a process is not finished before a next sampling time if there are many track correlation matrices and hypotheses when a new hypothesis is to be generated at the next sampling time. SOLUTION: Clusters are separated by a previous data rejection judgment cluster-separating part 13. A size of in-gate judgment matrices generated by an in-gate judgment matrix-calculating part 14 corresponding to each of the clusters is reduced. The number of track correlation matrices generated by a track correlation matrix-calculating part 15 from the in-gate judgment matrices is reduced. The hypothesis is updated by a hypothesis-updating part 16 on the basis of the track correlation matrices.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention performs tracking of a target based on observation data indicating the position of the target obtained by observation means having a sensor such as a radar system, and estimates the wakes of a plurality of targets. And a target tracking method.

[0002]

2. Description of the Related Art In a target tracking process for estimating a trajectory as a target trajectory, first, observation data obtained from a sensor is determined as to whether it is a target signal or an unnecessary signal. When it is determined that the target is the target, it is determined whether the observation data is for any target that is already being tracked or for a newly detected target.

In determining whether or not observation data is a target signal, a spatial area centered on a position (predicted value) at a corresponding time from a speed of an existing target is used as a gate, and whether or not observation data exists in the gate is determined. It is determined whether or not. The size of the gate is normally set so that the probability that observation data from the tracking target is observed in the gate has a high probability of 0.9 or more.

It can be said that observation data existing in a gate is likely to be from an existing target, and observation data not entering any gate is likely to be from an unnecessary signal or a new target.

[0005] If the observation data is for any of the targets that are already being tracked, the predicted value calculated from the existing trajectory for that target is used to calculate the target at the time when the observation data is obtained. Estimate true position and speed.

FIG. 18 shows an example of a relationship between a wake and observation data. In FIG. 18, “30” at the current time (t = 3),
Four observation data of “31”, “32” and “33” are obtained. The observation data is represented by the obtained time and the number at that time. That is, for example, observation data “30” represents the 0th observation data obtained at time t = 3.

[0007] The observation data "3
“0” and “31” exist, the observation data “32” exists at the gate of the wake of the target 1, and the observation data “33” does not enter any of the gates.

At this time, as a method of assigning each observation data to the existing track, for example, observation data “30” is to target 0, observation data “31” is to an unnecessary signal, observation data “32” is to target 1, It is conceivable that the observation data “33” is assigned to a new target. Observation data “30” is converted to an unnecessary signal, and observation data “31” is set to target 0.
To observation data "32" and to observation data "3"
It is conceivable that "3" is assigned to an unnecessary signal.

However, if the assignment is determined to be one and the tracking is continued, the possibility that the target tracking will fail if the error is incorrect increases.

Therefore, a method of allocating each observation data to an existing track, that is, a plurality of hypotheses, which are combinations of tracks after allocation that do not contradict each other, coexist, and the relation between the observation data and the target is determined for each hypothesis. By selecting the track based on the best hypothesis, accurate target tracking becomes possible. The present applicant has previously proposed a target tracking device described in Japanese Patent Application Laid-Open No. 8-271617, assuming that the target tracking is performed in this manner.

FIG. 19 is a block diagram showing a conventional target tracking device described in Japanese Patent Application Laid-Open No. 8-271617.
In FIG. 19, 101 is a target tracking device, 2 is a target observation device having a sensor such as a radar system and acquiring observation data, and 3 is a target display device displaying a track of a target being tracked. .

In the target tracking device 101, the observation data 111 existing in the possible area of any track in each cluster is regarded as intra-cluster observation data of the existing cluster, and does not correspond to any track in each cluster. An observation data selection unit 112 uses observation data as independent observation data, and creates a new cluster including the independent observation data. If any observation data is intra-cluster observation data of a plurality of clusters, the Cluster integration / integration unit that integrates the clusters.

A gate 113 indicates whether or not the observation data in each cluster may be an unnecessary signal, whether or not it corresponds to an existing track, and whether or not there is a possibility that it is a new target. An in-gate judgment matrix calculation unit that calculates an inside judgment matrix for each cluster, 114 is one or a combination indicating an unnecessary signal, a signal corresponding to an existing wake, and a combination for allocating each observation data to any one of a new target. A track correlation matrix calculation unit that calculates a plurality of track correlation matrices based on the in-gate determination matrix, and a hypothesis update unit 115 that updates a hypothesis, which is a combination of tracks that do not contradict each other, based on the track correlation matrix.

Reference numeral 116 denotes an intra-system cluster table 121, an intra-cluster observation data table 122, an intra-cluster judgment matrix table 123, an intra-cluster wake correlation matrix table 124, an intra-cluster hypothesis table 125, a hypothesis intra-track table 126, and an intra-cluster trajectory. / Observation data table 127.

Reference numeral 117 denotes a gate calculation unit for calculating an area where observation data can exist for each existing track included in each cluster. Reference numeral 118 denotes a track determination unit for determining a track based on one of hypotheses. Is a hypothesis reduction unit that reduces the number of hypotheses, and 120 is a cluster separation unit that separates clusters based on the degree of sharing of observation data between tracks when reducing the number of hypotheses.

Next, the operation will be described. FIG. 20 is a flowchart for explaining the operation of the conventional target tracking device. First, in step ST101, the observation data selection unit 111 acquires observation data at that time obtained by a sensor or the like in the target observation device 2.

Next, in step ST 102, the observation data selection unit 111 calculates the gate based on the gate for each track of each cluster calculated by the gate calculation unit 117.
It is determined whether the acquired observation data belongs to any cluster or does not belong to any cluster.

FIG. 21 shows an example of an existing track.
When the existing track until time t = 2 is as shown in FIG. 21, a gate as shown in FIG. 18 is set for each cluster, and based on whether each observation data exists in those gates. Thus, the observation data is sorted into clusters.

In step ST103, if there is observation data that does not belong to any cluster, the cluster new / integration unit 112 newly creates a cluster for the observation data, and any of the observation data is If the data is intra-cluster observation data of a plurality of clusters, those clusters are integrated.

Further, the cluster new / integration unit 112
A wake is created based on the positional relationship between the gate and observation data. The following three types of wakes are created. (1) Updated track: A track that can be obtained by adding observation data that has entered the gate to an existing track. (2) New track: A track starting from the observation data entered at that time. (3) Memory track track: A track indicating that "there was no corresponding observation data at the relevant time" with respect to the existing track.

Here, the operation of the cluster new / integration section 112 will be described in detail. FIG. 22 is a diagram illustrating an example of updating a track.

In a certain cluster, as shown in FIG. 21, observation data "00" at time t = 0, observation data "10" and "11" at time t = 1, and observation data "20" at time t = 2. , "21", the following six possible trajectories can be expressed by a set of observation data (one observation data in the case of a new trajectory). T1: "10"-"20" T2: "10"-"21" T3: "11"-"20" T4: "11"-"21" T5: "20" T6: "21"

In this case, a hypothesis that is a combination of wakes is expressed as “ID: {set of wakes} reliability”.
It looks like this: H1: ｛T1, T4｝ rel1 H2: ｛T1, T6｝ rel2 H3: ｛T2, T5｝ rel3 H4: ｛T3, T6｝ rel4 H5: ｛T4, T5｝ rel5 where rel is the reliability of the hypothesis Hi. Represent.

In this state, as shown in FIG.
= 3, the observation data “30”, “31”, “3”
2 was obtained, and the observation data "30" and "31"
When the observation track “32” exists in the gates of the wakes T1, T3, and T5 and the observation data “32” exists in the gates of the wakes T2, T4, and T6, T11: “10” — “20” — “30” T12: "10"-"20"-"31" T21: "10"-"21"-"32" T31: "11"-"20"-"30" T32: "11"-"20"- "31" T41: "11"-"21"-"32" T51: "20"-"30" T52: "20"-"31" T61: "21"-"32" as a new track , T71: "30" T72: "31" T73: "32" are obtained, and as the track of the memory track, T81: "10"-"20" T82: "10"-"21" T83: "11"-" 20 "T84:" 11 "-" 21 "T85:" 20 "T86: 21 "is obtained.

Next, in step ST104, the intra-gate judgment matrix calculation section 113 determines whether or not each of the intra-cluster observation data may be an unnecessary signal, and whether or not the data corresponds to an existing wake. An in-gate decision matrix indicating whether there is a possibility of being a target is calculated for each cluster. FIG. 23 is a diagram illustrating an example of the in-gate decision matrix.

Each row of the in-gate decision matrix corresponds to observation data, each column corresponds to an existing wake, and its element value is 1 or 0. However, the first column indicates whether there is a possibility of being an unnecessary signal, and the columns corresponding to the number of tracking targets (that is, the number of wakes) from the second column indicate whether or not they correspond to each existing wake. And the remaining columns indicate whether it is likely to be a new goal.

In the above example, when the observation data "30", "31", and "32" are obtained as shown in FIG. 18, the first line is the observation data "30" as shown in FIG.
The second row corresponds to observation data “31”, and the third row corresponds to observation data “32”.

The first column indicates unnecessary signals, and the second to seventh columns indicate whether there is a possibility that the signals correspond to the existing tracks T1 to T6, respectively.
The remaining three columns indicate whether there is a possibility of a new track.

First, since all of the observation data may be unnecessary signals, the elements in the first column are all 1.
Regarding the second column, the observation data corresponding to the existing track T1 are “30” and “31”, so that the elements in the first and second rows are “1” and the elements in the third row are “0”. For the remaining three columns, new wakes can be obtained only from the starting point, so that they are unit matrices of three rows and three columns.

Next, the wake correlation matrix calculation section 114 generates a wake correlation matrix from the in-gate decision matrix. FIG. 24 is a diagram showing a wake correlation matrix obtained from the in-gate decision matrix of FIG.

The wake correlation matrix is a matrix that satisfies all of the following three conditions. This matrix represents the correspondence between consistent tracks and observation data. The element of the in-gate decision matrix that is the same as the element satisfying the condition 1: 1 is always 1. The number of elements satisfying the condition 2: 1 is only one in each row. Condition 3: At most one element is 1 in each column. However, there may be any number of elements having a value of 1 in the first column.

Thus, for example, a plurality of wake correlation matrices shown in FIG. 24 can be obtained from the above-described in-gate judgment matrix shown in FIG. The wake correlation matrix obtained from the in-gate decision matrix shown in FIG. 23 is not limited to that shown in FIG.

Then, in step ST105, the hypothesis updating unit 115 updates the existing hypothesis based on the wake correlation matrix. At this time, in updating each hypothesis, the hypothesis updating unit 115 uses all wake correlation matrices except for the wake correlation matrix in which a wake other than the wake included in the hypothesis exists.

For example, in the hypothesis H1, the wake T1 and the wake T4 are selected. Therefore, the wake correlation matrix shown in FIG.
The wake correlation matrix of (c) is not used for updating the hypothesis H1. When the hypothesis H1 is updated based on the wake correlation matrix of FIG. 24A, the wake correlation matrix of FIG.
Indicate that all the observation data are unnecessary signals, the updated hypothesis H11 of the hypothesis H1 is as follows. H11: T81, T84 rel11

When the hypothesis H1 is updated based on the wake correlation matrix in FIG. 24D, the hypothesis H after the update of the hypothesis H1 is obtained.
12 is as follows. H12: T11, T41 rel12

When the hypothesis H1 is updated based on the wake correlation matrix shown in FIG. 24 (e), the hypothesis H1 after the update of the hypothesis H1 is obtained.
13 is as follows. H13: T81, T72, T41 rel13

As described above, each hypothesis is updated based on each wake correlation matrix except for the wake correlation matrix in which there is a wake other than the wake included in the hypothesis, and a new hypothesis is generated.

Next, in step ST106, the hypothesis reduction unit 119 executes a process of reducing the number of hypotheses by deleting hypotheses with low reliability or integrating similar hypotheses, and the cluster separation unit 120 causes the cluster Execute the separation process.

There are various methods for quasi-optimization. Here, a case will be described in which past data rejection processing for integrating hypotheses having the same observation data for the past N sampling times is applied. In this processing, the wakes and hypotheses in which the latest N samplings of observation data are the same are integrated without considering past observation data before N samplings before the latest time.

At this time, first, in step ST111, the hypothesis reducing unit 119 determines whether or not the observation data for the past N sampling times is the same between the wakes.

For example, assuming that N = 2, the current time is the time t =
3, the wakes T11, T12, T21,
T31, T32, T41, T51, T52, T61, T
71, T72, T73, T81, T82, T83, T8
Comparing 4, T85, and T86 except for the observation data before time t = 1, the following eight sets of wakes common to the observation data at times t = 2 and 3 are obtained. ｛T11, T31, T51｝ ｛T12, T32, T52｝ ｛T21, T41, T61｝ ｛T71｝ ｛T72｝ ｛T73｝ ｛T81, T83, T85｝ ｛T82, T84, T86｝

Next, in step ST112, the hypothesis reducing unit 119 integrates the tracks determined to be the same in step ST111, and generates a new track.

For example, according to the above example, the following eight new tracks after integration are generated. T91: "20"-"30" T92: "20"-"31" T93: "21"-"32" T94: "20" T95: "21" T96: "30" T97: "31" T98: " 32 "

Then, a motion specification such as a smooth value is calculated for a new track after the integration. The smoothed value of the track after integration is calculated as an average weighted by reliability.

Next, in step ST113, the hypothesis reducing unit 119 integrates the hypotheses having the original wakes based on the results of the integration of the wakes.

Then, in step ST114, the cluster separating section 120 separates the cluster into a set of a track and a hypothesis that do not share observation data with each other by reducing the track and the hypothesis.

All observation data constituting a track sharing any observation data must all belong to the same cluster. For example, in one cluster, T
a: “11” − “20” − “30” and Tb: “11” −
If there are only two tracks, "21"-"32",
Since both share observation data “11”, the observation data “11”, “20”, “2”
“1”, “30”, and “32” must belong to the same cluster (this relationship is called an equivalence relationship).

However, when the two data are compared with each other except for the observation data "11" due to the rejection of the past data, the observation data "2"
“0” and “30” and the observation data “21” and “32” belong to different clusters. Therefore, in such a case, clusters are separated.

FIG. 25 is a diagram showing an example of cluster separation. In the example shown in FIG. 25, the observation data “20”, “30”,
Cluster 1 to which “31” belongs, and observation data “21”,
It is separated into cluster 2 to which “32” belongs.

In this way, sub-optimization and cluster separation are performed in step ST106.

Thereafter, steps ST101 to ST101
The processing of 106 is performed on observation data obtained by sequentially sampling. Then, the optimum hypothesis is appropriately selected by the track determination unit 118, and the track based on the hypothesis is supplied to the target display device 3 and displayed.

As described above, in the conventional target tracking device, the observation data, the wake as a combination thereof, and the hypothesis as the combination of the wakes are independently processed for each cluster that does not share observation data with each other. By
The target tracking process is executed efficiently.

[0053]

The conventional target tracking device and target tracking method are configured as described above.
The integration of tracks and hypotheses after the generation of new hypotheses suppresses the increase in tracks and hypotheses, but the number of track correlation matrices and hypotheses becomes large when new hypotheses are generated at the next sampling time However, there is a problem that the time required for the processing increases and the processing may not be completed by the next sampling time.

The present invention has been made in order to solve the above-described problems. The present invention executes cluster separation, wake and integration of a hypothesis before updating a hypothesis, so that a series of observation data at each time can be obtained. It is an object of the present invention to obtain a target tracking device and a target tracking method capable of suppressing the amount of data generated in the middle of the processing and executing the processing at high speed.

[0055]

According to the present invention, there is provided a target tracking apparatus, comprising: a gate calculating section for calculating an area where observation data can exist for each existing track included in each cluster; Observation data that exists in the possible area of the cluster as observation data within the cluster of that cluster, observation data selection unit that makes observation data that does not correspond to any wake in each cluster as independent observation data, and a cluster containing independent observation data A newly created / integrated unit for newly creating a cluster and integrating the clusters when any of the observed data is intra-cluster observation data of a plurality of clusters; A cluster separation unit that separates clusters according to the determination result of whether or not the observation data is common; After processing, an in-gate determination data calculation unit that calculates in-gate determination data for each cluster; a wake correlation data calculation unit that calculates one or more wake correlation data based on the in-gate determination data; It includes a hypothesis updating unit that updates based on correlation data, a track determination unit that determines a track based on one of the hypotheses, and a hypothesis reduction unit that reduces the number of hypotheses.

In the target tracking device according to the present invention, the number of hypotheses is reduced by discarding the hypotheses in accordance with the result of the judgment by the cluster separating section.

The target tracking apparatus according to the present invention is provided with a parameter determining section for determining a predetermined period of past observation data determined by the cluster separating section.

In the target tracking apparatus according to the present invention, the parameter determination unit determines a predetermined period according to the number of observation data at the latest sampling time existing in each cluster.

In the target tracking device according to the present invention, the parameter determination unit determines a predetermined period according to the number of tracks existing in each cluster.

In the target tracking device according to the present invention, the parameter determining unit determines a predetermined period according to the number of observation data and the number of wakes at the latest sampling time existing in each cluster. .

In the target tracking method according to the present invention, a step of calculating an area where observation data can be present for each existing track included in each cluster is provided. Making the data into the intracluster observation data of the cluster, making the observation data that does not correspond to any wake in each cluster into independent observation data, and creating a new cluster containing the independent observation data, If the data is intra-cluster observation data of a plurality of clusters, the steps of integrating those clusters and separating the clusters according to whether or not the observation data of a predetermined period in the past included in each existing track are common. Calculating the in-gate decision data for each cluster; and Calculating one or more track correlation data; updating a hypothesis based on the track correlation data; determining a track based on one of the hypotheses; and reducing the number of hypotheses. It is provided.

The target tracking device according to the present invention includes a gate calculation unit for calculating an area where observation data can be present for each existing track included in each cluster, and a gate calculation unit for determining the existence area for any track in each cluster. Observation data to be used as intra-cluster observation data for the cluster, an observation data selector that makes observation data that does not correspond to any wake in each cluster an independent observation data, and a new cluster containing the independent observation data. If one of the observation data is in-cluster observation data of a plurality of clusters, a new cluster that integrates those clusters /
The integration unit, the track integration unit that integrates the trajectory of the common part of the observation data for the past predetermined period included in each existing track, and the in-gate judgment data is calculated for each cluster after processing by the wake integration unit An in-gate determination data calculation unit,
A wake correlation data calculation unit that calculates one or more wake correlation data based on the in-gate determination data, a hypothesis update unit that updates a hypothesis based on the wake correlation data, and determines a wake based on one of the hypotheses And a hypothesis reducing unit that reduces the number of hypotheses.

In the target tracking method according to the present invention, a step of calculating an area where observation data can exist for each existing track included in each cluster, and a step of calculating the area where observation exists in any of the tracks in each cluster Making the data into the intracluster observation data of the cluster, making the observation data that does not correspond to any wake in each cluster into independent observation data, and creating a new cluster containing the independent observation data, In the case of intra-cluster observation data of a plurality of clusters, integrating the clusters, and integrating the wakes of the common part of the observation data of the past predetermined period included in each existing wake, and thereafter, Calculating the in-gate determination data for each cluster; and Calculating the number of track correlation data, updating the hypothesis based on the track correlation data, determining the track based on any of the hypotheses, and reducing the number of the hypotheses. is there.

[0064]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below. Embodiment 1 FIG. FIG. 1 is a block diagram showing a configuration of a target tracking device according to Embodiment 1 of the present invention. In FIG. 1, reference numeral 1 denotes a target tracking device, reference numeral 2 denotes a target observation device (observation means) having a sensor such as a radar system and acquiring observation data, and reference numeral 3 denotes a target which displays a track of a target being tracked. A display device.

In the target tracking device 1, reference numeral 11 designates observation data existing in a possible area for any track in each cluster as intra-cluster observation data of the existing cluster, and does not correspond to any track in each cluster. An observation data selection unit 12 uses observation data as independent observation data, and newly creates a cluster including the independent observation data. If any of the observation data is observation data within a cluster of a plurality of clusters, these are selected. Cluster integration / integration unit that integrates the clusters.

Reference numeral 13 denotes a past data rejection determination cluster separation unit (cluster separation unit) that separates clusters according to the identity of past observation data.

Reference numeral 14 denotes whether or not the observed data in each cluster may be an unnecessary signal after processing by the past data rejection determination cluster separation unit 13 and whether or not the data corresponds to an existing wake. An in-gate judgment matrix calculation unit (in-gate judgment data calculation unit) that calculates an in-gate judgment matrix (in-gate judgment data) indicating whether there is a possibility of being a target for each cluster, 15 is an unnecessary signal, A track correlation matrix for calculating one or more track correlation matrices (track correlation data) indicating combinations corresponding to existing tracks and assigning each observation data to any of the new targets based on the in-gate decision matrix. A calculation unit (track correlation data calculation unit) 16 is a hypothesis update unit that updates hypotheses, which are combinations of tracks that do not contradict each other, based on the track correlation matrix.

Reference numeral 17 denotes an intra-system cluster table 31, an intra-cluster observation data table 32, an intra-cluster judgment matrix table 33, an intra-cluster track correlation matrix table 34, an intra-cluster hypothesis table 35, a hypothesis intra-track table 36, and an intra-cluster track. / A storage unit for storing the observation data table 37.

Reference numeral 18 denotes a gate calculation unit for calculating an area where observation data can exist (ie, a gate) for each existing track included in each cluster. Reference numeral 19 denotes a track determination for determining a track based on one of hypotheses. Department.

Next, the operation will be described. FIG. 2 is a flowchart for explaining the operation of the target tracking device according to the first embodiment of the present invention, FIG. 3 is a flowchart for explaining the details of the past data rejection determination cluster separation of FIG. 2, and FIG. 4 is a flowchart of FIG. It is a flowchart explaining the detail of a sub-optimization.

First, in step ST1, the observation data selection section 11 acquires observation data at that time obtained by a sensor or the like in the target observation device 2.

Next, in step ST 2, based on the gate for each track of each cluster calculated by the gate calculation unit 18, the observation data selection unit 11 determines which cluster the acquired observation data belongs to, Determine whether it belongs to a cluster. The gate calculator 18 calculates the gate based on the intra-cluster track / observation data table 37.

In step ST3, if there is observation data that does not belong to any of the clusters, the cluster new / integration unit 12 newly creates a cluster for the observation data, and any of the observation data is If the data is intra-cluster observation data of a plurality of clusters, those clusters are integrated. The cluster information after the processing is stored in the in-system cluster table 31.

Further, the cluster new / integration unit 12 creates a track based on the positional relationship between the gate and the observation data, and stores the track in the intra-cluster track / observation data table 37. The following three types of wakes are created. (1) Updated track: A track that can be obtained by adding observation data that has entered the gate to an existing track. (2) New track: A track starting from the observation data entered at that time. (3) Memory track track: A track indicating that "there was no corresponding observation data at the relevant time" with respect to the existing track.

Here, the operation of the cluster new / integration unit 12 will be described in detail. FIGS. 5 and 6 are diagrams for explaining an example of the processing of the cluster new / integration unit 12.

In a certain cluster, as shown in FIG.
When the observation data “00” at time t = 0, the observation data “10” and “11” at time t = 1, and the observation data “20” and “21” at time t = 2, the observation data When expressed as a set (one observation data for a new track),
There are six possible wakes: T1: "10"-"20" T2: "10"-"21" T3: "11"-"20" T4: "11"-"21" T5: "20" T6: "21"

In this case, a hypothesis that is a combination of wakes is expressed as “ID: {set of wakes} reliability”.
It looks like this: At this time, the tracks of each hypothesis are selected without contradiction so that the same observation data is not included in a plurality of tracks. H1: ｛T1, T4｝ rel1 H2: ｛T1, T6｝ rel2 H3: ｛T2, T5｝ rel3 H4: ｛T3, T6｝ rel4 H5: ｛T4, T5｝ rel5 where rel is the reliability of the hypothesis Hi. Represent. The reliability of the hypothesis is calculated from the reliability of the wake included in the hypothesis. The reliability of the wake is calculated based on, for example, the distance from the center of the gate to the target position based on the observation data.

In this state, as shown in FIG.
In 3, observation data "30", "31", "32"
Are obtained, and the observation data "30" and "31"
It exists in the gate of the wake of T3 and T5, and the observation data "3
When “2” exists in the gates of the wakes T2, T4, and T6, T11: “10”-“20”-“30” T12: “10”-“20”-“31” T21: “ 10 "-" 21 "-" 32 "T31:" 11 "-" 20 "-" 30 "T32:" 11 "-" 20 "-" 31 "T41:" 11 "-" 21 "-" 32 "T51 : "20"-"30" T52: "20"-"31" T61: "21"-"32", and as new tracks, T71: "30" T72: "31" T73: "32" T81: "10"-"20" T82: "10"-"21" T83: "11"-"20" T84: "11"-"21" T85: "20" T86 : “21” is obtained.

Next, in step ST4, the past data rejection determination cluster separation unit 13 separates clusters according to whether or not observation data of a predetermined past period included in each wake is common. FIG. 7 is a diagram illustrating an example of a process of the past data rejection determination cluster separation unit 13.

At this time, the past data rejection determination cluster separating unit 13 first refers to the intra-cluster track / observation data table 37 in step ST11 to determine whether the observation data for the past N sampling times is the same between the tracks. Is determined, the wakes are classified based on the result of the determination, and clusters are separated according to the presence or absence of the equivalence relation of the observation data of the common part.

For example, assuming that N = 2, the current time is changed to time t =
3, and the wakes T11, T12, T21, T31,
T32, T41, T51, T52, T61, T71, T
72, T73, T81, T82, T83, T84, T8
Comparing T5 and T86 except for the observation data before time t = 1, the following eight sets of wakes common to the observation data at times t = 2 and 3 are obtained. ｛T11, T31, T51｝ ｛T12, T32, T52｝ ｛T21, T41, T61｝ ｛T71｝ ｛T72｝ ｛T73｝ ｛T81, T83, T85｝ ｛T82, T84, T86｝

At this time, the wake pair {T11, T31, T
In 51｝, "20"-"30" are common,
In the wake set {T12, T32, T52}, "20"-
"31" is common, and the wake pair {T21, T41,
In T61 #, "21"-"32" are common.

Further, a set of wakes {T81, T83, T8
In 5｝, the part of “20” is common, and the set of wakes ｛T
82, T84, and T86 #, the portion "21" is common.

The track {T71}, track {T72} and track {T73} have only observation data "30", only observation data "31" and only observation data "32", respectively.

Therefore, there is no equivalence between observation data "20" and observation data "32", and between observation data "21" and observation data "30" and "31".
Thereby, as shown in FIG. 7, the observation data “20”,
It is separated into cluster 1 to which “30” and “31” belong and cluster 2 to which observation data “21” and “32” belong.

Next, in step ST12, the past data rejection determination cluster separation unit 13 distributes the wake to each cluster separated in step ST11. That is, since the clusters are separated based on the presence or absence of the equivalence relation of the observation data of the common part, the observation data belonging to each cluster has already been obtained. Therefore, wake distribution is performed by including wakes having observation data belonging to each cluster in the cluster. As a result, the intra-cluster hypothesis table 35, the intra-hypothesis track table 36, and the intra-cluster track / observation data table 37 in the storage unit 17 are updated.

Therefore, cluster 1 in the above example
The existing track T1, T3, T5 and the updated track T1
1, T12, T31, T32, T51, T52, T7
1, T72, T81, T83, T85 are sorted,
Cluster 2 includes existing tracks T2, T4, T6 and updated tracks T21, T41, T61, T73, T82, T8.
4, T86 is sorted.

Next, in step ST13, the past data rejection determination cluster separation unit 13 integrates the hypotheses based on the wakes constituting the hypotheses. Thereby, the intra-cluster hypothesis table 35 and the intra-hypothesis wakeup table 36 in the storage unit 17 are updated. At this time, the hypotheses are integrated as follows. (1) Based on a predetermined cluster A, tracks that do not belong to the cluster A are excluded from tracks constituting each hypothesis. (2) Integrate hypotheses whose wakes are the same.
Note that the reliability of the hypotheses after integration is the sum of the reliability of a plurality of hypotheses to be integrated.

Based on the above cluster 1, the hypothesis H1: {T1, T4} rel1, H2: {T1, T6} rel2, H3: {T2, T5} rel3, H4: {T3, T6rel4, H5:} When T4, T5｝ rel5 are integrated, T2, T4, T
With the exception of 6, these hypotheses are as follows: H1 ': \ T1 \ rel1 H2': \ T1 \ rel2 H3 ': \ T5 \ rel3 H4': \ T3 \ rel4 H5 ': \ T5 \ rel5 Hypothesis H1' and hypothesis H2 ', and hypothesis H3' and hypothesis H
5 'is integrated because the wakes are identical, and the hypothesis for cluster 1 is as follows. H21: \ T1 \ rel1 + rel2 H22: \ T3 \ rel4 H23: \ T5 \ rel3 + rel5

Similarly, the hypothesis regarding cluster 2 is as follows. H31: ｛T2 rel3 H32: ｛T4 rel1 + rel5 H33: ｛T6 rel2 + rel4

In this manner, the cluster is separated by the past data rejection determination cluster separation unit 13 in step ST4, and the integration of the hypotheses is executed accordingly.

Then, in step ST5, the in-gate judgment matrix calculation section 14 refers to the in-cluster observation data table 32 and the in-cluster track / observation data table 37, and there is a possibility that each in-cluster observation data is an unnecessary signal. Whether or not, whether or not it corresponds to the existing track, whether or not there is a possibility that it is a new target is calculated for each cluster, the in-gate determination matrix in the cluster gate in-gate determination matrix table 33 save. FIG. 8 is a diagram showing an example of the in-gate decision matrix.

Each row of the in-gate judgment matrix corresponds to observation data, each column corresponds to an existing wake, and its element value is 1 or 0. However, the first column indicates whether there is a possibility of being an unnecessary signal, and the columns corresponding to the number of tracking targets (that is, the number of wakes) from the second column indicate whether or not they correspond to each existing wake. And the remaining columns indicate whether it is likely to be a new goal.

For example, the in-gate decision matrix for cluster 1 described above is as shown in FIG. 8A, and the in-gate decision matrix for cluster 2 is as shown in FIG. 8B. In the in-gate decision matrix for cluster 1, the first row corresponds to observation data "30" and the second row corresponds to observation data "31". In the in-gate decision matrix for cluster 2, the first row corresponds to observation data. This corresponds to “32”.

In the in-gate decision matrix for cluster 1, the first column indicates unnecessary signals, and the second to fourth columns correspond to existing trajectories T1, T3, and T5, respectively. Indicates whether there is a possibility, and the remaining two columns indicate whether there is a possibility of a new track. In the in-gate decision matrix for cluster 2, the first column indicates an unnecessary signal, and the second to fourth columns may correspond to the existing trajectories T2, T4, and T6, respectively. And the remaining one column indicates whether there is a possibility of a new track.

First, since all observation data may be unnecessary signals, all elements in the first column of the in-gate decision matrix for cluster 1 are 1. As shown in FIG. 7, the observation data corresponding to the existing wakes T1, T3, and T5 are “30” and “31”, so that the elements in the first and second rows are included in the second to fourth columns. Becomes 1. As for the remaining two columns, new wakes can be obtained only from the starting point, and therefore, are each a unit matrix of 2 rows and 2 columns. Similarly, since any observation data may be an unnecessary signal, the element in the first column is 1 in the in-gate judgment matrix for cluster 2. As shown in FIG.
Since the observation data corresponding to 2, T4, and T6 is "32", the elements are all 1 in the second to fourth columns. For the remaining one row, the element is 1 because there is a possibility that a new track may be formed.

Next, the wake correlation matrix calculation unit 15 generates a wake correlation matrix from the in-gate judgment matrix with reference to the intra-gate judgment matrix in-gate table 33. FIG. 9 is a diagram showing wake correlation matrices obtained from the in-gate judgment matrices of FIGS. 8A and 8B, respectively.

The wake correlation matrix is a matrix that satisfies all of the following three conditions. This matrix represents the correspondence between consistent tracks and observation data. The element of the in-gate decision matrix that is the same as the element satisfying the condition 1: 1 is always 1. The number of elements satisfying the condition 2: 1 is only one in each row. Condition 3: At most one element is 1 in each column. However, there may be any number of elements having a value of 1 in the first column.

Thus, for example, from the in-gate decision matrix shown in FIGS. 8A and 8B,
And a plurality of track correlation matrices shown in FIG. 9B are obtained. The plurality of track correlation matrices respectively obtained from the in-gate judgment matrices shown in FIGS. 8A and 8B are not limited to those shown in FIG.

Although the number of clusters is increased by separating clusters, each in-gate decision matrix (FIG. 8)
The matrix of 2 rows and 6 columns shown in FIG.
Since the size of the column matrix is small, the total number of wake correlation matrices generated from the in-gate decision matrices is smaller than when the wake correlation matrix is generated without separating the clusters.

Then, in step ST6, the hypothesis updating unit 16 generates the intra-cluster wake correlation matrix table 34, the intra-cluster hypothesis table 35, the hypothesis intra-track table 36, and the intra-cluster wake /
With reference to the observation data table 37, the hypothesis is updated based on the wake correlation matrix. At this time, in updating each hypothesis, the hypothesis updating unit 16 uses all the wake correlation matrices except for the wake correlation matrix in which there is a wake other than the wake included in the hypothesis.

For example, in the above-mentioned hypothesis H21, the wake T1 is selected. Therefore, the lower one of the wake correlation matrices shown in FIG. 9A where the wake T3 exists is not used for updating the hypothesis H1. When the hypothesis H21 is updated based on the wake correlation matrix shown in the upper part of FIG.
Since the wake correlation matrix indicates that all observation data at t = 3 are unnecessary signals, the hypothesis H41 after the update of the hypothesis H21 is as follows. H41: T81 rel41

When the hypothesis H21 is updated based on the wake correlation matrix shown in the middle part of FIG. 9A, the hypothesis H42 after the update of the hypothesis H21 is as follows. H42: T11 rel42

As described above, each hypothesis is updated based on each wake correlation matrix except for the wake correlation matrix in which there is a wake other than the wake included in the hypothesis, and a new hypothesis is generated.

Next, in step ST7, the hypothesis reducing section 20 refers to the intra-cluster hypothesis table 35, the hypothesis intra-track table 36, and the intra-cluster track / observation data table 37 to reduce the number of hypotheses. The hypothesis reduction unit 20 executes a past data rejection process of integrating hypotheses having the same observation data for the past N sampling times. In this processing, N
Hypotheses in which the latest observation data for N samplings are the same without considering past observation data before sampling are integrated.

That is, in step ST21, the hypothesis reducing unit 20 first determines whether or not the observation data for the past N sampling times is the same between each track.
Tracks determined to be the same are integrated to generate a new track. Accordingly, in step ST22, motion specifications such as smooth values are calculated for the new wake after integration. Note that the smoothed value of the track after integration is calculated as an average using the reliability as a weight. And step ST23
In, the hypothesis reducing unit 20 integrates the hypotheses having the integration source wakes based on the integration results of the wakes. That is,
When the tracks included in the existing hypothesis are used as the tracks after integration, the same hypotheses are integrated.

Thus, in step ST7, the number of hypotheses is reduced by the hypothesis reduction unit 20.

Thereafter, the processing of steps ST1 to ST7 is performed on observation data obtained by sequentially sampling. Then, an optimal hypothesis is appropriately selected by the wake determination unit 19, and a wake based on the hypothesis is supplied to the target display device 3 and displayed. Note that when selecting a hypothesis, the hypothesis is selected based on, for example, the reliability of the hypothesis and the reliability of each track constituting the hypothesis.

As described above, according to the first embodiment, cluster separation is executed before updating the hypothesis, so that the in-gate decision matrix becomes smaller and the number of wake correlation matrices decreases. The effect that processing can be performed at high speed is obtained.

Embodiment 2 FIG. 10 is a block diagram showing a configuration of a target tracking device according to Embodiment 2 of the present invention. In FIG. 10, 13A operates in the same manner as the past data rejection determination cluster separation unit 13, and determines whether or not the observation data for the past N sampling times between the wakes is the same as the past rejection determination data 38. Is a past data rejection determination cluster separation unit (cluster separation unit) that is stored in the storage unit 17 as a storage unit. The storage unit 20A does not determine whether the observation data for the past N sampling times is the same between the wakes. The hypothesis reducing unit reads out the past rejection determination data 38 stored in 17 and reduces the number of hypotheses in the same manner as the hypothesis reducing unit 20. The other components in FIG. 10 are the same as those according to the first embodiment, and a description thereof will not be repeated.

Next, the operation will be described. FIG. 11 is a flowchart illustrating details of the sub-optimization in the target tracking device according to the second embodiment.

The past data rejection determination cluster separation unit 13A
Determines whether or not the observation data for the past N sampling times is the same between each track, stores the determination result in the storage unit 17 as the past rejection same determination data 38, and based on the determination result, Are classified, and clusters are separated according to the presence or absence of the equivalence relation of the observation data of the common part. Thereafter, the past data rejection determination cluster separation unit 13A performs the above-described steps ST12 and S12.
The process of T13 is performed by the past data rejection determination cluster separation unit 13
Execute in the same way as

For example, the updated tracks assigned to the above-mentioned cluster 1 T11: “10” − “20” − “30”, T12: “10” − “20” − “31”, T31: “11” − "20"-"30", T32: "11"-"20"-"31", T51: "20"-"30", T52: "20"-"31", T71: "30", T72: Of the "31", T81: "10"-"20", T83: "11"-"20", T85: "20", the wakes T11, T31, T51 and the wakes T12, T
32 and T52 have the same observation data at the past two sampling times, so that two wake pairs {T11, T
31, T51}, {T12, T32, T52} are stored as the past rejection determination data 38.

On the other hand, the hypothesis reducing section 20A determines in step ST
At 31, the past rejection identical determination data 38 stored in the storage unit 17 is read out, and based on the past rejection identical determination data, the wakes having the same observation data for the past N sampling times are integrated to generate a new wake. I do. Accordingly, in step ST22, motion specifications such as smooth values are calculated for the new track after integration, and in step ST23, hypotheses having the original track are integrated based on the result of integration of the tracks. .

The other operations are the same as those in the first embodiment, and the description thereof will not be repeated.

As described above, according to the second embodiment, the number of hypotheses is reduced by discarding the number of hypotheses based on the determination result by the past data rejection determination cluster separation unit 13A. The effect is obtained that the hypothesis reduction processing can be performed at high speed without performing the same determination again.

Embodiment 3 FIG. 12 is a block diagram showing a configuration of a target tracking device according to Embodiment 3 of the present invention. In FIG. 12, reference numeral 41 denotes a parameter determination unit that determines the number of samplings of past observation data, that is, the parameter N described above, in which the past data rejection determination cluster separation unit 13B determines whether or not each track is the same. Reference numeral 13B denotes a past data rejection determination cluster separation unit that separates clusters according to the sameness of past observation data of the sampling number N determined by the parameter determination unit 41. The other components in FIG. 12 are the same as those in the first embodiment,
The description is omitted.

Next, the operation will be described. FIG. 13 is a flowchart illustrating the operation of the target tracking device according to the third embodiment of the present invention.

In the target tracking device according to the third embodiment, in step ST41, the parameter determination unit 41 determines whether or not the past data rejection determination cluster separation unit 1 according to the number of observation data at the latest sampling time existing in each cluster.
3B determines the number of samplings N of past observation data, which is determined to be the same between the wakes, and supplies the same to the past data rejection determination cluster separation unit 13B. At this time, for example, the parameter determination unit 41 sets the sampling number N of the past observation data to be small when the number of observation data is large, and sets the sampling number N of the past observation data to be large when the number of observation data is small.

In the past data rejection determination, the smaller the value of the parameter N, the easier the wakes are to be integrated, and the higher the probability of cluster separation. As for the in-gate decision matrix, the number of rows increases as the number of observation data corresponding to the cluster at the corresponding time increases, and the number of columns increases as the number of existing tracks increases. Therefore, when it is determined that the in-gate judgment matrix is large based on the number of observation data, the number of existing wakes, and the like, the above-described parameter N is set to be small so that the size of each in-gate judgment matrix is small. Clusters are easily separated.

In step ST42, the past data rejection determination cluster separation unit 13B separates clusters according to the identity of the past observation data of the sampling number N determined by the parameter determination unit 41. In addition,
The sampling number N is supplied by the parameter determination unit 41, and the operation of the past data rejection determination cluster separation unit 13B other than using the sampling number N is described in the past data rejection determination cluster separation unit 1B in the first embodiment.
Same as 3.

Note that the other operations are the same as those in the first embodiment, and a description thereof will not be repeated.

As in the second embodiment, the past data rejection determination cluster separation unit 13B stores the past rejection identity determination data in the storage unit 17, and the hypothesis reduction unit 20 uses the past rejection identity determination data. The number of hypotheses may be reduced.

As described above, according to the third embodiment, the sampling number N of the past observation data considered in the past data rejection determination is determined according to the observation situation. Therefore, when it is determined that the amount of data being processed is large, clusters can be easily separated, the amount of data being processed can be reduced, and the effect of updating the hypothesis at high speed can be obtained. .

Embodiment 4 In the target tracking device according to the fourth embodiment of the present invention, the parameter determination unit 41 determines the sampling number N according to the number of wakes existing in each cluster.
Is determined. The rest is the same as the target tracking device according to the third embodiment.

As described above, according to the fourth embodiment, the same effects as in the third embodiment can be obtained.

Embodiment 5 FIG. In the target tracking device according to the fifth embodiment of the present invention, the parameter determination unit 41 determines the above-described sampling number N according to the number of observation data and the number of tracks at the latest sampling time existing in each cluster. It was done. At this time, the above-mentioned sampling number N is determined according to, for example, the product of the number of observation data and the number of existing tracks. The rest is the same as the target tracking device according to the third embodiment.

As described above, according to the fifth embodiment, the same effects as in the third embodiment can be obtained.

Embodiment 6 FIG. The target tracking device according to the sixth embodiment of the present invention integrates tracks that become the same when discarding past observation data before N sampling times before generating an in-gate decision matrix, and forms a new track. It is like that. Therefore, the target tracking device according to the sixth embodiment is the same as the target tracking device according to the first embodiment except for the past data rejection determination cluster separation unit (track integration unit) 13.

Next, the operation will be described. FIG. 14 is a flowchart illustrating the operation of the target tracking device according to the sixth embodiment of the present invention, and FIG. 15 is a flowchart illustrating details of the past data rejection determination cluster separation of FIG.

In the target tracking device according to the sixth embodiment, in step ST51, the past data rejection determination cluster separating unit 13 performs the same operation as that performed when the past observation data before the N sampling time is discarded in the past data rejection determination. Are integrated into a new track. FIG. 16 is a diagram illustrating an example of the processing of the past data rejection determination cluster separation unit 13.

At this time, in step ST61, the past data rejection determination cluster separation unit 13 determines whether or not the observation data for the past N sampling times is the same between each track, and After merging tracks that are identical if data is discarded,
Clusters are separated based on the combined tracks.

For example, assuming that N = 2, the current time is the time t =
3, and the wakes T11, T12, T21, T31,
T32, T41, T51, T52, T61, T71, T
72, T73, T81, T82, T83, T84, T8
Comparing T5 and T86 except for the observation data before time t = 1, the following eight sets of wakes having the same observation data at times t = 2 and 3 are obtained. ｛T11, T31, T51｝ ｛T12, T32, T52｝ ｛T21, T41, T61｝ ｛T71｝ ｛T72｝ ｛T73｝ ｛T81, T83, T85｝ ｛T82, T84, T86｝

At this time, the wake pair {T11, T31, T
In 51｝, "20"-"30" are common,
In the wake set {T12, T32, T52}, "20"-
"31" is common, and the wake pair {T21, T41,
In T61 #, "21"-"32" are common.

Also, the wake pair {T81, T83, T8
In 5｝, the part of “20” is common, and the set of wakes ｛T
82, T84, and T86 #, the portion "21" is common.

The track {T71}, track {T72} and track {T73} have only observation data "30", only observation data "31" and only observation data "32", respectively.

Accordingly, the above-mentioned track is integrated into the next new track. T111: "20"-"30" T112: "20"-"31" T113: "21"-"32" T114: "20" T115: "21" T116: "30" T117: "31" T118: " 32 "

In the integrated wakes T111 to T118, there is no equivalence between the observation data "20" and the observation data "32" and between the observation data "21" and the observation data "30" and "31". . As a result, as shown in FIG. 16, the cluster is divided into the cluster 1 to which the observation data “20”, “30”, and “31” belongs and the cluster 2 to which the observation data “21” and “32” belong.

Next, in step ST62, the past data rejection determination cluster separation unit 13 distributes the new wake (integrated wake) generated in step ST61 to each cluster. At this time, a wake having observation data belonging to each cluster is included in the cluster, so that the integrated wake is distributed.

Therefore, cluster 1 in the above example
Include integrated wakes T111, T112, T114, T11
6, T117, and the integrated wakes T113, T115, and T118 are allocated to the cluster 2.

Next, the past data rejection determination cluster separation unit 1
3 integrates existing tracks according to whether or not the past (N-1) observation data from the immediately preceding time is the same. The integration of existing tracks only unifies the observation data to be referenced, and does not specifically integrate the motion data.

In the example described above, the existing track is shifted one time t
= 2, the existing tracks T1 to T6 are appropriately integrated based on the observation data of the past (N-1) times. Tracks T1, T3, and T5 having the same observation data "20" at time t = 2 are integrated into an integrated existing track Ta (Ta: "20"), allocated to cluster 1, and observed at time t = 2. Data "2
Tracks T2, T4, and T6 having the same “1” are integrated into an integrated existing track Tb (Tb: “21”) and allocated to cluster 2.

[0143] In step ST63, the past data rejection determination cluster separation unit 13 integrates the existing hypotheses based on the wakes that constitute the existing hypotheses. At this time, the existing hypotheses are integrated as follows. (1) Replace the tracks that make up each existing hypothesis with integrated tracks. (2) Based on a predetermined cluster A, tracks that do not belong to the cluster A are excluded from tracks that constitute each existing hypothesis. (3) Integrate hypotheses whose wakes are the same.
Note that the reliability of the integrated hypothesis is the sum of the reliability of a plurality of integrated hypotheses.

In the above example, first, the existing hypothesis H1: {T1, T4} rel1, H2: {T1, T6} rel2, H3: {T2, T5} rel3, H4: {T3, T6} rel4, H5:} Each track of T4, T5｝ rel5 is replaced with the integrated existing track as follows. H1 ': ｛Ta, Tb｝ rel1 H2': ｛Ta, Tb｝ rel2 H3 ': ｛Ta, Tb｝ rel3 H4': ｛Ta, Tb｝ rel4 H5 ': ｛Ta, Tb｝ rel5

Next, when integrating these hypotheses, except for Tb which does not belong to cluster 1, these existing hypotheses are as follows. H111: {Ta} rel1 + rel2 + rel3 + r
el4 + rel5

Similarly, excluding Ta which does not belong to cluster 2, these existing hypotheses are as follows. H112: {Tb} rel1 + rel2 + rel3 + r
el4 + rel5

As described above, the cluster is separated by the past data rejection determination cluster separation unit 13 in step ST51, and along with this, integration of existing tracks and integration of existing hypotheses are executed.

The other operations are the same as those in the first embodiment, and the description thereof will not be repeated.

By integrating existing tracks as described above, the size of the in-gate decision matrix for each cluster is reduced, and the number of generated track correlation matrices is reduced. FIG. 17 is a diagram illustrating an example of the in-gate decision matrix. In the case of the above example, the in-gate decision matrix for cluster 1 and cluster 2 is as shown in FIGS. 17 (a) and 17 (b). In FIG. 17, the in-gate decision matrix of cluster 1 is 2 rows and 4 columns, and the in-gate decision matrix of cluster 2 is 1 row and 3 columns.

As described above, according to the sixth embodiment, integration of existing wakes and hypotheses is executed, so that the in-gate decision matrix is reduced, the number of wake correlation matrices is reduced, and high speed is achieved. The effect that processing can be performed is acquired.

In the target tracking device according to the sixth embodiment, both cluster separation and integration of existing trajectories and hypotheses are performed, but only integration of existing trajectories and hypotheses is performed. Can also perform processing at high speed, and can perform processing at higher speed by performing both cluster separation and integration of existing wakes and hypotheses.

The past data rejection determination cluster separation unit 13 and the hypothesis reduction unit 20 in the target tracking device according to the sixth embodiment are the same as the past data rejection determination cluster separation unit 13A and the hypothesis reduction unit 20A in the second embodiment.
May be changed to Thereby, an effect similar to that of the second embodiment can be obtained.

Furthermore, the target tracking device according to the sixth embodiment may be provided with the parameter determination unit 41 according to the third to fifth embodiments. Thereby, effects similar to those of the third to fifth embodiments can be obtained.

[0154]

As described above, according to the present invention, the possible area of the observation data is calculated for each existing track included in each cluster, and the existing area is determined in the possible area of any track in each cluster. A new cluster containing independent observation data is created as observation data that does not correspond to any wake in each cluster, and the observation data that does not correspond to any track in each cluster is newly created. In the case of observation data within a cluster, the clusters are integrated, and the clusters are separated according to the result of determination as to whether or not the observation data of the past predetermined period included in each existing track is common, and thereafter, In addition, the in-gate judgment data is calculated for each cluster, and one or more wake correlation data are calculated based on the in-gate judgment data. Calculate, update the hypothesis based on the track correlation data, determine the track based on one of the hypotheses, and reduce the hypothesis, so it occurs in the middle of a series of processing on the observation data at each time This has the effect of suppressing an increase in the amount of data to be performed and enabling high-speed processing.

According to the present invention, the hypotheses are discarded in accordance with the result of the judgment by the cluster separation unit, and the number of the hypotheses is reduced. There is an effect that reduction processing can be performed.

According to the present invention, since the parameter separation unit for determining a predetermined period of past observation data determined by the cluster separation unit is provided, it is determined that the amount of data being processed is large according to the observation situation. In this case, clusters can be easily separated, the amount of data during processing can be reduced, and the effect of updating the hypothesis at high speed can be obtained.

According to the present invention, the possible area of the observation data for each existing track included in each cluster is calculated, and the observation data existing in the possible area for any track in each cluster is calculated for that cluster. A new cluster containing independent observation data is created using the observation data that does not correspond to any track in each cluster as the intra-cluster observation data, and any of the observation data is In some cases, those clusters are integrated, the wakes of the portion where the observation data in the past predetermined period included in each existing wake are common, and then the in-gate determination data is calculated for each cluster, One or more track correlation data is calculated based on the in-gate determination data, and a hypothesis is calculated based on the track correlation data. Update Te, determines a track based on any hypothesis, and then, is cut hypothesis,
There is an effect that it is possible to suppress an increase in the amount of data generated during a series of processing on observation data at each time and execute processing at high speed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a target tracking device according to a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating an operation of the target tracking device according to the first embodiment of the present invention.

FIG. 3 is a flowchart illustrating details of past data rejection determination cluster separation in FIG. 2;

FIG. 4 is a flowchart illustrating details of the sub-optimization in FIG. 2;

FIG. 5 is a diagram illustrating an example of processing of a cluster new / integration unit.

FIG. 6 is a diagram illustrating an example of processing of a cluster new / integration unit.

FIG. 7 is a diagram illustrating an example of processing of a past data rejection determination cluster separation unit.

FIG. 8 is a diagram showing an example of an in-gate decision matrix.

9 is a diagram showing a wake correlation matrix obtained from each of the in-gate judgment matrices shown in FIGS. 8A and 8B. FIG.

FIG. 10 is a block diagram showing a configuration of a target tracking device according to a second embodiment of the present invention.

FIG. 11 is a flowchart illustrating details of sub-optimization in the target tracking device according to the second embodiment.

FIG. 12 is a block diagram showing a configuration of a target tracking device according to a third embodiment of the present invention.

FIG. 13 is a flowchart illustrating an operation of the target tracking device according to the third embodiment of the present invention.

FIG. 14 is a flowchart illustrating an operation of the target tracking device according to the sixth embodiment of the present invention.

FIG. 15 is a flowchart illustrating details of past data rejection determination cluster separation in FIG. 14;

FIG. 16 is a diagram illustrating an example of processing of a past data rejection determination cluster separation unit.

FIG. 17 is a diagram illustrating an example of an in-gate decision matrix.

FIG. 18 shows an example of a relationship between a track and observation data.

FIG. 19 is a block diagram showing a conventional target tracking device.

FIG. 20 is a flowchart illustrating the operation of a conventional target tracking device.

FIG. 21 is a diagram illustrating an example of an existing track.

FIG. 22 is a diagram illustrating an example of updating a track.

FIG. 23 is a diagram illustrating an example of an in-gate decision matrix.

24 is a diagram showing a wake correlation matrix obtained from the in-gate decision matrix of FIG. 23.

FIG. 25 is a diagram illustrating an example of cluster separation.

[Explanation of symbols]

2 Target observation device (observation means), 11 observation data selection unit, 12 cluster new / integration unit, 13, 13A past data rejection determination cluster separation unit (cluster separation unit, wake integration unit), 14 in-gate determination matrix calculation unit ( In-gate determination data calculation unit), 15 track correlation matrix calculation unit (track correlation data calculation unit), 16 hypothesis update unit, 18 gate calculation unit, 19 track determination unit, 20, 20A hypothesis reduction unit,
41 Parameter determination unit.

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Yoshio Kosuge 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Mitsubishi Electric Corporation F-term (reference) 5J070 AC02 AC06 AC13 AE04 AH19 AK15 AK22 BB06 BB12 BB15 BB30

Claims (9)

[Claims] 1. A target tracking device that performs a target tracking process based on observation data obtained by an observation means, wherein a gate calculation unit that calculates a region where observation data can exist for each existing track included in each cluster. And observation data that exists in the possible area for any track in each cluster as observation data within the cluster of that cluster, and observation data that does not correspond to any track in each cluster is independent observation data. A selecting unit; and a new cluster / integrating unit for newly creating a cluster including the independent observation data and integrating the clusters when any of the observation data is intra-cluster observation data of a plurality of clusters. It is determined whether or not the observation data of the past specified period included in the existing wake is common, and the result of the determination is determined. A cluster separating unit that separates clusters according to the following: after the processing by the cluster separating unit, whether or not each of the observed data in the cluster may be an unnecessary signal and whether or not it corresponds to an existing wake Or an in-gate determination data calculation unit that calculates in-gate determination data indicating whether there is a possibility of being a new target for each cluster, an unnecessary signal, one corresponding to an existing wake, and a new target A wake correlation data calculation unit that calculates one or more wake correlation data indicating a combination of allocating each observation data based on the in-gate determination data; and a wake correlation that is a combination of wakes that are not inconsistent with each other. A hypothesis updating unit that updates based on data; a track determination unit that determines a track based on any of the hypotheses; and a hypothesis that reduces the number of the hypotheses. Target tracking device, characterized in that it comprises a reduction unit. 2. The target tracking apparatus according to claim 1, wherein the hypothesis reducing unit discards the hypotheses in accordance with a result of the determination by the cluster separation unit and reduces the number of hypotheses. 3. The target tracking device according to claim 1, further comprising a parameter determination unit that determines the predetermined period of the past observation data determined by the cluster separation unit. 4. The parameter determination unit according to claim 3, wherein the predetermined period is determined according to the number of observation data at the latest sampling time existing in each cluster.
A target tracking device as described. 5. The target tracking device according to claim 3, wherein the parameter determination unit determines the predetermined period according to the number of tracks existing in each cluster. 6. The target according to claim 3, wherein the parameter determination unit determines the predetermined period according to the number of observation data and the number of wakes at the latest sampling time existing in each cluster. Tracking device. 7. A target tracking method for performing a target tracking process based on observation data obtained by observation means, wherein a region where observation data can exist for each existing wake included in each cluster is calculated; Observation data existing in the possible area for any wake in each cluster as observation data in the cluster of the cluster, and observation data not corresponding to any wake in each cluster as independent observation data; A step of newly creating a cluster including independent observation data, and integrating any of the clusters if any of the observation data is intracluster observation data of a plurality of clusters; Whether or not the observation data during the period of time is common, and separate clusters according to the judgment result After that, a gate that indicates whether the observation data in each cluster may be an unnecessary signal, whether it corresponds to an existing track, or whether it is a new target Calculating determination data for each of the clusters; and one or more track correlation data indicating a combination of allocating each observation data to one of an unnecessary signal, a signal corresponding to an existing track, and a new target. Calculating based on the inside determination data; updating a hypothesis, which is a combination of tracks that do not contradict each other, based on the track correlation data; determining a track based on any of the hypotheses; A target tracking method. 8. A target tracking device for executing a target tracking process based on observation data obtained by an observation means, wherein a gate calculation unit for calculating a region where observation data can exist for each existing wake included in each cluster. And observation data that exists in the possible area for any track in each cluster as observation data within the cluster of that cluster, and observation data that does not correspond to any track in each cluster is independent observation data. A selecting unit; and a new cluster / integrating unit for newly creating a cluster including the independent observation data and integrating the clusters when any of the observation data is intra-cluster observation data of a plurality of clusters. A track track that integrates the tracks of the common part of the observation data for the past specified period included in the existing track After processing by the track integration unit, whether or not each cluster observation data may be an unnecessary signal, whether or not it corresponds to an existing track, the possibility of a new target A combination of assigning each observation data to any of an unnecessary signal, an existing wake, and a new target, and an in-gate judgment data calculation unit for calculating the in-gate judgment data indicating whether or not there is any of the clusters, A track correlation data calculation unit that calculates one or a plurality of track correlation data indicating the following based on the in-gate determination data; and a hypothesis update unit that updates a hypothesis that is a combination of tracks that do not conflict with each other based on the track correlation data. And a track determination unit that determines a track based on any of the hypotheses; and a hypothesis reduction unit that reduces the number of the hypotheses. Location. 9. A target tracking method for performing a target tracking process based on observation data obtained by observation means, wherein a region where observation data can exist for each existing wake included in each cluster is calculated; Observation data existing in the possible area for any wake in each cluster as observation data in the cluster of the cluster, and observation data not corresponding to any wake in each cluster as independent observation data; A step of newly creating a cluster including independent observation data, and integrating any of the clusters if any of the observation data is intracluster observation data of a plurality of clusters; Of the common part of the observation data for the period of Is calculated for each cluster, indicating whether there is a possibility that the signal is an unnecessary signal, whether or not the signal corresponds to an existing track, and whether or not there is a possibility that the signal is a new target. Calculating, based on the in-gate determination data, one or more track correlation data indicating a combination of assigning each observation data to one of an unnecessary signal, an existing track, and a new target, respectively. And updating a hypothesis, which is a combination of tracks that do not contradict each other, based on the track correlation data; determining a track based on any of the hypotheses; and reducing the number of the hypotheses. A target tracking method characterized by the following.