JP5082394B2 - Multi-target tracking device - Google Patents

Description

A technique for tracking a target based on position information of a target such as a flying object obtained by observation with a sensor such as a radar, and in particular, a technique for tracking a target based on observation values of a plurality of types of sensors. It is about.

Many techniques for tracking a target using observation values obtained by a radar have already been cited in literatures such as papers and patents, and various proposals have been made for tracking devices and tracking methods. Also, when tracking multiple nearby targets, in order to reduce the possibility of mistracking caused by incorrect correlation between the target track and the observed value, multiple hypotheses are generated for the correlation between the target track and the observed value, A method for determining the final correlation while calculating the reliability has been proposed (see, for example, Patent Document 1).
Japanese Patent No. 3145893

However, since such a tracking method generates a correlation hypothesis of a target wake based only on the target position information obtained from the radar, it has a complicated wake such as the wakes of a plurality of targets approaching each other. However, there is still a possibility that erroneous correlation results are likely to be generated, and erroneous tracking is likely to occur. The present invention has been made to solve this problem, and an object of the present invention is to obtain a multi-target tracking device that can perform target tracking with higher reliability.

The multi-target tracking device according to the present invention receives a first observation value that represents the target position in terms of elevation angle, azimuth angle, and distance, and a second observation value that represents the target position in terms of elevation angle and azimuth angle. , A position information tracking processing unit that generates a hypothesis group including a target track using the first observation value and the second observation value, and a target of the second observation value and the second observation value. Identification information is input, and the second observation value is compared with a predicted position range of a target calculated from an angle wake including the second observation value observed in the past, and within the predicted position range the correspondence of the second observation values the input to the angle track having the same identification information and the second observation value included in the angle track, the target identification tracking processing unit for generating a new angle trajectory with some And the second view included in the wake in the hypothesis included in the hypothesis group. A degree of matching between the hypothesis and the angle wake by comparing the value with the second observation value included in the track in the angle wake, and the matching exceeding a predetermined threshold in the hypothesis group A hypothesis having a low degree of reliability among a hypothesis group including a hypothesis matching unit that selects a hypothesis having a degree and deletes a hypothesis having the matching degree that does not exceed a predetermined threshold; and a hypothesis group selected by the hypothesis matching part And a hypothesis selection integration unit that reduces the hypothesis group by integrating similar hypotheses.

According to the present invention, the angle wake generated based on the observation value of the target identification information and the position information is utilized, and the hypothesis of the target wake generated based only on the observation value of the target position information is selected. Therefore, the possibility of leading to an incorrect correlation result is reduced, the reliability can be determined with higher reliability, and an accurate correlation can be determined, and the possibility of erroneous tracking can be reduced.

Embodiment 1 FIG.
FIG. 1 shows the configuration of the multi-target tracking device according to the first embodiment. The multi-target tracking device 1 is connected to a radar device 2 and a target identification sensor device 3 installed outside, and inputs observation values about a target observed by the radar device 2 and the target identification sensor device 3, and based on these observation values. The target tracking process is performed.

The radar apparatus 2 emits a radio wave by itself, and detects the position of the target by measuring a reflected wave reflected by the target (for example, a flying object), and the detected target position is represented by spatial coordinates (elevation angle, azimuth angle, The position information expressed in (distance) is output as observation values in time series.

Further, the target identification sensor device 3 detects the target position and identifies the target by exclusively measuring the electromagnetic wave radiated from the target without emitting a radio wave by itself, and the detected target position is represented by spatial coordinates (elevation angle, Position information expressed in (azimuth)) and identification information regarding the detected target are output as observation values in time series. The position information output from the target identification sensor device 3 is different from the position information output from the radar device 2 in that it does not include a distance. The target identification information is information related to target identification (for example, the model of the flying object) determined by comparing observation data such as the pulse shape of the electromagnetic pulse emitted by the target with previously collected library data. That is.

FIG. 2 shows an operation flow of the multi-target tracking device 1. Hereinafter, the operation of the multi-target tracking device 1 will be described with reference to the drawings. The multi-target tracking device 1 inputs an observation value from the radar device 2 or the target identification sensor device 3 (step A1). Next, it is determined whether the device that outputs the input observation value is the radar device 2 or the target identification sensor device 3 (step A2).

First, a process when the observation value output device is the radar device 2 will be described. When the output device is the radar device 2, the position information tracking processing unit 4 first performs processing. The position information tracking processing unit 4 includes an observation value selection unit 5, a wake generation unit 6, a correlation hypothesis generation unit 7, and a wake database 8 and a hypothesis database 9 used by these units.

When the observation value of the radar apparatus 2 is input to the position information tracking processing unit 4, the observation value selection unit 5 reads the data of the existing wake from the wake database 8, and uses the motion parameters of the existing wake at the observation time. Calculate the predicted position range (this predicted position range is called a gate), and calculate the gate, which is a spatial region where observation values may be obtained, from the predicted position and the estimated value of the error covariance of the motion specifications. To do. Then, by comparing the calculated gate with the observed value, it is checked whether or not the input observed value is in the gate of each existing wake, and it is determined to which wake the observed value can be correlated (step) A3).

FIG. 3 is an example showing the relationship between existing tracks and observed values. The observation value output by the radar device 2 is a target position represented by spatial coordinates (elevation angle, azimuth angle, distance). In FIG. It is shown on the coordinate space consisting of In this example, there are T1 and T2 as existing wakes, and three observation values O1, O2, and O3 are obtained. It can be seen that the observation values O1 and O2 are included in the gates of the existing tracks T1 and T2. Note that the observed value O3 is not included in any gate. In this example, the observed value O1 can be correlated with the existing tracks T1 and T2, and the observed value O2 can be correlated with the existing tracks T1 and T2. Further, the observed value O3 cannot be correlated with any of the existing wakes T1 and T2. The observation value selection unit 5 outputs all of the existing wake and observation value pairs that can be correlated (for example, (T1, O2)) to the wake generation unit 6.

Next, the wake generation unit 6 creates a wake based on the pair of existing wake and observation values that can be correlated obtained in step A3 (step A4). There are three types of wakes created here:
(1) Renewed wake: A wake that is generated by adding observations that entered the gate to the existing wake. In the example of FIG. 3, a track in which the existing track T1 is updated using the observed value O1, a track in which the existing track T1 is updated using the observed value O2, a track in which the existing track T2 is updated using the observed value O1, and the observed value O2 A total of four updated wakes are generated by updating the existing wake T2 using.
(2) New wake: A wake starting from the observed value at that time. In the example of FIG. 3, a total of three tracks are generated: a new track starting from the observation value O1, a new track starting from the observation value O2, and a new track starting from the observation value O3.
(3) Memory track track: A track indicating that “there were no correlated observations at the corresponding time” with respect to the existing track, which is also referred to as a track that caused a missing detection. In the example of FIG. 3, if the observation values O1 and O2 do not exist, there are no observation values that can be correlated with the existing tracks T1 and T2, and two memory track tracks based on the existing tracks T1 and T2 are generated.

The motion specifications at the observation time of these generated wakes are calculated, and the likelihood of the wake corresponding to the correlation result is calculated. When the generated wake is an updated wake, a specific method for calculating the wake likelihood is to calculate the following equation assuming that the probability distribution of the observed values is a Gaussian distribution centered on the predicted position: It has been known.

Here, L _k is the likelihood of track in the case of correlation is observed value z _k at time k, z _k is the observed value, z _kp predicted observation position, S _k is the residual covariance matrix, n represents the observations Dimension.

When the generated wake is a new wake or a memory track wake, a specified value determined based on a parameter such as an observation error of the observed value is used as the likelihood of the wake. The generated wake and the likelihood pair of each wake obtained as described above are output to the correlation hypothesis generation unit 7.

The correlation hypothesis generation unit 7 first reads data of existing hypotheses from the hypothesis database 9. The data of the existing hypothesis is data including a wake constituting the hypothesis and the reliability of the hypothesis. For example, it is data indicating that the existing hypothesis H1 includes the existing wakes T1 and T2 and the reliability is R1.

Next, a new hypothesis is generated based on the pair of generated wakes and likelihood of each wake output from the wake generation unit 6 and the existing hypothesis read out from the hypothesis database 9 (step A5). This newly generated hypothesis (herein called the correlation hypothesis) must satisfy the following conditions.

(A1) The correlation hypothesis must be an extension of one of the existing hypotheses generated at the previous observation time. FIG. 4 shows two examples of correlation hypotheses. Correlation hypothesis 1 adopts two tracks, an updated track in which the existing track T1 is updated with the observed value O2 and an updated track in which the existing track T2 is updated with the observed value O1, and the observed value O3 is regarded as an unnecessary signal. Since this correlation hypothesis 1 is premised on the existence of T1 and T2 as existing wakes, it is generated by the development of the existing hypothesis 1 in FIG. Therefore, if the existing hypothesis 1 is not included in the existing hypothesis group, the correlation hypothesis 1 cannot be generated. Correlation hypothesis 2 adopts two tracks, an updated track obtained by updating the existing track T1 with the observed value O1 and a new track T3 starting from the observed value O3, and regards the observed value O2 as an unnecessary signal. Since this correlation hypothesis 2 is based on the premise that only T1 exists as an existing wake, it is generated by the development of the existing hypothesis 2 in FIG.

(A2) For the selected combination of observation value and wake, the observation value must be within the wake gate. For example, a hypothesis for adopting a track obtained by updating the existing track T1 with the observation value O3 is not generated.

(A3) The observation is associated with any one wake or regarded as an unnecessary signal. For example, in one hypothesis, the observed value O1 is associated with both the existing wake T1 and the existing wake T2, or the observed value O2 does not correspond to any wake or is considered not to be an unnecessary signal. There is nothing.

(A4) There is at most one observation value that can be correlated to one existing track. For example, in one hypothesis, both the observed value O1 and the observed value O2 are not associated with the existing wake T1.

Next, a reliability is calculated as an evaluation value for each generated correlation hypothesis. The reliability is the reliability of each existing hypothesis used at the time of hypothesis generation, the likelihood of the track used (update track, new track, memory track track) and the observed value considered as an unnecessary signal in the hypothesis. Calculate by the product of degrees. The reliability of each hypothesis calculated in this way is normalized so that the sum of the reliability of all correlation hypotheses is 1.0.

The generated hypothesis and its reliability are output to the hypothesis selection integration unit 11 of the common processing unit 10. The processing in steps A3 to A5 described above is processing in the position information tracking processing unit 4.

Next, the hypothesis selection integration unit 11 reduces the hypothesis group by deleting hypotheses with low reliability from the hypotheses generated in step A5 and integrating similar hypotheses (step A6). The following methods are well known for reducing this hypothesis group.

(B1) A threshold is set for the reliability, and all hypotheses having a reliability less than that are deleted.
(B2) An upper limit is set for the number of hypotheses, and only the set number of hypotheses are left in order of higher reliability, and other hypotheses are deleted.
(B3) The hypotheses having the same correlation contents of observation values for the past several observation times are integrated retroactively from the latest observation time point.

The hypothesis group thus obtained by selective integration is output to the wake display unit 12 and the hypothesis database 9. The hypothesis database 9 accumulates this hypothesis group in cooperation with existing hypotheses. In addition, since the existing track that is not included in any of the hypothesis groups is not necessary for the subsequent processing, a process of deleting from the track database 8 is performed.

The wake display unit 12 displays the wake included in the hypothesis on the hypothesis having the highest reliability value from the hypothesis group output by the hypothesis selection / integration unit 11 (step A7).

Steps A1 to A7 described above are a series of operation flows when one observation value is input from the radar apparatus 2. Thereafter, the process waits until the next observation value is input. When the next observation value is input, the processing of the next cycle is executed from step A1.

The operation when the observation value from the radar device 2 is input has been described above. Next, the operation when the observation value from the target identification sensor device 3 is input will be described.

If it is determined in step A2 that the observation value output device is the target identification sensor device 3, the position information portion of the observation values output by the target identification sensor device 3 is the position information tracking processing unit 4 and the target identification. It is output to both of the tracking processing unit 13, and the identification information portion of the observed value is output only to the target identification tracking processing unit 13. Hereinafter, the process of the target identification tracking processing unit 13 will be described first. The target identification and tracking processing unit 13 includes an observation value selection unit 14, a correlation determination unit 15, an angle wake generation unit 16, and an angle wake database 17 used by these units.

When the observation value of the target identification sensor device 3 is input to the target identification tracking processing unit 13, the observation value selection unit 14 calculates the predicted position range (gate) of the target at the observation time from the existing angular track, and this gate Based on this, we investigate the possibility of correlation between the observed value and the existing angle wake. The specific procedure is the same as the operation of the observation value selection unit 5 in the radar information processing unit 4. That is, by comparing the observation value with the gate calculated from the existing angle wake read from the angle wake database 17, it is determined whether or not the input observation value is within the gate of each existing angle wake. It is determined whether it is possible to correlate with the angle track, and all of the pairs of existing angle tracks and observation values that can be correlated are output to the correlation determination unit 15. (Step A8).

Next, the correlation determining unit 15 uniquely associates the existing angle track with the observed value (step A9). At this time, the policy for determining the correspondence is as follows.
(C1) Only the observation value that entered the gate is associated with the existing angle track.
(C2) When a plurality of observation values enter the gate, the observation value having the identification information that matches the identification information included in the observation value used in the latest update of the existing angle track is associated with the highest priority.

FIG. 6 shows an example of correspondence between existing angle tracks and observed values. In this example, the observation values O1_PASS and O2_PASS are entered in the gate of the existing angle track T1_PASS, and the observation values O1_PASS and O2_PASS are entered in the gate of the existing angle track T2_PASS. Then, the observation information OT1_PASS used for the latest update of the existing angle track T1_PASS matches the identification information of the current observation value O1_PASS, and the observation value OT2_PASS used for the latest update of the existing angle track T2_PASS and the current observation value The identification information of O2_PASS matches. Therefore, the observed value O1_PASS is associated with the existing angular track T1_PASS, and the observed value O2_PASS is associated with the existing angular track T2_PASS.

Next, the angle track generation unit 16 updates the angle track using the associated observation values (step A10). The updated angle track is stored in the angle track database 17 and output to the hypothesis matching unit 18 of the common processing unit. Thus, the update of the angle track is performed without creating a correlation hypothesis.

On the other hand, the position information tracking processing unit 4 generates a correlation hypothesis using the position information of the observation value input from the target identification sensor device 3 and the information of the wake database 8 and the hypothesis database 9 (steps A11 to A13). . The generation of the correlation hypothesis is the same as the operation (steps A3 to A5) of the position information tracking processing unit 4 based on the observation value input from the radar device 2 described above, and thus detailed operation description is omitted. . Note that the position information of the observation value input from the target identification sensor device 3 is expressed in spatial coordinates composed of an elevation angle and an azimuth angle, whereas the existing track in the wake database 8 and the existing hypothesis in the hypothesis database 9 are used. The included position information is represented by spatial coordinates of elevation angle, azimuth angle, and distance. For this reason, when examining the correspondence between the observed value and the existing track or existing hypothesis, the predicted value of the distance component is used as an alternative to the observed value of the distance component, and the association is performed. The position information tracking processing unit 4 outputs the generated hypothesis and its reliability to the hypothesis matching unit 18 of the common processing unit 10.

Next, the hypothesis collating unit 18 collates the angle track generated by the target identification tracking processing unit 13 with the hypothesis generated by the position information tracking processing unit 4, and leaves only a hypothesis similar to the angle track (step A14). This collation is performed by calculating the degree of matching with the angle track generated by the target identification tracking processing unit 13 for all hypotheses generated by the position information tracking processing unit 4, and selecting and leaving only hypotheses with a high matching degree. FIG. 7 shows a further detailed procedure for hypothesis matching in step A14. The hypothesis matching method will be described below using an example.

First, the similarity is calculated for all combinations of the wake included in the generated hypothesis and the wake included in the angle wake (step B1). Here, the similarity is defined as “a ratio of the number of observation values included in both the wake in the hypothesis and the wake in the angle wake to the number of observation values included in the wake in the angle wake”. The similarity calculation will be described by taking the hypothesis of FIG. 8 and the angle wake of FIG. 9 as examples. The hypothesis in FIG. 8 includes two tracks, a wake T1_FUS and a wake T2_FUS, whereas the angle wake in FIG. 9 includes three angle tracks, a wake T1_PASS, a wake T2_PASS, and a wake T3_PASS. Similarity is calculated for 3 × 2 = 6 combinations. For example, the similarity in the combination of the wake T1_PASS and the wake T1_FUS is five observation values in the wake T1_PASS, of which four observation values are included in the wake T1_FUS, and the similarity is 0.8 (= 4 / 5). Table 1 below shows the similarity calculated for each combination.

Next, on the basis of the calculated similarity, the trail on the hypothesis side that realizes the highest similarity is associated with each track in the angle track (step B2). In the case of the similarity in Table 1, the correspondence is as follows.
Action 1: T1_PASS Ｔ T1_FUS
Correspondence 2: T2_PASS ⇔ T2_FUS
Action 3: T3_PASS Ｔ T1_FUS

Next, the matching degree of the hypothesis is calculated as a product of the similarity corresponding to this hypothesis (step B3). In this example, the similarity of correspondence 1 is 0.8, and the similarity of correspondence 2 and correspondence 3 are both 1.0, so 0.8 × 1.0 × 1.0 = 0.8 is the hypothesis. It becomes the collation degree. The hypothesis matching unit 18 calculates the matching degree of each hypothesis by repeating the above procedure.

Next, a threshold is set for the hypothesis matching degree, a hypothesis having a matching degree equal to or higher than the threshold is left, and a hypothesis having a matching degree lower than the threshold is deleted (step B4). As an example, assume that an observed value as shown in FIG. 10 is obtained. In this example, judging from the identification information that is the observation value of the target identification sensor device 3, it is considered that there is a high possibility that two targets exist and draw a trajectory that becomes an approaching intersection. On the other hand, it is assumed that there are three hypotheses shown in the upper part of FIG. 11 as hypothesis groups before hypothesis matching corresponding to these observation values. However, it is assumed that the reliability of the first hypothesis depicting the intersecting trajectory is the highest, and the reliability decreases in the order of the second hypothesis and the third hypothesis. The middle part of FIG. 9 shows the degree of matching between each hypothesis in this hypothesis group and the angle track by the target identification sensor device alone. Comparing each hypothesis with the matching degree, the matching degree of the hypothesis with the highest reliability (the first hypothesis before hypothesis matching) is 0.25, and the second hypothesis before hypothesis matching and the third hypothesis matching degree are both 1.0. Assuming that the threshold value of the matching degree is 0.5, as shown in the lower part of FIG. 9, the hypothesis (second hypothesis before hypothesis matching) that draws the locus of approach and separation becomes the highest (first hypothesis after hypothesis matching). The third hypothesis before hypothesis matching becomes the second hypothesis after hypothesis matching.

In this way, by considering the hypothesis matching degree, two hypotheses having a matching degree equal to or higher than the threshold can be selected from the three hypotheses before hypothesis matching. The hypothesis selected in this way is output to the hypothesis selection integration unit 11.

In step B1 “Calculation of wake similarity”, all observation values included in the wake are targeted for similarity calculation, but past observation values more than a certain time ago are ignored and within a certain time. The similarity may be calculated only for the latest observation value. Alternatively, the similarity may be calculated only for observation values for the latest several observation times.

In the above description, the similarity is calculated based on the degree of matching of the observation values included in the wake. However, the degree of similarity may be calculated based on the degree of matching of the smooth values. For example, the angle component of the wake motion specification estimate included in the hypothesis is x _hypo , its error covariance matrix is P _hypo , the angle wake motion specification estimate is x _pass , and its error covariance matrix is P _pass The similarity Λ _ang is calculated as follows.

Further, the similarity between the hypothesis and the angle wake may be calculated as a product of the similarity calculated based on the observed value and the similarity calculated based on the smooth value.

Next, the hypothesis selection integration unit 11 reduces the hypothesis group by deleting hypotheses with low reliability and integrating similar hypotheses (step A15). Since the method for reducing the hypothesis group is the same as that described in step A6, detailed description thereof is omitted. The hypothesis group thus obtained by selective integration is output to the wake display unit 12 and the hypothesis database 9. The hypothesis database 9 stores selected and integrated hypotheses in cooperation with existing hypotheses. In addition, since the existing track that is not included in any of the hypothesis groups is not necessary for the subsequent processing, a process of deleting from the track database 8 is performed.

Finally, the wake display unit 12 displays the wake included in the hypothesis on the hypothesis having the highest reliability value from the hypothesis group output by the hypothesis selection / integration unit 11 on a screen or the like (step A16).

Steps A8 to A16 described above are a series of operation flows when one observation value is input from the target identification sensor device 3. Thereafter, the process waits until the next observation value is input. When the next observation value is input, the processing of the next cycle is executed from step A1.

In the above description, the position information tracking processing unit (steps A11 to A13) is executed after the processing in the target identification tracking processing unit 13 (steps A8 to A10). The processing (steps A8 to A10) in the target identification tracking processing unit 13 may be executed after the processing (steps A11 to A13), or the processing in the target identification tracking processing unit 13 (steps A8 to A10). ) And the processing of the position information tracking processing unit (steps A11 to A13) may be performed in parallel.

As described above, the multi-target tracking device according to Embodiment 1 uses the angular track generated based on the target identification information and the observed value of the position information, and generates it based only on the observed value of the target position information. Since the target track hypothesis is selected, it is less likely to lead to an incorrect correlation result, and it is possible to execute multi-target tracking with higher reliability and accurate correlation determination.

Embodiment 2. FIG.
It is conceivable to fly while changing the form of radiated radio waves emitted by a target such as a flying object. Although the identification information of the observed value obtained from such a target may change in the middle of the wake, it is desirable that it can be recognized as the same target regardless of the transition of the identification information. In the multi-target tracking device according to the second embodiment, when the transition of identification information is the same as the transition observed in the past in the correlation hypothesis selection and integration process, the reliability of the hypothesis including such transition is high. By processing as a thing, a correct correlation result can be obtained even when the identification information of the observation value obtained from one target changes, and a more accurate tracking process is performed.

FIG. 12 shows the configuration of the multi-target tracking device according to the first embodiment. FIG. 13 shows an operation flow of the multi-target tracking device 1 according to the second embodiment. 12 and 13, the same or corresponding parts as those in FIGS. 1 and 2 are denoted by the same reference numerals, and detailed description thereof is omitted. Hereinafter, the operation of the multi-target tracking device 1 will be described with reference to the drawings.

Since the difference from the multi-target tracking device of the first embodiment in the operation flow is the operation of the common processing unit 10 when the observation value is input from the target identification sensor device 3, the hypothesis matching (step A14) and the subsequent steps The processing from will be described. As in the case of the first embodiment, the hypothesis collating unit 18 collates the correlation hypothesis generated by the position information tracking processing unit 4 with the angular track generated by the target identification tracking processing unit 13. Then, the selected hypothesis that has been verified is output to the identification information transition determination unit 19 (step A14).

The identification information transition determination unit 19 does not change the identification information of the correlated observation value in any of the wakes constituting the hypothesis for the hypothesis having the highest reliability among the hypothesis group output from the hypothesis matching unit 18. Find out. When a change in the identification information of the observed value is detected, the identification information before and after the transition is stored in the identification information transition history database 20. For example, if the hypothesis shown in FIG. 8 is the hypothesis with the highest reliability, the identification information of the observation value of the target identification sensor device 3 correlated with the track T1_FUS changes midway (rectangle to pentagon). The transition determination unit 19 stores the change in the identification information transition history database 20. In the processing of the observation time after the next time, if another hypothesis includes a hypothesis that includes a track where a transition similar to this transition has occurred, the reliability of that hypothesis is multiplied by a constant and increased in a direction that predominates over others. Correct it. Thereby, the reliability of the hypothesis can be corrected in consideration of the transition state of the identification information. The identification information transition determination unit 19 outputs the hypothesis whose reliability has been corrected to the hypothesis selection integration unit 11 (step A17).

As in the case of the first embodiment, the hypothesis selection integration unit 11 reduces hypotheses by deleting hypotheses with low reliability and integrating similar hypotheses (step A15). At this time, since the transition of the identification information occurring in the existing track that is not included in any of the integrated hypotheses is unnecessary for the subsequent processing, the process of deleting the corresponding transition information from the identification information transition history database 20 I do.

Finally, the wake display unit 12 displays the wake included in the hypothesis for the hypothesis having the highest reliability value from the hypothesis group output by the hypothesis selection / integration unit 11 (step A16).

As described above, the multi-target tracking device according to the second embodiment records the transition of the identification information in the correlation hypothesis, and corrects the hypothesis reliability so as to prioritize the hypothesis that has caused the transition similar to the past. Even when the identification information of the observation value obtained from one target changes, the probability of obtaining a correct correlation result is high.

It is a figure which shows the structure of the multi-target tracking apparatus by Embodiment 1 of this invention. It is a figure which shows the operation | movement flow of the multi-target tracking apparatus by Embodiment 1 of this invention. It is explanatory drawing of the existing wake, a gate, and an observed value. It is an example of a correlation hypothesis. This is an example of an existing hypothesis. It is an example of the existing track and the observed value in the tracking by the target identification sensor device. It is a figure which shows the procedure of hypothesis collation. It is an example of existing tracks and observations in the hypothesis. It is an example of the existing wake and observation value in an angle wake. It is an example of the observed value of a radar apparatus, and the observed value of a target identification sensor apparatus. It is an example of a hypothesis group before and after hypothesis collation. It is a figure which shows the structure of the multi-target tracking apparatus by Embodiment 2 of this invention. It is a figure which shows the operation | movement flow of the multi-target tracking apparatus by Embodiment 2 of this invention.

Explanation of symbols

1: multi-target tracking device, 2: radar device, 3: target identification sensor device, 4: position information tracking processing unit, 5: observation value selection unit, 6: wake generation, 7: correlation hypothesis generation unit, 8: wake database , 9: Hypothesis database, 10: Common processing unit, 11: Hypothesis selection integration unit, 12: Track display unit 12, 13: Target identification tracking processing unit, 14: Observation value selection unit, 15: Correlation determination unit, 16: Angle Wake generation unit, 17: Angular wake database, 18: Hypothesis matching unit

Claims (8)

A first observation value that represents the target position as an elevation angle, an azimuth angle, and a distance and a second observation value that represents the target position as an elevation angle and an azimuth angle are input, and the first observation value and the first observation value are input. A position information tracking processing unit that generates a hypothesis group including the target track using the observation values of 2,
The second observation value and identification information of the target possessed by the second observation value are input, and calculated from the second observation value and an angle wake including the second observation value observed in the past. was compared with the predicted position range of the target, the angle of the second observation values the input having the same identification information and the second observation value included in the angle track with lies within the predicted position range A target identification tracking processing unit that associates with a wake and generates a new angle wake;
Collating the hypothesis with the angle wake by comparing the second observation value included in the wake in the hypothesis included in the hypothesis group and the second observation value included in the wake in the angle wake A hypothesis matching unit that selects a hypothesis having the matching degree that exceeds a predetermined threshold from the hypothesis group, and deletes a hypothesis having the matching degree that does not exceed the predetermined threshold;
A hypothesis selection integration unit that reduces the hypothesis group by deleting hypotheses with low reliability from hypotheses consisting of hypotheses selected by the hypothesis matching unit and integrating similar hypotheses is provided. Multi-target tracking device. The hypothesis verification unit, the angle with respect to the number of the second observed value contained in track in track, the number of the second observed value included in both the track of track and the angle of trajectory in the hypothesis The multi-target tracking device according to claim 1, wherein the similarity is calculated from the ratio of and the matching is obtained from a product of the similarities. The multi-target tracking device according to claim 2, wherein the hypothesis matching unit calculates the similarity only for the second observation values for the latest several observation times. The hypothesis matching unit includes a smooth value calculated based on the first observation value and the second observation value included in the wake in the hypothesis and the second observation value included in the wake in the angle wake. The multi-target tracking device according to claim 1, wherein similarity is calculated by error covariance with a smooth value calculated based on, and the matching degree is obtained from a product of the similarity. The hypothesis verification unit, to the number of observations included in the track in the angle track, the value calculated from the ratio of the number of observations included in both the track in track and the angle track in the hypothesis,
Smoothness calculated based on the first observation value and the second observation value included in the wake in the hypothesis and the second observation value included in the wake in the angle wake The multi-target tracking device according to claim 1, wherein a similarity is calculated from a product of a value calculated by error covariance with the value, and the matching degree is obtained from the product of the similarity. A first observation value that represents the target position as an elevation angle, an azimuth angle, and a distance and a second observation value that represents the target position as an elevation angle and an azimuth angle are input, and the first observation value and the first observation value are input. A position information tracking processing unit that generates a hypothesis group including the target track using the observation values of 2,
The second observation value and identification information of the target possessed by the second observation value are input, and calculated from the second observation value and an angle wake including the second observation value observed in the past. was compared with the predicted position range of the target, the angle of the second observation values the input having the same identification information and the second observation value included in the angle track with lies within the predicted position range A target identification tracking processing unit that associates with a wake and generates a new angle wake;
Collating the hypothesis with the angle wake by comparing the second observation value included in the wake in the hypothesis included in the hypothesis group and the second observation value included in the wake in the angle wake A hypothesis matching unit that selects a hypothesis having the matching degree that exceeds a predetermined threshold from the hypothesis group, and deletes a hypothesis having the matching degree that does not exceed the predetermined threshold;
Identification information that increases the reliability of the hypothesis when the change in the identification information of the observation value correlated with the wake included in the hypothesis selected by the hypothesis matching unit is the same as the change in the identification information detected in the past A transition determination unit;
A hypothesis selection integration unit that reduces the hypothesis group by deleting hypotheses with low reliability from hypotheses consisting of hypotheses selected by the hypothesis matching unit and integrating similar hypotheses is provided. Multi-target tracking device. A first observation value that represents the target position as an elevation angle, an azimuth angle, and a distance and a second observation value that represents the target position as an elevation angle and an azimuth angle are input, and the first observation value and the first observation value are input. Generating a hypothesis group including the target track using the observation values of 2,
The second observation value and identification information of the target possessed by the second observation value are input, and calculated from the second observation value and an angle wake including the second observation value observed in the past. was compared with the predicted position range of the target, the angle of the second observation values the input having the same identification information and the second observation value included in the angle track with lies within the predicted position range Associating with a wake and generating a new angle wake;
Collating the hypothesis with the angle wake by comparing the second observation value included in the wake in the hypothesis included in the hypothesis group and the second observation value included in the wake in the angle wake Determining a degree, selecting a hypothesis having the matching degree that exceeds a predetermined threshold from the hypothesis group, and deleting a hypothesis having the matching degree that does not exceed the predetermined threshold;
Of the hypotheses made of pre-hexene-option hypotheses, the multi-target tracking method characterized by a step of reducing the hypotheses deletion of low reliability hypothesis, and the integration hypothesis similar. A first observation value that represents the target position as an elevation angle, an azimuth angle, and a distance and a second observation value that represents the target position as an elevation angle and an azimuth angle are input, and the first observation value and the first observation value are input. Generating a hypothesis group including the target track using the observation values of 2,
The second observation value and identification information of the target possessed by the second observation value are input, and calculated from the second observation value and an angle wake including the second observation value observed in the past. was compared with the predicted position range of the target, the angle of the second observation values the input having the same identification information and the second observation value included in the angle track with lies within the predicted position range Associating with a wake and generating a new angle wake;
Collating the hypothesis with the angle wake by comparing the second observation value included in the wake in the hypothesis included in the hypothesis group and the second observation value included in the wake in the angle wake Determining a degree, selecting a hypothesis having the matching degree that exceeds a predetermined threshold from the hypothesis group, and deleting a hypothesis having the matching degree that does not exceed the predetermined threshold;
A step change in the identity of the observations that correlate the track included in the pre-hexene-option hypotheses is, when the same change in the previously detected identification information, to increase the reliability of the hypothesis,
Of the hypotheses made of pre-hexene-option hypotheses, the multi-target tracking method characterized by a step of reducing the hypotheses deletion of low reliability hypothesis, and the integration hypothesis similar.

Images