JP2007292553A - Multiple target tracking system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To realize a multiple target tracking system capable of obtaining an accurately tracking result in the allowable processing time. <P>SOLUTION: The multiple target tracking system comprises an observation value selecting section 1 for determining a combination of observed values and temporary tracks used for performing predicting computation; a predicting section 22 for computing prediction of a temporary track, based on the observed value acquired time about the combination of observed values and temporary tracks determined by the observation value selecting section 1; a gate inside/outside determining section 23 for determining as to whether the temporary tracks can be associated with the observed values, by comparing the observed values with the predicted values of the temporary tracks computed by the predicting section 22; a track likelihood evaluating section 24 for computing the likelihood of the temporary track, when it is associated with the observed value; a correlation determining section 25 for determining the association of the temporary tracks and the observed values, by using the likelihood as an index which is computed by the track likelihood evaluating section 24; a smoothing section 26 for computing smoothing parameters, by using a predicted parameter of the temporary tracks and the observed value associated by the correlation determining section 25; and a track determining section 27 which determines a temporary track, having a high likelihood among target tracks. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  In a sensor system such as a search radar, a tracking start technique for generating a temporary track using a target observation value from a sensor that observes a target position within a certain area is required. The present invention relates to a multi-target tracking device that starts tracking a target track in a realistic processing time when there is a difference in observation time within a sensor search at the start of tracking.

  There are many papers, patents, and other literature already on the technology to start tracking a target using observation values obtained by searching a certain area with a sensor. Various devices and methods are available. Proposals have been made.

  In target tracking, it is determined whether or not it is possible to associate the temporary track with the observed value by comparing the predicted position of the temporary track and the observed position (whether there is a correlation). In the calculation of the predicted position of a normal tentative track, a calculation is performed by extrapolating the smoothing parameters at the latest time of the tentative track to the time of the observed value. In the case of observation by a search radar, the time at which an observation value is obtained differs depending on the time when the beam arrives even within one search as in the example of FIG. Because of this difference, the predicted time in the temporary track prediction process is generated by the number of observation values.

  A conventional multi-target tracking device in the case where tracking start processing is performed using such observation values will be described with reference to FIGS. FIG. 21 is a block diagram showing a configuration of a conventional multi-target tracking device. FIG. 22 is a flowchart showing the operation of the conventional multi-target tracking device. FIG. 23 is a diagram showing a search area in a conventional multi-target tracking device. FIG. 24 is a timing chart showing the relationship between the observation time of the conventional multi-target tracking device and the predicted time of the temporary track.

  In FIG. 21, a conventional multi-target tracking device 20 includes a temporary track file 21 that stores a plurality of temporary tracks including time and prediction specifications, a prediction processing unit 22, a gate inside / outside determination unit 23, and a track likelihood evaluation. A unit 24, a correlation determining unit 25, a smoothing processing unit 26, a track determining unit 27, and a target track file 28 for storing the target track are provided.

  In the “observation value input” step 900, when an observation value is input from the search radar (sensor) 10, first, in a “prediction process” step 901, the prediction processing unit 22 predicts the predicted position at the time of the observation value for the temporary track. And calculate the prediction error covariance.

  Next, in the “gate inside / outside determination” step 902, the gate inside / outside determination unit 23 uses the prediction error covariance of the temporary track and the observation error covariance of the observation value to determine the possibility that the temporary track and the observed value are associated with each other. Determined by chi-square test.

  Next, in the “wake likelihood evaluation” step 903, the wake likelihood evaluation unit 24 tentatively calculates an observation value based on the combination of the tentative track and the observation value determined to be in the gate in the previous step. Calculate the likelihood of a temporary wake when assigned to a wake.

  Next, in the “correlation determination” step 904, the correlation determination unit 25 determines which of the combination of the observed value and the provisional track is finally selected based on the likelihood calculated in the previous step.

  Next, in the “smoothing process” step 905, the smoothing processing unit 26 calculates the smooth specification using the observation value combined with the prediction specification of the temporary track based on the correlation result determined in the previous step. . That is, the temporary track time stored in the temporary track file 21 is updated to the latest time value.

  Then, in a “track determination” step 906, the track determination unit 27 determines whether or not to consider the temporary track as the target track based on the likelihood of the temporary track and the number of correlations with the observed value. The determined target track is stored in the target track file 28.

  In addition, another conventional target tracking device for solving the problem of an increase in calculation load caused by a time lag of observation values in one search assuming a search radar has been proposed (see, for example, Patent Document 1). . In this target tracking device, the observation region is divided in advance, and the tracking processing is performed for each divided region to reduce the calculation load.

JP 2004-233136 A

  In the conventional multi-target tracking device as described above, if an attempt is made to process observation values for one search with a time lag, the number of prediction processes required in the tracking start processing of one search is “the number of temporary tracks × the number of observation values”. It becomes. In the observation example shown in FIG. 23, a total of 25 prediction processes are required as in the timing chart shown in FIG. The calculation load resulting from the prediction process increases as the target number increases, making it difficult to start tracking within a realistic processing time.

  In addition, other conventional target tracking devices designed to reduce the processing time of the time lag observation values have had the problem that accurate tracking cannot be performed when the target track moves across the boundary of the divided areas. .

  The present invention has been made to solve the above-described problems, and an object thereof is to obtain a multi-target tracking device capable of obtaining an accurate tracking result in an allowable processing time.

  The multi-target tracking device according to the present invention is a multi-target tracking device that creates a target track using observation values obtained by searching an observation area with a sensor, and is a combination of a temporary track and an observation value for performing prediction calculation An observation value selection unit that determines the tentative value, a prediction processing unit that predicts and calculates the temporary track according to the time of the observation value for the combination of the temporary track and the observation value determined by the observation value selection unit, and the provisional track and the observation value A gate internal / external determination unit that determines whether the predicted value calculated by the prediction processing unit is compared with an observed value, and a track that calculates the likelihood of the temporary track when the observed value is associated A likelihood evaluation unit, a correlation determination unit that determines which temporary track is associated with which observation value, using the likelihood calculated by the track likelihood evaluation unit as an index, a prediction parameter of the temporary track, and the correlation determination unit Associated The observed values with the smoothing processing unit for calculating a smoothed specifications was one in which those likelihood of tentative track is high is provided a track determiner that the target track.

  The multi-target tracking device according to the present invention has an effect that an accurate tracking result can be obtained in an allowable processing time.

Embodiment 1 FIG.
A multi-target tracking device according to Embodiment 1 of the present invention will be described with reference to FIGS. FIG. 1 is a block diagram showing the configuration of the multitarget tracking apparatus according to Embodiment 1 of the present invention. In the following, in each figure, the same reference numerals indicate the same or corresponding parts.

  In FIG. 1, the multi-target tracking device 20 according to the first embodiment includes an observation value selection unit 1, a temporary track file 21 that stores a plurality of temporary tracks including time and prediction specifications, and a prediction processing unit 22. A gate inside / outside determination unit 23, a track likelihood evaluation unit 24, a correlation determination unit 25, a smoothing processing unit 26, a track determination unit 27, and a target track file 28 for storing the target track are provided.

  Next, the operation of the multi-target tracking device according to the first embodiment will be described with reference to the drawings. FIG. 2 is a flowchart showing the operation of the multitarget tracking apparatus according to Embodiment 1 of the present invention. FIG. 3 is a diagram showing a search area in the multi-target tracking device according to Embodiment 1 of the present invention. FIG. 4 is a timing chart showing the relationship between the observation time of the multi-target tracking device according to Embodiment 1 of the present invention and the predicted time of the temporary track.

  In the “observation value input” step 100, when an observation value is input from the search radar (sensor) 10, first, in the “observation value selection process” step 101, the observation value selection unit 1 performs a prediction process on the temporary track. Select observations to be performed. In this selection, it is determined whether or not all combinations of the temporary track and the observed value are candidates for correlation. The determination of a combination of a certain temporary track and a certain observed value is any one of the following four methods.

  (1-1) If the difference between the observation time of the observation value and the smoothing time in the previous search of the temporary track does not exceed the threshold value (first threshold value), the observation value is set as a candidate. That is, it is determined as a candidate when the following inequality (1) holds.

Here, ΔT is the difference between the observation time and the smooth time of the temporary track, T _search is the time required for the search, and th _T is a threshold value (first threshold value) set in advance. That is, the observation value selection unit 1 obtains ΔT from the observation time included in the input observation value and the smoothing time included in the temporary track read from the temporary track file 21. Further, from the obtained ΔT and T _search obtained from the search radar (sensor) 10, the difference between the observation time of the observed value and the smooth time in the previous search of the temporary track is obtained. Next, when the difference between the observation time of the obtained observation value and the smoothing time in the previous search of the temporary track is less than the threshold th _T , the observation value is selected.

  (1-2) As shown in FIG. 3, the difference between the time of the observed value and the smoothing time in the search before the temporary track does not exceed the allowable time difference obtained from the angular velocity of the search, the time required for the search, and the target maximum angular velocity. In this case, the observed value is a candidate. The judgment formula (2) in this case is as follows.

Here, ΔT is the difference between the observation time and the smooth time of the tentative track, T _search is the time required for the _search , v _{Az_max} is the target maximum angular velocity set in advance, and v _search is the angular velocity of the _search . That is, the observation value selection unit 1 obtains ΔT from the observation time included in the input observation value and the smoothing time included in the temporary track read from the temporary track file 21. Further, from the obtained ΔT and T _search obtained from the search radar (sensor) 10, the difference between the observation time of the observed value and the smooth time in the previous search of the temporary track is obtained. Further, the value on the right side of the equation (2) is obtained from the target maximum angular velocity and T _search and v _search obtained from the search radar (sensor) 10. Next, if the difference between the observed time of the observed value and the smoothing time in the previous search of the temporary track is less than the value on the right side of Equation (2), the observed value is selected.

  (1-3) In addition to the determination of (1-2) above, the distance difference between the observed value and the temporary track is also added to the determination. That is, the observation value is a candidate only when the following equation (3) is satisfied in addition to the equation (2).

Here, R _T is the distance from the sensor calculated from the position after T _search (time required for one search) from the smoothing time of the previous search of the tentative track, R _O is the distance from the sensor of the observed value, and th _R Is a threshold (second threshold) set in advance. That is, the observed value selecting unit 1 obtains a smoothing time of the search before included in the tentative track read from the tentative track file 21, the R _T and a T _search obtained from search radar (sensor) 10. Further, from the R _T obtained, the R _O contained in the observation value input, obtains the distance difference observations and tentative track. Next, when the difference between the observed time of the observed value and the smoothing time in the previous search for the temporary track is less than the value on the right side of Equation (2), and the distance difference between the observed value and the temporary track is the threshold th If it is less than _R , the observed value is selected.

(1-4) In the above-described determination formula (2), v _{Az_max} is not the preset parameter, but the maximum value of the speed of the temporary track calculated in the previous search. That is, the observation value selection unit 1 obtains the maximum value among the speeds included in the temporary track read from the temporary track file 21. Further, ΔT is obtained from the observation time included in the input observation value and the smoothing time included in the temporary track read from the temporary track file 21. Further, from the obtained ΔT and T _search obtained from the search radar (sensor) 10, the difference between the observation time of the observed value and the smooth time in the previous search of the temporary track is obtained. Further, the value on the right side of the equation (2) is obtained from the maximum value of the obtained tentative track speed and the T _search and v _search obtained from the search radar (sensor) 10. Next, if the difference between the observed time of the observed value and the smoothing time in the previous search of the temporary track is less than the value on the right side of Equation (2), the observed value is selected.

  Next, in the “prediction process” step 102, the prediction processing unit 22 calculates the predicted position and the prediction error covariance at the observation time of the observed value for the temporary track. However, prediction calculation is not performed for a certain tentative track and an observed value that is not determined as a candidate in the previous step for the tentative track.

  Next, in the “gate inside / outside determination” step 103, the gate inside / outside determination unit 23 uses the prediction error covariance of the temporary track and the observation error covariance of the observation value as a possibility of associating the temporary track with the observation value. Determined by chi-square test.

  Next, in the “wake likelihood evaluation” step 104, the wake likelihood evaluation unit 24 tentatively determines an observation value based on the combination of the temporary wake and the observation value determined to be in the gate in the previous step. Calculate the likelihood of a temporary wake when assigned to a wake.

  Next, in a “correlation determination” step 105, the correlation determination unit 25 determines which of the combination of the observed value and the provisional track is finally selected based on the likelihood calculated in the previous step.

  Next, in the “smoothing process” step 106, the smoothing processing unit 26 calculates the smooth specification using the observation value combined with the prediction specification of the temporary track based on the correlation result determined in the previous step. . That is, the temporary track time stored in the temporary track file 21 is updated to the latest time value.

  In “track determination” step 107, the track determination unit 27 determines whether or not to consider the temporary track as the target track based on the likelihood of the temporary track and the number of correlations with the observed value. The determined target track is stored in the target track file 28.

  When the multi-target tracking device according to the first embodiment is applied to the observation example shown in FIG. 23 (the first to fifth temporary tracks and the first to fifth observation values from the left in the figure). The first to third observation values for the first temporary track satisfy the above-described determination formula (2), and the second to fourth observation values for the second temporary track satisfy the above-described determination formula (2). The third to fourth observation values for the third temporary track satisfy the above-described determination formula (2), and the fourth to fifth observation values for the fourth temporary track satisfy the above-described determination formula (2). If the third to fourth observation values for the fifth temporary track satisfy the above-described determination formula (2), a total of twelve prediction calculations are required as shown in the time chart of FIG. This is less than the required number of prediction calculations of 25 in the conventional method.

  That is, the multi-target tracking device according to the first embodiment is a multi-target tracking device that creates a target wake using observation values obtained by searching an observation region with the sensor 10, and is a temporary target that performs prediction calculation. An observation value selection unit 1 that determines a combination of a wake and an observation value, and a prediction processing unit 22 that predicts and calculates the temporary track according to the time of the observation value for the combination of the tentative track and the observation value determined by the observation value selection unit 1. And the gate internal / external determination unit 23 that determines whether the temporary track and the observed value can be associated with each other by comparing the predicted value of the temporary track calculated by the prediction processing unit 22 with the observed value. The track likelihood evaluation unit 24 for calculating the likelihood of the temporary track in the case, and the correlation determination unit 25 for determining which temporary track and which observation value are associated with each other using the likelihood calculated by the track likelihood evaluation unit 24 as an index. When A smoothing processing unit 26 that calculates a smooth specification using the predicted value of the provisional wake and the observation value associated with the correlation determination unit 25, and a wake determination that uses a tentative wake with a high likelihood as a target wake. A portion 27 is provided.

  As described above, according to the multi-target tracking device according to the first embodiment, the preprocessing for selecting the observation value for which the prediction process is performed for each temporary track is performed, so that the processing load is reduced. Therefore, even when the target number is large, it can be dealt with.

Embodiment 2. FIG.
A multi-target tracking device according to Embodiment 2 of the present invention will be described with reference to FIGS. FIG. 5 is a block diagram showing the configuration of the multi-target tracking device according to Embodiment 2 of the present invention.

  In FIG. 5, the multi-target tracking device 20 according to the second embodiment includes an observation value grouping unit 2, a group center of gravity calculation unit 3, and a temporary track file that stores a plurality of temporary tracks including time and prediction specifications. 21, prediction processing unit 22, gate inside / outside determination unit 23, track likelihood evaluation unit 24, correlation determination unit 25, smoothing processing unit 26, track determination unit 27, and target track file that stores the target track 28 are provided.

  Next, the operation of the multi-target tracking device according to the second embodiment will be described with reference to the drawings. FIG. 6 is a flowchart showing the operation of the multitarget tracking apparatus according to Embodiment 2 of the present invention. FIG. 7 is a diagram showing a search area in the multi-target tracking device according to Embodiment 2 of the present invention. FIG. 8 is a diagram for explaining the same group determination of the observation value grouping unit of the multitarget tracking apparatus according to Embodiment 2 of the present invention. FIG. 9 is a timing chart showing the relationship between the observation time of the multi-target tracking device according to Embodiment 2 of the present invention and the predicted time of the temporary track.

  In the “observation value input” step 200, when an observation value is input from the search radar (sensor) 10, first, in the “observation value grouping” step 201, the observation value grouping unit 2 determines that observation values having a short distance are Group by. The same group determination for a combination of a certain observation value and another observation value is any one of the following three methods.

  (2-1) If an observation time difference between two observation values does not exceed a threshold value (third threshold value), the observation value is set as a candidate. That is, it is determined as a candidate when the following inequality (4) holds.

Here, ΔT is a difference between observation times of two observation values, and th _T is a threshold value (third threshold value) set in advance. That is, the observation value grouping unit 2 obtains the difference between the observation times of the two observation values from the observation times included in the input observation values. Next, when the difference between the observed times of the two observed values is less than the threshold th _T , these two observed values are determined as the same group.

  (2-2) As shown in FIG. 8, when the difference between the observation times of the two observation values does not exceed the allowable time difference obtained from the angular velocity of the search, the time required for the search, and the target maximum angular velocity, the observation value Is a candidate. The judgment formula (5) in this case is as follows.

Here, ΔT is the time difference between the two observation values, T _search is the time required for the _search , v _{Az_max} is the target maximum angular velocity set in advance, and v _search is the angular velocity of the _search . That is, the observation value grouping unit 2 obtains the difference between the observation times of the two observation values from the observation times included in the input observation values. Further, the right side of Expression (5) is obtained from the maximum angular velocity of the target and T _search and v _search obtained from the search radar (sensor) 10. Next, when the difference between the observation times of the two obtained observation values is less than the value on the right side of Equation (5), these two observation values are determined to be the same group.

  (2-3) In addition to the determination of (2-2) above, the distance difference between the two observation values is also added to the determination. That is, the observation value is a candidate only when the following equation (6) is satisfied in addition to the equation (5).

Here, ΔR is a distance difference between two observation values, and th _R is a threshold (fourth threshold) set in advance. That is, the observation value grouping unit 2 obtains the difference between the observation times of the two observation values from the observation times included in the input observation values. Further, the right side of Expression (5) is obtained from the maximum angular velocity of the target and T _search and v _search obtained from the search radar (sensor) 10. Further, the distance difference between the two observation values is obtained from the distance included in the input observation value. Next, when the difference in observation time between the two observed values is less than the value on the right side of Equation (5) and the distance difference between the two observed values is less than the threshold th _R These two observation values are determined as the same group.

  Next, in the “group centroid calculation” step 202, the group centroid calculation unit 3 calculates the observation time corresponding to the position of the centroid of the observation value group belonging to the group (hereinafter referred to as the centroid time). The group center-of-gravity calculation unit 3 obtains the position of the center of gravity of the observation value group belonging to the same group from the position and the observation time included in the input observation value by averaging, and the observation time corresponding to the position of the center of gravity, that is, Find the centroid time.

  Next, in the “prediction process” step 203, the prediction processing unit 22 calculates the predicted position and the prediction error covariance at the barycentric time of the group to which the observation value belongs for the temporary track.

  Next, in the “gate inside / outside determination” step 204, the gate inside / outside determination unit 23 uses the prediction error covariance of the temporary track and the observation error covariance of the observation value as a possibility of associating the temporary track with the observation value. Determined by chi-square test.

  Next, in the “track likelihood evaluation” step 205, the track likelihood evaluation unit 24, based on the combination of the temporary track and the observation value determined to be in the gate in the previous step, temporarily calculates the observation value. Calculate the likelihood of a temporary wake when assigned to a wake.

  Next, in a “correlation determination” step 206, the correlation determination unit 25 determines which of the combination of the observed value and the provisional track is finally selected based on the likelihood calculated in the previous step.

  Next, in the “smoothing process” step 207, the smoothing processing unit 26 calculates the smooth specification using the observation value combined with the prediction specification of the temporary track based on the correlation result determined in the previous step. . That is, the temporary track time stored in the temporary track file 21 is updated to the latest time value.

  Then, in a “track determination” step 208, the track determination unit 27 determines whether to consider the temporary track as the target track based on the likelihood of the temporary track and the number of correlations with the observed value. The determined target track is stored in the target track file 28.

  When the multi-target tracking device according to the second embodiment is applied to the observation example shown in FIG. 23, assuming that two groups can be formed as shown in FIG. 7, the number of provisional tracks ×× as shown in the time chart of FIG. A total of 10 prediction calculations for the number of groups (5 × 2) are required. This is less than the required number of prediction calculations of 25 in the conventional method.

  That is, the multi-target tracking device according to the second embodiment is a multi-target tracking device that creates a target wake using observation values obtained by searching an observation region with the sensor 10, and is obtained in one search. The observed value grouping unit 2 for grouping the observed value groups, the group centroid calculating unit 3 for calculating the centroid time which is the observation time corresponding to the centroid position of the group for each group, and the combination of the temporary track and the observed value The prediction processing unit 22 that predicts and calculates the temporary track based on the centroid time of the group to which the observation value belongs, and the prediction value of the temporary track that the prediction processing unit 22 calculates whether the temporary track and the observation value can be associated with each other. Gate inside / outside determination unit 23 that is determined by comparing observation values, track likelihood evaluation unit 24 that calculates the likelihood of a temporary track when the observation values are associated, and the track likelihood evaluation unit 2 The correlation determination unit 25 that determines which temporary track and which observation value are associated with each other using the likelihood calculated by the above as an index, and smoothing using the predicted values of the temporary track and the observation values associated with the correlation determination unit 25 A smoothing processing unit 26 that calculates specifications and a track determination unit 27 that uses a temporary track having a high likelihood as a target track are provided.

  As described above, according to the multi-target tracking device according to the second embodiment, the observation values are grouped and the prediction process for each temporary track is performed at the center of gravity time, so the processing load is reduced. Therefore, even when the target number is large, it can be dealt with.

Embodiment 3 FIG.
A multi-target tracking device according to Embodiment 3 of the present invention will be described with reference to FIGS. FIG. 10 is a block diagram showing the configuration of the multitarget tracking apparatus according to Embodiment 3 of the present invention.

  In FIG. 10, the multi-target tracking device 20 according to the third embodiment includes a centroid calculating unit 4, an error covariance expanding unit 5, and a temporary track file 21 that stores a plurality of temporary tracks including time and prediction specifications. A prediction processing unit 22, a gate inside / outside determination unit 23, a track likelihood evaluation unit 24, a correlation determination unit 25, a smoothing processing unit 26, a track determination unit 27, and a target track file 28 that stores a target track. And are provided.

  Next, the operation of the multi-target tracking device according to the third embodiment will be described with reference to the drawings. FIG. 11 is a flowchart showing the operation of the multitarget tracking apparatus according to Embodiment 3 of the present invention. FIG. 12 is a diagram showing search areas in the multitarget tracking apparatus according to Embodiment 3 of the present invention. FIG. 13 is a timing chart showing the relationship between the observation time of the multi-target tracking device according to Embodiment 3 of the present invention and the predicted time of the temporary track.

  When an observation value is input from the search radar (sensor) 10 in the “observation value input” step 300, first, in a “centroid calculation” step 301, the centroid calculation unit 4 determines the position of the centroid of all the obtained observation values. Is calculated (hereinafter referred to as the total centroid time). The center-of-gravity calculation unit 4 obtains the position of the center of gravity of all the observed values from the position and the observation time included in the input observation value by averaging, and obtains the observation time corresponding to the position of the center of gravity, that is, the total center-of-gravity time.

  Next, in the “prediction process” step 302, the prediction processing unit 22 calculates the predicted position and the prediction error covariance at the total center-of-gravity time for the temporary track.

  Next, in an “error covariance expansion” step 303, the error covariance expansion unit 5 expands the prediction error covariance matrix in consideration of the error due to the prediction processing performed at the total centroid time. The calculation for this enlargement is either:

(3-1) Assuming that the target maximum angular velocity, which is a preset parameter, is v _max , and the maximum value of the difference between the overall centroid time and the observation time of each observation value is Δt _max , the expansion of the prediction error covariance is Let v _max Δt _max . That is, the error covariance enlarging unit 5 obtains the maximum value of the difference between the total centroid time obtained in step 301 and the observation time of each observation value from the observation time included in the input observation value. Next, an enlargement of the prediction error covariance is obtained from the product of the target maximum angular velocity v _max and the maximum value Δt _max .

(3-2) Assuming that the maximum value of the speed of the temporary track is v _Tmax and the maximum value of the difference between the overall center of gravity time and the observation time of each observation value is Δt _max , the expanded portion of the prediction error covariance is v _Tmax Δt _max To do. That is, the error covariance enlarging unit 5 obtains the maximum value of the speed of the temporary track from the speed included in the temporary track read from the temporary track file 21. Next, an enlargement of the prediction error covariance is obtained from the product of the maximum value v _Tmax and the maximum value Δt _max of the speed of the temporary track.

  Next, in the “gate inside / outside determination” step 304, the gate inside / outside determination unit 23 uses the prediction error covariance of the temporary track and the observation error covariance of the observation value to determine whether the temporary track and the observed value are associated with each other. Determined by chi-square test.

  Next, in “Wake Likelihood Evaluation” step 305, the wake likelihood evaluation unit 24 tentatively determines an observation value based on the combination of the tentative track and the observation value determined to be in the gate in the previous step. Calculate the likelihood of a temporary wake when assigned to a wake.

  Next, in the “correlation determination” step 306, the correlation determination unit 25 determines which of the combination of the observed value and the provisional track is to be finally selected based on the likelihood calculated in the previous step.

  Next, in the “smoothing process” step 307, the smoothing processing unit 26 calculates the smooth specification using the observation value combined with the prediction specification of the temporary track based on the correlation result determined in the previous step. . That is, the temporary track time stored in the temporary track file 21 is updated to the latest time value.

  In a “wake determination” step 308, the wake determination unit 27 determines whether the temporary wake is regarded as the target wake based on the likelihood of the tentative wake and the number of correlations with the observed value. The determined target track is stored in the target track file 28.

  When the multi-target tracking device according to the third embodiment is applied to the observation example shown in FIG. 23, the error ellipse is enlarged as shown in FIG. 12, and the total number of tentative tracks is five times as shown in the time chart of FIG. Prediction calculation is required. This is less than the required number of prediction calculations of 25 in the conventional method.

  That is, the multi-target tracking device according to the third embodiment is a multi-target tracking device that creates a target wake using observation values obtained by searching an observation region with the sensor 10, and is obtained in one search. A center-of-gravity calculation unit 4 for calculating the total center-of-gravity time, which is an observation time corresponding to the position of the center of gravity of all observed values, and a prediction processing unit for predicting and calculating the temporary track based on the total center-of-gravity time for a combination of the temporary track and the observed value 22 and the error covariance expansion unit 5 that expands the prediction error covariance matrix in consideration of the error due to the prediction processing at the total center-of-gravity time, and whether the temporary track and the observation value can be associated with each other A gate inside / outside determination unit 23 that determines by comparing the predicted value of the temporary track calculated by the processing unit 22 and the observed value, and a track likelihood evaluation unit 24 that calculates the likelihood of the temporary track when the observed value is associated. When, Correlation determining unit 25 that determines which temporary track and which observation value are associated with each other using the likelihood calculated by recorded track likelihood evaluation unit 24 as an index, and the prediction parameters of temporary track and the correlation determining unit 25 are associated with each other. The smoothing processing unit 26 that calculates the smooth specification using the observed values, and the wake determination unit 27 that uses the tentative track having a high likelihood as the target track is provided.

  As described above, according to the multi-target tracking device according to the third embodiment, prediction processing is performed for each temporary track at the entire center of gravity time of the entire observation value, and further, the error of the covariance matrix generated by time approximation is compensated. Therefore, the processing load is reduced without degrading the tracking performance. Therefore, even when the target number is large, it can be dealt with.

Embodiment 4 FIG.
A multi-target tracking device according to Embodiment 4 of the present invention will be described with reference to FIGS. FIG. 14 is a block diagram showing the configuration of the multitarget tracking apparatus according to Embodiment 4 of the present invention.

  In FIG. 14, the multi-target tracking device 20 according to the fourth embodiment includes an observation value grouping unit 2, a group center-of-gravity calculation unit 3, an error covariance expansion unit 5, and a temporary track including time and prediction specifications. Tentative track file 21, a prediction processing unit 22, a gate inside / outside determination unit 23, a track likelihood evaluation unit 24, a correlation determination unit 25, a smoothing processing unit 26, a track determination unit 27, A target track file 28 for storing the target track is provided.

  Next, the operation of the multi-target tracking device according to the fourth embodiment will be described with reference to the drawings. FIG. 15 is a flowchart showing the operation of the multitarget tracking apparatus according to Embodiment 4 of the present invention. FIG. 16 is a diagram showing search areas in the multi-target tracking device according to Embodiment 4 of the present invention.

  When an observation value is input from the search radar (sensor) 10 in the “observation value input” step 400, first, in an “observation value grouping” step 401, the observation value grouping unit 2 determines that observation values having a short distance are Group by. About the same group determination about the combination of a certain observation value and another observation value, it is set as the same method as the step of the same name of the said Embodiment 2. FIG.

  Next, in a “group centroid calculation” step 402, the group centroid calculation unit 3 calculates an observation time corresponding to the position of the centroid of the observation value group belonging to the group (hereinafter referred to as the centroid time). The calculation method of the centroid time is the same as the step of the same name in the second embodiment.

  Next, in a “prediction process” step 403, the prediction processing unit 22 calculates the predicted position and the prediction error covariance at the centroid time of the group to which the observation value belongs for the temporary track.

  Next, in the “error covariance expansion” step 404, the error covariance expansion unit 5 expands the prediction error covariance matrix in consideration of the error due to the prediction processing at the centroid time. The enlargement is calculated in the same manner as the step of the same name in the third embodiment. However, the center of gravity time is defined as the center of gravity time.

  Next, in the “gate inside / outside determination” step 405, the gate inside / outside determination unit 23 uses the prediction error covariance of the temporary track and the observation error covariance of the observation value to determine whether the temporary track and the observed value are associated with each other. Determined by chi-square test.

  Next, in the “wake likelihood evaluation” step 406, the wake likelihood evaluation unit 24 tentatively determines an observation value based on the combination of the temporary track and the observation value determined to be in the gate in the previous step. Calculate the likelihood of a temporary wake when assigned to a wake.

  Next, in the “correlation determination” step 407, the correlation determination unit 25 determines which of the combination of the observed value and the provisional track is finally selected based on the likelihood calculated in the previous step.

  Next, in the “smoothing process” step 408, the smoothing processing unit 26 calculates the smooth specification using the observation value combined with the prediction specification of the temporary track based on the correlation result determined in the previous step. . That is, the temporary track time stored in the temporary track file 21 is updated to the latest time value.

  Then, in a “track determination” step 409, the track determination unit 27 determines whether or not to consider the temporary track as the target track based on the likelihood of the temporary track and the number of correlations with the observed value. The determined target track is stored in the target track file 28.

  When the multi-target tracking device according to the fourth embodiment is applied to the observation example shown in FIG. 23, if two observation value groups can be formed as shown in FIG. 16, the same as in FIG. 9 of the second embodiment described above. Thus, a total of 10 prediction calculations of the number of temporary tracks × the number of groups (5 × 2) are required. This is less than the required number of prediction calculations of 25 in the conventional method.

  That is, the multi-target tracking device according to the fourth embodiment is a multi-target tracking device that creates a target wake using observation values obtained by searching an observation region with the sensor 10, and is obtained in one search. The observed value grouping unit 2 for grouping the observed value groups, the group centroid calculating unit 3 for calculating the centroid time which is the observation time corresponding to the centroid position of the group for each group, and the combination of the temporary track and the observed value The prediction processing unit 22 that predicts and calculates the tentative track based on the centroid time of the group to which the observation value belongs, and the error co-variance matrix that expands the prediction error covariance matrix in consideration of the error caused by the prediction processing at the centroid time. In the gate, the dispersion enlarging unit 5 determines whether the temporary track and the observed value can be associated with each other by comparing the predicted value and the observed value of the temporary track calculated by the prediction processing unit 22 The determination unit 23, the track likelihood evaluation unit 24 for calculating the likelihood of the temporary track when the observation value is associated, and the temporary track and which of the temporary track, using the likelihood calculated by the track likelihood evaluation unit 24 as an index Correlation determining unit 25 for determining whether to associate observation values, smoothing processing unit 26 for calculating smooth specifications using predicted values of temporary track and observation values associated with correlation determining unit 25, Of these, a track determination unit 27 is provided which sets a target track having a high likelihood.

  As described above, according to the multi-target tracking device according to the fourth embodiment, the observation values are grouped, the prediction process is performed for each temporary track at the center of gravity time of the group, and the covariance matrix generated by time approximation is further calculated. Since the error is compensated, the processing load is reduced without degrading the tracking performance. Therefore, even when the target number is large, it can be dealt with.

Embodiment 5 FIG.
A multi-target tracking device according to Embodiment 5 of the present invention will be described with reference to FIGS. FIG. 17 is a block diagram showing a configuration of a multitarget tracking apparatus according to Embodiment 5 of the present invention.

  In FIG. 17, the multi-target tracking device 20 according to the fifth embodiment includes an observation value grouping unit 2, a group correlation calculation unit 6, and a wake group that stores tentative track group information describing a tentative track of the same group. Information file 7, track grouping unit 8, temporary track file 21 for storing a plurality of temporary tracks including time and prediction specifications, prediction processing unit 22, gate inside / outside determination unit 23, track likelihood evaluation unit 24, a correlation determining unit 25, a smoothing processing unit 26, a track determining unit 27, and a target track file 28 for storing the target track.

  Next, the operation of the multi-target tracking device according to the fifth embodiment will be described with reference to the drawings. FIG. 18 is a flowchart showing the operation of the multitarget tracking apparatus according to Embodiment 5 of the present invention. FIG. 19 is a diagram showing search areas in the multi-target tracking device according to Embodiment 5 of the present invention. FIG. 20 is a timing chart showing the relationship between the observation time of the multi-target tracking device according to Embodiment 5 of the present invention and the predicted time of the temporary track.

  When an observation value is input from the search radar (sensor) 10 in the “observation value input” step 500, first, in the “observation value grouping” step 501, the observation value grouping unit 2 determines that observation values having a short distance are Group by. About the same group determination about the combination of a certain observation value and another observation value, it is set as the same method as the step of the same name of the said Embodiment 2. FIG.

  Next, in a “group correlation process” step 502, the group correlation calculation unit 6 performs a correlation process between the observed value group obtained in the previous step and the temporary track group read from the track group information file 7. This is based on the distribution calculated by mixing the distribution of the observed value and the distribution due to the observation error covariance and the distribution calculated by mixing the distribution based on the temporary track position and the error covariance matrix. The presence or absence of correlation is determined by testing the hypothesis that “is the same”.

  Next, in the “prediction process” step 503, the prediction processing unit 22 determines the predicted position of the temporary track and the prediction error at the observation time for the combination of the temporary track and the observed value of the group determined to be correlated in the previous step. Calculate the variance.

  Next, in the “gate inside / outside determination” step 504, the gate inside / outside determination unit 23 uses the prediction error covariance of the temporary track and the observation error covariance of the observation value as a possibility of associating the temporary track with the observation value. Determined by chi-square test.

  Next, in the “wake likelihood evaluation” step 505, the wake likelihood evaluation unit 24 tentatively calculates an observation value for the combination of the temporary track and the observation value determined to be in the gate in the previous step. Calculate the likelihood of a temporary wake when assigned to a wake.

  Next, in the “correlation determination” step 506, the correlation determination unit 25 determines which of the combination of the observed value and the provisional track is finally selected based on the likelihood calculated in the previous step.

  Next, in a “smoothing process” step 507, the smoothing processing unit 26 calculates a smooth specification using the observation value combined with the predicted specification of the temporary track based on the correlation result determined in the previous step. . That is, the temporary track time stored in the temporary track file 21 is updated to the latest time value.

  Next, in a “track grouping” step 508, the track grouping unit 8 groups the temporary tracks according to the group to which the observation value used for the smoothing process belongs. Then, it is stored in the wake group information file 7 as temporary wake group information.

  In a “wake determination” step 509, it is determined whether to consider the temporary track as the target track based on the likelihood of the temporary track and the number of correlations with the observed value. The determined target track is stored in the target track file 28.

  When the multi-target tracking device according to the fifth embodiment is applied to the observation example shown in FIG. 23, there are two observation value groups and two temporary track groups as shown in FIG. Assume that wake group 1 has a correlation, and observation value group 2 and wake group 2 have a correlation. At this time, as shown in the time chart of FIG. 20, in order to perform the prediction process within the correlated group, the number of temporary tracks for group 1 × the number of observed values (3 × 3) is 9 times, and the number of temporary tracks for group 2 X Prediction calculation of 4 times of the number of observed values (2 x 2) is required, and the total number of prediction calculations is 13 (= 9 + 4). This is less than the required number of prediction calculations of 25 in the conventional method.

  That is, the multi-target tracking device according to the fifth embodiment is a multi-target tracking device that creates a target wake using observation values obtained by searching an observation region with the sensor 10, and is obtained in one search. Observation group grouping unit 2 for grouping observation groups, group correlation calculation unit 6 for performing correlation processing between the observation group and the temporary track group, and observing the temporary track for the combination of the temporary track and the observation value A prediction processing unit 22 that performs prediction calculation according to the observation time of the value, and a gate that determines whether the temporary track and the observation value can be associated by comparing the predicted value of the temporary track calculated by the prediction processing unit 22 with the observation value The inside / outside determination unit 23, the track likelihood evaluation unit 24 for calculating the likelihood of the temporary track when the observation values are associated, and the temporary track with which the likelihood calculated by the track likelihood evaluation unit 24 is used as an index Which A correlation determining unit 25 for determining whether to associate the measurement values, a smoothing processing unit 26 for calculating a smooth specification using the predicted values of the temporary track and the observation values associated with the correlation determining unit 25, and smoothing processing According to the group to which the observation value used belongs, there is provided a wake grouping unit 8 for grouping the tentative track, and a track determination unit 27 for setting a target track having a high likelihood among the tentative tracks.

  As described above, according to the multi-target tracking device according to the fifth embodiment, the observation value and the temporary track are grouped and the correlation between the groups is performed, and then the temporary track and the observation value in the correlated group are correlated. Since the prediction process is performed on the set, the processing load is reduced. Therefore, even when the target number is large, it can be dealt with.

It is a block diagram which shows the structure of the multi-target tracking apparatus which concerns on Embodiment 1 of this invention. It is a flowchart which shows operation | movement of the multi target tracking apparatus which concerns on Embodiment 1 of this invention. It is a figure which shows the search area | region in the multi-target tracking apparatus which concerns on Embodiment 1 of this invention. It is a timing chart which shows the relationship between the observation time of the multitarget tracking device concerning Embodiment 1 of this invention, and the prediction time of a temporary track. It is a block diagram which shows the structure of the multi-target tracking apparatus which concerns on Embodiment 2 of this invention. It is a flowchart which shows operation | movement of the multi-target tracking apparatus which concerns on Embodiment 2 of this invention. It is a figure which shows the search area | region in the multi-target tracking apparatus which concerns on Embodiment 2 of this invention. It is a figure for demonstrating the same group determination of the observation value grouping part of the multitarget tracking device which concerns on Embodiment 2 of this invention. It is a timing chart which shows the relationship between the observation time of the multitarget tracking device concerning Embodiment 2 of this invention, and the prediction time of a temporary track. It is a block diagram which shows the structure of the multi-target tracking apparatus which concerns on Embodiment 3 of this invention. It is a flowchart which shows operation | movement of the multi-target tracking apparatus which concerns on Embodiment 3 of this invention. It is a figure which shows the search area | region in the multi-target tracking apparatus which concerns on Embodiment 3 of this invention. It is a timing chart which shows the relationship between the observation time of the multitarget tracking device concerning Embodiment 3 of this invention, and the prediction time of a temporary track. It is a block diagram which shows the structure of the multi-target tracking apparatus which concerns on Embodiment 4 of this invention. It is a flowchart which shows operation | movement of the multi-target tracking apparatus which concerns on Embodiment 4 of this invention. It is a figure which shows the search area | region in the multi-target tracking apparatus which concerns on Embodiment 4 of this invention. It is a block diagram which shows the structure of the multi-target tracking apparatus which concerns on Embodiment 5 of this invention. It is a flowchart which shows operation | movement of the multitarget tracking device which concerns on Embodiment 5 of this invention. It is a figure which shows the search area | region in the multi-target tracking apparatus which concerns on Embodiment 5 of this invention. It is a timing chart which shows the relationship between the observation time of the multitarget tracking device concerning Embodiment 5 of this invention, and the prediction time of a temporary track. It is a block diagram which shows the structure of the conventional multi-target tracking apparatus. It is a flowchart which shows operation | movement of the conventional multi-target tracking apparatus. It is a figure which shows the search area | region in the conventional multi-target tracking apparatus. It is a timing chart which shows the relationship between the observation time of the conventional multi-target tracking apparatus and the prediction time of a temporary track.

Explanation of symbols

  1 observation value selection unit, 2 observation value grouping unit, 3 group centroid calculation unit, 4 centroid calculation unit, 5 error covariance expansion unit, 6 group correlation calculation unit, 7 track group information file, 8 track grouping unit, 10 Search radar (sensor), 20 multi-target tracking device, 21 temporary track file, 22 prediction processing unit, 23 gate inside / outside determination unit, 24 track likelihood evaluation unit, 25 correlation determination unit, 26 smoothing processing unit, 27 track determination unit, 28 Target track file.

Claims (16)

A multi-target tracking device that creates a target wake using observations obtained by searching an observation area with a sensor,
An observation value selector for determining a combination of the temporary track and the observation value for which the prediction calculation is performed;
For the combination of the temporary track and the observation value determined by the observation value selection unit, a prediction processing unit that predicts and calculates the temporary track according to the time of the observation value;
A gate inside / outside determination unit that determines whether the temporary track and the observed value can be associated with each other by comparing the predicted value and the observed value of the temporary track calculated by the prediction processing unit,
A wake likelihood evaluation unit that calculates the likelihood of a tentative wake when an observation value is associated,
Using the likelihood calculated by the track likelihood evaluation unit as an index, a correlation determination unit that determines which temporary track and which observation value are associated with each other,
A smoothing processing unit for calculating a smoothing specification using the predicted values of the provisional wake and the observation values associated with the correlation determination unit;
A multi-target tracking device, comprising: a track determination unit that sets a target track with a high likelihood among temporary tracks. The observation value selection unit is characterized in that the observation value is a candidate when the difference between the time of the observation value and the smoothing time in the previous search of the temporary track does not exceed the first threshold value set in advance. The multi-target tracking device according to claim 1. The observed value selection unit determines the observed value when the difference between the time of the observed value and the smoothing time in the preliminary search before the temporary track does not exceed the allowable time difference obtained from the angular velocity of the search, the time required for the search, and the target maximum angular velocity. The multi-target tracking device according to claim 1, wherein The observation value selection unit is the case where the difference between the time of the observation value and the smoothing time in the previous search of the temporary track does not exceed the allowable time difference obtained from the angular velocity of the search and the time required for the search and the target maximum angular velocity, and The multi-target tracking device according to claim 1, wherein the observation value is set as a candidate when a difference in distance between the observation value and the provisional track does not exceed a preset second threshold value. 5. The multi-target tracking device according to claim 3, wherein the maximum angular velocity of the target is a maximum value among the angular velocities of the tentative track calculated in the previous search. A multi-target tracking device that creates a target wake using observations obtained by searching an observation area with a sensor,
An observation value grouping unit for grouping observation value groups obtained in one search;
A group centroid calculation unit for calculating a centroid time which is an observation time corresponding to the centroid position of the group for each group;
About the combination of the temporary track and the observed value, a prediction processing unit that predicts and calculates the temporary track based on the centroid time of the group to which the observed value belongs,
A gate inside / outside determination unit that determines whether the temporary track and the observed value can be associated with each other by comparing the predicted value and the observed value of the temporary track calculated by the prediction processing unit,
A wake likelihood evaluation unit that calculates the likelihood of a tentative wake when an observation value is associated,
Using the likelihood calculated by the track likelihood evaluation unit as an index, a correlation determination unit that determines which temporary track and which observation value are associated with each other,
A smoothing processing unit for calculating a smoothing specification using the predicted values of the provisional wake and the observation values associated with the correlation determination unit;
A multi-target tracking device, comprising: a track determination unit that sets a target track with a high likelihood among temporary tracks. A multi-target tracking device that creates a target wake using observations obtained by searching an observation area with a sensor,
A center-of-gravity calculation unit that calculates an overall center-of-gravity time that is an observation time corresponding to the position of the center of gravity of all observation values obtained in one search;
For a combination of the temporary track and the observed value, a prediction processing unit that predicts and calculates the temporary track based on the total centroid time,
An error covariance expansion unit that expands the prediction error covariance matrix in consideration of the error due to the prediction processing at the overall centroid time,
A gate inside / outside determination unit that determines whether the temporary track and the observed value can be associated with each other by comparing the predicted value and the observed value of the temporary track calculated by the prediction processing unit,
A wake likelihood evaluation unit that calculates the likelihood of a tentative wake when an observation value is associated,
Using the likelihood calculated by the track likelihood evaluation unit as an index, a correlation determination unit that determines which temporary track and which observation value are associated with each other,
A smoothing processing unit for calculating a smoothing specification using the predicted values of the provisional wake and the observation values associated with the correlation determination unit;
A multi-target tracking device, comprising: a track determination unit that sets a target track with a high likelihood among temporary tracks. A multi-target tracking device that creates a target wake using observations obtained by searching an observation area with a sensor,
An observation value grouping unit for grouping observation value groups obtained in one search;
A group centroid calculation unit for calculating a centroid time which is an observation time corresponding to the centroid position of the group for each group;
About the combination of the temporary track and the observed value, a prediction processing unit that predicts and calculates the temporary track based on the centroid time of the group to which the observed value belongs,
In consideration of the error due to the prediction processing at the centroid time, an error covariance expansion unit that expands the prediction error covariance matrix;
A gate inside / outside determination unit that determines whether the temporary track and the observed value can be associated with each other by comparing the predicted value and the observed value of the temporary track calculated by the prediction processing unit,
A wake likelihood evaluation unit that calculates the likelihood of a tentative wake when an observation value is associated,
Using the likelihood calculated by the track likelihood evaluation unit as an index, a correlation determination unit that determines which temporary track and which observation value are associated with each other,
A smoothing processing unit for calculating a smoothing specification using the predicted values of the provisional wake and the observation values associated with the correlation determination unit;
A multi-target tracking device, comprising: a track determination unit that sets a target track with a high likelihood among temporary tracks. A multi-target tracking device that creates a target wake using observations obtained by searching an observation area with a sensor,
An observation value grouping unit for grouping observation value groups obtained in one search;
A group correlation calculation unit that performs correlation processing between the observation value group and the temporary track group;
A prediction processing unit that predicts and calculates the temporary track based on the observation time of the observation value for the combination of the temporary track and the observation value;
A gate inside / outside determination unit that determines whether the temporary track and the observed value can be associated with each other by comparing the predicted value and the observed value of the temporary track calculated by the prediction processing unit,
A wake likelihood evaluation unit that calculates the likelihood of a tentative wake when an observation value is associated,
Using the likelihood calculated by the track likelihood evaluation unit as an index, a correlation determination unit that determines which temporary track and which observation value are associated with each other,
A smoothing processing unit for calculating a smoothing specification using the predicted values of the provisional wake and the observation values associated with the correlation determination unit;
A track grouping unit for grouping temporary tracks according to the group to which the observation values used for smoothing belong,
A multi-target tracking device, comprising: a track determination unit that sets a target track with a high likelihood among temporary tracks. The observation value grouping unit makes the two observation values the same group when a difference between observation times of the two observation values does not exceed a preset third threshold value. The multi-target tracking device according to 6, 8, or 9. The observation value grouping unit assigns the two observation values to the same group when the difference between the observation times of the two observation values does not exceed the allowable time difference obtained from the angular velocity of the search, the time required for the search, and the target maximum angular velocity. The multi-target tracking device according to claim 6, 8 or 9. The observation value grouping unit is configured so that the difference between the observation times of the two observation values does not exceed the allowable time difference obtained from the angular velocity of the search, the time required for the search, and the target maximum angular velocity, and the two observation values The multi-target tracking device according to claim 6, wherein the two observation values are set to be in the same group when the difference in distance does not exceed a preset fourth threshold value. The error covariance enlarging unit sets the product of the target maximum angular velocity set in advance and the maximum value of the difference between the overall centroid time and each observed value as an enlargement of the prediction error covariance matrix. The multi-target tracking device according to claim 7, wherein: The error covariance enlarging unit is characterized in that a product of the maximum value of the speed of the tentative track and the maximum value of the difference between the total centroid time and the time of each observation value is an expansion of the prediction error covariance matrix. The multi-target tracking device according to claim 7. The error covariance expansion unit uses a product of a target maximum angular velocity set in advance and the maximum value of the difference between the center of gravity time and each observation value as an expansion of the prediction error covariance matrix. The multi-target tracking device according to claim 8. The error covariance expansion unit is characterized in that a product of the maximum value of the speed of the tentative track and the maximum value of the difference between the center of gravity time and the time of each observation value is an expansion of the prediction error covariance matrix. The multi-target tracking device according to claim 8.

Images