KR101833238B1 - System and method for tracking target using asynchronous sensors - Google Patents

Abstract

The present invention proposes a target tracking system and method using asynchronous sensors that match local track information in each sensor in the asynchronous sensor environment to the same time zone and then perform distributed fusion of information to track the target. The system according to the present invention includes detection information acquisition units for acquiring detection information on a target using sensors having different sampling periods; Track generating means for generating tracks indicative of a movement trajectory of the target based on the detection information; A track for correcting a second track associated with second detection information obtained by at least one other sensor of the sensors, excluding the reference sensor, based on time information when the reference sensor selected from the sensors acquires the first detection information, Complementary government; And a target tracking unit for tracking the target based on the first track and the corrected second track associated with the first detection information.

Description

[0001] The present invention relates to an asynchronous sensor,

The present invention relates to a system and method for tracking a target. More particularly, the present invention relates to a system and method for tracking a target using a plurality of sensors.

Due to the limitations of single sensor such as low detection probability and high clutter, the research on target tracking method in multi - sensor environment is continuously increasing.

However, in the automatic track tracking method proposed in the related art, fusion between local tracks is performed under the assumption of a synchronous environment in which the target information is detected at the same time with the sampling periods of the respective sensors.

However, since the actual multi-sensor system has a different sampling period and is often detected asynchronously, it is somewhat difficult to apply the conventional automatic target tracking method to an actual system.

Korean Laid-Open Patent Application No. 2012-0137314 proposes a device for tracking a target. However, this device proposes a method of tracking a moving target using an image, and can not solve the above problem because it does not take into consideration multiple sensors, in particular, multiple sensors operating in an asynchronous environment.

SUMMARY OF THE INVENTION The present invention has been made in order to solve the above-mentioned problems, and it is an object of the present invention to provide an asynchronous sensor system, Tracking system and method.

However, the objects of the present invention are not limited to those mentioned above, and other objects not mentioned can be clearly understood by those skilled in the art from the following description.

In order to achieve the above object, the present invention provides detection information acquisition units for acquiring detection information on a target using sensors having different sampling periods. Track generating units for generating tracks for displaying the movement trajectory of the target based on the detection information; A second track related to the second detection information obtained by at least one other sensor other than the reference sensor among the sensors based on time information when the reference sensor selected from the sensors acquires the first detection information, A track corrector for correcting the track; And a target tracking unit for tracking the target based on a first track and a corrected second track associated with the first detection information.

Preferably, the track generators include: a track information predictor for predicting information on the tracks using a state transition matrix; A track information reforming unit for reforming information on the tracks based on the detection information; And a final trajectory extractor for generating final trajectories selected from temporary trajectories of the target on the basis of the information on the tracks.

Preferably, the track generators include: a trajectory association determination unit that determines whether or not there are similar trajectories in which the degree of similarity is equal to or greater than a reference value among the final trajectories; And a final trajectory merging unit for merging the similar trajectories to generate a single track if the similar trajectories are determined to exist.

Preferably, the track information predicting unit uses an integrated probabilistic data association (IPDA) filter to estimate an estimated value of a state variable related to the target, an error covariance associated with estimation of the state variable, Predicts the probability that the target exists.

Preferably, the track information predicting unit predicts an estimated value of the state variable at a current time based on the state transition matrix and the estimated value of the state variable renewed at a previous time, and updates the state transition matrix, Predicting the error covariance at the current time based on the process noise covariance associated with the error covariance and dynamics modeling, and estimating the error probability of the target at the previous time and the current time, Predicts the probability that the target for the track exists at the current time based on a second probability that the target exists for the track and a third probability that the target will be present at the previous time.

Preferably, the track information predicting unit may calculate a first calculation value by multiplying the third probability by the first probability, multiplying a value obtained by subtracting the third probability from 1 by the second probability, and calculating a second calculation value And adds the first calculation value and the second calculation value to predict a probability that the target exists for the track in the current time.

Preferably, the track information predicting unit predicts a value obtained by estimating the position of the target with the estimated value of the state variable and a value obtained by estimating the speed of the target.

Preferably, the final trajectory extractor uses the probability that the target exists for a corresponding track when the final trajectories are selected from among the temporary trajectories.

Preferably, the detection information obtaining units obtain the distance to the target and the orientation of the target with each detection information.

Advantageously, the track compensator uses the kinetic modeling to correct the second track such that the second detection information is obtained at the same time as the first detection information.

Preferably, the track corrector corrects the second track by correcting information on the second track, and estimates the state variable associated with the target, the state variable, The associated error covariance, and the probability that the target exists for that track.

Preferably, the track compensator may include a first matrix including a time difference between a time when the reference sensor acquires the first detection information and a time when the other sensor acquires the second detection information as a component, And the second matrix obtained from the Kronecker product between the unitary matrix.

Preferably, the track compensating unit may include a second matrix, a transposed matrix of the second matrix, a third matrix including a value obtained by dividing the time difference by the time difference by 2, And corrects the error covariance based on the transpose matrix.

Preferably, the target tracking unit tracks the target based on a fusion track obtained by fusing the first track and the second track.

The present invention also provides a method of detecting a target, the method comprising: acquiring detection information on a target using sensors having different sampling periods; Generating tracks that display the movement trajectory of the target based on the detection information; A second track related to the second detection information obtained by at least one other sensor other than the reference sensor among the sensors based on time information when the reference sensor selected from the sensors acquires the first detection information, Correcting; And tracking the target based on the first track and the corrected second track associated with the first detection information. The present invention also provides a method of tracking a target using asynchronous sensors.

Advantageously, the step of generating the tracks comprises: predicting information on the tracks using a state transition matrix; Renewing information on the tracks based on the detection information; And generating the final trajectories selected from the temporary trajectories of the target as the tracks based on the information about the tracks.

Preferably, the step of generating the tracks may include: determining whether there are similar trajectories among the final trajectories, the similar trajectories having a similarity value higher than a reference value, after the step of generating the final trajectories as the tracks; And merging the similar trajectories into a single track if the similar trajectories are determined to exist.

Advantageously, the step of predicting comprises using an IPDA (Integrated Probabilistic Data Association) filter to estimate an estimate of the state variable associated with the target, an error covariance associated with the estimation of the state variable, Predicts the probability that the target exists.

Preferably, the predicting step predicts an estimated value of the state variable at the current time based on the state transition matrix and the estimated value of the state variable renewed at a previous time, and updates the state transition matrix, Predicting the error covariance at the current time based on the process noise covariance associated with the error covariance and dynamics modeling, and estimating the error covariance at the current time with the first probability that the target exists in the track, Predicts a probability that the target exists for the track at the current time based on a second probability of the target being present on the track and a third probability of the target being renewed at the previous time.

Preferably, the predicting step may include calculating a first calculation value by multiplying the third probability by the first probability, multiplying a value obtained by subtracting the third probability from 1 and the second probability to calculate a second calculation value And adds the first calculation value and the second calculation value to predict a probability that the target exists for the track in the current time.

Preferably, the predicting step predicts a value obtained by estimating the position of the target with the estimated value of the state variable and a value obtained by estimating the speed of the target.

Advantageously, the step of generating the final trajectories as the tracks utilizes the probability that the target exists for the track as information about each track when selecting the final trajectory from the temporary trajectories.

Advantageously, said acquiring acquires the distance to said target and the orientation of said target with each piece of detection information.

Advantageously, said correcting step corrects said second track such that said second detection information is obtained at the same time as said first detection information using kinetic modeling.

Advantageously, said correcting step corrects said second track by correcting information about said second track, and estimates the state variable associated with said target with said information about said second track, And the probability that the target for the track is present.

Preferably, the correcting step may include a first time interval including a time difference between a time when the reference sensor acquires the first detection information and a time when the other sensor acquires the second detection information, And corrects the estimate of the state variable based on the second matrix obtained from the Kronecker product between the matrix and the unit matrix.

Advantageously, the correcting step comprises: a second matrix, a transposed matrix of the second matrix, a third matrix containing as a component a value obtained by dividing the time difference by the time difference by 2, The error covariance is corrected based on the transposed matrix of the error covariance matrix.

Advantageously, said tracking tracks said target based on a fusion track obtained by fusing said first track and said second track.

The present invention can achieve the following effects through the above-described configurations.

First, flexible information fusion is possible in a multi-sensor environment with asynchronous multiple sampling periods.

Second, the accuracy of the fusion track can be improved by correcting the confirmed local track of the transmitted sensor to fit the fusion track.

1 is a conceptual diagram of an automatic target tracking system according to an embodiment of the present invention.
2 is an internal block diagram of a local track processing unit interlocked with each sensor in the automatic target tracking system of FIG.
3 is an internal configuration diagram of a track fusion unit constituting the automatic target tracking system of FIG.
4 is a conceptual diagram of a target tracking system according to a preferred embodiment of the present invention.
FIG. 5 is an internal configuration diagram of a track generation unit constituting the target tracking system of FIG. 4;
6 is a flowchart schematically showing a target tracking method according to a preferred embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the drawings, the same reference numerals are used to designate the same or similar components throughout the drawings. In the following description of the present invention, a detailed description of known functions and configurations incorporated herein will be omitted when it may make the subject matter of the present invention rather unclear. In addition, the preferred embodiments of the present invention will be described below, but it is needless to say that the technical idea of the present invention is not limited thereto and can be variously modified by those skilled in the art.

Automatic target tracking and information fusion in a synchronous multiple sensor environment are applied to a synchronous sensor environment and operate in the following order.

First, local tracking is performed using the information (measured values) detected in each single sensor, and a track management technique for automatic target tracking is applied in local tracking. And then transmits the confirmed local tracks to the fusion center. It then fuses the determined local tracks in each single sensor and calculates the fused track score for track evaluation. It then stores this information and uses it for fusion track management.

However, in the case of applying the above technique to a multi-sensor environment, there is a limitation that the transmitted definite local tracks need to be synchronized with each other, and there is also a problem that synchronization of sensors in a real multi-sensor system is difficult.

The limitations and problems of the above-described technique can be solved by the present invention. BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, the present invention will be described in detail with reference to the drawings.

1 is a conceptual diagram of an automatic target tracking system according to an embodiment of the present invention.

In actual multi-sensor systems, asynchronous multi-sampling cycle environment is often used. In this environment, there is not proposed a method of automatically tracking a target using detection information. The automatic target tracking system 100 proposed by the present invention is intended to solve such a problem and automatically tracks a target in a multi-sensor environment having an asynchronous multiple sampling period and fuses the information accordingly.

The automatic target tracking system 100 is required to match the determined local track information in each sensor to the same time zone in order to perform distributed information fusion (track fusion) in the asynchronous sensor environment. The automatic target tracking system 100 attempts to solve the existing information fusion problem in the asynchronous sensor environment by correcting the confirmed local track information using the dynamic model.

Referring to Figure 1, an automatic target tracking system 100 includes a first sensor 110, a second sensor 150, a first local track processing unit 120, a second local track processing unit 160, A correction unit 130 and a track fusion unit 140. [

The first sensor 110 and the second sensor 150 perform the function of acquiring detection information on the target. The first sensor 110 and the second sensor 150 operate in the asynchronous environment 170 in this embodiment. That is, the first sensor 110 and the second sensor 150 have different sampling periods.

Detection information obtained by the first sensor 110 and the second sensor 150 may be composed of two-dimensional distance and orientation with respect to the target. This can be expressed as follows.

Z _k = [? _K ,? _K ] ^T

In the above, Z _k denotes a detection information vector obtained from the sensor at scan k. γ _k denotes distance information in the detection information vector, and θ _k denotes azimuth information in the detection information vector.

In FIG. 1, two sensors are provided for convenience. However, in the present invention, three or more sensors may be provided. Of course, when three or more sensors are provided, a local track processing unit is added correspondingly.

The first local track processing unit 120 and the second local track processing unit 160 generate the confirmed local track based on the detection information obtained by the first sensor 110 and the second sensor 150 that are interlocked Function. In the present embodiment, a track means a trajectory in which an object regarded as a target moves.

The first local track processing unit 120 and the second local track processing unit 160 may have the same internal structure to perform the above functions. Hereinafter, the internal structure of the first local track processing unit 120 will be described. However, it goes without saying that the internal structure is applied to the second local track processing unit 160 as well.

2 is an internal block diagram of a local track processing unit interlocked with each sensor in the automatic target tracking system of FIG.

2, the first local track processing unit 120 includes a track initialization module 121, a prediction module 122, a modification module 123, a track removal module 124, a track determination module 125, Module 126. < / RTI >

The track initialization module 121 performs the function of initializing the available tracks using the detection information obtained by the first sensor 110. [

The prediction module 122 predicts track information using a state transition matrix. The track information predicted by the prediction module 122 includes state variable estimates (position, velocity), error covariance, and track existence probability.

The prediction module 122 may estimate a state variable estimate, an error covariance, a track presence probability, etc. using an algorithm based on track presence probability. The prediction module 122 may use an integrated probabilistic data association (IPDA) filter, for example, as a track existence probability based algorithm.

The prediction module 122 may estimate a state variable estimate, an error covariance, a track existence probability, etc. using the following equation (1).

Referring to Equation (1), the prediction module (122) can predict a state variable estimate using Equation (1). In Equation (1), 401 denotes a state variable estimation value predicted from time t _{i -1} (previous time) to t _i (current time). F means a state variable transition matrix (4 × 4 matrix). 402 denotes a state variable estimate renewed at time t _{i -1} .

The prediction module 122 can predict the error covariance using Equation (1) (b). In (b) of equation (1), 403 denotes a state variable error covariance predicted from time t _{i -1} to t _i . 404 denotes the covaried state variable error covariance at time t _{i -1} . Q is the process noise covariance.

The prediction module 122 can predict the track existence probability using Equation (1) (c). In (c) of equation (1), 405 denotes an event in which a target exists at time t _i . Reference numeral 406 denotes a pre-track existence probability at time t _i . p ₁₁ is the probability that the target was present at time t _{i -1} and that the target exists at time t _i . And 407 denotes a probability of a post track existence at time t _{i -1} . p ₁₂ means the probability that the target does not exist at time t _{i -1} , but the target exists at time t _i .

Meanwhile, in the present embodiment, the state variable can be expressed by a vector composed of two-dimensional position and speed with respect to the target. The state variable vector can be expressed as shown in equation (2).

Where X _k denotes the state variable vector of the target at scan k. x _k denotes x-axis position information among state variables, and y _k denotes y-axis position information among state variables. 408 denotes x-axis speed information among state variables, and 409 denotes y-axis speed information among state variables.

The modification module 123 performs a function of replacing the track information using the detection information in the current scan. In FIG. 2, the information input to the modification module 123 indicates detection information obtained by the first sensor 110.

The modification module 123 can rewrite the state variable estimation value, the error covariance, and the like among the track information using Equation (3).

Referring to Equation (3), the modification module (123) can rewrite the state variable estimate using Equation (3). In Equation (3) (a), 410 denotes a state variable estimate renewed at time t _i . 411 denotes probability that the kth measurement value at time t _i is the target information. And 412 denotes a state variable estimate renewed from the time t _i to the k-th measurement value.

The modification module 123 can rewrite the error covariance using Equation (3) (b). In (b) of equation (3), 413 denotes the covaried state variable error covariance at time t _i . 414 denotes a state variable estimation error covariance reconstructed from the time t _i to the k-th measurement value.

Equation (4) below represents the condition when the transformation module 123 uses the equation (3) to reform the state variable estimation value and the error covariance.

Where 415 denotes the k-th measurement at time t _i . Reference numeral 416 denotes a measurement equation. H denotes a measurement matrix for linearizing a nonlinear measurement model. R denotes a measurement error covariance matrix.

On the other hand, in the above description, 415 means the k-th measurement value at time t _i . The measurement modeling associated with the k-th measurement at time t _i will be described with reference to the following equation (5).

In the above, ω _{ti, r} means measurement noise (assuming white Gaussian noise) for distance information, and ω _{ti, θ} means measurement noise (assuming white Gaussian noise) for azimuth information. x _ti denotes the x-axis position of the target at time t _i , and x ^s denotes the x-axis position of the sensor. y _ti denotes the y-axis position of the target at time t _i , and y ^s denotes the y-axis position of the sensor.

In Equation 4, 415 represents a measurement, and Equation 5 becomes a related measurement modeling if the measurement is information on a target. However, when the measured value is false information (clutter), it is generally generated in the form of uniform distribution for the surveillance area in the simulation.

Assuming that the distance information and the azimuth information are obtained as the detection information, the measurement modeling example assumed in this embodiment assumes that the measurement noise is mixed in the relative distance and azimuth between the target and the sensor in the x-axis and y- Lt; / RTI >

The modification module 123 may rewrite the track existence probability among the track information using Equation (6).

In the above, 417 denotes the probability of existence of a track after time t _i . Reference numeral 418 denotes a pre-track existence probability at time t _i . Λ _ti denotes the likelihood ratio at time t _i , and P _D denotes the detection probability of the target. P _G denotes the gating probability, and m _ti denotes the number of valid measurements at time t _i . λ denotes the clutter spatial density, and p _{t, k} denotes the likelihood of the track for the k th measurement at time t _i .

The track removal module 124 performs a function of removing tracks having a value less than or equal to the removal threshold value (threshold value) using the track existence probability in the track information. The reason that the track clear module 124 removes tracks having a value below the removal threshold value is that these tracks are unlikely to be tracks for the target.

The track determination module 125 performs a function of determining tracks having a value equal to or greater than a predetermined threshold value by using a track existence probability in the track information. It is because the track confirmation module 125 confirms the tracks having a value equal to or larger than the determination threshold value because these tracks are likely to be tracks for the target.

The track merge module 126 merges similar tracks into one track. In the above, similar tracks mean tracks for the same target.

There can be many tracks for the same target by performing every scan track initialization step. In the present invention, the track merge module 126 is further provided in consideration of this point. The track merge module 126 merges similar tracks into one track and then transfers the merged track to the local track correction unit 130. [

In the present embodiment, however, there may be only one track for the same target. In this case, the track determination module 125 performs the function of transmitting the determined track to the local track correction unit 130. [ In this embodiment, it is also possible that the track merge module 126 is not provided in consideration of such a case.

In FIG. 2, the track initialization module 121, the prediction module 122, the modification module 123, the track removal module 124, the track determination module 125, and the track merge module 126 have a structure of circulating. The reason for this structure is that when the detection information is input through each scan sensor, the track information is rewritten using the information.

Referring back to FIG.

The local track correction unit 130 performs a function of correcting the determined local track when a definite local track is generated from the local track determination unit interlocked with the remaining sensors except the reference sensor, i.e., other sensors. In the example of Fig. 1, when the reference sensor is the second sensor 150, the local track determination unit associated with the reference sensor means the second local track processing unit 160. [ And the local track determination unit associated with other sensors and other sensors means the first sensor 110 and the first local track processing unit 120, respectively.

The local track correction unit 130 corrects the determined local track generated from the local track determination unit that is interlocked with other sensors based on the time information at which the reference sensor acquired the detection information. In the present embodiment, the sensors (the first sensor 110 and the second sensor 150 in FIG. 1) provided in the sensing field are those operating in an asynchronous environment and having different sampling periods. Therefore, when the sampling periods of the first sensor 110 and the second sensor 150 are different from each other, the track information time is different for each sensor.

The local track correction unit 130 may be provided in correspondence with each local track determination unit interlocked with other sensors in this embodiment, but may be connected to all local track determination units provided with only one sensor It is also possible.

When the first sensor 110 and the second sensor 150 are provided in the sensor field as shown in FIG. 1, the determined local track information (hereinafter referred to as first local track information) by the first local track processing unit 120, (Hereinafter, referred to as second local track information) by the second local track processing unit 160 are different from each other, the local track correction unit 130 corrects the local track information in the same time zone as the second local track information 1 Correct local track information. This is because the second sensor 150 is assumed to be the reference sensor. The local track correction unit 130 can correct the same time zone using a dynamic model.

When the first local track information (state variable estimation value, error covariance, etc.) and second local track information (state variable estimation value, error covariance, etc.) are expressed by the following Equation (7) The covariance and the like can be corrected by the following formula (8).

(A) represents a change state variable estimation value among the confirmed local track information of the first sensor 110 at time t _i , and may be represented by a 4 × 1 matrix. (b) represents a change state variable error covariance among the confirmed local track information of the first sensor 110 at time t _i , and may be represented by a 4 × 4 matrix. (c) represents a change state variable estimation value among the confirmed local track information of the second sensor 150 at time t _i , and may be represented by a 4 × 1 matrix. (d) means a covariance state variable error covariance among the definite local track information of the second sensor 150 at time t _i , and can be represented by a 4 × 4 matrix.

In the above, t _i denotes the determined local track information time of the first sensor 110, and t _j denotes the confirmed local track information time of the second sensor 150. I denotes the Kronecker product, and I _{2 × 2} denotes the unit matrix (2 × 2 matrix). Q _{tj, ti} denote the process noise covariance, and F _{tj and t i} denote the state variable transition matrix (4 × 4 matrix). σ _p means the process noise standard deviation.

Equation (8) shows a process of correcting the state variable estimate, error covariance, track existence probability, and the like in the first local track information. In Equation (8), the first equation shows the process of correcting the state variable estimate, and the second equation shows the process of correcting the error covariance. And the last equation corrects the track presence probability.

The state variable estimate and the error covariance are corrected to the corresponding time to be corrected using the state variable transition matrix of the dynamics modeling in a form similar to the prediction process of the IPDA filter. In the case of the track existence probability, .

The track fusing unit 140 performs a function of receiving a confirmed local track from a local track determining unit (e.g., a second local track processing unit 160) interlocked with a reference sensor (e.g., a second sensor 150) . The track fusing unit 140 also performs the function of receiving the corrected local track corrected from the local track correction unit 130. [

The track fusing unit 140 performs the function of fusing the received definite local tracks. The definite local track corrected by the local track correcting unit 130 is corrected to be generated in the same time zone as the definite local track generated by the local track confirming unit interlocked with the reference sensor. Thus, the track fusing unit 140 becomes able to fuse the received definite local tracks.

The track fusing unit 140 may include a track fusing module 141, a fusing track removing module 142, and a fusing track determining module 143 as shown in FIG. 3 is an internal configuration diagram of a track fusion unit constituting the automatic target tracking system of FIG.

The track fusing module 141 performs a track fusing function based on the track presence probability. At this time, the track fusing module 141 fuses the definite local track corrected by the local track correcting unit 130 with the determined local track generated by the local track processing unit, which is used as a measurement value and interlocked with the reference sensor.

In the embodiment of FIG. 1, the second sensor 150 is described as a reference sensor. Thus, the definite local track of the second local track processing unit 160 is used as a convergence local track by the determined local track, which is generated by the first local track processing unit 120 and corrected by the local track correcting unit 130, Create a track. The track fusing module 141 may calculate a fusion result such as a track presence probability, a state variable estimate, and an error covariance, and generate a fusion track using the result as a final estimate.

The track fusing module 141 may calculate a state variable estimate, an error covariance, a track existence probability, and the like for the fusion track when fusing the tracks based on the track presence probability. This will be described below.

The track fusing module 141 may calculate the state variable estimate and the error covariance for the fusion track using the following equation (9).

In Equation (9), the above equation is used to calculate the state variable estimate, and the following equation is used to calculate the error covariance. In equation (9), 419 represents the state variable estimate for the fusion track at time t _j . Reference numeral 420 denotes a probability that the nth local track of the first sensor 110 at time t _j is the target information, and 421 denotes a state variable estimate updated at time t _j to the nth local track of the first sensor 110 . In the following equation (9), 422 denotes a state variable error covariance for the fusion track at time t _j , and 423 denotes a state variable error covariance _{reconstructed} from the first sensor 110 to the n-th local track at time t _j .

Meanwhile, when calculating the state variable estimation value and the error covariance for the fusion track according to Equation (9), the conditional expression is expressed by Equation (10).

In equation (10) 424 refers to the first sensor (110) n-th track in the local time t _j.

On the other hand, the track fusing module 141 can calculate the track presence probability (that is, the post track existence probability for the fusion track) using the following equation (11).

In Equation (11), 425 denotes a post track existence probability for the fusion track at time t _j . Reference numeral 426 denotes a post track existence probability of the local track of the second sensor 150 at time t _j . 427 denotes a likelihood ratio of the fusion track at time t _j , and 428 denotes a post track existence probability of the n-th local track of the first sensor 110 at time t _j . P _C denotes a track detection probability, and n denotes a local track index of the first sensor 110 at time t _j . P _{t + i, n} means the likelihood for the nth local track of the first sensor 110 at time t _j , and λ _{t + i, n} means the false track spatial density.

The fusion track removal module 142 performs a function of removing tracks having a value equal to or less than the removal threshold value (threshold value) using the track presence probability in the fusion track. It is because the fused track removal module 142 removes tracks having a value below the removal threshold value because these tracks are less likely to be tracks for the target.

The fusing track determining module 143 performs a function of determining tracks having a value equal to or larger than a predetermined threshold value by using the track existence probability in the fusing track. It is because the convergence track confirmation module 143 confirms the tracks having a value equal to or larger than the determination threshold value because these tracks are likely to be tracks for the target.

The confirmed local tracks are filtered by each local track processing unit (i.e., the track removal module 124 and the track determination module 125) before being input to the track fusing unit 140. Therefore, in the present embodiment, it is not necessary that the fusing track removing module 142 and the fusing track determining module 143 are provided in the track fusing unit 140 in consideration of this point.

The performance of the track convergence technique in the multi - sensor environment is somewhat lower than that of the measurement convergence technique known as the optimal information fusion technique. However, there is a disadvantage that the measurement convergence technique is highly dependent on the convergence center, and the failure of the convergence center leads to the reliability of the whole network and the overload due to a large amount of traffic can be caused. Therefore, there are many studies on the track convergence technique, have.

Recently, the proposed track fusion method manages and fuses tracks automatically based on the assumption that the sensors are synchronized with each other. However, since there is a problem that the actual sensor environment is difficult to synchronize, the present invention proposes a solution do.

In each sensor, an automatic target tracking method using a track management technique in a single sensor environment is performed using the detected information. In this way, in the sensor environment where not only the target information but also the false information are detected, tracks generated as false information can be removed and the tracks for the target can be identified and tracked.

In order to solve the problem that the determined local tracks of each sensor are asynchronous with each other, it is necessary to fuse the confirmed local tracks formed by the respective sensors in order to perform the track fusion. In the present invention, Respectively. In the calibration process, a method of predicting the target information at a desired time using a dynamic model is applied.

Thus, track fusion is possible by synchronizing the deterministic local tracks of each sensor in a multi-sensor environment with asynchronous multiple sampling periods.

The present invention described above can be applied to an automatic target tracking and information fusion system based on multiple sensors. The present invention can be applied to surveillance and reconnaissance target (ex. Enemy group) using a sonar sensor or the like. For example, the present invention can be used in a system requiring improved target estimation results with less traffic in a multi-sensor environment consisting of multiple Sonar or radars.

1 to 3, an embodiment of the present invention has been described. Best Mode for Carrying Out the Invention Hereinafter, preferred forms of the present invention that can be inferred from the above embodiment will be described.

4 is a conceptual diagram of a target tracking system according to a preferred embodiment of the present invention.

4, the target tracking system 200 includes detection information obtaining units 210, track generating units 220, a track correcting unit 230, a target tracking unit 240, and a main control unit 250.

The main control unit 250 controls the overall operation of each configuration of the target tracking system 200.

The detection information acquisition units 210 acquire detection information on a target using sensors having different sampling periods. In this embodiment, the detection information acquisition units 210 include a first detection information acquisition unit 210a, a second detection information acquisition unit 210b, N-th detection information acquisition unit 210n, and the like, and it is preferable that at least two detection units are provided. The sensor may be implemented as a sonar sensor, for example.

Detection information obtained by the detection information acquisition units 210 may include not only information about a target but also information about an object that is not a target. The detection information acquisition units 210 correspond to the first sensor 110, the second sensor 150, and the like in FIG.

The detection information acquisition units 210 can acquire the distance to the target and the orientation of the target with each piece of detection information.

The track generators 220 generate tracks that display the movement trajectory of the target based on the detection information obtained by the detection information acquiring units 210. [ In this embodiment, the track generators 220 include a first track generator 220a, a second track generator 220b, , The n-th track generating unit 220n, and the like, and it is preferable that at least two of them are provided.

In this embodiment, the number of track generators 220 corresponding to the detection information obtainers 210 may be provided. The track generators 220 are concepts corresponding to the first local track processing unit 120, the second local track processing unit 160, and the like in FIG.

The track generating units 220 may include a track information predicting unit 221, a track information creating unit 222 and a final trajectory extracting unit 223 as shown in FIG. FIG. 5 is an internal configuration diagram of a track generation unit constituting the target tracking system of FIG. 4;

The track information predicting unit 221 performs a function of predicting information on tracks using a state transition matrix. The track information prediction unit 221 corresponds to the prediction module 122 of FIG.

The track information predicting unit 221 can predict information on tracks using a track existence probability based algorithm. For example, the track information predicting unit 221 can predict information on tracks using an IPDA (Integrated Probabilistic Data Association) algorithm.

The track information predicting unit 221 estimates information about each track when estimating information about each track, an error covariance associated with estimation of state variables associated with the target, an error covariance associated with estimation of a state variable, Can be predicted. The track information predicting unit 221 can predict a value obtained by estimating the position of the target and an estimated value of the speed of the target with the estimated value of the state variable.

The track information predicting unit 221 can estimate an estimated value of the state variable at the current time based on the state transition matrix and the estimated value of the state variable renewed at the previous time. In the above, the current time means the time at which the detection information is obtained.

Also, the track information predicting unit 221 can predict the error covariance at the current time based on the state transition matrix, the error covariance reconstructed at the previous time, and the process noise covariance associated with the dynamic modeling.

The track information predicting unit 221 estimates a first probability that a target exists in the track at the previous time and the current time, a second probability that the target exists in the track only at the current time, and a third probability that the target, The probability that a target exists in the track at the present time can be predicted based on the base.

The track information predicting unit 221 calculates a first calculation value by multiplying a third probability by a first probability, multiplies a value obtained by subtracting the third probability from 1 and a second probability to calculate a second calculation value, 1 < / RTI > computed value and the second computed value may be added to predict the probability that a target for the track will exist in the current time.

The track information creating unit 222 performs a function of replacing information on tracks predicted by the track information predicting unit 221 based on the detection information. The track information creating unit 222 corresponds to the modification module 123 of FIG.

The final trajectory extracting unit 223 performs a function of generating final trajectories selected from the temporary trajectories of the target on the basis of the information about the tracks that have been changed by the track information creating unit 222 as tracks. The final trajectory extractor 223 is a concept corresponding to the track elimination module 124 and the track determination module 125 in Fig.

The final trajectory extracting unit 223 can use the probability that a target for the corresponding track exists as information about each track when selecting the final trajectory from the temporary trajectories. The final trajectory extractor 223 removes temporary trajectories less than the threshold value based on the probability that a target for the track exists, and selects temporary trajectories with a threshold value or more as final trajectories.

The track generating units 220 may further include a trajectory association determining unit 224 and a final trajectory merging unit 225.

The trajectory association determination unit 224 performs a function of determining whether or not there are similar trajectories in which the degree of similarity is equal to or greater than the reference value among the final trajectories. In this embodiment, the trajectory association determination unit 224 can detect similar trajectories by comparing two trajectories selected from the final trajectories.

The final trajectory merging unit 225 merges similar trajectories to generate a single track if the trajectory association determining unit 224 determines that similar trajectories exist in the final trajectories. The final trajectory merge unit 226 corresponds to the track merge module 126 of FIG.

Meanwhile, the track generators 220 may further include a temporary locus generator (not shown). The temporary trajectory generator performs a function of generating temporary trajectories of the target based on the detection information about the target. When the track generators 220 further include a temporary locus generator, the final locator extractor 223 may perform a function of extracting final trajectories with respect to the temporary loci generated by the temporary locus generator. The temporary locus generator is a concept corresponding to the track initialization module 121 of FIG.

Referring back to FIG.

The track corrector 230 may be configured to determine whether the reference sensor selected from the sensors is associated with the second detection information obtained by at least one other sensor other than the reference sensor among the sensors based on the time information when the reference sensor selected from the sensors acquires the first detection information And performs a function of correcting the second track. The track correcting unit 230 corresponds to the local track correcting unit 130 of FIG.

The track corrector 230 may correct the second track so that the second detection information is obtained at the same time as the first detection information using kinetic modeling.

The track corrector 230 may correct the second track by correcting information on the second track. At this time, the track corrector 230 can correct the estimated value of the state variable related to the target, the error covariance associated with the estimation of the state variable, and the probability that the target for the track exists, with information on the second track.

The track corrector 230 may correct the estimated value of the state variable based on the first matrix and the second matrix obtained from the Kronecker product between the unit matrix. The first matrix means a matrix including as a component a time difference between a first time when the reference sensor acquires the first detection information and a second time when the other sensor acquires the second detection information.

The track corrector 230 may correct the error covariance based on the transpose matrix of the second matrix, the transposed matrix of the second matrix, the third matrix, and the transpose matrix of the third matrix. The third matrix is a matrix including a time difference between a first time and a second time, and a value obtained by dividing the time difference by two.

The target tracking unit 240 performs tracking of the target based on the first track and the corrected second track associated with the first detection information. In this case, the target tracking unit 240 may track the target using the fusion track after fusing the first track and the second track.

The target tracking unit 240 can track the target for the purpose of monitoring and scouting a specific target (ex. Enemy group). The target tracking unit 240 is a concept including the track fusion unit 140 of FIG.

Next, an operation method of the target tracking system 200 will be described. 6 is a flowchart schematically showing a target tracking method according to a preferred embodiment of the present invention.

First, the detection information acquisition units 210 acquire detection information on the target using sensors having different sampling periods (S310).

Thereafter, the track generators 220 generate tracks indicative of the movement trajectory of the target based on the detection information (S320).

Then, the track correcting unit 230 corrects the second detection information obtained by at least one other sensor other than the reference sensor based on the time information when the reference sensor selected from the sensors acquires the first detection information The associated second track is corrected (S330).

Thereafter, the target tracking unit 240 tracks the target based on the first track and the corrected second track associated with the first detection information (S340).

It is to be understood that the present invention is not limited to these embodiments, and all elements constituting the embodiment of the present invention described above are described as being combined or operated in one operation. That is, within the scope of the present invention, all of the components may be selectively coupled to one or more of them. In addition, although all of the components may be implemented as one independent hardware, some or all of the components may be selectively combined to perform a part or all of the functions in one or a plurality of hardware. As shown in FIG. In addition, such a computer program may be stored in a computer readable medium such as a USB memory, a CD disk, a flash memory, etc., and read and executed by a computer to implement an embodiment of the present invention. As the recording medium of the computer program, a magnetic recording medium, an optical recording medium, a carrier wave medium, and the like can be included.

Furthermore, all terms including technical or scientific terms have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs, unless otherwise defined in the Detailed Description. Commonly used terms, such as predefined terms, should be interpreted to be consistent with the contextual meanings of the related art, and are not to be construed as ideal or overly formal, unless expressly defined to the contrary.

It will be apparent to those skilled in the art that various modifications, substitutions and substitutions are possible, without departing from the scope and spirit of the invention as disclosed in the accompanying claims. will be. Therefore, the embodiments disclosed in the present invention and the accompanying drawings are intended to illustrate and not to limit the technical spirit of the present invention, and the scope of the technical idea of the present invention is not limited by these embodiments and the accompanying drawings . The scope of protection of the present invention should be construed according to the following claims, and all technical ideas within the scope of equivalents should be construed as falling within the scope of the present invention.

Claims (14)

Detection information acquisition units for acquiring detection information on a target using sensors having different sampling periods;
Track generating units for generating tracks for displaying the movement trajectory of the target based on the detection information;
A second track related to the second detection information obtained by at least one other sensor other than the reference sensor among the sensors based on time information when the reference sensor selected from the sensors acquires the first detection information, A track corrector for correcting the track; And
A target tracking unit for tracking the target based on a first track associated with the first detection information and a corrected second track,
Wherein the asynchronous sensors are used for tracking the target. The method according to claim 1,
Wherein the detection information obtaining units obtain the distance to the target and the orientation of the target with each detection information. The method according to claim 1,
The track generators include:
A track information predicting unit for predicting information on the tracks using a state transition matrix;
A track information reformer for updating information on the tracks predicted by the track information predictor based on the detection information; And
A final trajectory extracting unit for generating final trajectories selected from the temporary trajectories of the target on the basis of the information about the tracks updated by the track information creating unit,
Wherein the asynchronous sensors are used for tracking the target. The method of claim 3,
The track information predicting unit may use an IPDA (Integrated Probabilistic Data Association) filter to estimate an estimated value of the state variable related to the target, an error covariance related to the estimation of the state variable, And estimating the probability of the presence of the asymmetric sensor. The method of claim 3,
The track generators include:
A trajectory association determining unit for determining whether or not there are similar trajectories in which the degree of similarity is equal to or greater than a reference value among the final trajectories; And
If it is determined that the similar trajectories exist, merges the similar trajectories to generate a single track,
Further comprising an asynchronous sensor. The method according to claim 1,
Wherein the track correction unit corrects the second track so that the second detection information is obtained at the same time as the first detection information using kinetic modeling. . The method according to claim 1,
Wherein the track corrector corrects the second track by correcting information about the second track and uses the information about the second track as an estimate of the state variable associated with the target, an error covariance associated with the estimation of the state variable, And corrects the probability that the target for the track is present. 8. The method of claim 7,
The track compensator may include a first matrix and a unit matrix including a time difference between a time when the reference sensor acquires the first detection information and a time when the other sensor acquires the second detection information, And corrects the estimate of the state variable based on a second matrix obtained from a Kronecker product. 9. The method of claim 8,
Wherein the track corrector comprises: a second matrix, a transposed matrix of the second matrix, a third matrix including a value obtained by dividing the time difference by the time difference by 2, and a transpose matrix of the third matrix, And correcting the error covariance by using the asynchronous sensors. The method according to claim 1,
Wherein the target tracking unit tracks the target based on a fusion track obtained by fusing the first track and the second track. Acquiring detection information on a target using sensors having different sampling periods;
Generating tracks that display the movement trajectory of the target based on the detection information;
A second track related to the second detection information obtained by at least one other sensor other than the reference sensor among the sensors based on time information when the reference sensor selected from the sensors acquires the first detection information, Correcting; And
Tracking the target based on a first track associated with the first detection information and a corrected second track
Wherein the asynchronous sensors are used for tracking the target. 12. The method of claim 11,
Wherein the correcting step corrects the second track by correcting information about the second track, and estimates an estimated value of the state variable associated with the target with the information about the second track, an error covariance And corrects the probability that the target exists for the track. 13. The method of claim 12,
Wherein the correcting step includes a first matrix including a time difference between a time when the reference sensor acquires the first detection information and a time when the other sensor acquires the second detection information as a component, Wherein the estimate of the state variable is corrected based on a second matrix obtained from a Kronecker product. 14. The method of claim 13,
Wherein the step of correcting comprises the steps of: transforming the second matrix, a transposed matrix of the second matrix, a third matrix including a value obtained by dividing the time difference by the time difference by 2, and a transpose matrix of the third matrix, And correcting the error covariance based on the asymmetric sensors.

Images