CN106526584A - Target detection and tracking combined processing method in multi-radar system - Google Patents
Target detection and tracking combined processing method in multi-radar system Download PDFInfo
- Publication number
- CN106526584A CN106526584A CN201610817863.7A CN201610817863A CN106526584A CN 106526584 A CN106526584 A CN 106526584A CN 201610817863 A CN201610817863 A CN 201610817863A CN 106526584 A CN106526584 A CN 106526584A
- Authority
- CN
- China
- Prior art keywords
- target
- radar
- represent
- moment
- detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
- G01S13/66—Radar-tracking systems; Analogous systems
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a target detection and tracking combined processing method in a multi-radar system. The method comprises: (1), constructing a target motion model; (2), constructing a target measuring model; (3), carrying out initialization; (4), selecting a radar for target detection (5), determining a constant false alarm detection threshold value; (6), determining effective measurement; (7), determining whether all target detection radars in a multi-radar system are selected; (8), calculating the number of combined joint events; (9) updating a current target tracking state; (10) updating a covariance of a target tracking state; (11), determining whether a target tracking track is in a divergence state; and (12) ending target tracking. Compared with the prior art, the method has the following beneficial effects: the average target detection probability and the target tracking performance are improved on the premise of constant false alarm inside a port door; and the calculating complexity is low. The method disclosed by the invention can be applied to target detection and tracking of a multi-radar system.
Description
Technical field
The present invention relates to communication technical field, further relates to a kind of many radars in Radar Signal Processing Technology field
Target detection tracking combination treatment method in system.The present invention can be used to realize that multiple radar system combines the detecting and tracking of target
Process.
Background technology
Target following plays important role in many applications of radar and Sonar system, in battlefield monitoring, air defence,
The aspect such as air traffic control and fire control all plays an important role.But target following is carried out under dense clutter environment, every
During secondary scanning, a large amount of metric data can be obtained, but the source of each data be it is unknown (be probably derived from target,
It is probably derived from false-alarm).
The paper that B K Habtemariam, R Tharmarasa, and T Kirubarajan et al. are delivered at which
“Multiple detection probabilistic data association filter for multistatic
target tracking”(Proceedings of the,International Conference on Information
Fusion.IEEE,2011:1-6.) with paper " A Multiple-Detection Joint Probabilistic Data
Association Filter”(IEEE Journal of Selected Topics in Signal Processing,
2013,7(3):A kind of many detection probability data interconnection (Multiple Detection are proposed in 461-471)
Probabilistic data association, MD-PDA) method, multiple measurement sources are processed with the joint event of combination
In the probability of same target.The method for being tracked to single goal in multiple radar system, in the tracking gate one
Individual target may produce the situation of multiple measurements, and calculate the probability of each joint event, and the probability for being then based on obtaining will be measured
Survey and target association.This method solve probability data interconnection method (Probabilistic Data Association, PDA) to exist
Deficiency when being tracked in multiple radar system.There is a kind of scene, same target meeting in interconnection door in multiple radar system
Multiple measurements are produced, and PDA methods assume that, in scanning every time, be up to one measures with which as source for target.But, should
The weak point that method is present is that MD-PDA assumes that a target may produce multiple measurements, but does not account for a radar
A restriction for measuring is obtained at most from a target, it is therefore desirable to process a large amount of combinatorial association events;And MD-PDA methods
Detector be based on how graceful Pearson came (Neyman-Pearson, NP) criterion, employ fixed threshold.In such case
Under, information can only pass to tracker from detector, it is impossible to be adaptively adjusted detection threshold according to target movement position.
The patent that Xian Electronics Science and Technology University applies at which is " at the FUSION WITH MULTISENSOR DETECTION tracking joint based on bayesian theory
Reason method " (number of patent application:201110003111.4, publication No.:CN 102147468B) in disclose a kind of based on Bayes
Theoretical FUSION WITH MULTISENSOR DETECTION tracking combination treatment method.The method is the detecting and tracking Combined Treatment side based on bayesian theory
Method, directly processes the observation data of original non-thresholding, using the motion model of target between frame data in the form of probability pair
Signal is accumulated, and can effectively utilizes prior information, so as to improve the detection performance of system.Using prior information and all
Observation data calculate dbjective state posterior probability density function, and in the dbjective state posterior probability density function obtained by calculating
On the basis of realize to target detection and tracking.The method is deduced the relation of detection threshold value and given false alarm rate, but, still
The weak point of presence is that the detection criteria of the method foundation is NP criterions, and in tracking gate, the thresholding of all detector units is identical,
No reasonable feedback information of the application from tracker to detector, it is impossible to according to the position of target following come self-adaptative adjustment ripple door
The thresholding of interior each detector unit.
The content of the invention
It is an object of the invention to overcome the shortcomings of above-mentioned existing method, it is proposed that target detection in a kind of multiple radar system
Tracking combination treatment method (Joint Detection and Tracking Processing, JDTP) method, to strengthen many thunders
Up to detecting and tracking performance of the system to target.
Realize that basic thought of the invention is that the motion model and multiple radar system for initially setting up target is measured to target
Model, using Bayesian detection device, according to the tracking mode for obtaining target from tracker feedback, CFAR detection door is set
Limit, on the premise of the average false alarm rates of Bo Mennei are constant, lifts the average detected probability of target and the tracking performance of system;Simultaneously
The quantity of combinatorial association event is reduced on the basis of MD-PDA, and reduction computation complexity is low, more meets the true feelings of radar application
Condition.
The present invention's comprises the following steps that:
(1) according to the following formula, build target movement model:
ξTk=F ξTk-1+uTk-1
Wherein, ξTkThe motion state of Tk moment targets is represented, F represents the transfer square of dbjective state in the case of uniform motion
Battle array, ξTk-1Represent the motion state of Tk-1 moment targets, uTk-1It is that zero, covariance is Q to represent that average is obeyed in distributionTk-1Gauss
The motion process noise of distribution;
(2) according to the following formula, build target measurement model:
Wherein,Represent in t multiple radar system that n-th radar obtained j-th of threshold value and contain to target measurement
The measuring value of noise, hn,t(ξt) represent t multiple radar system in n-th radar to dbjective state ξtMeasuring value, wn,tRepresent
It is that zero, covariance is Σ that in t multiple radar system, average is obeyed in n-th radar distributionn,tGaussian Profile measurement noise,
vn,tRepresent that n-th radar tracking Bo Mennei obeys equally distributed false measuring value in multiple radar system in t;
(3) initialize:
(3a) using the prior information of dbjective state as target initial time to be tracked tracking mode;
(3b) according to the following formula, calculate the initial covariance of target state:
C0=CRLB (y0)
Wherein, C0Represent the initial covariance of target state, CRLB (y0) represent the tracking shape that initial time target is moved
State y0Carat Metro limit;
(4) radar of selection target detection:
From multiple radar system, radar of the optional radar as target detection;
(5) determine CFAR detection threshold value:
(5a) according to the following formula, calculate running parameter of the detector of target detection radar in constant false alarm rate condition:
Wherein, ηkRepresent running parameter of the detector of k moment target detection radars in constant false alarm rate condition, VkRepresent k
The size of moment target detection radar tracking ripple door, μkRepresent that the radar of k moment target detections obtains the average signal-to-noise ratio of echo,
P represents the false alarm rate required by the radar of target detection;nzRepresent the dimension that target detection radar is measured to target, ΣkRepresent k
The measurement noise covariance of moment target detection radar;
(5b) according to the following formula, calculate the CFAR detection threshold value of target detection radar:
Wherein, γkRepresent the detection threshold value of k moment target detection radars, μkRepresent that the radar of k moment target detections is obtained
To the average signal-to-noise ratio of echo, ln () is represented and is taken from right log operations, ηkRepresent that the detector of k moment target detection radars exists
Running parameter during constant false alarm rate condition, VkThe size of the radar tracking ripple door of k moment target detections is represented,
RepresentObeying average isVariance isNormal distribution,Represent l of the radar of k moment target detections in tracking gate
Measuring value at individual detector unit,Measuring value of the radar of k moment target detections to target prediction state is represented,Represent and sit
Measurement covariance value of the radar of k moment target detections to target prediction state after mark conversion;
(6) determine and effectively measure:
Using all measurements more than detection threshold value in target detection radar measurement value as effectively measurement;
(7) in judging multiple radar system, whether all of target detection radar has selected, if so, then execution step (8), no
Then, execution step (4);
(8) according to the following formula, calculate the quantity of combinatorial association event:
Wherein, N 'eThe quantity of e moment combinatorial association events is represented, S represents the sum of radar in multiple radar system, ∏ tables
Show that company takes advantage of operation, s to represent the label of each radar in multiple radar system,Represent from Ms,r1 is selected to measure in individual measurement
The permutation and combination operation of target is come from,Represent all M of s-th radar in r moment multiple radar systemss,rIndividual measurement is all originated
In false-alarm, Ms,rEffective measurement sum of s-th radar in r moment multiple radar systems is represented, the value of r is identical with e;
(9) according to the following formula, update current goal tracking mode:
Wherein, ydRepresent the d moment update after target following state, ∑ represents sum operation, yJ,dRepresent that the d moment combines
State in joint event after j-th event update, J represent the label in all combinatorial association events,Represent d moment institutes
There is the probability of j-th event generation in combinatorial association event;
(10) according to the following formula, update the covariance of target following state:
Wherein, CdThe dbjective state covariance after updating is represented, ∑ represents sum operation, and J is represented in all combinatorial associations
Label in event,Represent the probability of j-th event generation in c moment all combinatorial association events, CJ,dRepresent yJ,dAssociation
Variance yields, yJ,dState in expression d moment combinatorial association events after j-th event update, ()TRepresenting matrix transposition is operated,
ydRepresent the d moment update after target following state;
(11) judge whether target following flight path dissipates, if so, then execution step (12), otherwise, current time adds 1, will
The current tracking mode of tracker feeds back to execution step (4) after inspection center;
(12) target following terminates.
The present invention is had the advantage that compared with prior art:
First, as the present invention is by feeding back to inspection center by tracker current tracking mode, calculate target detection
The CFAR detection threshold value of radar, overcomes prior art each come self-adaptative adjustment Bo Mennei according to the position of target following
The thresholding of detector unit so that the present invention design on the premise of the average false alarm rates of Bo Mennei are constant, lifted target it is average
The tracking performance of detection probability and system.
Second, as the present invention is according to the quantity for calculating combinatorial association event in many detections of radar tracking systems, updates and work as
Front target following state, overcoming prior art needs to process the deficiency of a large amount of combinatorial association events so that the present invention has drop
Low computation complexity is low, more meets the advantage of radar application truth.
Description of the drawings
Fig. 1 is the flow chart of the present invention;
Schematic diagrames of the Fig. 2 for target measurement model;
Radar and target bearing schematic diagram of the Fig. 3 for multiple radar system;
Fig. 4 is detection probability correlation curve of each radar of multiple radar system using distinct methods detector;
Fig. 5 is the schematic diagram of NP detectors and Bayesian detection device thresholding;
Fig. 6 is that multiple radar system adopts accuracy comparison figure of the distinct methods to target following.
Specific embodiment
The present invention will be further described below in conjunction with the accompanying drawings:
With reference to Fig. 1, the specific implementation step of the present invention is as follows:
Step 1, builds target movement model.
According to the following formula, build target movement model:
ξTk=F ξTk-1+uTk-1
Wherein, ξTkThe motion state of Tk moment targets is represented, F represents the transfer square of dbjective state in the case of uniform motion
Battle array, ξTk-1Represent the motion state of Tk-1 moment targets, uTk-1It is that zero, covariance is Q to represent that average is obeyed in distributionTk-1Gauss
The motion process noise of distribution.
In the case of uniform motion, the transfer matrix F of dbjective state is given by:
Wherein,Representing matrix direct product is operated, T0Represent the sampling interval.
Motion artifacts covariance matrix Qk-1It is given by:
Wherein q1Represent the parameter of the process noise intensity of control targe dynamic model.
Step 2, builds target measurement model.
According to the following formula, build target measurement model:
Wherein,Represent in t multiple radar system that n-th radar obtained j-th of threshold value and contain to target measurement
The measuring value of noise, hn,t(ξt) represent t multiple radar system in n-th radar to dbjective state ξtMeasuring value, wn,tRepresent
It is that zero, covariance is Σ that in t multiple radar system, average is obeyed in n-th radar distributionn,tGaussian Profile measurement noise,
vn,tRepresent that n-th radar tracking Bo Mennei obeys equally distributed false measuring value in multiple radar system in t.
Referring to the drawings two radar observation object delineations shown in 2, wherein, the measurement source of box indicating radar 1
In target, triangle represents the measurement source of radar 1 in false-alarm, and round dot represents the measurement source of radar 2 in target, and rhombus is represented
The measurement source of radar 2 in false-alarm, star-like expression target predicted position, V1,tAnd V2,tThe tracking of radar 1 and radar 2 is represented respectively
Ripple door.
Step 3, initialization.
Using the prior information of dbjective state as target initial time to be tracked tracking mode.
Using prior information as the initial dbjective state of targetpath in embodiments of the invention.
According to the following formula, calculate the initial covariance of target state:
C0=CRLB (y0)
Wherein, C0Represent the initial covariance of target state, CRLB (y0) represent the tracking shape that initial time target is moved
State y0Carat Metro limit.
Step 4, the radar of selection target detection.
From multiple radar system, radar of the optional radar as target detection.
Step 5, determines CFAR detection threshold value.
According to the following formula, calculate running parameter of the detector of target detection radar in constant false alarm rate condition:
Wherein, ηkRepresent running parameter of the detector of k moment target detection radars in constant false alarm rate condition, VkRepresent k
The size of moment target detection radar tracking ripple door, μkRepresent that the radar of k moment target detections obtains the average signal-to-noise ratio of echo,
P represents the false alarm rate required by the radar of target detection;nzRepresent the dimension that target detection radar is measured to target, ΣkRepresent k
The measurement noise covariance of moment target detection radar.
In embodiments of the invention, the probability density of the echo amplitude that radar is received is represented by:
Wherein, H0Represent that the position does not have target, the data at this moment receiving come solely from noise.H1Represent that the position is present
Target, the data for obtaining by be target echo and noise superposition.Represent that the k moment detects radar in l-th resolution cell
Echo amplitude (on the premise of target fluctuation model is Swerling I types, the quadratic sum of I, Q two-way is output as index point
Cloth), μkRepresent that the k moment detects that radar obtains the average signal-to-noise ratio of echo.
According to the following formula, provide the prior probability that target whether there is:
Wherein,The non-existent prior probability of target is represented,Represent the prior probability that target is present, VkTable
Show that the k moment detects the size of radar tracking ripple door,Represent the k moment detect radar to target prediction measure withRepresent the k moment
Radar is to target prediction covariance.hk(ξk) represent that the k moment detects measuring value of the radar to dbjective state,RepresentObeying average isVariance isNormal distribution.
According to the prior information obtained in tracking process feedback, Bayes's form that discriminate is written as:
WhereinWithRepresent the prediction distribution that detection radar is measured at l-th resolution cell, H0Representing should
The aimless situation in position, H1Represent that the position has the situation of target, ηkRepresent that the k moment detects radar Bayesian detection device
Constant in constant false alarm rate condition.
Calculate average false alarm rate to be calculated by following formula:
Wherein, P represents average false alarm rate, VkRepresent that the k moment detects the size of radar tracking ripple door, exp () is represented and referred to
Number operation, μkRepresent that the k moment detects that radar obtains the average signal-to-noise ratio of echo, ηkRepresent that the k moment detects radar Bayesian detection device
Constant in constant false alarm rate condition, P represent the false alarm rate required by detection radar;nzRepresent the dimension that radar is measured to target
Number, ΣkRepresent that the k moment detects the measurement noise covariance of radar.
So, average false alarm rate is given, constant η of the Bayesian detection device in constant false alarm rate condition can be obtainedk。
According to the following formula, calculate the CFAR detection threshold value of target detection radar.
Wherein, γkRepresent the detection threshold value of k moment target detection radars, μkRepresent that the radar of k moment target detections is obtained
To the average signal-to-noise ratio of echo, ln () is represented and is taken from right log operations, ηkRepresent that the detector of k moment target detection radars exists
Running parameter during constant false alarm rate condition, VkThe size of the radar tracking ripple door of k moment target detections is represented,
RepresentObeying average isVariance isNormal distribution,Represent l of the radar of k moment target detections in tracking gate
Measuring value at individual detector unit,Measuring value of the radar of k moment target detections to target prediction state is represented,Represent and sit
Measurement covariance value of the radar of k moment target detections to target prediction state after mark conversion;The detection threshold of multiple radar system
Can be obtained by Bayesian Decision formula is entered line translation:
Wherein,Represent that the k moment detects echo amplitude of the radar in l-th resolution cell, by itself and detection threshold γkThan
Compared with, it can be determined that measurement source is in target or false-alarm.As can be seen from the above equation, ifThe closer toThenMore
Greatly, detection threshold γkIt is lower.In other words, the detection probability of Bayes' theorem detector is different at different positions.
Step 6, it is determined that effectively measure.
Using all measurements more than detection threshold value in target detection radar measurement value as effectively measurement.
Step 7, in judging multiple radar system, whether all of target detection radar has selected, if so, then execution step 8,
Otherwise, execution step 4.
Step 8, according to the following formula, calculates the quantity of combinatorial association event.
Wherein, N 'eThe quantity of e moment combinatorial association events is represented, S represents the sum of radar in multiple radar system, ∏ tables
Show that company takes advantage of operation, s to represent the label of each radar in multiple radar system,Represent from Ms,r1 is selected to measure in individual measurement
The permutation and combination operation of target is come from,Represent all M of s-th radar in r moment multiple radar systemss,rIndividual measurement is all originated
In false-alarm, Ms,rEffective measurement sum of s-th radar in r moment multiple radar systems is represented, the value of r is identical with e.
In embodiments of the invention, combinatorial association event measures the institute that model measurement derives from target or false-alarm for target
It is possible to event.
Be generally used for solve in multiple radar system the tracking problem of single goal method be MD-PDA methods, joint event
Quantity be defined as Nq。
Wherein, sums of the N for radar in multiple radar system, min () are represented and take minimum Value Operations, MsRepresent many radar systems
The measurement sum that system is received, a represent measurement source in the measurement of target, and ∑ represents sum operation, and C represents that permutation and combination is grasped
Make, ()!Represent factorial operation.However, in actual applications, more real situation is a radar from a target to multipotency
Receive a measurement.In this case, the quantity of joint event is defined as N 's:
Wherein, N 'eThe quantity of e moment combinatorial association events is represented, S represents the sum of radar in multiple radar system, ∏ tables
Show that company takes advantage of operation, s to represent the label of each radar in multiple radar system,Represent from Ms,r1 is selected to measure in individual measurement
The permutation and combination operation of target is come from,Represent all M of s-th radar in r moment multiple radar systemss,rIndividual measurement is all originated
In false-alarm, Ms,rRepresent effective measurement sum of s-th radar in r moment multiple radar systems.
Step 9, according to the following formula, updates current goal tracking mode.
Wherein, ydRepresent the d moment update after target following state, ∑ represents sum operation, yJ,dRepresent that the d moment combines
State in joint event after j-th event update, J represent the label in all combinatorial association events,Represent d moment institutes
There is the probability of j-th event generation in combinatorial association event.
State y in embodiments of the invention in d moment combinatorial associations event after j-th event updateJ,dObtained by following formula
Arrive:
Wherein, yJ,dState in expression d moment combinatorial association events after j-th event update,Represent d moment many thunders
Up to predicted state of the system to target,Represent that the Kalman of the individual measurements of J (b) of b-th target detection radar of d moment increases
Benefit,Represent that the correlation of the individual measurements of b-th radar J (b) of d moment is new to cease.
The probability that j-th event occurs in d moment all combinatorial association events in embodiments of the inventionIt is by following formula
Obtain:
Wherein,The probability of j-th event generation in d moment all combinatorial association events is represented, ∏ represents that company takes advantage of operation,
I represents the radar label in multiple radar system, and N represents the sum of radar in multiple radar system, and p { } is represented and asked probability to operate,Represent that the individual measurement sources of J (i) of i-th radar in m moment multiple radar systems, in the event of target, represent i-th radar
Measurement all derive from false-alarm.
Step 10, according to the following formula, updates the covariance of target following state:
Wherein, CdThe dbjective state covariance after updating is represented, ∑ represents sum operation, and J is represented in all combinatorial associations
Label in event,Represent the probability of j-th event generation in c moment all combinatorial association events, CJ,dRepresent yJ,dAssociation
Variance yields, yJ,dState in expression d moment combinatorial association events after j-th event update, ()TRepresenting matrix transposition is operated,
ydRepresent the d moment update after target following state.
Step 11, judges whether target following flight path dissipates, if so, then execution step 12, and otherwise, current time adds 1, will
The current tracking mode of tracker feeds back to execution step 4 after inspection center.
In multiple radar system, when target not in the detection range of multiple radar system when, the flight path of target following will
Diverging.
Step 12, target following terminate.
Below in conjunction with the accompanying drawings the effect of the present invention is further described.
1. simulated conditions:
The present invention simulated running system be Intel (R) Core (TM) i7-4590CPU@3.30GHz, 32-bit Windows
Operating system, simulation software adopt MATLAB (R2014b).
2. emulation content and interpretation of result:
The emulation experiment of the present invention sets a radar fence being made up of the radar 6 different locations, each radar
Signal effective bandwidth and effective duration are respectively B=2MHz and T=1ms, and simulation sequence data are 60 frames, observation interval T0
=2s, door coefficient g=8;Target initial position (0,0) km, and with speed (- 30,0) m/s fly at a constant speed, work as R0=10km
When signal to noise ratio μ0=8dB.
With reference to the radar and target bearing schematic diagram of Fig. 3 multiple radar systems, emulation experiment of the present invention is directed to a kind of scene:One
Individual target is close to and is deployed in borderline radar fence.
Radar in Fig. 3 in box indicating multiple radar system, dotted line represent the track of target motion, and arrow represents that target is transported
Dynamic direction.Embodiments of the invention are emulated for following two situation:(1) the false alarm rate P of the first situationi=10-2;
False-alarm P of (2) second situationi=10-3, i=1,2 ..., N.
In above-mentioned two situations, each radar of multiple radar system is detected the employing present invention with reference to shown in Fig. 4 using NP
Detection probability correlation curve under device and Bayesian detection device.Ordinate in Fig. 4 represents detection probability, and abscissa is represented to mesh
The moment of mark tracking.With the radar 1 in 3 deployment diagram of graphical representation of pecked line sign in Fig. 4, indicated with plus line in Fig. 4
3 deployment diagram of graphical representation in radar 2, the radar 3 in 3 deployment diagram of graphical representation for being indicated in Fig. 4 in dash-dot line, figure
With the radar 4 in 3 deployment diagram of graphical representation of solid line sign in 4, the graphical representation 3 indicated with cross line in Fig. 4 is disposed
Radar 5 in figure, with the radar 6 in 3 deployment diagram of graphical representation of dotted line sign in Fig. 4.Fig. 4 (a) show many radar systems
In the first scenario using the detection probability curve of NP detectors, Fig. 4 (b) show multiple radar system in the first situation to system
The detection probability curve of lower employing Bayesian detection device.Fig. 4 (c) is shown multiple radar system and is examined using NP in the latter case
The detection probability curve of device is surveyed, Fig. 4 (d) show multiple radar system in the latter case using the detection of Bayesian detection device
Probability curve.
Relatively Fig. 4 (a) and Fig. 4 (b) can see, in the first scenario, using the detection performance of Bayesian detection device
It is better than NP detectors.Relatively Fig. 4 (c) and Fig. 4 (d), it will also be seen that excellent using the detection performance of Bayesian detection device
In NP detectors.Therefore, comprehensive Fig. 4 can be seen that the Bayesian detection device using the present invention, and false alarm rate is higher, to detection property
The lifting of energy is better.
With reference to shown in Fig. 5, as an example, comparison is of the invention for the radar 1 taken in Fig. 3 deployment diagrams at the 10th moment of the invention
Bayesian detection device and NP detectors detection threshold.Ordinate in Fig. 5 represents detection threshold of the radar to target, horizontal seat
Mark represents the detector unit scope in significant wave door.Solid line in Fig. 5 represents the detection threshold curve using NP detectors, dotted line
The detection threshold curve of the Bayesian detection device using the present invention is represented, Filled Rectangle represents target in the detection residing for Bo Mennei
Cell position.Fig. 5 (a) show the present embodiment in the first scenario using Bayesian detection device and NP detector detection thresholds
Comparison diagram, Fig. 5 (b) show the present embodiment in the latter case using Bayesian detection device and NP detector detection thresholds
Comparison diagram.
It is relatively low in the prediction immediate vicinity region of feedback that comprehensive Fig. 5 can be seen that Bayesian detection thresholding;Further away from prediction
Center, detection threshold are higher.
Table 1 is the quantitative comparison one of many detection probability data interconnection MD-PDA methods and JDTP methods joint event of the present invention
Look at table.As can be seen from Table 1, in the first scenario, the relatively many detection probability data interconnection MD-PDA methods of JDTP methods are obtained
To less joint event;In the latter case, the joint event number that two methods are obtained is almost identical.In high false alarm rate
In the case of, compared to original many detection probability data interconnection MD-PDA methods, the present invention can substantially reduce joint thing
The quantity of part;If false alarm rate is too low, compare with original many detection probability data interconnection MD-PDA methods, the present invention cannot
The quantity of joint event is reduced substantially.
The quantity list of 1 joint event of table
It is mutual that the employing present invention with reference to shown in Fig. 6 compares normal probability data interconnection PDA methods, many detection probability data
The tracking accuracy of linked method MD-PDA and JDTP methods of the present invention, precision are come using root-mean-square error (Root MSEs, RMSE)
Weigh.In Fig. 6, ordinate represents root-mean-square error, and abscissa represents the moment to target following, and Fig. 6 plants pecked line and represents standard
Probability data interconnects root-mean-square error RMSE of PDA methods, and the dotted line with cross represents many detection probability data interconnection MD-PDA
Root-mean-square error RMSE of method, dotted line represent Bayes's carat Metro lower limit of many detection probability data interconnection MD-PDA methods
BCRLB, the dotted line with plus sige represent root-mean-square error RMSE of JDTP methods of the present invention, and solid line represents JDTP methods of the present invention
Bayes carat Metro lower limit BCRLB.Fig. 6 (a) represents multiple radar system in the first scenario using distinct methods to target
The accuracy comparison figure of tracking, Fig. 6 (b) represent that multiple radar system adopts essence of the distinct methods to target following in the latter case
Degree comparison diagram.
Comprehensive Fig. 6 is can be seen that in the case of high false alarm rate, and the present invention is relative to other two methods in estimated accuracy
Aspect has higher raising;In the case of low false alarm rate, it is only capable of tracking essence using the Bayesian detection device of the present invention
Degree is slight to be improved.
Claims (6)
1. target detection tracking combination treatment method in a kind of multiple radar system, comprises the steps:
(1) according to the following formula, build target movement model:
ξTk=F ξTk-1+uTk-1
Wherein, ξTkThe motion state of Tk moment targets is represented, F represents the transfer matrix of dbjective state in the case of uniform motion,
ξTk-1Represent the motion state of Tk-1 moment targets, uTk-1It is that zero, covariance is Q to represent that average is obeyed in distributionTk-1Gaussian Profile
Motion process noise;
(2) according to the following formula, build target measurement model:
Wherein,In expression t multiple radar system, n-th radar obtained j-th Noise of threshold value to target measurement
Measuring value, hn,t(ξt) represent t multiple radar system in n-th radar to dbjective state ξtMeasuring value, wn,tWhen representing t
In carving multiple radar system, n-th radar distribution obedience average is that zero, covariance is Σn,tGaussian Profile measurement noise, vn,t
Represent that n-th radar tracking Bo Mennei obeys equally distributed false measuring value in multiple radar system in t;
(3) initialize:
(3a) using the prior information of dbjective state as target initial time to be tracked tracking mode;
(3b) according to the following formula, calculate the initial covariance of target state:
C0=CRLB (y0)
Wherein, C0Represent the initial covariance of target state, CRLB (y0) represent tracking mode y that initial time target is moved0
Carat Metro limit;
(4) radar of selection target detection:
From multiple radar system, radar of the optional radar as target detection;
(5) determine CFAR detection threshold value:
(5a) according to the following formula, calculate running parameter of the detector of target detection radar in constant false alarm rate condition:
Wherein, ηkRepresent running parameter of the detector of k moment target detection radars in constant false alarm rate condition, VkRepresent the k moment
The size of target detection radar tracking ripple door, μkRepresent that the radar of k moment target detections obtains the average signal-to-noise ratio of echo, P tables
Show the false alarm rate required by the radar of target detection;nzRepresent the dimension that target detection radar is measured to target, ΣkRepresent the k moment
The measurement noise covariance of target detection radar;
(5b) according to the following formula, calculate the CFAR detection threshold value of target detection radar:
Wherein, γkRepresent the detection threshold value of k moment target detection radars, μkRepresent that the radar of k moment target detections is returned
The average signal-to-noise ratio of ripple, ln () are represented and take from right log operations, ηkRepresent the detector of k moment target detection radars permanent empty
Running parameter during alert rate condition, VkThe size of the radar tracking ripple door of k moment target detections is represented,Represent
Obeying average isVariance isNormal distribution,Represent l-th detection of the radar of k moment target detections in tracking gate
Measuring value at unit,Measuring value of the radar of k moment target detections to target prediction state is represented,Denotation coordination is converted
Measurement covariance value of the radar of k moment target detections to target prediction state afterwards;
(6) determine and effectively measure:
Using all measurements more than detection threshold value in target detection radar measurement value as effectively measurement;
(7) in judging multiple radar system, whether all of target detection radar has selected, if so, then execution step (8), otherwise,
Execution step (4);
(8) according to the following formula, calculate the quantity of combinatorial association event:
Wherein, N 'eThe quantity of e moment combinatorial association events is represented, S represents the sum of radar in multiple radar system, and ∏ represents that company takes advantage of
Operation, s represent the label of each radar in multiple radar system,Represent from Ms,r1 measurement source is selected in individual measurement in mesh
Target permutation and combination is operated,Represent all M of s-th radar in r moment multiple radar systemss,rIndividual measurement all derives from false-alarm,
Ms,rEffective measurement sum of s-th radar in r moment multiple radar systems is represented, the value of r is identical with e;
(9) according to the following formula, update current goal tracking mode:
Wherein, ydRepresent the d moment update after target following state, ∑ represents sum operation, yJ,dRepresent d moment combinatorial association things
State in part after j-th event update, J represent the label in all combinatorial association events,Represent all combinations of d moment
The probability that j-th event occurs in joint event;
(10) according to the following formula, update the covariance of target following state:
Wherein, CdThe dbjective state covariance after updating is represented, ∑ represents sum operation, and J is represented in all combinatorial association events
Label,Represent the probability of j-th event generation in c moment all combinatorial association events, CJ,dRepresent yJ,dCovariance value,
yJ,dState in expression d moment combinatorial association events after j-th event update, ()TRepresenting matrix transposition is operated, ydRepresent d
Target following state after moment renewal;
(11) judge whether target following flight path dissipates, if so, then execution step (12), otherwise, current time adds 1, will tracking
The current tracking mode of device feeds back to execution step (4) after inspection center;
(12) target following terminates.
2. in multiple radar system according to claim 1, target detection tracks combination treatment method, it is characterised in that step
(5b) measuring value of the radar of the k moment target detections described in target prediction stateObtained by following formula:
Wherein,Represent measuring value of the k moment target detection radars to target prediction state, hk() represents k moment target detections
Measuring value to dbjective state, F represent the transfer matrix of dbjective state in the case of uniform motion, yvRepresent v moment many radar systems
Tracking mode of the system to target.
3. in multiple radar system according to claim 1, target detection tracks combination treatment method, it is characterised in that step
(5b) measurement covariance value of the radar of k moment target detections to target prediction state after the coordinate transform described inBe by
What following formula was obtained:
Wherein,Measurement covariance value of the i-th target detection radar of k moment to target prediction state after denotation coordination conversion,
Hi,kThe Jacobian matrix that i-th target detection radar of k moment is measured to target is represented, F represents target-like in the case of uniform motion
The transfer matrix of state,Represent prediction covariance value of the current time v multiple radar system to target, QvRepresent current time v mesh
Mark motion process noise covariance value, ()TRepresenting matrix transposition is operated.
4. in multiple radar system according to claim 1, target detection tracks combination treatment method, it is characterised in that step
(8) the combinatorial association event described in is all Possible events that target measures that model measurement derives from target or false-alarm.
5. in multiple radar system according to claim 1, target detection tracks combination treatment method, it is characterised in that step
(9) probability that j-th event occurs in the d moment all combinatorial association events described inObtained by following formula:
Wherein,The probability of j-th event generation in d moment all combinatorial association events is represented, ∏ represents that company takes advantage of operation, i tables
Show the radar label in multiple radar system, N represents the sum of radar in multiple radar system, and p { } is represented and asked probability to operate,
Represent that the individual measurement sources of J (i) of i-th radar in m moment multiple radar systems, in the event of target, represent i-th radar
Measurement all derives from false-alarm.
6. in multiple radar system according to claim 1, target detection tracks combination treatment method, it is characterised in that step
(11) the target following flight path diverging described in is referred to, in multiple radar system, when target is not in the detection model of multiple radar system
When enclosing interior, the flight path of target following will dissipate.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610817863.7A CN106526584A (en) | 2016-09-12 | 2016-09-12 | Target detection and tracking combined processing method in multi-radar system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610817863.7A CN106526584A (en) | 2016-09-12 | 2016-09-12 | Target detection and tracking combined processing method in multi-radar system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106526584A true CN106526584A (en) | 2017-03-22 |
Family
ID=58343892
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610817863.7A Pending CN106526584A (en) | 2016-09-12 | 2016-09-12 | Target detection and tracking combined processing method in multi-radar system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106526584A (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106970370A (en) * | 2017-04-25 | 2017-07-21 | 西安电子科技大学 | Radar Multi Target tracking optimization method based on chaotic neural network |
CN107390180A (en) * | 2017-07-18 | 2017-11-24 | 中国人民解放军国防科学技术大学 | The multifunction radar signal estimation method of model is represented based on predicted state |
CN107479051A (en) * | 2017-07-18 | 2017-12-15 | 中国人民解放军国防科学技术大学 | The Operating Modes of Multi-function Radar discrimination method of model is represented based on predicted state |
CN108089183A (en) * | 2017-11-28 | 2018-05-29 | 西安电子科技大学 | A kind of detecting and tracking integral method for asynchronous multi-static radar system |
CN108802720A (en) * | 2018-04-27 | 2018-11-13 | 西安电子科技大学 | The cooperation detection and power distribution method of target following in a kind of multiple radar system |
CN109901152A (en) * | 2019-03-25 | 2019-06-18 | 西安电子科技大学 | The asynchronous more base sensing system target detections of moving platform track combined estimation method |
CN110673134A (en) * | 2019-10-18 | 2020-01-10 | 南京市德赛西威汽车电子有限公司 | Track tracking optimization method under radar long-short alternating range mode |
CN111142100A (en) * | 2020-01-06 | 2020-05-12 | 南京鹰目电子科技有限公司 | Radar target tracking method |
CN112381076A (en) * | 2021-01-18 | 2021-02-19 | 西南石油大学 | Method for preprocessing picture in video significance detection task |
CN113534133A (en) * | 2021-07-21 | 2021-10-22 | 西安电子科技大学 | Radar multi-target joint detection and tracking method based on expectation maximum iteration algorithm |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6246718B1 (en) * | 1997-09-08 | 2001-06-12 | Tut Systems, Inc. | Method and apparatus for decoding a bit sequence for transmission over pots wiring |
CN102628936B (en) * | 2012-04-12 | 2013-06-26 | 杭州电子科技大学 | Method for integrally detecting and tracking motorized dim target based on information mutual feedback |
CN105929391A (en) * | 2016-04-24 | 2016-09-07 | 西安电子科技大学 | Variable structure interacting multiple model radar maneuvering target tracking method based on posterior probability |
-
2016
- 2016-09-12 CN CN201610817863.7A patent/CN106526584A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6246718B1 (en) * | 1997-09-08 | 2001-06-12 | Tut Systems, Inc. | Method and apparatus for decoding a bit sequence for transmission over pots wiring |
CN102628936B (en) * | 2012-04-12 | 2013-06-26 | 杭州电子科技大学 | Method for integrally detecting and tracking motorized dim target based on information mutual feedback |
CN105929391A (en) * | 2016-04-24 | 2016-09-07 | 西安电子科技大学 | Variable structure interacting multiple model radar maneuvering target tracking method based on posterior probability |
Non-Patent Citations (3)
Title |
---|
JUNKUN YAN 等: ""Joint Detection and Tracking Processing Algorithm or Target Tracking in Multiple Radar System"", 《IEEE SENSORS JOURNAL》 * |
JUNKUN YAN 等: ""Power Allocation Algorithm for Target Tracking in Unmodulated Continuous Wave Radar Network"", 《IEEE SENSORS JOURNAL》 * |
严俊坤: ""认知雷达中的资源分配算法研究"", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106970370B (en) * | 2017-04-25 | 2019-10-01 | 西安电子科技大学 | Radar Multi Target based on chaotic neural network tracks optimization method |
CN106970370A (en) * | 2017-04-25 | 2017-07-21 | 西安电子科技大学 | Radar Multi Target tracking optimization method based on chaotic neural network |
CN107390180A (en) * | 2017-07-18 | 2017-11-24 | 中国人民解放军国防科学技术大学 | The multifunction radar signal estimation method of model is represented based on predicted state |
CN107479051A (en) * | 2017-07-18 | 2017-12-15 | 中国人民解放军国防科学技术大学 | The Operating Modes of Multi-function Radar discrimination method of model is represented based on predicted state |
CN108089183A (en) * | 2017-11-28 | 2018-05-29 | 西安电子科技大学 | A kind of detecting and tracking integral method for asynchronous multi-static radar system |
CN108089183B (en) * | 2017-11-28 | 2021-10-08 | 西安电子科技大学 | Detection and tracking integrated method for asynchronous multi-base radar system |
CN108802720A (en) * | 2018-04-27 | 2018-11-13 | 西安电子科技大学 | The cooperation detection and power distribution method of target following in a kind of multiple radar system |
CN108802720B (en) * | 2018-04-27 | 2022-02-22 | 西安电子科技大学 | Cooperative detection and power distribution method for target tracking in multi-radar system |
CN109901152A (en) * | 2019-03-25 | 2019-06-18 | 西安电子科技大学 | The asynchronous more base sensing system target detections of moving platform track combined estimation method |
CN110673134A (en) * | 2019-10-18 | 2020-01-10 | 南京市德赛西威汽车电子有限公司 | Track tracking optimization method under radar long-short alternating range mode |
CN111142100A (en) * | 2020-01-06 | 2020-05-12 | 南京鹰目电子科技有限公司 | Radar target tracking method |
CN111142100B (en) * | 2020-01-06 | 2022-08-09 | 南京鹰目电子科技有限公司 | Radar target tracking method |
CN112381076A (en) * | 2021-01-18 | 2021-02-19 | 西南石油大学 | Method for preprocessing picture in video significance detection task |
CN113534133A (en) * | 2021-07-21 | 2021-10-22 | 西安电子科技大学 | Radar multi-target joint detection and tracking method based on expectation maximum iteration algorithm |
CN113534133B (en) * | 2021-07-21 | 2022-10-21 | 西安电子科技大学 | Radar multi-target joint detection and tracking method based on expectation maximum iteration algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106526584A (en) | Target detection and tracking combined processing method in multi-radar system | |
CN106249219B (en) | SAR moving target detection methods based on adaptive matched filter response | |
US8502731B2 (en) | System and method for moving target detection | |
Metcalf et al. | A machine learning approach to cognitive radar detection | |
CN108089183A (en) | A kind of detecting and tracking integral method for asynchronous multi-static radar system | |
Liang-Qun et al. | Intuitionistic fuzzy joint probabilistic data association filter and its application to multitarget tracking | |
Chen et al. | Multi-target tracking algorithm based on PHD filter against multi-range-false-target jamming | |
CN107942308A (en) | Adaptive Rao detection methods based on gamma texture under complex Gaussian environment | |
JP6773606B2 (en) | Radar device and its radar signal processing method | |
CN104714225A (en) | Dynamic programming tracking-before-detection method based on generalized likelihood ratios | |
CN105301584A (en) | IPPHDF maneuvering multi-target tracking method of simultaneously solving range ambiguities | |
Guo et al. | Passive tracking in heavy clutter with sensor location uncertainty | |
CN106353743B (en) | It is matched with the nearly optimal radar target detection method of equivalent shapes parameter | |
Pérez et al. | Range detection on time-domain FMCW radar signals with a deep neural network | |
Wei et al. | A novel weak target detection strategy for moving active sonar | |
CN105353353A (en) | Multi-target tracking method through multi-search particle probability hypothesis density filter | |
Bae et al. | Adaptive PRF selection technique for multiple targets in track-before-detect | |
Pearce et al. | Regional trajectory analysis through multi-person tracking with mmWave radar | |
Kouba et al. | Recurrent neural networks for radar target identication | |
Luo et al. | Recovery of moving targets for a novel super-resolution imaging radar with PSO-SRC | |
Cheng et al. | Novel polarimetric detector for target detection in heterogeneous clutter | |
Hua et al. | Track initiation in monostatic-bistatic composite high frequency surface wave radar network based on NFE model | |
Kabakchiev et al. | Multisensor detection in randomly arriving impulse interference using the Hough transform | |
Ma et al. | Distributive JPDAF for multi-target tracking in wireless sensor networks | |
Kabakchiev et al. | Netted radar Hough detector in randomly arriving impulse interference |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170322 |
|
WD01 | Invention patent application deemed withdrawn after publication |