Background technique
Radar track start algorithm could can be divided into two major classes, i.e. sequential processes technology and batch system.In general sequenceProcessing technique, which is used primarily in the less environment of clutter, to be started to originate track, there is direct-vision method, logical approach etc.；And in strong clutter ringUnder border, batch system is particularly suited for processing track initiation, and batch system mainly includes Hough transform method and others oneA little innovatory algorithms.Studies have shown that sequential processes technology can preferably handle track initiation problem in relatively quiet environment；And in a noisy environment, Hough transform method can processing track initiation problem relatively accurately, but in such caseUnder, Hough transform method wants computation burden larger, unsuitable when also just causing to use in engineering because its calculation amount is excessive.CauseThis provides a kind of accurate and rapidly track initiation method, is of great significance.
There is patent CN106054150A (" a kind of first to originate the radar track initial mode confirmed afterwards ") introduction in the country at presentA kind of radar track initial mode, by setting actually detected judgement false-alarm probability, so that it is determined that the method for detection threshold.It shouldTargetpath initial probability can be improved compared with existing track initiation method in method, but algorithm realize it is complicated, scene adaptability andProject Realization is poor.
Patent CN107436434A (" the track initiation method based on two-way Doppler estimation ") describes a kind of based on moreThe general track initiation method for strangling information can assist track initiation association algorithm by echo Doppler information, effective to rejectFalseness measures vector, it is suppressed that the formation of false track, and then under conditions of same false track probability, improve targetDetection probability has opened up far detection range.But track initiation stage doppler information, which calculates, to be obscured, and engineering calculation amount is big.
Patent CN106896363A (" a kind of submarine target active tracing track initiation method ") describes a kind of underwater meshTrack initiation method is marked, data are acquired by navigation sensor and sonar sensor, assessed value is obtained and calculates wave door thresholding, moreNew potential track.This method can be suitable for different marine environment can be accurately and efficiently right particularly with high clutter environmentTarget trajectory carries out track initiation, but this method is higher to signal processing system requirement of real-time, and engineering calculation amount is big.
Open source literature " airborne early warning radar sea multiple target boat in " radar science and technology " periodical of 6th phase in 2016The research of mark start algorithm could " describe a kind of sea multi-target traces initial mode.The algorithm is constructed using three frame amount measured dataIt with the primary wave door for deferring sentence structure, overcomes traditional primary wave door and track division and composition problem easily occurs, and pass throughAngle restriction improves traditional related wave door to modified Hough transform.But its limitation is that the algorithm is only applicable toUnder the conditions of airborne radar is to sea, application space has limitation, and project amount is larger, and real-time performance is poor.
" probability based on heuristic logic is false for open source literature in " Shanghai communications university's journal " periodical of 1st phase in 2018If density filtering high-efficiency track initial mode " a kind of high-efficiency track initial mode is described, this method is in regular length sliding windowThe constraint conditions such as speed, acceleration and angle are set, validation checking and confirmation are carried out to the potential measurement for generating newborn target,With the false newborn target of reduction.This method is compared with traditional PHD filtering for surveying starting based on full dose, and the method proposed is substantiallyImprove computational efficiency.But its limitation is that this method track initiation precision is not high, and algorithm is also and remarkable, thus engineeringRealization has little significance.
Summary of the invention
The purpose of the present invention is to provide a kind of improved track initiation methods, with target position, speed, acceleration, angleIt spends variable and calculates thresholding to echo progress preliminary screening, carry out measurement echo in the association algorithm for carrying out target signature auxiliary informationWith being associated with for target, the purpose of accurate initial target track is realized.
In order to achieve the above object, the invention is realized by the following technical scheme:
A kind of improved track initiation method comprising the steps of:
S1, track initiation point mark is established, into S2；
S2, direct-vision method determine wave door thresholding, and lock second starting point mark, enter S3；
S3, it is extrapolated according to the first two starting point mark, centered on predicted value, using course error covariance value as tracking waveThe thresholding of door, into S4；
If not fallen in wave door thresholding in S4, scanning element mark, track revocation will have been originated, has returned to step S1 continuationIt carries out.If being fallen into wave door thresholding in scanning element mark, following scanning is dropped into the point mark in tracking threshold according to target spyThe association algorithm of reference breath auxiliary is associated, into S5；
After S5, the association of scanning element mark, step S3 is continued to execute, until meeting M/N logical approach, determines track initiation success.
Compared with prior art, the present invention with target position, speed, acceleration, angle variables calculate thresholding to echo intoRow preliminary screening measure in the association algorithm for carrying out target signature auxiliary information the associated method of echo and target, purportIt is suitable for the simultaneous complex scene of dense clutter noise providing one kind, computation complexity is low, and real-time is high, track initiationAccuracy is high, and software algorithm is realized simply, convenient for the track initiation method of the application in engineering in practice.
Specific embodiment
Below based on specific embodiment, next the present invention is described in detail.
In the present embodiment, it is assumed that in radar sensor scanning area, there are 5 targets to do linear uniform motion, it is assumed that itsIn two-dimensional surface, the initial position of this 5 targets is respectively as follows: (55000m, 55000m), (45000m, 45000m), (35000m,35000m), (45000m, 25000m), (55000m, 15000m), the initial velocity of this 5 targets are (500m/s, 0m/s).Assuming that radar sensor scan period T is 5 seconds, and the distance of radar sensor observation standard deviation and azimuth observation standard differenceNot are as follows: σr=40m and σθ=0.3 °.Assuming that 4 periods of radar sensor continuous scanning, the number for the clutter that each intermittent scanning arrivesAmount is to obey Poisson distribution, the number average value of clutter be by the square simulating area that is considered size andIn unit area determined by the product of clutter number.
Assuming that limiting the maximum speed in correlation rule in Track initialization algorithm are as follows:And minimum speedAre as follows:Peak acceleration amax=50m/s2The angle continuously scanned three timesAssuming that based on patrolling3/4 logical approach is used in the Track initialization algorithm for the method for collecting, each radar sensor scans preceding 5 points in resulting trace setIt is set to target point, the thresholding value 4 of parameter space enables Nθ=90, Nρ=90.
As shown in Figure 1, method proposed by the present invention specifically executes following steps:
S1, using preliminary sweep point mark as first starting point mark, track initiation point mark is established with this, is transferred to S2.
S2, threshold value is determined using direct-vision method, the point mark fallen into thresholding that second is scanned is as potential trackInitially set up.
Assuming that ri(i=1,2), ti(i=1,2) is the position measuring value of target twice sweep, time value, the amount scannedThe movement velocity of survey should be between maximum speed VmaxWith minimum speed VminBetween, it may be assumed that
R in formulaiFor target span, tiFor target run duration.The constraint of speed greatly limits the range of wave door, especiallyIt is in the environment of strong clutter, clutter number is more, leverages the data correlation of target.
S3, in step S2 as a result, carrying out linear extrapolation obtains each possible track, it is pre- with this momentPoint centered on measured value, the threshold value size by course error covariance value as tracking gate.
It is as follows that course error covariance calculates wave door threshold value method:
If zi(k) be i-th of k moment measurement information, wherein i=1,2 ..., mk.Then measure Zi(k) and Zj(k+1)Distance vector dijAre as follows:
dij(t)=max [0, zj(k+1)-zi(k)-vmaxt]+max[0,-zj(k+1)+zi(k)+vmint] (1)
In formula, t is the time interval of radar scanning twice, generally assumes that error in measurement is zero-mean, independent white GaussianNoise profile, enabling its covariance is Ri(k), then normalized cumulant square are as follows:
D in formulaijIt (k) is obedience χ2The stochastic variable of distribution, freedom degree are that (p is according to Practical Project concrete condition system by pIt is fixed).Distance threshold value γ can be obtained by freedom degree p inquiry, if Dij(k)≤γ, then measure Zi(k) and ZjIt (k+1) is phaseIt is associated.
Similarly, speed, acceleration and angle threshold value calculating method can ibid obtain, wherein measuring Zi(k) and Zj(k+1)Velocity vector are as follows:
Acceleration are as follows:
Angle vector are as follows:
If not fallen in wave door thresholding in S4, scanning element mark, track revocation will have been originated, has returned to step S1 continuationIt carries out.If being fallen into wave door thresholding in scanning element mark, following scanning is dropped into the point mark in tracking threshold according to target spyThe association algorithm of reference breath auxiliary is associated.
Assuming that a is the amplitude information of i-th of k moment measurement, the probability for the amplitude information for enabling it measure from real goalDensity function is p1(a), the probability density function of the false amplitude information measured is p0(a).Then:
D is signal-to-noise ratio (or signal to noise ratio) SNR in formula.
Enabling τ is amplitude information detection threshold, then false-alarm probability and detection probability are respectivelyWithThen:
Increase detection probabilityThen need to reduce amplitude information detection threshold τ, but this also results in false-alarm probabilityIncrease, therefore to reasonably select amplitude information detection threshold τ, the selection of general τ is by false-alarm probabilityIt obtains, it may be assumed that
It is measured by the real goal exported after amplitude information detection threshold τ and the probability of the false amplitude information measured is closeSpending function isWithThen:
It can to sum up obtain:
Enabling amplitude likelihood ratio ρ is the probability density function for the amplitude information that real goal measuresIt is measured with falsenessThe probability density function of amplitude informationThe ratio between, it may be assumed that
Assuming that i-th of measuring value of k+1 moment is zi(k+1), target prediction value is z (k+1 | k), target signature information auxiliaryAssociation algorithm principle be by tracking gate predicted position value of the echo from target and echo amplitude likelihood ratio it is minimumMeasurement as Targets Dots, i.e.,Reach minimum, which then can be used for dbjective state moreIn new.
After S5, the association of scanning element mark, step S3 is continued to execute, (M/N has according to Practical Project until meeting M/N logical approachBody situation is formulated), determine track initiation success.
Although the contents of the present invention are discussed in detail through the above steps, but it should be appreciated that the description above is notIt should be considered as limitation of the present invention.After those skilled in the art have read above content, a variety of repaired for of the inventionChanging and substituting all will be apparent.Therefore, protection scope of the present invention should be limited to the appended claims.