CN105223559A - A kind of long-range radar track initiation method switched that walks abreast - Google Patents

Abstract

The invention discloses a kind of long-range radar track initiation method switched that walks abreast, correction Hough transform track initiation method based on a step time delay is revised the parallel running of Hough transform track initiation method with quick, wherein introduce discount judgement factor Dis to switch the judgement of two kinds of Track initialization algorithms, and the comprehensive evaluation to track initiation success ratio, false track inception rate and initial time.Use when lower hybrid wave density revise fast Hough transform can track initiation method fast, the used time is shorter.Adopt the correction Hough transform based on a step time delay when high clutter density, the method can good initial flight path in such a case.The present invention could the fastest the most initial flight path by solving to the perfection of the weight of principal element coupling when this switch.

Description

A kind of long-range radar track initiation method switched that walks abreast

Technical field

The invention belongs to radar signal processing field, be specifically related to a kind of long-range radar track initiation method switched that walks abreast.

Background technology

Track initiation is as the first step of following the tracks of, and can realize the quick initial of flight path be obviously one of key factor that realize following the tracks of fast.Track initiation all exists when various flight path processing, and in single goal quiet environment and multiple goal quiet environment situation, track initiation process is relatively simple, and the track initiation process under single goal noisy environment and multiple goal noisy environment is relatively complicated.Multi-target traces is managed, track initiation is the first step, it is the basis of carrying out Track In Track (maintenance), in recent years along with the appearance of complicated new technology, war environment becomes increasingly complex, also occurred a lot of achievement in research at multi-target traces management domain, but mainly for the aspect such as Track Fusion, target following, the achievement in research of track initiation aspect is little.And for engineering project, relate generally to the track initiation method of probability, likelihood function calculating, as the discussion in method, can only can not be used for engineering project.How to find a kind of algorithm, a compromise preferably can be found between fast Track Initiation and correct track initiation probability, be convenient to Project Realization, become the emphasis of research.

Standard Hough transform method---the method belongs to batch processing method, and the metric data of a past TotalScan scan period processes by it simultaneously, to determine possible targetpath.The ultimate principle of Hough transform will a bit transform to a curve in parameter space or curved surface in measurement space; And the point with same parametric characteristics can intersect by parameter space after the conversion; By judging that the accumulation degree of point of intersection has been detected characteristic curve, thus judge whether true flight path exists.Because Hough transform has the insensitive of SOL and the robustness to random noise, and be applicable to the advantages such as parallel processing, become the study hotspot of current track initiation method.Random Hough transformation method (RHT)---RHT is a kind of Probabilistic Hough Transform method proposed by LeiXu etc.Compared with standard Hough transform, random Hough transformation uses three new operation mechanisms: map in the convergence of the dynamic link list of image space random sampling, parameter space, connection layout image space and parameter space.RHT adopts many to a mapping, avoids the huge calculated amount caused by standard Hough transform one to map more; Adopt dynamic link table structure, reduce memory requirements.RHT is widely used in track initiation field.

Hough transform---the one that the method is proposed on standard Hough transform basis by ChenJ is improved one's methods in correction.Middle by the continuous T otalScan received by radar sweep time Transformed Measurement become parameter space, calculate oblique distance (parameter value) difference function; Judge according to track initiation two criterions:

(1) angle crossing zero cross point must be closely.

(2) slope sign crossing zero cross point place angle must be identical.

If satisfied condition, judge whether the angle between aimed acceleration and flight path meets certain constraint, as satisfied constraint, then these measurements just can form a flight path.Correction Hough transform basis adds again a condition, exactly measure must meet speedgate condition could use correction Hough transform transform to parameter space, correction Hough transform is made to transform to parameter space like this, make the gauge point quantity of correction Hough transform greatly reduce like this, reach the object of quick initial flight path.

But to the correction Hough transform Track initialization algorithm based on a step time delay, initial umber of beats TotalScan is more, and the track initiation time is longer, and track initiation success ratio is also not improved, false track inception rate improves.Based on the correction Hough transform Track initialization algorithm of a step time delay track initiation success ratio only when detection probability Pd=1 effect relatively good.For quick random Hough transformation Track initialization algorithm, initial umber of beats TotalScan is more, and the track initiation time there is no significant change, and track initiation success ratio improves, and false track inception rate improves.Clutter density Lambda is larger, and false track inception rate is higher.

Summary of the invention

The object of the present invention is to provide a kind of long-range radar track initiation method switched that walks abreast, to overcome the defect that above-mentioned prior art exists, the present invention adopts the weight of self-adaptation switching function to the principal element of track initiation to mate, and carry out the switching of track initiation method by discount judgement factor, can the fastest the most initial flight path.

For achieving the above object, the present invention adopts following technical scheme:

Walk abreast the long-range radar track initiation method switched, and comprises the following steps:

Step one: gathering the two-dimentional energy datum that radar returns, and calculate radar detedtion probability Pd, calculating clutter density Lambda from obtaining TotalScan beat of data;

Step 2: by following formulae discovery discount judgement factor Dis:

Dis＝0.2*3/TotalScan+0.6*Pd+0.2*Lambda/30；

Wherein, Dis represents discount judgement factor, and TotalScan represents initial umber of beats, and Pd represents radar detedtion probability, and Lambda represents clutter density;

Step 3: adopt quick random Hough transformation track initiation method as Dis<0.8, adopts the correction Hough transform track initiation method based on a step time delay when Dis >=0.8.

Further, also comprise step 4: arithmetical discount evaluation coefficient β, and adopt discount evaluation coefficient β to evaluate track initiation effect, β value less then track initiation effect is better;

$\mathrm{\&beta;}=\frac{P_F\&times;t}{P_T}$

Wherein, P _ttrack initiation success ratio, P _fbe false track inception rate, t is initial time.

Further, described quick random Hough transformation track initiation method comprises the following steps:

(1) select distance by radar plane (x, y) as measuring space, definition flight path parameter set is P=[p _c, score] dynamic link list, initialization flight path parameter set P=NULL, sampling number k=1, arranges threshold values T;

(2) data point D in definition of T otalScan bat _i=(x _i, y _i), wherein i=1,2,3...n, random selecting gauge point z from kth is clapped _i(k)=(x _i, y _i), random selecting gauge point z from kth+1 is clapped _j(k+1)=(x _j, y _j);

(3) the distance dist=|z between two gauge points is calculated _i(k)-z _j(k+1) |, obtain the speed v=dist/T between two gauge points _sif, v _min≤ v≤v _max, then enter step (4), otherwise, enter step (2); Wherein, T _srepresent the sampling interval of two gauge points, v _maxand v _minbe respectively maximal rate and the minimum speed of target;

(4) two determined straight line parameter p of gauge point are calculated _c=(θ ₀, ρ ₀);

(5) if k=1, go to step (6); Otherwise, in P, look for a p _c=(θ _c, ρ _c), if | θ _c-θ ₀|≤Δ θ and | ρ _c-ρ ₀|≤Δ ρ, then by p _cscore add 1; Otherwise by p _cinsert P, and make p _cscore be 1; Wherein, Δ θ and Δ ρ is allowable error;

(6) use sampling termination rules limited samples number of times, if satisfied rule, then kth bat and kth+1 clap sampling end, and k=k+1, goes to step (7); If do not meet rule, then go to step (2);

(7) if k>TotalScan, go to step (2); Otherwise, by p corresponding for score>=T in P _cextract, be kept in matrix Para, Para is the flight path parameter detected, and track initiation terminates.

Further, the described correction Hough transform track initiation method based on a step time delay comprises the following steps:

(1) data point D in definition of T otalScan bat _i=(x _i, y _i), wherein i=1,2,3...n, the maximal rate V of Offered target _maxwith minimum speed V _min;

(2) i-th gauge point z that the kth of getting radar is clapped _ik (), as the starting point of track initiation, carries out One-step Extrapolation, the jth gauge point z clapped with kth+1 _j(k+1) carry out the association of traversal formula, obtain kth clap and kth+1 clap the speed V between measuring _ij, will satisfy condition V _min≤ V _ij≤ V _maxtwo clap gauge points preserve;

(3) get meet step (2) conditional gauge point as the intermediate point of track initiation, carry out One-step Extrapolation, m that claps with k+2 measures z _m(k+2) carry out the association of traversal formula, obtain kth+1 and to clap and kth+2 claps speed V between gauge point _jm;

(4) calculate kth according to formula ρ=xcos θ+ysin θ and clap the parameter θ that gauge point and kth+1 clap gauge point _ijand ρ _ij, and be kept in candidate's flight path matrix P (k);

(5) calculate kth+1 according to formula ρ=xcos θ+ysin θ and clap the parameter θ that gauge point and kth+2 clap gauge point _jmand ρ _jm, compare with the parameter in P (k) respectively;

(6) if | θ _jm-θ _ij|≤Δ θ, and θ _ijand θ _jmthe slope at place is identical, and wherein Δ θ is permissible error, and be so saved in flight path matrix Para by this gauge point, Para is the flight path parameter detected, and track initiation terminates triple time simultaneously.

Compared with prior art, the present invention has following useful technique effect:

Correction Hough transform track initiation method based on a step time delay is revised the parallel running of Hough transform track initiation method with quick by the present invention, introduce discount judgement factor Dis to switch the judgement of two kinds of Track initialization algorithms, use when lower hybrid wave density revise fast Hough transform can track initiation method fast, the used time is shorter.Adopt the correction Hough transform based on a step time delay when high clutter density, the method can good initial flight path in such a case.The present invention could the fastest the most initial flight path by solving to the perfection of the weight of principal element coupling when this switch.Achieve when complicated and changeable by the switching of above two kinds of algorithms, can accomplish that two kinds of algorithms are arbitrated automatically and realize at different clutter density, different detection probability, different initial flight path more rapidly and accurately when detecting umber of beats.Under this can only adapt to specific environment than existing single calculation, initial flight path has had better effect.

Accompanying drawing explanation

Fig. 1 is track initiation method flow diagram of the present invention;

Fig. 2 is correction Hough transform track initiation result figure (TotalScan=3) based on a step time delay, and wherein (a) represents the discount judgement factor value under different clutter density; B () represents the discount evaluation coefficient value under different clutter density; C () represents the value of the track initiation success ratio under different clutter density; D () represents the value of the false track inception rate under different clutter density;

Fig. 3 is quick random Hough transformation track initiation result figure (TotalScan=3), and wherein (a) represents the discount judgement factor value under different clutter density; B () represents the discount evaluation coefficient value under different clutter density; C () represents the value of the track initiation success ratio under different clutter density; D () represents the value of the false track inception rate under different clutter density;

Fig. 4 is correction Hough transform track initiation result figure (TotalScan=10) based on a step time delay, and wherein (a) represents the discount judgement factor value under different clutter density; B () represents the discount evaluation coefficient value under different clutter density; C () represents the value of the track initiation success ratio under different clutter density; D () represents the value of the false track inception rate under different clutter density;

Fig. 5 is quick random Hough transformation track initiation result figure (TotalScan=10), and wherein (a) represents the discount judgement factor value under different clutter density; B () represents the discount evaluation coefficient value under different clutter density; C () represents the value of the track initiation success ratio under different clutter density; D () represents the value of the false track inception rate under different clutter density.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described in further detail:

See Fig. 1, because track initiation problem is different from static image procossing, it is a dynamic process, is apparent that target most or clutter all has velocity information no matter wherein show.Radar scanning image is not to the whole sampling measuring set, but samples by shooting, uses sliding window method cumulative information, uses Hough transform parallel data processing, realize track initiation.By finding the research based on the correction Hough transform Track initialization algorithm of a step time delay and quick random Hough transformation Track initialization algorithm, the track initiation success ratio of two kinds of algorithms, false track inception rate and initial time and initial umber of beats TotalScan, clutter density Lambda and radar detedtion probability Pd closely related.To the correction Hough transform Track initialization algorithm based on a step time delay, initial umber of beats TotalScan is more, and the track initiation time is longer, and track initiation success ratio is also not improved, and false track inception rate improves.Based on the correction Hough transform Track initialization algorithm of a step time delay track initiation success ratio only when detection probability Pd=1 effect relatively good.For quick random Hough transformation Track initialization algorithm, initial umber of beats TotalScan is more, and the track initiation time there is no significant change, and track initiation success ratio improves, and false track inception rate improves.Clutter density Lambda is larger, and false track inception rate is higher, but converts the fast of raising based on the correction Hough transform Track initialization algorithm of a step time delay than quick random Hough transformation Track initialization algorithm.But when detection probability Pd ≠ 1, this algorithm still has good track initiation effect.

Discount judgement factor Dis and discount evaluation coefficient β is introduced for above analysis the present invention.For switching the judgement of two kinds of Track initialization algorithms, the comprehensive evaluation of track initiation success ratio, false track inception rate and initial time.Discount judgement factor adopts following formula to calculate: Dis=0.2*3/TotalScan+0.6*Pd+0.2*Lambda/30.When Dis >=0.8 by algorithm changeover to the correction Hough transform based on a step time delay, so also can good initial correct flight path under high clutter density.As Dis<0.8 by algorithm changeover to revising Hough transform fast, like this can initial flight path faster.

1) random Hough transformation Track initialization algorithm

any analytic curve in image space, wherein, n parameter of curve.Can know, if obtain the parameter of a suite line, a curve can be determined by them so accordingly.For straight line, from binary edge map, extract 2 points randomly at every turn, obtain coordinate figure and be respectively (x ₁, y ₁) and (x ₂, y ₂), by solving an equation

$\left\{\begin{array}{c}{x}_{1}cos\mathrm{\&theta;}+y_{1}sin\mathrm{\&theta;}=\mathrm{\&rho;}\\ x_{2}cos\mathrm{\&theta;}+y_{2}sin\mathrm{\&theta;}=\mathrm{\&rho;}\end{array}\right.$

Can obtain one group (ρ, θ), and it is stored in dynamic link list and accumulates.For each group data, dynamic link list is made up of parameter value p and accumulating value score.Similarly, through several times stochastic sampling, obtain the parameter value in parameter space and accumulating value, the parameter thinking corresponding to the accumulating value reaching threshold value is exactly the parameter of image space cathetus.

Random Hough transformation Track initialization algorithm is as follows:

Input: data point D, D in all measurement spaces _i=(x _i, y _i) i=1,2,3...n.

(1) flight path parameter set is P=[p _c, score] and dynamic link list, initialization flight path parameter set P=NULL, sampling number k=1, define maximum sampling number K _max;

(2) random selecting two gauge point (x from D _i, y _i), (x _j, y _j);

(3) 2 determined straight line parameter p=(θ are calculated ₀, ρ ₀);

(4) if k=1, go to step (5); Otherwise, in P, look for a p _c=(θ _c, ρ _c), if | θ _c-θ ₀|≤Δ θ and | ρ _c-ρ ₀|≤Δ ρ (Δ θ and Δ ρ is allowable error, Δ θ=π/180*5, Δ ρ=1), then by p _cscore add 1; Otherwise p is inserted P, makes its score be 1;

(5) k=k+1, if k < is K _maxgo to step (2), otherwise sampling completes;

(6) by p corresponding for score>=T in P _cextract, be kept in matrix Para.

Export: Para is exactly the flight path parameter detected.

2) quick random Hough transformation Track initialization algorithm

Hypothetical target does linear uniform motion, selects distance by radar plane (x, y) as measuring space, sampling interval T _s.If z _i(k)=(x _i, y _i), i=1 ... l, _kbe i-th gauge point mark in k moment, measurement itself is containing velocity information.Different types of target has himself velocity range, assuming that maximal rate and minimum speed are respectively v _maxand v _min.The transforming function transformation function of Hough transform is:

ρ＝x _icosθ+y _isinθ

First, stochastic sampling utilize target travel information to reduce invalid sampling.Clap measurement from measure space adjacent two and randomly draw two some z _i(k) and z _j(k+1).Calculate the distance dist=|z between them _i(k)-z _j(k+1|) speed v=dist/T between 2, is obtained _s.If meet speed restriction v _min≤ v≤v _max, then sampling effectively; If do not meet speed restriction, be then considered as invalid sampled point, resampling, until meet speed restriction.

Secondly, calculate flight path parameter and accumulate.Calculate the flight path parameter p between 2 _c=(ρ, θ), and left in the middle of parameter chained list P, record the accumulating value score of this parameter simultaneously.If the flight path parameter p that certain calculates _cidentical with the parameter value in P, so only upgrading its corresponding accumulating value, depositing this parameter without the need to repeating again.

Again, termination rules limited samples number of times is used.Sampling termination rules is utilized to limit sampling number, if met some requirements, then end of sampling, otherwise continue sampling.Finally, flight path parameter is extracted.Setting threshold value T, for all accumulating value score in P, if score>=T, then extracts its corresponding parameter p _c, as the flight path parameter detected.

Quick random Hough transformation Track initialization algorithm step is as follows:

Input: data point D, D in all measurement spaces (comprising TotalScan to clap) _i=(x _i, y _i) i=1,2,3..n..

(1) flight path parameter set is P=[p _c, score] dynamic link list, initialization flight path parameter set P=NULL, sampling number k=1, arrange close value T;

(2) random selecting two gauge point z from D _i(k)=(x _i, y _i) and z _j(k+1)=(x _j, y _j);

(3) the distance dist=|z between them is calculated _i(k)-z _j(k+1|) speed v=dist/T between 2, is obtained _s.If v _min≤ v≤v _max, go to step (4); Otherwise, go to step (2);

(4) 2 determined straight line parameter p are calculated _c=(θ ₀, ρ ₀);

(5) if k=1, go to step (6); Otherwise, in P, look for a p _c=(θ _c, ρ _c), if | θ _c-θ ₀|≤Δ θ and | ρ _c-ρ ₀|≤Δ ρ (Δ θ and Δ ρ is allowable error, Δ θ=π/180*5, Δ ρ=1), then by p _cscore add 1; Otherwise by p _cinsert P, make p _cscore be 1;

(6) termination rules limited samples number of times is used.If meet rule, then kth bat and kth+1 are clapped to sample and are terminated, and k=k+1, goes to step (7); If do not meet rule, then go to step (2);

Sampling termination rules is defined as follows:

Assuming that the unknown parameter in sample is θ, parameter space is Θ.The hypothesis put forward as the prerequisite of inspection is called null hypothesis, is designated as hypothesis H ₀: θ ∈ Θ ₀.When null hypothesis is rejected, just means logically and the hypothesis that acceptance one is different be with it called alternative hypothesis, be designated as hypothesis H ₁: θ ∈ Θ ₁, one with H ₀for null hypothesis, H ₁for the Hypothesis Testing Problem of alternative hypothesis is designated as:

$H_0:\mathrm{\&theta;}\&Element;{\mathrm{\&Theta;}}_0\&LeftRightArrow;H_1:\mathrm{\&theta;}\&Element;{\mathrm{\&Theta;}}_1---\left(1\right)$

Wherein work as Θ ₁=Θ-Θ ₀time, alternative hypothesis is called the logic alternative hypothesis of null hypothesis.If Θ ₀, Θ ₁only containing a value, claim H ₀, H ₁meet hypothesis.

When (1) formula is tested, likely violate Type Ⅰ Ⅱ error.Error type I is: when null hypothesis is set up, because sample drops on the refusal null hypothesis of mistake in region of rejection, be called " abandoning very mistake ", probability is P (D=H ₁| H ₀).Error type II is: when null hypothesis is false, and that drops on mistake in acceptance domain due to sample accepts null hypothesis, and be called " depositing pseudo-mistake ", probability is P (D=H ₀| H ₁).In Hypothesis Testing Problem, this situation can run into usually: when wanting the probability of the criminal's of minimizing Error type I, and the probability violating error type II will increase; Vice versa.For this double-barreled question, according to the principle of protection null hypothesis, Newman-Pearson came proposes Newman-Pearson criterion: specify the positive number α that little in advance, P (D=H ₁| H ₀under the condition of)=α, make P (D=H ₀| H ₁) little as much as possible.The hypothesis accepted is determined by (2) formula:

$D=\left\{\begin{array}{ccc}{H}_0& if& \frac{P(X_1,X_2,\mathrm{...},X_n|H_0)}{P(X_1,X_2,\mathrm{...},X_n|H_1)}\&GreaterEqual;\mathrm{\&lambda;}\\ H_1& if& \frac{P(X_1,X_2,\mathrm{...},X_n|H_0)}{P(X_1,X_2,\mathrm{...},X_n|H_1)}<\mathrm{\&lambda;}\end{array}\right.---\left(2\right)$

Wherein, X ₁, X ₂..., X _nbe n sampling in sample space, the selection of λ should meet the requirement of given positive number α.

More than the describing of Hypothesis Testing Problem when number of samples is fixed.In many cases, if uncertain observation time or number of samples before observation, and deciding when stopping observing this problem be placed in observation process according to the situation of required result, likely obtaining and better printing result.Sequence Detection refers to, when making judgement, observation sample number or observation time are not fixing test of hypothesis, what namely observation was each time made sentences two anything but, but three, namely accept null hypothesis, accept alternative hypothesis, continue observation, Wa Erte Sequence Detection that Here it is.The decision made is described by (3) formula:

$\left\{\begin{array}{ccc}{H}_0& if& \frac{P(X_1,X_2,\mathrm{...},X_n|H_0)}{P(X_1,X_2,\mathrm{...},X_n|H_1)}>A\\ H_1& if& \frac{P(X_1,X_2,\mathrm{...},X_n|H_0)}{P(X_1,X_2,\mathrm{...},X_n|H_1)}<B\\ {\mathrm{SampleX}}_{n+1}& Otherwise& \end{array}\right.---\left(3\right)$

Wherein, A, B are the P (D=H according to specifying respectively ₀| H ₁) and P (D=H ₁| H ₀) the threshold value calculated of value.

In random Hough transformation Track initialization algorithm, the determination of optional sampling number can be realized by Sequence Detection.Assuming that in front n stochastic sampling, have recorded altogether the individual different flight path parameter value of k in dynamic link list, X _jrepresent a jth flight path parameter, X _j∈ { X ₁, X ₂..., X _k; X _jcorresponding flight path parameter accumulating value is n _jwhole random sampling procedure should observe following three rules:

L () uses the least possible stochastic sampling to complete the Detection task of flight path parameter;

(2) number of stochastic sampling should be more than or equal to 2;

(3) a jth flight path parameter X in dynamic link list _jprobability can be described as P (X=X _j)=p _j, j=1,2 ..., k.J _max=argmaxn _jrepresent the flight path parameter with maximum accumulating value sequence number in dynamic link list, so the probability of the flight path parameter of its correspondence obeys distribution. ${\mathrm{\&Theta;}}_{j_{max}}=\left\{\mathrm{\&theta;}\right|P_{j_{max}}>P_j,\&ForAll;j\&NotEqual;j_{\max }\}.$

In front n stochastic sampling, if probability with probability P _jmeet certain relation, so just can be considered to final flight path parameter.Now, can terminate sampling.Due to with entirety it is unpractical for comparing, and problem can be made more complicated, so only need more maximum accumulating value and the secondary parameter probability that accumulating value is corresponding greatly.

Make the following assumptions:

H ₀: the parameter that in dynamic link list, maximum accumulating value is corresponding it is finally initial flight path parameter.

H ₁: the parameter that in dynamic link list, secondary large accumulating value is corresponding it is finally initial flight path parameter.

Random Hough transformation must carry out according to the sampling rule of (3) formula when sampling.If error probability given is in advance P (D=H ₁| H ₀)=α, P (D=H ₀| H ₁)=β is by deriving, and can draw α, the relation of β and A, B is as follows:

$A=\frac{1-\mathrm{\&beta;}}{\mathrm{\&alpha;}}---\left(4\right)$

$B=\frac{\mathrm{\&beta;}}{1-\mathrm{\&alpha;}}---\left(5\right)$

According to protection null hypothesis H ₀thought, the selection of α, β must satisfy condition A>1>B>0.Describe according to above, (3) formula, (4) formula and (5) formula just constitute the sampling termination rules of random Hough transformation track initiation.

(7) if k>TotalScan, go to step (2); Otherwise, by p corresponding for score>=T in P _cextract, be kept in matrix Para, track initiation terminates.

Export: Para is exactly the flight path parameter detected.

3) based on the correction Hough transform Track initialization algorithm of a step time delay

This algorithm can overcome a step time delay Track initialization algorithm cannot be applied to dense clutter environment, revises the shortcoming that Hough transform Track initialization algorithm traversal calculated amount is unfavorable for greatly realizing, rapidly and efficiently realize track initiation.This algorithm adopts the many hypothesis algorithm process associated data improved, and adopts sliding window method and multiple filter to carry out flight path information accumulation simultaneously, the final flight path by Hough transform algorithm determination target.

Algorithm steps is as follows:

Input: data point (x in all measurement spaces _i, y _i), x _iand y _ithe positional information that representative measures, i=1,2,3...n.

(1) maximal rate V is set _maxwith minimum speed V _min, continuous triple time is measured process below.

(2) i-th measurement z that the kth of getting radar is clapped _ik (), as the starting point of track initiation, carries out One-step Extrapolation, the jth of clapping with kth+1 measures z _j(k+1) carry out the association of traversal formula, obtain kth clap and kth+1 clap the speed V between measuring _ij.To be satisfied condition V _min≤ V _ij≤ V _maxtwo clap measurements preserve.

(3) get the intermediate point of measurement as track initiation of satisfy condition (2), carry out One-step Extrapolation, m that claps with k+2 measures z _m(k+2) carry out the association of traversal formula, obtain kth+1 clap and kth+2 clap the speed V between measuring _jm.

(4) calculate kth bat measurement according to formula ρ=xcos θ+ysin θ and clap with kth+1 the parameter θ measured _ij, ρ _ijbe kept in candidate's flight path matrix P (k).

(5) calculate kth+1 bat measurement according to formula ρ=xcos θ+ysin θ and clap with kth+2 the parameter θ measured _jm, ρ _jm, compare with the parameter in P (k) respectively, if meet criterion one: | θ _jm-θ _ij|≤Δ θ and criterion two: θ _ijand θ _jmthe slope at place is identical, and wherein Δ θ is permissible error, is so measured this triple time and is saved in flight path matrix Para simultaneously, the initial results that Here it is measures triple time;

Export: what preserve in Para is exactly initial flight path, and track initiation is complete.

Fig. 1 is the process flow diagram of this algorithm, therefrom can find out that how the present invention carries out the switching of two kinds of algorithms by the judgement of discount judgement factor Dis intuitively, wherein discount evaluation coefficient β is that discount factor is generally less than 1, as can be seen from Fig. 2 to Fig. 5, track initiation success ratio is higher, the initial success ratio of false track is lower, and initial time is shorter, and track initiation effect is better.Discount evaluation coefficient is lower, and track initiation effect is better.When Dis >=0.8, the correction Hough transform Track initialization algorithm based on a step time delay is lower than the discount evaluation coefficient revising Hough transform Track initialization algorithm fast, and track initiation is effective; When Dis < 0.8, quick random Hough transformation Track initialization algorithm is lower than the discount evaluation coefficient based on the correction Hough transform Track initialization algorithm of a step time delay, and track initiation is effective.

Claims (4)

1. can walk abreast the long-range radar track initiation method switched, and it is characterized in that, comprise the following steps:

Step one: gathering the two-dimentional energy datum that radar returns, and calculate radar detedtion probability Pd, calculating clutter density Lambda from obtaining TotalScan beat of data;

Step 2: by following formulae discovery discount judgement factor Dis:

Dis＝0.2*3/TotalScan+0.6*Pd+0.2*Lambda/30；

Wherein, Dis represents discount judgement factor, and TotalScan represents initial umber of beats, and Pd represents radar detedtion probability, and Lambda represents clutter density;

Step 3: adopt quick random Hough transformation track initiation method as Dis<0.8, adopts the correction Hough transform track initiation method based on a step time delay when Dis >=0.8.

2. a kind of long-range radar track initiation method switched that walks abreast according to claim 1, it is characterized in that, also comprise step 4: arithmetical discount evaluation coefficient β, and adopt discount evaluation coefficient β to evaluate track initiation effect, β value less then track initiation effect is better;

$\mathrm{\&beta;}=\frac{P_F\&times;t}{P_T}$

Wherein, P _ttrack initiation success ratio, P _fbe false track inception rate, t is initial time.

3. a kind of long-range radar track initiation method switched that walks abreast according to claim 1, it is characterized in that, described quick random Hough transformation track initiation method comprises the following steps:

(1) select distance by radar plane (x, y) as measuring space, definition flight path parameter set is P=[p _c, score] dynamic link list, initialization flight path parameter set P=NULL, sampling number k=1, arranges threshold values T;

(2) data point D in definition of T otalScan bat _i=(x _i, y _i), wherein i=1,2,3...n, random selecting gauge point z from kth is clapped _i(k)=(x _i, y _i), random selecting gauge point z from kth+1 is clapped _j(k+1)=(x _j, y _j);

(3) the distance dist=|z between two gauge points is calculated _i(k)-z _j(k+1) |, obtain the speed v=dist/T between two gauge points _sif, v _min≤ v≤v _max, then enter step (4), otherwise, enter step (2); Wherein, T _srepresent the sampling interval of two gauge points, v _maxand v _minbe respectively maximal rate and the minimum speed of target;

(4) two determined straight line parameter p of gauge point are calculated _c=(θ ₀, ρ ₀);

(5) if k=1, go to step (6); Otherwise, in P, look for a p _c=(θ _c, ρ _c), if | θ _c-θ ₀|≤Δ θ and | ρ _c-ρ ₀|≤Δ ρ, then by p _cscore add 1; Otherwise by p _cinsert P, and make p _cscore be 1; Wherein, Δ θ and Δ ρ is allowable error;

(6) use sampling termination rules limited samples number of times, if satisfied rule, then kth bat and kth+1 clap sampling end, and k=k+1, goes to step (7); If do not meet rule, then go to step (2);

(7) if k>TotalScan, go to step (2); Otherwise, by p corresponding for score>=T in P _cextract, be kept in matrix Para, Para is the flight path parameter detected, and track initiation terminates.

4. a kind of long-range radar track initiation method switched that walks abreast according to claim 1, it is characterized in that, the described correction Hough transform track initiation method based on a step time delay comprises the following steps:

(1) data point D in definition of T otalScan bat _i=(x _i, y _i), wherein i=1,2,3...n, the maximal rate V of Offered target _maxwith minimum speed V _min;

(2) i-th gauge point z that the kth of getting radar is clapped _ik (), as the starting point of track initiation, carries out One-step Extrapolation, the jth gauge point z clapped with kth+1 _j(k+1) carry out the association of traversal formula, obtain kth clap and kth+1 clap the speed V between measuring _ij, will satisfy condition V _min≤ V _ij≤ V _maxtwo clap gauge points preserve;

(3) get meet step (2) conditional gauge point as the intermediate point of track initiation, carry out One-step Extrapolation, m that claps with k+2 measures z _m(k+2) carry out the association of traversal formula, obtain kth+1 and to clap and kth+2 claps speed V between gauge point _jm;

(4) calculate kth according to formula ρ=xcos θ+ysin θ and clap the parameter θ that gauge point and kth+1 clap gauge point _ijand ρ _ij, and be kept in candidate's flight path matrix P (k);

(5) calculate kth+1 according to formula ρ=xcos θ+ysin θ and clap the parameter θ that gauge point and kth+2 clap gauge point _jmand ρ _jm, compare with the parameter in P (k) respectively;

(6) if | θ _jm-θ _ij|≤Δ θ, and θ _ijand θ _jmthe slope at place is identical, and wherein Δ θ is permissible error, and be so saved in flight path matrix Para by this gauge point, Para is the flight path parameter detected, and track initiation terminates triple time simultaneously.