CN101126806A - Method for revising maximum likelihood registration based information infusion - Google Patents

Abstract

The utility model relates to a modified maximum likelihood registration method based on information fusion, belonging to the technical field of target tracking, which comprises the following steps: step one, a passive sensor in multi-platform system is responsible for measuring the target position within observation region, and obtaining redundant information of the target within blind area; step two, with the use of the redundant information in the multi-platform system, the target position in the blind area is calculated; step three, the deviation of the passive sensor is estimated with the maximum likelihood method according to the target position calculated in step two; step four, the estimated deviation is adopted to convergence criterion; step five, the deviation values meeting the convergence requirement is registered to the passive sensor to estimate the target state further. The utility model has the advantages of small calculation quantity, simplicity and effectiveness, and easy operation; thus, the method can be widely used in robotics, intelligent transportation, air traffic control and aerospace, aviation, navigation and other fields.

Description

Revised Maximum Likelihood method for registering based on information fusion

Technical field

The present invention relates to a kind of method for parameter estimation of target following technical field, specifically is a kind of revised Maximum Likelihood method for registering based on information fusion.

Background technology

Pure angle tracking is a kind of important tracking in the tracking field, and it has, and operating distance is far away, lobe-on-receive, difficult by advantages such as the other side realize.The angle information about target that this method utilizes passive sensor to obtain is estimated the state of target.Thereby the guiding armament systems are implemented to hide to radiation source and are attacked, play an important role for the survivability and warfighting capabilities of raising system under the electronic warfare environment, in navigation, Aeronautics and Astronautics, scouting, observing and controlling, rescue and geophysics's research, also playing the part of important role simultaneously.But because the unintentional nonlinearity of Dan Zhanchun angle tracking own, the imperfection of metric data and the polytrope of maneuvering target motion state, the tracking effect that makes pure angle follow the tracks of can not be satisfactory.In order to improve the effect that pure angle is followed the tracks of, often adopt two angular transducers that are arranged on the different platform that the motion state of target is estimated in actual use.When target and two sensors not during conllinear, the metric data of target is completely, and promptly target can be observed, and system can estimate preferably to maneuvering target.

In the Target Tracking System of multi-platform multisensor, information fusion technology can improve detection, the Identification And Traceability to target.Simultaneously, the use of multisensor also brings some problems, for example the registration of sensor.Sensor combinations without registration may cause system performance also poorer than single-sensor, tracking effect is worsened, even can produce false target.Therefore, before the metric data to multisensor merges, need carry out registration to sensor.The process of registration is the deviation of first estimated sensor normally, and the estimated value with deviation compensates in next measurement data constantly again, calculates the estimated value of dbjective state at last.The key of registration is the deviation of estimated sensor; The purpose of registration is to eliminate the adverse effect of sensor bias, and then the estimating target state.The deviation of sensor is normally fixed, or changes very slow.The method for registering of common droop has least square method, generalized least square method and maximum-likelihood method etc.

After to the prior art literature search, find, Nickens Okello and Branko Ristic are at " IEEETransaction on Aerospace and Electronic Systems " (" aerospace and electronic system ") (pp.1074-1083,2003) delivered " Maximum Likelihood Registration for MultipleDissimilar Sensors " (the maximum likelihood method for registering of foreign peoples's sensor), the document has proposed a kind of droop method for registering that can be used for multi-platform passive sensor.Under the situation that does not have any prior imformation of target, droop and dbjective state that this method simultaneously can estimated sensor.But the document is not considered problem that target in the multi-platform passive sensor system is unobservable and to the influence of deviation registration.And one of gordian technique that the pure exactly angle-tracking system of this problem need solve.

Summary of the invention

The present invention is directed to above-mentioned prior art problems, a kind of revised Maximum Likelihood method for registering based on information fusion is provided, target unobservable problem in the blind area when making its redundant information be used to solve passive sensor deviation registration with multisensor, solve target location in the unobservable zone to the influence of deviation registration, improved the estimation of deviation precision of passive sensor.

The present invention is achieved by the following technical solutions, the present invention includes following concrete steps:

Step 1, the passive sensor in the multiple platform system is responsible for measuring the target location in may observe zone, and obtains the redundant information information of blind area internal object;

Described multiple platform system, each passive sensor wherein all is a relatively independent system, the position of radiation source (target) in can the independent measurement monitor area, multiple platform system adopts benchmark unified time based on GPS (GPS), guaranteed in the multiple platform system signal between each passive sensor synchronously.A scan period finishes, and each passive sensor is sent to fusion center by tactical data link with data and carries out subsequent treatment.

The target location in described measurement may observe zone under the observable situation of target, according to taking measurement of an angle of two passive sensors, utilizes cross bearings can obtain the position of target.

In the unobservable blind area of target, obtain redundant information, the calculating of internal object position, two pairs of blind areas of waiting step.Described redundant information is meant the measurement of angle information of all the other sensors beyond necessary two passive sensors of target passive positioning in the multiple platform system.

Described blind area is meant the unobservable scope of target that causes owing to the passive sensor registration bias, and this scope is expanded into a zone by straight line, and this zone is called as the unobservable blind area of target.

Step 2 is utilized the target location in the multiple platform system redundant information calculating blind area;

Target location in the described calculating blind area, be meant when target enters the observation blind area of two passive sensors, the situation of intersection point can appear not having in two observation angles, target is unobservable, the situation of real solution appears not having in the measuring distance of target, utilize the redundant information of multiple platform system, the use Cross Location Method is obtained the measuring distance between target and the passive sensor, replace the measuring distance that utilizes redundant information " non-real number " before with it, and then the preceding measuring position to target of definite research station registration.

Step 3 according to the target location that step 2 is calculated, utilizes maximum likelihood method to estimate the deviation of passive sensor.

Described maximum likelihood method is estimated the deviation of passive sensor, is meant obtaining under the situation about taking measurement of an angle that passive sensor contains deviation, and the numerical value that makes the probability maximum that the maximum likelihood function of deviation occurs is promptly as the estimated value of deviation.

Step 4 is carried out convergence to the deviation of estimating gained and is differentiated.

According to the requirement of system, set deviation convergent threshold value to the estimation of deviation precision.

Described convergence is differentiated, be for each sampling period, need repeat repeatedly step 3, phasor difference to the twice estimated bias value in front and back is asked the second order norm, if this norm, thinks that deviation does not restrain greater than the threshold value of setting, return step 3 and continue estimated bias, satisfy the condition of convergence up to it, promptly this norm is less than or equal to the threshold value of setting.

Step 5 will satisfy the deviate registration passive sensor that convergence requires, and further estimating target state.

Described further estimating target state, be meant will satisfy the deviate registration passive sensor that requires of convergence, the numerical value that makes the probability maximum that the dbjective state maximum likelihood function occurs is promptly as the estimated value of dbjective state.

Described registration is meant that the deviate that current time is estimated compensates in the measurement data of next passive sensor constantly.

Behind the passive sensor registration, passive sensor is used it for the estimating target state to the measuring position of target behind further definite registration, obtains position, speed and the acceleration information of target.

Compared with prior art, the present invention has following beneficial effect: target unobservable problem in blind area when the present invention is used to the redundant information of multiple platform system to solve passive sensor deviation registration.Under the unobservable situation of target, this method utilizes the redundant information of multiple platform system to calculate the preceding measuring distance to target of passive sensor registration, and " non-real number " measuring distance before the replacement information compensation does not increase calculated amount.The inventive method can estimate registration bias apace in 15 sampling periods, and the error criterion difference of estimation of deviation has reached the theoretical next time of carat Metro.This method is simply effective, easy to implement, can be widely used in robot, intelligent transportation, each field such as air traffic control and space flight, aviation, navigation.

Description of drawings

Fig. 1 has the unobservable synoptic diagram of target under the registration bias situation for analytic system of the present invention;

Fig. 2 is the synoptic diagram that the present invention is based on the system redundancy information compensation;

Fig. 3 is the workflow diagram that the present invention is based on the system redundancy information compensation;

Fig. 4 is a targetpath comparison diagram in the embodiment of the invention;

Fig. 5 is the Monte-Carlo Simulation comparison diagram of the error mean of estimation of deviation in the embodiment of the invention;

Fig. 6 is the Monte-Carlo Simulation comparison diagram of the error criterion difference of estimation of deviation in the embodiment of the invention.

Embodiment

Below in conjunction with accompanying drawing embodiments of the invention are elaborated: present embodiment is being to implement under the prerequisite with the technical solution of the present invention, provided detailed embodiment and concrete operating process, but protection scope of the present invention is not limited to following embodiment.

The true flight path state equation of the target of present embodiment is: x (k)=130+150sin (0.06k), and y (k)=30-5k, k=1 wherein ... N (N=100), unit: km, the coordinate of passive sensor A, B, C is respectively: (300 ,-200), (80,75), (0 ,-200).

Observation platform and radiation source be all in same plane X Y in the present embodiment, only the position angle of measurement target.System's droop vector beta=[4 °, 7 ° ,-7 °] ^T, the covariance matrix R=diag[(0.5 of systematic survey noise °) ², (1 °) ², (1 °) ²], initial estimation of deviation value ${\widehat{\mathrm{\β}}}_0={\left[\mathrm{0,0,0}\right]}^T.$

As shown in Figure 3, present embodiment comprises following concrete steps:

Step 1, the passive sensor in the multiple platform system are responsible for measuring the target location in the may observe zone, and obtain the redundant information information of blind area internal object;

The individual passive sensor of n (n 〉=3) is arranged in the multiple platform system, the position angle of measurement target only, the stationkeeping of each sensor, and be known.The measurement equation that sensor contains deviation is:

z _i(k)＝h _i(x(k))+β _i+w _i(k) (1)

Wherein, i=1 ..., n; K=1 ..., N, k are sampling number, z _i(k) be the measured value of i sensor, the time of day of x (k) expression target, h _i(x (k)) is the nonlinear measurement function of dbjective state, β _iBe the registration bias of i sensor, w _i(k) be the white noise of zero-mean, Gaussian distributed, covariance matrix is R _iMeasured value vector z (k)=[z ₁(k) ..., z _n(k)] ^T, bias vector β=[β ₁..., β _n] ^T, sub-covariance matrix R=diag[R ₁..., R _n].

Under the situation that target can be observed, pure angle-tracking system utilizes cross bearings can obtain distance between target and the sensor.As shown in Figure 1, be that example is calculated its measuring position to target with passive sensor A.When target was in the normal observation scope of passive sensor A and B, the position of target is expressed as tri with trigonometric function, and (α, φ), α was the measured angular of passive sensor A, the measured angular of φ passive sensor B.

Described blind area is meant the unobservable scope of target that causes owing to the passive sensor registration bias, and this scope is expanded into a zone by straight line, and this zone is called as the unobservable blind area of target.

Described redundant information, the indirect measured angular α ' of the measured angular γ ' of sensor C and sensors A.

Step 2 is utilized the target location in the multiple platform system redundant information calculating blind area;

As shown in Figure 2, when target was positioned at the observation blind area of passive sensor A and B, two measured angular α and φ laid respectively at the both sides of straight line AB, do not have the point of crossing, and the target range that calculates with cross bearings is an imaginary number.Utilize the redundant information in the system, determine the measuring distance of target, be expressed as tri (α ', γ ') with trigonometric function, and then obtain the measuring position of sensors A target.

The measuring position of target

Wherein, M is the observation blind area of two station passive sensor A and B, and this blind area is that blind area in the present embodiment is (10 °, 10 °) between the special section of observation angle scope of passive sensor.

The concrete scope of observation blind area can determine that the research station can not be imaginary number to the measuring distance of target when the principle of determining was the passive location of two stations according to the actual conditions of system, and promptly target will be within the may observe scope of research station all the time.Because in the time of near the gtoal setting blind area, the research station will obviously increase the difference of the measuring distance of the measuring distance of the current time of target measurement and previous moment.Therefore, in implementation process, usually the blind area scope is enlarged a little.

Step 3 according to the target location that step 2 is calculated, utilizes maximum likelihood method to estimate the deviation of passive sensor;

Described maximum-likelihood method is estimated the deviation of passive sensor, and is specific as follows:

Under to the situation of target without any prior imformation, given one group of measured value { z (k); K=1 ..., N}.

$p(Z/X,\mathrm{\β})=\arg \underset{\mathrm{\β}}{\max }\left\{\underset{k=1}{\overset{N}{\mathrm{\Π}}}\underset{x\left(k\right)}{\max }[p\left(z\left(k\right)\right|x\left(k\right),\mathrm{\β}\left]\right)\right\}---\left(3\right)$

To braces outside maximizing in (3) formula, that is:

$p(z\left(1\right),\·\·\·,z\left(N\right)|\hat{X},\mathrm{\β})$

$=\stackrel{\‾}{K}\exp \{-\frac{1}{2}\left({(\mathrm{\β}-\widehat{\mathrm{\β}})}^T\right[\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\Ω}}^T\left(k\right){\mathrm{\Ψ}}^{-1}\left(k\right)\mathrm{\Ω}\left(k\right)](\mathrm{\β}-\widehat{\mathrm{\β}})+C)\}---\left(4\right)$

To (4) formula maximizing, can obtain the estimated value of sensor bias.

Wherein,

$\widehat{\mathrm{\β}}={\left[\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\Ω}}^T\left(k\right){\mathrm{\Ψ}}^{-1}\left(k\right)\mathrm{\Ω}\left(k\right)\right]}^{-1}\left[\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\Ω}}^T\left(k\right){\mathrm{\Ψ}}^{-1}\left(k\right){\stackrel{\‾}{X}}_0\left(k\right)\right]---\left(5\right)$

$\mathrm{\Ω}\left(k\right)=\mathrm{diag}[H_1^{-R}\left(k\right),\·\·\·,H_n^{-R}\left(k\right)]---\left(6\right)$

$R_{\mathrm{xi}}=H_i^T{R}_i{H}_i---\left(7\right)$

${\mathrm{\Ψ}}^{-1}\left(k\right)=\mathrm{diag}[R_{x1}^{-1}\left(k\right),\·\·\·,R_{\mathrm{xn}}^{-1}\left(k\right)]-\left[{\left\{R_{\mathrm{xi}}^{-1}\left(k\right){\left[\underset{l=1}{\overset{n}{\mathrm{\Σ}}}{R}_{\mathrm{xl}}^{-1}\left(k\right)\right]}^{-1}{R}_{\mathrm{xj}}^{-1}\left(k\right)\right\}}_y\right]---\left(8\right)$

${\stackrel{\‾}{X}}_0\left(k\right)=X_0\left(k\right)+\mathrm{\Ω}\left(k\right){\widehat{\mathrm{\β}}}_0---\left(9\right)$

In the formula (5)

Exactly formula (4) is got the estimated value of the deviation that maximum value obtains.

Wherein, H _iBe the nonlinear measurement function h in the formula (1) _i(x (k)) is to the Jacobian matrix of x (k), H _i ^-RBe H _iRight inverse matrix, Ω (k) is by H _i ^-RThe diagonal matrix of forming, R _XiThe covariance matrix of the dbjective state of expression equivalence is by measuring the covariance matrix R of noise _iObtain Ψ by matrixing ^-1Be a kind of description form after the mathematical operation, Ψ does not have physical meaning.X ₀(k) the initial estimated value of expression dbjective state,

It is the dbjective state that calculates by the measurement data that contains deviation.

Maximum likelihood method of estimation in this step has in " the maximum likelihood method for registering of foreign peoples's sensor " document of retrieving in background technology at length and introduces.

Step 4 is carried out convergence to the deviation of estimating gained and is differentiated.

At first set deviation convergent threshold value ε, the size of threshold value can determine that to the requirement of estimation of deviation precision threshold value is more little according to real system, and estimated accuracy is high more, and operation times is also many more.But when threshold value during less than certain numerical value, estimated accuracy does not significantly improve.ε in the present embodiment=|| β || * λ, wherein || and β || be the second order norm of system deviation vector, λ is a coefficient, and λ＞0.In the present embodiment, λ=0.001, ε=0.0107.

Described convergence is differentiated, and is meant:

$\left|\right|\widehat{\mathrm{\β}}-{\widehat{\mathrm{\β}}}_0\left|\right|\≤\mathrm{\ϵ},\mathrm{\ϵ}>0---\left(10\right)$

Wherein,

Be the estimated value of the deviation of current time,

Estimation of deviation value for the previous moment.After differentiating convergence, order ${\widehat{\mathrm{\β}}}_0=\widehat{\mathrm{\β}},$ So that use when differentiating next time, if (10) formula is false, return step 3 again to estimation of deviation, set up up to (10) formula.

Step 5 will satisfy the deviation registration passive sensor that convergence requires, further the estimating target state.

Described registration is meant that the deviate that current time is estimated compensates in the measurement data of next passive sensor constantly.

Described further estimating target state is meant deviate is compensated in measurement data passive sensor next time, obtains the target location behind the sensor registration:

$X\left(k\right)\≈{\stackrel{\‾}{X}}_0\left(k\right)-\mathrm{\Ω}\left(k\right)\widehat{\mathrm{\β}}---\left(11\right)$

Estimated value for the deviation of current time.

Then to braces the inside part maximizing in (3) formula, that is:

$p({\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="A20071004614700131.png"\; state="\{\{state\}\}">z</figure-callout>}}_1,{\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="A20071004614700131.png"\; state="\{\{state\}\}">z</figure-callout>}}_2,\&CenterDot;\&CenterDot;\&CenterDot;z_n|x,\mathrm{\&beta;})$

$\&ap;\mathrm{Kexp}\{-\frac{1}{2}(x-{\hat{x}}^T)[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{R}_{\mathrm{xi}}^{-1}](x-\hat{x})-\frac{1}{2}\left(\right[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{x}_i^T{R_{\mathrm{xi}}^{-1}x}_i]-{\left[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{R}_{\mathrm{xi}}^{-1}{x}_i\right]}^T\&CenterDot;\hat{x})\}---\left(12\right)$

Wherein, R _XiThe covariance matrix of the dbjective state of expression equivalence,

The position vector of expression target, x (dbjective state) is the variable that formula (12) remains to be found the solution,

Be equation (12) separating about x.To (12) formula maximizing, can obtain the estimated value of dbjective state.

Wherein, $\hat{x}{=\left[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{R}_{\mathrm{xi}}^{-1}\right]}^{-1}\left[\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{R}_{\mathrm{xi}}^{-1}{x}_i\right]---\left(13\right)$

$x_i=h_i^{-1}(z_i-{\mathrm{\&beta;}}_i)---\left(14\right)$

Formula (13) is illustrated in the estimated value of the dbjective state that k constantly obtains (12) formula maximizing behind the sensor registration, and formula (14) obtains x by formula (1) by the inverse operation in the mathematics _iRepresent the dbjective state that i passive sensor measures, h _i ^-1Expression nonlinear measurement function h _iThe inverse operation of (x (k)).

In embodiments of the present invention, vector

The position coordinates that only comprises target is asked single order, second derivative respectively to the position coordinates of target, just obtains speed, the acceleration of target.

As shown in Figure 4, be targetpath comparison diagram in the present embodiment, the curve of representing with asterisk " * " is the simulation curve of the inventive method, coincide with the true flight path of target; The curve of with dashed lines "--" expression is the simulation curve of the used maximum-likelihood method of cited literature 2 in this instructions.Fig. 4 shows, the situation of dispersing has appearred in the simulation curve of maximum likelihood method in the observation blind area of passive sensor, departed from the true flight path of target, and the inventive method eliminated the unobservable influence to registration of blind area internal object.

As shown in Figure 5, be the Monte-Carlo Simulation comparison diagram of the error mean of estimation of deviation in the present embodiment, the simulation curve of modification method among solid line "-" expression the present invention among the figure, the simulation curve of dotted line "--" expression maximum likelihood method.Modification method among the present invention is owing to eliminated the unobservable adverse effect of blind area internal object, can in 15 sampling periods, estimate the deviation of passive sensor, the deviation of passive sensor is general to need 40 sampling periods and maximum likelihood method will estimate, the inventive method has shifted to an earlier date 25 cycles than maximum likelihood method, and the needed time of estimated bias has improved 60%.

As shown in Figure 6, be the Monte-Carlo Simulation comparison diagram of the error criterion difference of estimation of deviation in the embodiment of the invention.The curve of solid line "-" expression carat Metro next time among the figure, the simulation curve of modification method among dotted line "--" expression the present invention, dotted line " ... " the simulation curve of expression maximum likelihood method.The Monte-Carlo Simulation number of times is 100 times in the embodiment of the invention.Equal at 100 o'clock at sampling number, each curve among the figure has all reached steady state (SS).The estimated error criterion difference of modification method is 0.1 degree, and the estimated error criterion difference of maximum likelihood method is 0.17 degree.The inventive method is compared with maximum likelihood method, and the estimated accuracy of deviation has improved 40%.Simultaneously, this figure shows that the estimated result of the inventive method has reached the theoretical next time---the OK a karaoke club Metro next time, promptly the inventive method has reached optimum estimation effect, has verified the validity of the inventive method.

Fig. 4, Fig. 5 and Fig. 6 are the simulation result of sensors A, and the simulation result of other sensors is similar to the simulation result of sensors A.

Claims (10)

1. the revised Maximum Likelihood method for registering based on information fusion is characterized in that, comprises following concrete steps:

Step 1, the passive sensor in the multiple platform system are responsible for measuring the target location in the may observe zone, and obtain the redundant information information of blind area internal object;

Step 2 is utilized the target location in the multiple platform system redundant information calculating blind area;

Step 3 according to the target location that step 2 is calculated, utilizes maximum likelihood method to estimate the deviation of passive sensor;

Step 4 is carried out convergence to the deviation of estimating gained and is differentiated;

Step 5 will satisfy the deviate registration passive sensor that convergence requires, and further estimating target state.

2. the revised Maximum Likelihood method for registering based on information fusion according to claim 1, it is characterized in that, described multiple platform system, each passive sensor wherein all is a relatively independent system, position that can independent measurement monitor area internal object, multiple platform system adopts benchmark unified time based on GPS, guaranteed in the multiple platform system signal between each passive sensor synchronously, a scan period finishes, and each passive sensor is sent to fusion center by tactical data link with data and carries out subsequent treatment.

3. the revised Maximum Likelihood method for registering based on information fusion according to claim 1, it is characterized in that, target location in the described measurement may observe zone, under the observable situation of target, according to taking measurement of an angle of two passive sensors, utilize cross bearings can obtain the position of target.

4. the revised Maximum Likelihood method for registering based on information fusion according to claim 1, it is characterized in that, described blind area, be meant because the unobservable scope of target that the passive sensor registration bias causes, this scope is expanded into a zone by straight line, and this zone is called as the unobservable blind area of target.

5. the revised Maximum Likelihood method for registering based on information fusion according to claim 1, it is characterized in that, target location in the described calculating blind area, be meant when target enters the observation blind area of two passive sensors, the situation of intersection point can appear not having in two observation angles, target is unobservable, the situation of real solution appears not having in the measuring distance of target, utilize the redundant information of multiple platform system, the use Cross Location Method is obtained the measuring distance between target and the sensor, replace the measuring distance that utilizes redundant information " non-real number " before with it, and then the preceding measuring position to target of definite research station registration.

6. according to claim 1 or 5 based on the revised Maximum Likelihood method for registering of information fusion, it is characterized in that, described redundant information is meant the measurement of angle information of all the other sensors beyond necessary two passive sensors of target passive positioning in the multiple platform system.

7. the revised Maximum Likelihood method for registering based on information fusion according to claim 1, it is characterized in that, described convergence is differentiated, and is for each sampling period, need repeat repeatedly step 3, phasor difference to the twice estimated bias value in front and back is asked the second order norm, if this norm, thinks that deviation does not restrain greater than the threshold value of setting, return step 3 and continue estimated bias, satisfy the condition of convergence up to it, promptly this norm is less than or equal to the threshold value of setting.

8. the revised Maximum Likelihood method for registering based on information fusion according to claim 1, it is characterized in that, described maximum likelihood method is estimated the deviation of passive sensor, be meant that under the situation of the deviation that obtains passive sensor the numerical value that makes the probability maximum that deviation occurs is promptly as the estimation of deviation value of true value.

9. the revised Maximum Likelihood method for registering based on information fusion according to claim 1 is characterized in that, described registration is meant that the deviate that current time is estimated compensates in the measurement data of next passive sensor constantly.

10. the revised Maximum Likelihood method for registering based on information fusion according to claim 1, it is characterized in that, described further estimating target state, be meant will satisfy the deviate registration passive sensor that convergence requires after, the numerical value that makes the probability maximum that the maximum likelihood function of dbjective state occurs is promptly as the estimated value of dbjective state.

Images