Multi-target tracking method and system for flicker noise

A multi-target tracking and flicker noise technology, which is applied in the field of multi-target tracking methods and systems to solve tracking problems and improve tracking accuracy.

Inactive Publication Date: 2017-05-31
SHENZHEN UNIV
5 Cites 12 Cited by

AI-Extracted Technical Summary

Problems solved by technology

[0003] The technical problem to be solved by the present invention is to provide a multi-target tracking method and trackin...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Method used

The embodiment of the present invention predicts the contract distribution and the existence probability of each target of the new measurement data received at the current moment according to the contract distribution and the existence probability of each target of the measurement data received at the previous moment; according to the predicted contract distribution and existence probability Probability, use the variational Bayesian method to sequentially process each measurement at t...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Abstract

The invention is applicable to the field of multi-sensor information fusion and provides a multi-target tracking method and system for flicker noise. The method comprises a prediction step, an updating step, a generation step and an extraction step. According to the technical scheme provided by the invention, while the real-time performance of data processing is guaranteed, a tracking problem of a nonlinear moving target under flashing noise is solved effectively.

Application Domain

Technology Topic

Nonlinear motionData processing +4

Image

  • Multi-target tracking method and system for flicker noise
  • Multi-target tracking method and system for flicker noise
  • Multi-target tracking method and system for flicker noise

Examples

  • Experimental program(1)

Example Embodiment

[0021] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the following further describes the present invention in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, but not to limit the present invention.
[0022] The embodiment of the present invention predicts the contract distribution and the existence probability of each target of the new measurement data received at the current time based on the contract distribution and the existence probability of each target of the measurement data received at the previous moment; according to the predicted contract distribution and the existence probability, use The variational Bayesian method sequentially processes each measurement at the current moment to obtain the updated contract distribution and existence probability of each target; respectively combines the updated contract distribution and existence probability with the contract distribution and existence probability of the new target to generate each current time The contract distribution and the existence probability of the target, so that the embodiment of the present invention can well solve the problem of tracking non-linear moving targets under flicker noise.
[0023] Based on the foregoing principles, the embodiments of the present invention provide figure 1 A multi-target tracking method suitable for flicker noise is shown, including:
[0024] S101: According to the contract distribution and existence probability of each target at the previous moment and the time difference between the current moment and the previous moment, a heuristic method is used to generate the shape parameter and scale parameter of the gamma distribution, thereby obtaining the predicted contract distribution of each target at the current moment And the predicted probability of existence;
[0025] S102: According to the predicted contract distribution and predicted existence probability of each target at the current moment, use the variational Bayes method to sequentially process the measurement data at the current moment to obtain the updated contract distribution and updated existence probability of each target at the current moment;
[0026] S103. Use the measurement data at the current moment to generate the contract distribution of the newborn target, and specify the existence probability for the newborn target, and separate the contract distribution and the existence probability of the newborn target with the updated contract distribution and the updated existence probability at the current moment. Merging to get the contract distribution and existence probability of each target at the current moment;
[0027] S104. Cut targets whose existence probability is less than a first threshold from each target at the current moment, and use the contract distribution and existence probability of the remaining targets after the reduction as the next recursive input of the filter, from the remaining targets after the reduction The target whose existence probability is greater than the second threshold is extracted from the target, the contract distribution of the extracted target is used as the output at the current moment, and the average value of the output contract distribution is used as the state estimation of the target at the current moment.
[0028] Specifically, in step S101, k-1 represents the previous moment, k represents the current moment, and t k-1 Indicates the time of the previous moment, t k Represents the time at the current moment, and the measurement noise at the current moment obeys the ξ-dimensional t distribution, with S(z k; H k x k ,R k ,r k ) Represents the probability density function measured at the current moment, where H k x k Represents the mean value of t distribution, R k Represents the precision matrix, r k Represents the degrees of freedom of the t distribution, and The multivariate contract distribution of target i at the previous moment is The existence probability of target i is ρ i,k-1 , Where N represents Gaussian distribution, g represents gamma distribution, x i,k-1 Represents the state vector of the i-th contract distribution at the previous moment, m i,k-1 Represents the mean value of the Gaussian distribution in the i-th contract distribution at the previous moment, P i,k-1 Represents the variance of the Gaussian distribution in the i-th contract distribution at the previous moment, Means R k Diagonal elements, And γ i,k-1 Represents the shape parameter of the gamma distribution in the i-th contract distribution at the previous moment, And η i,k-1 Represents the scale parameter of the gamma distribution in the i-th contract distribution at the previous moment, ξ is the dimension of the state vector, i=1, 2,...,N k-1 , N k-1 Is the total number of targets at the previous moment;
[0029] According to the contract distribution and existence probability of each target at the previous moment, and the time difference between the current moment and the previous moment, the predicted contract distribution of each target at the current moment is obtained The predicted existence probability of each target at the current moment is ρ i,k|k-1 =P s,k (t k -t k-1 )ρ i,k-1; Among them, i=1,2,...,N k-1 , Is the mean value of the Gaussian distribution in the i-th contract distribution at the current moment, Is the variance of the Gaussian distribution in the i-th contract distribution at the current moment, Sigma point x i,0 =m i,k-1 , Sigma point weight l=1,...,ξ, Is the survival probability of the target, γ i,k|k-1 =ρ γ γ i,k-1 Is the shape parameter of the gamma distribution in the i-th contract distribution at the current moment, η i,k|k-1 =ρ η η i,k-1 Is the scale parameter of the gamma distribution in the i-th contract distribution at the current moment, f is a nonlinear function, Q k-1 Is the process noise variance matrix at the receiving moment, the superscript T represents the transposition of the matrix or vector, T is the sampling period, δ is a known constant, ρ α , Ρ β , Ρ γ , Ρ η Is the spread factor, the value range is (0,1], r k Indicates the degree of freedom, which is a known constant, ξ is the dimension of the state vector, and k is a scale parameter.
[0030] In step S102, set the observation set received at the current moment as y k =(y 1,k ,...,Y M,k ), where M is the total number of measurement data received at the current moment, then according to the predicted contract distribution and predicted existence probability of each target at the current moment, the measurement data at the current moment are sequentially processed using the variational Bayes method to obtain The update contract distribution and update existence probability of each target at the current moment, including:
[0031] S1021, taking the predicted contract distribution and predicted existence probability of each target at the current moment as the initial contract distribution and initial existence probability of each target at the current moment, that is, the initial contract distribution is taken as
[0032] The initial probability of existence is taken as Where i=1,2,...,N k-1 ,
[0033] S1022, using the variational Bayes method to sequentially process the 1st to the Mth measurement data in sequence;
[0034] Suppose the contract distribution and existence probability of each target before the jth measurement data processing are respectively with Among them, i=1,2,...,N k-1 ,1≤j≤M; by with The existence probability of each target when updated with the jth measurement is among them
[0035]
[0036] The contract distribution of each target when updated with the jth measurement is
[0037] among them, Represents the gamma function, tr represents the trace of the matrix, Represents the mean vector, Represents the covariance matrix, Represents the filter gain; where Sigma point The shape parameter of the gamma distribution is The scale parameter of the gamma distribution is H k Is the observation matrix, R k Is the observed noise variance matrix, P D,k Is the detection probability of the target, λ c,k Is the clutter density, I represents the identity matrix, y j,k Is the j-th measurement data received at the current moment, the superscript T represents the transpose of a matrix or vector, and ξ is the total dimension of the state vector;
[0038] If Then the contract distribution of target i after the jth measurement data processing is
[0039] Its existence probability is among them
[0040] If Then the contract distribution of target i after the jth measurement data processing is
[0041] Its existence probability is among them
[0042] S1023, the contract distribution and existence probability of each target after the M-th measurement data processing are respectively with Where i=1,2,...,N k-1; After processing the M-th measurement data, the contract distribution and the existence probability of each target are respectively regarded as the update contract distribution of each target at the current moment, and the update contract distribution of each target at the current moment is obtained as
[0043] And the update existence probability of each target at the current moment Where i=1,...,N k-1 ,
[0044] In step S103, it is assumed that the update contract distribution of each target at the current moment is
[0045] The existence probability of each target is ρ i,k; Where i=1,2,...,N k-1 , Using the M measured data at the current moment to generate the contract distribution of the new student target at the current moment is And specify the existence probability of each freshman target at the current moment as Where j=1,2,...,M, Is the covariance of the Gaussian distribution in the contract distribution of the given j-th freshman goal, Is the mean value of the Gaussian distribution in the contract distribution of the jth freshman goal, From the j-th measurement data y at the current moment j,k =[x j,k y j,k ] T Produce, and with Is the shape parameter of the gamma distribution in the contract distribution of the j-th newborn target at the current moment, with Is the scale parameter of the gamma distribution in the contract distribution of the j-th freshman target at the current moment.
[0046] Combine the updated contract distribution of each target at the current moment with the contract distribution of the new-born target at the current moment, and obtain the contract distribution of each target at the current moment as
[0047] Combine the existence probability of each target at the current moment with the existence probability of the new-born target at the current moment to obtain the existence probability of each target at the current moment as Where N k =N k-1 +M.
[0048] Combine below Figure 2 to Figure 5 To further explain this embodiment:
[0049] In this embodiment, consider the two-dimensional space [-1000m,1000m] X Non-linear moving targets in [-1000m,1000m]. The state of the target is composed of position, speed and turning rate, expressed as x=[x &x y &y ω], where x and y represent position components, &x and &y represent speed components, ω represents turning rate, and superscript T represents The transpose of the vector, the state transition matrix is The process noise variance matrix is △t k = T k -t k-1 Is the time difference between the current moment and the previous moment, σ v And σ w Is the process noise standard deviation; the radar observation equation is:
[0050] Observation noise variance matrix σ r And σ θ To observe the standard deviation of noise, measure noise v k Assume to obey r k = 10 t distribution.
[0051] In order to generate simulation data, take the survival probability p S,k =1.0, detection probability p D,k =0.9, the standard deviation of process noise σ v =1ms -2 , Σ w =0.1rads -2 And the standard deviation of the observation noise σ r = 2m, σ θ = 0.0003rad. The simulation observation data of an experiment is as figure 2 Shown. In order to process the simulation data, the relevant parameters of the embodiment of the present invention and the unscented Kalman Gaussian mixture probability hypothesis density filter under flicker noise are set to p S,k =1.0, p D,k =0.9、σ v =1ms -2 , Σ w =0.1rads -2 , Σ r = 2m, σ θ =0.0003rad, the first threshold is 10 -3 , The second threshold is 0.5, the propagation factor ρ α =ρ β =ρ γ =ρ η =0.75, the initial value of the shape parameter of the gamma distribution Gamma distribution scale parameter initial value Weight w of the newly born target of UK-PHD filter under flicker noise γ =0.1. The existence probability of the new target in the embodiment of the present invention ρ γ =0.1, the covariance of the newly generated target image 3 with Figure 4 To compare the filter with the results produced by the multi-target tracking method provided by the embodiment of the present invention. Compare the embodiment of the present invention with the existing UK-PHD filter under flicker noise figure 2 The simulation data is processed, and the average OSPA (Optimal Subpattern Assignment, optimal subpattern assignment) distance is obtained from 100 Monte Carlo experiments. Figure 5 Shown. Comparing the existing UK-PHD filter based on flicker noise with the present invention, the multi-target tracking method of the present invention can obtain more accurate tracking of non-linear moving targets with uncertain correlation and uncertain detection under flicker noise. And reliable target state estimation, its OSPA distance is smaller than the existing OSPA distance obtained by this method.
[0052] The present invention also provides Image 6 The illustrated embodiment, a multi-target tracking system, includes:
[0053] The prediction module 601 is used to generate the shape parameter and scale parameter of the gamma distribution by heuristic method according to the contract distribution and existence probability of each target at the previous moment and the time difference between the current moment and the previous moment, and then obtain the various targets at the current moment The predicted contract distribution and the predicted probability of existence;
[0054] The update module 602 is used to sequentially process the measurement data at the current time using the variational Bayes method according to the predicted contract distribution and predicted existence probability of each target at the current time to obtain the updated contract distribution and update existence of each target at the current time Probability
[0055] The generating module 603 is used to generate the contract distribution of the new target by using the measurement data at the current moment, and specify the existence probability for the new target, and compare the contract distribution and the existence probability of the new target with the updated contract distribution at the current moment. And update the existence probability to merge to obtain the contract distribution and existence probability of each target at the current moment;
[0056] The extraction module 604 is used to cut the targets whose existence probability is less than the first threshold from each target at the current moment, and use the contract distribution and existence probability of the remaining targets after the reduction as the input of the next recursive filter, from the reduction From the remaining targets, the target whose existence probability is greater than the second threshold is extracted, the contract distribution of the extracted target is used as the output at the current moment, and the average value of the output contract distribution is used as the state estimate of the target at the current moment.
[0057] Further, the prediction module 601 is specifically used for:
[0058] Let k-1 represent the previous moment, k represents the current moment, t k-1 Indicates the time of the previous moment, t k Represents the time at the current moment, and the measurement noise at the current moment obeys the ξ-dimensional t distribution, with S(z k; H k x k ,R k ,r k ) Represents the probability density function measured at the current moment, where H k x k Represents the mean value of t distribution, R k Represents the precision matrix, r k Represents the degrees of freedom of the t distribution, and The multivariate contract distribution of target i at the previous moment is The existence probability of target i is ρ i,k-1 , Where N represents Gaussian distribution, g represents gamma distribution, x i,k-1 Represents the state vector of the i-th contract distribution at the previous moment, m i,k-1 Represents the mean value of the Gaussian distribution in the i-th contract distribution at the previous moment, P i,k-1 Represents the variance of the Gaussian distribution in the i-th contract distribution at the previous moment, Means R k Diagonal elements, And γ i,k-1 Represents the shape parameter of the gamma distribution in the i-th contract distribution at the previous moment, And η i,k-1 Represents the scale parameter of the gamma distribution in the i-th contract distribution at the previous moment, ξ is the dimension of the state vector, i=1, 2,...,N k-1 , N k-1 Is the total number of targets at the previous moment;
[0059] According to the contract distribution and existence probability of each target at the previous moment, and the time difference between the current moment and the previous moment, the predicted contract distribution of each target at the current moment is obtained The predicted existence probability of each target at the current moment is ρ i,k|k-1 =P s,k (t k -t k-1 )ρ i,k-1; Among them, i=1,2,...,N k-1 , Is the mean value of the Gaussian distribution in the i-th contract distribution at the current moment, Is the variance of the Gaussian distribution in the i-th contract distribution at the current moment, Sigma point x i,0 =m i,k-1 , Sigma point weight l=1,...,ξ, Is the survival probability of the target, γ i,k|k-1 =ρ γ γ i,k-1 Is the shape parameter of the gamma distribution in the i-th contract distribution at the current moment, η i,k|k-1 =ρ η η i,k-1 Is the scale parameter of the gamma distribution in the i-th contract distribution at the current moment, f is a nonlinear function, Q k-1 Is the process noise variance matrix at the receiving moment, the superscript T represents the transposition of the matrix or vector, T is the sampling period, δ is a known constant, ρ α , Ρ β , Ρ γ , Ρ η Is the spread factor, the value range is (0,1], r k Represents the degree of freedom, which is a known constant, ξ is the dimension of the state vector, and k is a scale parameter.
[0060] Further, the update module 602 is specifically used for:
[0061] Take the predicted contract distribution and predicted existence probability of each target at the current moment as the initial contract distribution and initial existence probability of each target at the current moment, that is, the initial contract distribution is taken as The initial probability of existence is taken as Where i=1,2,...,N k-1 ,
[0062] Use the variational Bayes method to sequentially process the 1st to Mth measurement data;
[0063] Suppose the contract distribution and existence probability of each target before the jth measurement data processing are respectively with Among them, i=1,2,...,N k-1 ,1≤j≤M; by with The existence probability of each target when updated with the jth measurement is among them
[0064]
[0065] The contract distribution of each target when updated with the jth measurement is
[0066] among them, Represents the gamma function, tr represents the trace of the matrix, Represents the mean vector, Represents the covariance matrix, Represents the filter gain; where Sigma point The shape parameter of the gamma distribution is The scale parameter of the gamma distribution is H k Is the observation matrix, R k Is the observed noise variance matrix, P D,k Is the detection probability of the target, λ c,k Is the clutter density, I represents the identity matrix, y j,k Is the j-th measurement data received at the current moment, the superscript T represents the transpose of a matrix or vector, and ξ is the total dimension of the state vector;
[0067] If Then the contract distribution of target i after the jth measurement data processing is
[0068] Its existence probability is among them
[0069] If Then the contract distribution of target i after the jth measurement data processing is
[0070] Its existence probability is among them
[0071] The contract distribution and existence probability of each target after the M-th measurement data processing are respectively with Where i=1,2,...,N k-1;
[0072] After the M-th measurement data is processed, the contract distribution and the existence probability of each target are respectively regarded as the update contract distribution of each target at the current moment, and the update contract distribution of each target at the current moment is obtained as
[0073] And the update existence probability of each target at the current moment Where i=1,...,N k-1 ,
[0074] Further, the generating module 603 is specifically used for:
[0075] Suppose the update contract distribution of each target at the current moment is The existence probability of each target is ρ i,k; Where i=1,2,...,N k-1 , Using the M measured data at the current moment to generate the contract distribution of the new student target at the current moment is And specify the existence probability of each freshman target at the current moment as Where j = 1, 2,..., M, Is the covariance of the Gaussian distribution in the contract distribution of the given j-th freshman goal, Is the mean value of the Gaussian distribution in the contract distribution of the jth freshman goal, From the jth measurement data y at the current moment j,k =[x j,k y j,k ] T Produce, and with Is the shape parameter of the gamma distribution in the contract distribution of the j-th newborn target at the current moment, with Is the scale parameter of the gamma distribution in the contract distribution of the j-th freshman target at the current moment.
[0076] Combine the updated contract distribution of each target at the current moment with the contract distribution of the new-born target at the current moment, and obtain the contract distribution of each target at the current moment as
[0077] Combine the existence probability of each target at the current moment with the existence probability of the new-born target at the current moment to obtain the existence probability of each target at the current moment as Where N k =N k-1 +M.
[0078] The above descriptions are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modification, equivalent replacement and improvement made within the spirit and principle of the present invention shall be included in the protection of the present invention. Within range.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Similar technology patents

Target positioning method, device and system

ActiveCN110133637AFast and precise positioningImprove tracking accuracyBeacon systems using electromagnetic wavesPosition fixationRadar measurementEnvironmental geology
Owner:CHANGCHUN INST OF OPTICS FINE MECHANICS & PHYSICS CHINESE ACAD OF SCI

Classification and recommendation of technical efficacy words

  • Troubleshoot tracking issues
  • Improve tracking accuracy
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products