CN104867163A - Marginal distribution passing measurement-driven target tracking method and tracking system thereof - Google Patents
Marginal distribution passing measurement-driven target tracking method and tracking system thereof Download PDFInfo
- Publication number
- CN104867163A CN104867163A CN201510284138.3A CN201510284138A CN104867163A CN 104867163 A CN104867163 A CN 104867163A CN 201510284138 A CN201510284138 A CN 201510284138A CN 104867163 A CN104867163 A CN 104867163A
- Authority
- CN
- China
- Prior art keywords
- marginal distribution
- current time
- probability
- target
- distribution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The invention is suitable for the field of multi-sensor information integration, and provides a marginal distribution passing measurement-driven target tracking method. The method comprises the steps that marginal distribution and existence probability thereof predicted at current moment are obtained according to marginal distribution and existence probability thereof of the former moment; sequential processing is performed on the measured data of the current moment by utilizing the Bayesian rules according to the predicted marginal distribution and existence probability thereof so that updated marginal distribution and existence probability thereof are obtained; marginal distribution of a newly generated target of the current moment is generated by utilizing the measured data of the current moment, existence probability is assigned for the marginal distribution, and the marginal distribution and existence probability thereof are respectively merged with the updated marginal distribution and existence probability thereof so that the marginal distribution and the existence probability thereof of the current moment are generated; and the marginal distribution of which existence probability is less than a first threshold value is cut off in the generated marginal distribution of the current moment, the cut-off marginal distribution and existence probability thereof act as input of recursion of the next moment, and the marginal distribution of which existence probability is greater than a second threshold value is extracted to act as output of the current moment simultaneously.
Description
Technical field
The invention belongs to multi-sensor information fusion technology field, particularly relate to a kind of measurement of transmitting marginal distribution and drive method for tracking target and tracker.
Background technology
Multi-objective Bayesian filtering method is the effective ways solving target detection and tracking.But, in actual application, we find that multi-objective Bayesian filtering method exists following two problems: one is could start process after needing all measurement data of one-period by the time all to receive during Measurement and Data Processing, like this, if the measuring period of sensor is long, the reception that measurement data is different in one-period arrives, and the measurement data received can not get timely process in latent period can cause serious message delay.Two is initial positions that the recurrence of wave filter needs to know target, and when target initial position message cannot obtain, wave filter is difficult to use.Multiple target tracking problem in the delay issue of information processing, unknown object initial position situation is the key technical problem that multi-objective Bayesian filtering method needs to explore and solve.
Summary of the invention
Technical matters to be solved by this invention is to provide a kind of measurement of transmitting marginal distribution to drive method for tracking target and tracker, is intended to solve the multiple target tracking problem under the message delay problem and unknown object initial position situation that the measurement data that newly receives can not produce by processing in time.The present invention is achieved in that
The measurement of transmitting marginal distribution drives a method for tracking target, comprises the following steps:
Step 1: when receiving new measurement data, calculates the mistiming of current time and previous moment, and according to the marginal distribution of this mistiming and previous moment and there is probability and obtain marginal distribution that current time predicts and there is probability;
Step 2: according to current time prediction marginal distribution and there is probability, utilize Bayes rule to carry out Sequential processing to the measurement data that current time receives, obtain current time upgrade marginal distribution and there is probability;
Step 3: utilize the measurement data of current time to generate the marginal distribution of the newborn target of current time, and there is probability for its appointment, simultaneously, by the marginal distribution of newborn for current time target and there is marginal distribution that probability upgrades with current time respectively and there is probability and merge, the marginal distribution of generation current time and there is probability;
Step 4: the marginal distribution that there is probability and be less than first threshold is reduced from the marginal distribution of current time generated after merging, and using reduce after marginal distribution and there is the input of probability as subsequent time wave filter recurrence, simultaneously, the output of marginal distribution as current time that there is probability and be greater than Second Threshold is extracted from the marginal distribution after reduction, and using the mean and variance of each output marginal distribution as the state estimation of survival target and estimation of error.
Further, in described step 1, represent previous moment with k-1, k represents current time, t
k-1represent the time of previous moment, t
krepresent the time of current time;
The marginal distribution of known previous moment is N (x
i, k-1; m
i, k-1, P
i, k-1), i=1,2 ..., n
k-1; The probability that exists of each marginal distribution of previous moment is ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, N is Gaussian distribution, x
i, k-1for the state of previous moment i-th marginal distribution, m
i, k-1for the average of previous moment i-th marginal distribution, P
i, k-1for the variance of previous moment i-th marginal distribution, n
k-1for the sum of previous moment target, i is call number;
By the marginal distribution of previous moment, each marginal distribution of previous moment there is probability, and the predicted edge that the mistiming of current time and previous moment obtains each target of current time is distributed as N (x
i,k; m
i, k|k-1, P
i, k|k-1), i=1,2 ..., n
k-1; The probability that exists of each target prediction marginal distribution of current time is ρ
i, k|k-1=p
s,k(t
k, t
k-1) ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, m
i, k|k-1=F
k-1m
i, k-1, be the average of current time i-th marginal distribution; P
i, k|k-1=Q
k-1+ F
k-1p
i, k-1f
k-1 t, be the variance of current time i-th marginal distribution;
For the probability of survival of target; Δ t=t
k-t
k-1, be the mistiming of current time and previous moment; δ is known constant; T is the sampling period;
For the state-transition matrix of previous moment; Q
k-1for the process noise covariance matrix of previous moment; The transposition of subscript T representing matrix.
Further, in described step 2, if the measurement data that current time receives is y
k=(y
1, k..., y
m,k), wherein, M is the measurement data sum that current time receives; Bayes rule is utilized to comprise the step that the measurement data that current time receives carries out Sequential processing:
Steps A: get marginal distribution
I=1,2 ..., n
k-1, and there is probability
i=1,2 ..., n
k-1; Wherein
Step B: utilize Bayes rule to process 1 to M measurement data successively: set jth measurement data marginal distribution before treatment as
i=1,2 ..., n
k-1, the probability that exists of a jth measurement data each marginal distribution is before treatment
i=1,2 ..., n
k-1, wherein, 1≤j≤M; By
with
there is probability in that tries to achieve a jth measurement data renewal
Filter gain
Mean vector
Covariance matrix
wherein, H
kfor observing matrix, R
kfor the variance matrix of observation noise, p
d,kfor the detection probability of target, λ
c,kfor clutter density, y
j,kfor the jth measurement data that current time receives, I representation unit matrix, the transposition of subscript T representing matrix or vector;
If
i-th marginal distribution then after a jth Measurement and Data Processing is
There is probability in it
wherein
If
i-th marginal distribution after a jth Measurement and Data Processing is
There is probability in it
wherein
I-th marginal distribution after step C: a M Measurement and Data Processing is
i=1,2 ..., n
k-1, there is probability and be in it
i=1,2 ..., n
k-1;
The marginal distribution that the described current time obtained upgrades is
I=1,2 ..., n
k-1, the probability that exists of each marginal distribution upgraded is
i=1,2 ..., n
k-1; Wherein
Further, described step 3 comprises:
M measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time
j=1,2 ..., M, meanwhile, the marginal distribution for the newborn target of each current time is specified exists probability
j=1,2 ..., M;
The marginal distribution of the marginal distribution upgrade current time and the newborn target of current time merges, and generates the marginal distribution of current time
The probability that exists that there is the marginal distribution of probability and the newborn target of current time of the marginal distribution upgraded by current time merges, and what generate current time marginal distribution exists probability
Wherein n
k=n
k-1+ M.
Further, in described step 4, the marginal distribution that current time upgrades is N (x
i,k; m
i,k, P
i,k), i=1,2 ..., n
k, the probability that exists of each marginal distribution is ρ
i,k, i=1,2 ..., n
k; Step 4 comprises: from the marginal distribution merging the rear current time generated, reduce the marginal distribution fallen to exist probability and be less than first threshold, marginal distribution after reduction and there is the input of probability as subsequent time wave filter recurrence, meanwhile, from the marginal distribution after reduction, extract the output of marginal distribution as current time that there is probability and be greater than Second Threshold.
The measurement of transmitting marginal distribution drives a Target Tracking System, comprising:
Prediction module, when receiving new measurement data, calculates the mistiming of current time and previous moment, and according to the marginal distribution of this mistiming and previous moment and there is probability and obtain marginal distribution that current time predicts and there is probability;
Update module, according to current time prediction marginal distribution and there is probability, utilize Bayes rule to carry out Sequential processing to the measurement data that current time receives, obtain current time upgrade marginal distribution and there is probability;
Current time marginal distribution generation module, the measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time, and there is probability for its appointment, simultaneously, by the marginal distribution of newborn for current time target and there is marginal distribution that probability upgrades with current time respectively and there is probability and merge, the marginal distribution of generation current time and there is probability;
Marginal distribution extraction module, from the marginal distribution of current time generated after merging, the marginal distribution that there is probability and be less than first threshold is reduced, and using reduce after marginal distribution and there is the input of probability as subsequent time wave filter recurrence, simultaneously, the output of marginal distribution as current time that there is probability and be greater than Second Threshold is extracted from the marginal distribution after reduction, and using the mean and variance of each output marginal distribution as the state estimation of survival target and estimation of error.
Further, in described prediction module, represent previous moment with k-1, k represents current time, t
k-1represent the time of previous moment, t
krepresent the time of current time;
The marginal distribution of known previous moment is N (x
i, k-1; m
i, k-1, P
i, k-1), i=1,2 ..., n
k-1; The probability that exists of each marginal distribution of previous moment is ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, N is Gaussian distribution, x
i, k-1for the state of previous moment i-th marginal distribution, m
i, k-1for the average of previous moment i-th marginal distribution, P
i, k-1for the variance of previous moment i-th marginal distribution, n
k-1for the sum of previous moment target, i is call number;
By the marginal distribution of previous moment, each marginal distribution of previous moment there is probability, and the predicted edge that the mistiming of current time and previous moment obtains each target of current time is distributed as N (x
i,k; m
i, k|k-1, P
i, k|k-1), i=1,2 ..., n
k-1; The probability that exists of each target prediction marginal distribution of current time is ρ
i, k|k-1=p
s,k(t
k, t
k-1) ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, m
i, k|k-1=F
k-1m
i, k-1, be the average of current time i-th marginal distribution; P
i, k|k-1=Q
k-1+ F
k-1p
i, k-1f
k-1 t, be the variance of current time i-th marginal distribution;
For the probability of survival of target; Δ t=t
k-t
k-1, be the mistiming of current time and previous moment; δ is known constant; T is the sampling period;
For the state-transition matrix of previous moment; Q
k-1for the process noise covariance matrix of previous moment; The transposition of subscript T representing matrix.
Further, in described update module, if the measurement data that current time receives is y
k=(y
1, k..., y
m,k), wherein, M is the measurement data sum that current time receives; Described update module specifically for:
Get marginal distribution
I=1,2 ..., n
k-1, and there is probability
i=1,2 ..., n
k-1; Wherein
and
Utilize Bayes rule to process 1 to M measurement data successively: set jth measurement data marginal distribution before treatment as
i=1,2 ..., n
k-1, the probability that exists of a jth measurement data each marginal distribution is before treatment
i=1,2 ..., n
k-1, wherein, 1≤j≤M; By
with
there is probability in that tries to achieve a jth measurement data renewal
filter gain
Mean vector
Covariance matrix
wherein, H
kfor observing matrix, R
kfor the variance matrix of observation noise, p
d,kfor the detection probability of target, λ
c,kfor clutter density, y
j,kfor the jth measurement data that current time receives, I representation unit matrix, the transposition of subscript T representing matrix or vector;
If
i-th marginal distribution then after a jth Measurement and Data Processing is
There is probability in it
wherein
If
i-th marginal distribution after a jth Measurement and Data Processing is
There is probability in it
wherein
I-th marginal distribution after M Measurement and Data Processing is
i=1,2 ..., n
k-1, there is probability and be in it
i=1,2 ..., n
k-1;
The marginal distribution that the described current time obtained upgrades is
I=1,2 ..., n
k-1, the probability that exists of each marginal distribution upgraded is
i=1,2 ..., n
k-1; Wherein
Further, described current time marginal distribution generation module specifically for:
M measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time
j=1,2 ..., M, meanwhile, the marginal distribution for the newborn target of each current time is specified exists probability
j=1,2 ..., M; And
The marginal distribution of the marginal distribution upgrade current time and the newborn target of current time merges, and generates the marginal distribution of current time
The probability that exists that there is the marginal distribution of probability and the newborn target of current time of the marginal distribution upgraded by current time merges, and what generate current time marginal distribution exists probability
Wherein n
k=n
k-1+ M.
Further, the marginal distribution that current time upgrades is N (x
i,k; m
i,k, P
i,k), i=1,2 ..., n
k, the probability that exists of each marginal distribution is ρ
i,k, i=1,2 ..., n
k; Described marginal distribution extraction module specifically for: from the marginal distribution of current time generated after merging, reduce the marginal distribution fallen to exist probability and be less than first threshold, marginal distribution after reduction and there is the input of probability as subsequent time wave filter recurrence, meanwhile, from the marginal distribution after reduction, extract the output of marginal distribution as current time that there is probability and be greater than Second Threshold.
Compared with prior art, beneficial effect is in the present invention: the measurement data of Sequential processing current time, the data received can be processed in time, thus avoid the delay of information processing, improve the real-time of target following; Utilize the measurement data of current time to generate the marginal distribution of the newborn target of current time, without the need to master goal initial position message, expand applicability of the present invention.
Accompanying drawing explanation
Fig. 1 is the process flow diagram of the measurement driving method for tracking target of the transmission marginal distribution that the embodiment of the present invention provides;
Fig. 2 is the connection block diagram of the measurement driving Target Tracking System of the transmission marginal distribution that the embodiment of the present invention provides;
Fig. 3 is the measurement data of 50 scan periods of sensor that the embodiment of the present invention provides;
Fig. 4 is the result obtained according to method for tracking target process provided by the invention.
Fig. 5 is the average OSPA distance according to the present invention and existing GM-PHD and GM-CPHD filtering method.
Embodiment
In order to make object of the present invention, technical scheme and advantage clearer, below in conjunction with drawings and Examples, the present invention is further elaborated.Should be appreciated that specific embodiment described herein is only in order to the present invention, is not intended to limit the present invention.
The marginal distribution that the present invention obtains after utilizing previous moment Measurement and Data Processing and exist probability obtain current time prediction marginal distribution and there is probability, according to the marginal distribution of prediction and there is measurement data that probability utilizes Bayes rule Sequential processing current time to receive and obtain marginal distribution that current time upgrades and there is probability, and utilize the measurement data of current time to generate the marginal distribution of the newborn target of current time, make the present invention can process in time the measurement data that current time receives when non-master goal initial position like this.
As shown in Figure 1, a kind of measurement of transmitting marginal distribution drives method for tracking target, comprises the following steps:
Step 1: when receiving new measurement data, calculates the mistiming of current time and previous moment, and according to the marginal distribution of this mistiming and previous moment and there is the marginal distribution of probabilistic forecasting current time and there is probability.
In step 1, represent previous moment with k-1, k represents current time, t
k-1represent the time of previous moment, t
krepresent the time of current time;
The marginal distribution of known previous moment is N (x
i, k-1; m
i, k-1, P
i, k-1), i=1,2 ..., n
k-1; The probability that exists of each marginal distribution of previous moment is ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, N is Gaussian distribution, x
i, k-1for the state of previous moment i-th marginal distribution, m
i, k-1for the average of previous moment i-th marginal distribution, P
i, k-1for the variance of previous moment i-th marginal distribution, n
k-1for the sum of previous moment target, i is call number;
By the marginal distribution of previous moment, each marginal distribution of previous moment there is probability, and the predicted edge that the mistiming of current time and previous moment obtains each target of current time is distributed as N (x
i,k; m
i, k|k-1, P
i, k|k-1); The probability that exists of each marginal distribution of current time is ρ
i, k|k-1=p
s,k(t
k, t
k-1) ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, m
i, k|k-1=F
k-1m
i, k-1, be the average of current time i-th marginal distribution; P
i, k|k-1=Q
k-1+ F
k-1p
i, k-1f
k-1 t, be the variance of current time i-th marginal distribution;
For the probability of survival of target; Δ t=t
k-t
k-1, be the mistiming of current time and previous moment; δ is known constant; T is the sampling period;
For the state-transition matrix of previous moment; Q
k-1for the process noise covariance matrix of previous moment; Subscript T represents the transposition of state-transition matrix.
Step 2: according to prediction current time marginal distribution and there is probability, utilize Bayes rule to carry out Sequential processing to the measurement data that current time receives, obtain current time upgrade marginal distribution and there is probability.
In step 2, if the measurement data that current time receives is y
k=(y
1, k..., y
m,k), wherein, M is the measurement data sum that current time receives; Bayes rule is utilized to comprise the step that the measurement data that current time receives carries out Sequential processing:
Steps A: before Measurement and Data Processing, gets marginal distribution
I=1,2 ..., n
k-1, and there is probability
i=1,2 ..., n
k-1; Wherein
Step B: utilize Bayes rule to process 1 to M measurement data successively: set a jth measurement data before treatment (or after jth-1 Measurement and Data Processing) marginal distribution as
i=1,2 ..., n
k-1, the probability that exists of a jth measurement data each marginal distribution is before treatment
i=1,2 ..., n
k-1, wherein, 1≤j≤M; By
with
there is probability in that tries to achieve a jth measurement data renewal
Filter gain
Mean vector
Covariance matrix
wherein, H
kfor observing matrix, R
kfor the variance matrix of observation noise, p
d,kfor the detection probability of target, λ
c,kfor clutter density, y
j,kfor the jth measurement data that current time receives, I representation unit matrix, the transposition of subscript T representing matrix or vector;
If
i-th marginal distribution then after a jth Measurement and Data Processing is
There is probability in it
wherein
If
i-th marginal distribution after a jth Measurement and Data Processing is
There is probability in it
wherein
I-th marginal distribution after M Measurement and Data Processing is
i=1,2 ..., n
k-1, there is probability and be in it
i=1,2 ..., n
k-1;
The marginal distribution that the current time obtained upgrades is the marginal distribution after M Measurement and Data Processing, namely
I=1,2 ..., n
k-1, the probability that exists of each marginal distribution upgraded is
i=1,2 ..., n
k-1; Wherein
Step 3: utilize the measurement data of current time to generate the marginal distribution of the newborn target of current time, and there is probability for its appointment, simultaneously, by the marginal distribution of newborn for current time target and there is marginal distribution that probability upgrades with current time respectively and there is probability and merge, the marginal distribution of generation current time and there is probability.
Step 3 specifically comprises:
M measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time
j=1,2 ..., M, meanwhile, the marginal distribution for the newborn target of each current time is specified exists probability
j=1,2 ..., M; The marginal distribution of the marginal distribution upgrade current time and the newborn target of current time merges, and generates the marginal distribution of current time
The probability that exists that there is the marginal distribution of probability and the newborn target of current time of the marginal distribution upgraded by current time merges, and what generate current time marginal distribution exists probability
Wherein n
k=n
k-1+ M.
Step 4: the marginal distribution that there is probability and be less than first threshold is reduced from the marginal distribution of current time generated after merging, and using reduce after marginal distribution and there is the input of probability as subsequent time wave filter recurrence, simultaneously, the output of marginal distribution as current time that there is probability and be greater than Second Threshold is extracted from the marginal distribution after reduction, and using the mean and variance of each output marginal distribution as the state estimation of survival target and estimation of error.In step 4, the marginal distribution that current time upgrades is N (x
i,k; m
i,k, P
i,k), i=1,2 ..., n
k, the probability that exists of each marginal distribution is ρ
i,k, i=1,2 ..., n
k; Step 4 comprises: from the marginal distribution merging the rear current time generated, reduce the marginal distribution fallen to exist probability and be less than first threshold, marginal distribution after reduction and there is the input of probability as subsequent time wave filter recurrence, meanwhile, from the marginal distribution after reduction, extract the output of marginal distribution as current time that there is probability and be greater than Second Threshold.
As shown in Figure 2, the present invention also provides a kind of measurement of transmitting marginal distribution to drive Target Tracking System, comprise: prediction module 201, when receiving new measurement data, calculate the mistiming of current time and previous moment, and according to the marginal distribution of this mistiming and previous moment and there is the marginal distribution of probabilistic forecasting current time and there is probability; Update module 202, according to prediction current time marginal distribution and there is probability, utilize Bayes rule to carry out Sequential processing to the measurement data that current time receives, obtain current time upgrade marginal distribution and there is probability; Current time marginal distribution generation module 203, the measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time, and there is probability for its appointment, simultaneously, by the marginal distribution of newborn for current time target and there is marginal distribution that probability upgrades with current time respectively and there is probability and merge, the marginal distribution of generation current time and there is probability; Marginal distribution extraction module 204, from the marginal distribution of current time generated after merging, the marginal distribution that there is probability and be less than first threshold is reduced, and using reduce after marginal distribution and there is the input of probability as subsequent time wave filter recurrence, simultaneously, the output of marginal distribution as current time that there is probability and be greater than Second Threshold is extracted from the marginal distribution after reduction, and using the mean and variance of each output marginal distribution as the state estimation of survival target and estimation of error.
In prediction module 201, represent previous moment with k-1, k represents current time, t
k-1represent the time of previous moment, t
krepresent the time of current time;
The marginal distribution of known previous moment is N (x
i, k-1; m
i, k-1, P
i, k-1), i=1,2 ..., n
k-1; The probability that exists of each marginal distribution of previous moment is ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, N is Gaussian distribution, x
i, k-1for the state of previous moment i-th marginal distribution, m
i, k-1for the average of previous moment i-th marginal distribution, P
i, k-1for the variance of previous moment i-th marginal distribution, n
k-1for the sum of previous moment target, i is call number;
By the marginal distribution of previous moment, each marginal distribution of previous moment there is probability, and the predicted edge that the mistiming of current time and previous moment obtains each target of current time is distributed as N (x
i,k; m
i, k|k-1, P
i, k|k-1); The probability that exists of each marginal distribution of current time is ρ
i, k|k-1=p
s,k(t
k, t
k-1) ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, m
i, k|k-1=F
k-1m
i, k-1, be the average of current time i-th marginal distribution; P
i, k|k-1=Q
k-1+ F
k-1p
i, k-1f
k-1 t, be the variance of current time i-th marginal distribution;
For the probability of survival of target; Δ t=t
k-t
k-1, be the mistiming of current time and previous moment; δ is known constant; T is the sampling period;
For the state-transition matrix of previous moment; Q
k-1for the process noise covariance matrix of previous moment; Subscript T represents the transposition of state-transition matrix.
In update module 202, the predicted edge in k moment is distributed as N (x
i,k; m
i, k|k-1, P
i, k|k-1), i=1,2 ..., n
k-1, the probability that exists of each predicted edge distribution is ρ
i, k|k-1, i=1,2 ..., n
k-1; If the measurement data that current time receives is y
k=(y
1, k..., y
m,k), wherein, M is the measurement data sum that current time receives; Each measurement data that update module 202 utilizes Bayes rule sequentially to process current time to receive, disposal route is as follows:
(1) before Measurement and Data Processing, marginal distribution is got
I=1,2 ..., n
k-1, and there is probability
i=1,2 ..., n
k-1; Wherein
(2) utilize Bayes rule to process 1 to M measurement data successively: set a jth measurement data before treatment (or after jth-1 Measurement and Data Processing) marginal distribution as
i=1,2 ..., n
k-1, the probability that exists of a jth measurement data each marginal distribution is before treatment
i=1,2 ..., n
k-1, wherein, 1≤j≤M; By
with
there is probability in that tries to achieve a jth measurement data renewal
Filter gain
Mean vector
Covariance matrix
wherein, H
kfor observing matrix, R
kfor the variance matrix of observation noise, p
d,kfor the detection probability of target, λ
c,kfor clutter density, y
j,kfor the jth measurement data that current time receives, I representation unit matrix, the transposition of subscript T representing matrix or vector;
If
i-th marginal distribution then after a jth Measurement and Data Processing is
There is probability in it
wherein
If
i-th marginal distribution after a jth Measurement and Data Processing is
There is probability in it
wherein
I-th marginal distribution after M Measurement and Data Processing is
i=1,2 ..., n
k-1, there is probability and be in it
i=1,2 ..., n
k-1;
The marginal distribution that the current time obtained upgrades is the marginal distribution after M Measurement and Data Processing, namely
I=1,2 ..., n
k-1, the probability that exists of each marginal distribution upgraded is
i=1,2 ..., n
k-1; Wherein
In current time marginal distribution generation module 203, the marginal distribution that current time upgrades is N (x
i,k; m
i,k, P
i,k), i=1,2 ..., n
k-1, the probability that exists of each marginal distribution upgraded is ρ
i,k, i=1,2 ..., n
k-1.Current time marginal distribution generation module 203 utilizes M measurement data of current time to generate the marginal distribution of the newborn target of current time
j=1,2 ..., M, meanwhile, the marginal distribution for the newborn target of each current time is specified exists probability
j=1,2 ..., M; And
The marginal distribution of the marginal distribution upgrade current time and the newborn target of current time merges, and generates the marginal distribution of current time
The probability that exists that there is the marginal distribution of probability and the newborn target of current time of the marginal distribution upgraded by current time merges, and what generate current time marginal distribution exists probability
Wherein n
k=n
k-1+ M.
In marginal distribution extraction module 204, the marginal distribution that the k moment upgrades is N (x
i,k; m
i,k, P
i,k), i=1,2 ..., n
k, the probability that exists of each marginal distribution is ρ
i,k, i=1,2 ..., n
k.Marginal distribution extraction module 204 specifically for: from the marginal distribution of current time generated after merging, reduce the marginal distribution fallen to exist probability and be less than first threshold, marginal distribution after reduction and there is the input of probability as subsequent time wave filter recurrence, meanwhile, from the marginal distribution after reduction, extract the output of marginal distribution as current time that there is probability and be greater than Second Threshold.
A kind of measurement of transmitting marginal distribution of the present invention drive method for tracking target by the measurement data Sequential processing of current time to obtain the marginal distribution of current time renewal and to there is probability, and utilize the measurement data of current time to generate the marginal distribution of the newborn target of current time, make to process in time the measurement data that current time receives when non-master goal initial position, thus avoid the delay of information processing, improve the real-time to target following, expand applicability of the present invention.As one embodiment of the present of invention, consider the target of uniform motion in two-dimensional space [-1000m, 1000m] × [-1000m, 1000m].Dbjective state is made up of position and speed, is expressed as
wherein x and y represents location components respectively,
with
represent speed component respectively, subscript T represents the transposition of vector.State-transition matrix is
Process noise covariance matrix is
Wherein, Δ t
k=t
k-t
k-1represent the mistiming between current time and previous moment, σ
v=1ms
- 2for process noise standard deviation; Observing matrix
Observation noise variance matrix
σ
w=2m is observation noise standard deviation.
In order to produce emulated data, get probability of survival p
s,k=1.0, detection probability p
d,k=0.8, clutter density λ
c,k=5 × 10
-6m
-2, process noise standard deviation sigma
v=1m/s
2with observation noise standard deviation sigma
w=2m.The simulation observation data of once testing as shown in Figure 3.In order to process emulated data, we are by the present invention and Gaussian-mixture probability assumed density wave filter (Gaussian Mixtureprobability hypothesis density filter, GM-PHD wave filter) and the relative parameters setting of Gaussian Mixture gesture distribution probability assumed density wave filter (Gaussian Mixture cardinalized probabilityhypothesis density filter, GM-CPHD wave filter) be p
s,k=1.0, p
d,k=0.8, λ
c,k=5 × 10
-6m
-2, σ
v=1ms
-2, σ
w=2m, first threshold are 10
-3, Second Threshold is 0.5.Fig. 4 utilizes method of the present invention to the result after the emulated data process in Fig. 3.The present invention and existing GM-PHD wave filter and GM-CPHD wave filter are processed the emulated data shown in Fig. 3,100 Monte Carlo test the average OSPA (Optimal Subpattern Assignment, optimum sub-pattern is distributed) obtained, and distance as shown in Figure 5.As can be seen from Figure 5, compared with existing GM-PHD filtering and GM-CPHD filtering method, multi-object tracking method of the present invention when exist association uncertain, detect uncertain and clutter can obtain more accurate and reliable Target state estimator, in most cases, the OSPA that obtains than existing two kinds of methods of its OSPA distance is apart from little.
These are only preferred embodiment of the present invention, not in order to limit the present invention, all any amendments done within the spirit and principles in the present invention, equivalent replacement and improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. the measurement of transmitting marginal distribution drives a method for tracking target, it is characterized in that, comprises the following steps:
Step 1: when receiving new measurement data, calculates the mistiming of current time and previous moment, and according to the marginal distribution of this mistiming and previous moment and there is probability and obtain marginal distribution that current time predicts and there is probability;
Step 2: according to current time prediction marginal distribution and there is probability, utilize Bayes rule to carry out Sequential processing to the measurement data that current time receives, obtain current time upgrade marginal distribution and there is probability;
Step 3: utilize the measurement data of current time to generate the marginal distribution of the newborn target of current time, and there is probability for its appointment, simultaneously, by the marginal distribution of newborn for current time target and there is marginal distribution that probability upgrades with current time respectively and there is probability and merge, the marginal distribution of generation current time and there is probability;
Step 4: the marginal distribution that there is probability and be less than first threshold is reduced from the marginal distribution of current time generated after merging, and using reduce after marginal distribution and there is the input of probability as subsequent time wave filter recurrence, simultaneously, the output of marginal distribution as current time that there is probability and be greater than Second Threshold is extracted from the marginal distribution after reduction, and using the mean and variance of each output marginal distribution as the state estimation of survival target and estimation of error.
2. method for tracking target according to claim 1, is characterized in that, in described step 1, represent previous moment with k-1, k represents current time, t
k-1represent the time of previous moment, t
krepresent the time of current time;
The marginal distribution of known previous moment is N (x
i, k-1; m
i, k-1, P
i, k-1), i=1,2 ..., n
k-1; The probability that exists of each marginal distribution of previous moment is ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, N is Gaussian distribution, x
i, k-1for the state of previous moment i-th marginal distribution, m
i, k-1for the average of previous moment i-th marginal distribution, P
i, k-1for the variance of previous moment i-th marginal distribution, n
k-1for the sum of previous moment target, i is call number;
By the marginal distribution of previous moment, each marginal distribution of previous moment there is probability, and the predicted edge that the mistiming of current time and previous moment obtains each target of current time is distributed as N (x
i,k; m
i, k|k-1, P
i, k|k-1), i=1,2 ..., n
k-1; The probability that exists of each target prediction marginal distribution of current time is ρ
i, k|k-1=p
s,k(t
k, t
k-1) ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, m
i, k|k-1=F
k-1m
i, k-1, be the average of current time i-th marginal distribution; P
i, k|k-1=Q
k-1+ F
k-1p
i, k-1f
k-1 t, be the variance of current time i-th marginal distribution;
for the probability of survival of target; Δ t=t
k-t
k-1, be the mistiming of current time and previous moment; δ is known constant; T is the sampling period;
For the state-transition matrix of previous moment; Q
k-1for the process noise covariance matrix of previous moment; The transposition of subscript T representing matrix.
3. method for tracking target according to claim 2, is characterized in that, in described step 2, if the measurement data that current time receives is y
k=(y
1, k..., y
m,k), wherein, M is the measurement data sum that current time receives; Bayes rule is utilized to comprise the step that the measurement data that current time receives carries out Sequential processing:
Steps A: get marginal distribution
i=1,2 ..., n
k-1, and there is probability
I=1,2 ..., n
k-1; Wherein
Step B: utilize Bayes rule to process 1 to M measurement data successively: set jth measurement data marginal distribution before treatment as
i=1,2 ..., n
k-1, the probability that exists of a jth measurement data each marginal distribution is before treatment
i=1,2 ..., n
k-1, wherein, 1≤j≤M; By
with
try to achieve and there is probability when upgrading by jth measurement data
Filter gain
Mean vector
Covariance matrix
wherein, H
kfor observing matrix, R
kfor the variance matrix of observation noise, p
d,kfor the detection probability of target, λ
c,kfor clutter density, y
j,kfor the jth measurement data that current time receives, I representation unit matrix, the transposition of subscript T representing matrix or vector;
If
i-th marginal distribution then after a jth Measurement and Data Processing is
There is probability in it
Wherein
If
i-th marginal distribution after a jth Measurement and Data Processing is
There is probability in it
Wherein
I-th marginal distribution after step C: a M Measurement and Data Processing is
i=1,2 ..., n
k-1, there is probability and be in it
i=1,2 ..., n
k-1;
The marginal distribution that the described current time obtained thus upgrades is
i=1,2 ..., n
k-1, the probability that exists of each marginal distribution upgraded is
I=1,2 ..., n
k-1; Wherein
4. method for tracking target according to claim 3, is characterized in that, described step 3 comprises:
M measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time
j=1,2 ..., M, meanwhile, the marginal distribution for the newborn target of each current time is specified exists probability
j=1,2 ..., M; Wherein,
for the variance of a given jth newborn object edge distribution,
for the average of a jth newborn object edge distribution,
directly by a jth measurement data of current time
Produce, and
The marginal distribution of the marginal distribution upgrade current time and the newborn target of current time merges, and generates the marginal distribution of current time
The probability that exists that there is the marginal distribution of probability and the newborn target of current time of the marginal distribution upgraded by current time merges, and what generate current time marginal distribution exists probability
Wherein n
k=n
k-1+ M.
5. method for tracking target according to claim 4, is characterized in that, in described step 4, the marginal distribution that current time upgrades is N (x
i,k; m
i,k, P
i,k), i=1,2 ..., n
k, the probability that exists of each marginal distribution is ρ
i,k, i=1,2 ..., n
k; Step 4 comprises: from the marginal distribution merging the rear current time generated, reduce the marginal distribution fallen to exist probability and be less than first threshold, marginal distribution after reduction and there is the input of probability as subsequent time wave filter recurrence, meanwhile, from the marginal distribution after reduction, extract the output of marginal distribution as current time that there is probability and be greater than Second Threshold.
6. the measurement of transmitting marginal distribution drives a Target Tracking System, it is characterized in that, comprising:
Prediction module, when receiving new measurement data, calculates the mistiming of current time and previous moment, and according to the marginal distribution of this mistiming and previous moment and there is probability and obtain marginal distribution that current time predicts and there is probability;
Update module, according to current time prediction marginal distribution and there is probability, utilize Bayes rule to carry out Sequential processing to the measurement data that current time receives, obtain current time upgrade marginal distribution and there is probability;
Current time marginal distribution generation module, the measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time, and there is probability for its appointment, simultaneously, by the marginal distribution of newborn for current time target and there is marginal distribution that probability upgrades with current time respectively and there is probability and merge, the marginal distribution of generation current time and there is probability;
Marginal distribution extraction module, from the marginal distribution of current time generated after merging, the marginal distribution that there is probability and be less than first threshold is reduced, and using reduce after marginal distribution and there is the input of probability as subsequent time wave filter recurrence, simultaneously, the output of marginal distribution as current time that there is probability and be greater than Second Threshold is extracted from the marginal distribution after reduction, and using the mean and variance of each output marginal distribution as the state estimation of survival target and estimation of error.
7. Target Tracking System according to claim 6, is characterized in that, in described prediction module, represent previous moment with k-1, k represents current time, t
k-1represent the time of previous moment, t
krepresent the time of current time;
The marginal distribution of known previous moment is N (x
i, k-1; m
i, k-1, P
i, k-1), i=1,2 ..., n
k-1; The probability that exists of each marginal distribution of previous moment is ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, N is Gaussian distribution, x
i, k-1for the state of previous moment i-th marginal distribution, m
i, k-1for the average of previous moment i-th marginal distribution, P
i, k-1for the variance of previous moment i-th marginal distribution, n
k-1for the sum of previous moment target, i is call number;
By the marginal distribution of previous moment, each marginal distribution of previous moment there is probability, and the predicted edge that the mistiming of current time and previous moment obtains each target of current time is distributed as N (x
i,k; m
i, k|k-1, P
i, k|k-1), i=1,2 ..., n
k-1; The probability that exists of each target prediction marginal distribution of current time is ρ
i, k|k-1=p
s,k(t
k, t
k-1) ρ
i, k-1, i=1,2 ..., n
k-1; Wherein, m
i, k|k-1=F
k-1m
i, k-1, be the average of current time i-th marginal distribution; P
i, k|k-1=Q
k-1+ F
k-1p
i, k-1f
k-1 t, be the variance of current time i-th marginal distribution;
for the probability of survival of target; Δ t=t
k-t
k-1, be the mistiming of current time and previous moment; δ is known constant; T is the sampling period;
For the state-transition matrix of previous moment; Q
k-1for the process noise covariance matrix of previous moment; The transposition of subscript T representing matrix.
8. Target Tracking System according to claim 7, is characterized in that, in described update module, if the measurement data that current time receives is y
k=(y
1, k..., y
m,k), wherein, M is the measurement data sum that current time receives; Described update module specifically for:
Get marginal distribution
i=1,2 ..., n
k-1, and there is probability
I=1,2 ..., n
k-1; Wherein
And
Utilize Bayes rule to process 1 to M measurement data successively: set jth measurement data marginal distribution before treatment as
i=1,2 ..., n
k-1, the probability that exists of a jth measurement data each marginal distribution is before treatment
i=1,2 ..., n
k-1, wherein, 1≤j≤M; By
with
there is probability in that tries to achieve a jth measurement data renewal
k filter gain
Mean vector
Covariance matrix
wherein, H
kfor observing matrix, R
kfor the variance matrix of observation noise, p
d,kfor the detection probability of target, λ
c,kfor clutter density, y
j,kfor the jth measurement data that current time receives, I representation unit matrix, the transposition of subscript T representing matrix or vector;
If
i-th marginal distribution then after a jth Measurement and Data Processing is
There is probability in it
Wherein
If
i-th marginal distribution after a jth Measurement and Data Processing is
There is probability in it
Wherein
I-th marginal distribution after M Measurement and Data Processing is
i=1,2 ..., n
k-1, there is probability and be in it
i=1,2 ..., n
k-1;
The marginal distribution that the described current time obtained upgrades is
i=1,2 ..., n
k-1, the probability that exists of each marginal distribution upgraded is
I=1,2 ..., n
k-1; Wherein
9. Target Tracking System according to claim 8, is characterized in that, described current time marginal distribution generation module specifically for:
M measurement data of current time is utilized to generate the marginal distribution of the newborn target of current time
j=1,2 ..., M, meanwhile, the marginal distribution for the newborn target of each current time is specified exists probability
j=1,2 ..., M; And
The marginal distribution of the marginal distribution upgrade current time and the newborn target of current time merges, and generates the marginal distribution of current time
The probability that exists that there is the marginal distribution of probability and the newborn target of current time of the marginal distribution upgraded by current time merges, and what generate current time marginal distribution exists probability
Wherein n
k=n
k-1+ M.
10. Target Tracking System according to claim 9, is characterized in that, the marginal distribution that current time upgrades is N (x
i,k; m
i,k, P
i,k), i=1,2 ..., n
k, the probability that exists of each marginal distribution is ρ
i,k, i=1,2 ..., n
k; Described marginal distribution extraction module specifically for: from the marginal distribution of current time generated after merging, reduce the marginal distribution fallen to exist probability and be less than first threshold, marginal distribution after reduction and there is the input of probability as subsequent time wave filter recurrence, meanwhile, from the marginal distribution after reduction, extract the output of marginal distribution as current time that there is probability and be greater than Second Threshold.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510284138.3A CN104867163A (en) | 2015-05-28 | 2015-05-28 | Marginal distribution passing measurement-driven target tracking method and tracking system thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510284138.3A CN104867163A (en) | 2015-05-28 | 2015-05-28 | Marginal distribution passing measurement-driven target tracking method and tracking system thereof |
Publications (1)
Publication Number | Publication Date |
---|---|
CN104867163A true CN104867163A (en) | 2015-08-26 |
Family
ID=53912978
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510284138.3A Pending CN104867163A (en) | 2015-05-28 | 2015-05-28 | Marginal distribution passing measurement-driven target tracking method and tracking system thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104867163A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105719312A (en) * | 2016-01-19 | 2016-06-29 | 深圳大学 | Multi-target tracking method and tracking system based on sequential Bayes filtering |
CN106357275A (en) * | 2016-08-30 | 2017-01-25 | 国网冀北电力有限公司信息通信分公司 | Huffman compression method and device |
CN106405538A (en) * | 2016-09-13 | 2017-02-15 | 深圳大学 | Multi-target tracking method and tracking system suitable for clutter environment |
CN106772353A (en) * | 2016-11-29 | 2017-05-31 | 深圳大学 | A kind of multi-object tracking method and system suitable for flicker noise |
CN106952290A (en) * | 2017-04-07 | 2017-07-14 | 深圳大学 | A kind of method and system that turning maneuvering target is tracked for three dimensions |
WO2017124299A1 (en) * | 2016-01-19 | 2017-07-27 | 深圳大学 | Multi-target tracking method and tracking system based on sequential bayesian filtering |
WO2018010099A1 (en) * | 2016-07-12 | 2018-01-18 | 深圳大学 | Target tracking method for turn maneuver, and system for same |
WO2018049602A1 (en) * | 2016-09-14 | 2018-03-22 | 深圳大学 | Multi-target tracking method and tracking system applicable to cluttered environment |
CN110501671A (en) * | 2019-08-30 | 2019-11-26 | 深圳大学 | A kind of method for tracking target and device based on measurement distribution |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7636651B2 (en) * | 2003-11-28 | 2009-12-22 | Microsoft Corporation | Robust Bayesian mixture modeling |
US7698108B2 (en) * | 2006-10-10 | 2010-04-13 | Haney Philip J | Parameterization of non-linear/non-Gaussian data distributions for efficient information sharing in distributed sensor networks |
CN102110296A (en) * | 2011-02-24 | 2011-06-29 | 上海大学 | Method for tracking moving target in complex scene |
CN104318059A (en) * | 2014-09-24 | 2015-01-28 | 深圳大学 | Target tracking method applied to nonlinear gauss system and tracking system |
CN104333424A (en) * | 2014-10-16 | 2015-02-04 | 北京邮电大学 | Frequency spectrum detection and unknown noise variance tracking estimation method and device thereof |
-
2015
- 2015-05-28 CN CN201510284138.3A patent/CN104867163A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7636651B2 (en) * | 2003-11-28 | 2009-12-22 | Microsoft Corporation | Robust Bayesian mixture modeling |
US7698108B2 (en) * | 2006-10-10 | 2010-04-13 | Haney Philip J | Parameterization of non-linear/non-Gaussian data distributions for efficient information sharing in distributed sensor networks |
CN102110296A (en) * | 2011-02-24 | 2011-06-29 | 上海大学 | Method for tracking moving target in complex scene |
CN104318059A (en) * | 2014-09-24 | 2015-01-28 | 深圳大学 | Target tracking method applied to nonlinear gauss system and tracking system |
CN104333424A (en) * | 2014-10-16 | 2015-02-04 | 北京邮电大学 | Frequency spectrum detection and unknown noise variance tracking estimation method and device thereof |
Non-Patent Citations (1)
Title |
---|
ZONG-XIANG LIU ET AL.: "Sequential measurement-driven multi-target Bayesian filter", 《EURASIP JOURNAL ON ADVANCES IN SIGNAL PROCESSING》 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2017124299A1 (en) * | 2016-01-19 | 2017-07-27 | 深圳大学 | Multi-target tracking method and tracking system based on sequential bayesian filtering |
CN105719312A (en) * | 2016-01-19 | 2016-06-29 | 深圳大学 | Multi-target tracking method and tracking system based on sequential Bayes filtering |
CN105719312B (en) * | 2016-01-19 | 2018-07-27 | 深圳大学 | Multi-object tracking method based on sequential Bayesian filter and tracking system |
WO2018010099A1 (en) * | 2016-07-12 | 2018-01-18 | 深圳大学 | Target tracking method for turn maneuver, and system for same |
CN106357275B (en) * | 2016-08-30 | 2019-12-17 | 国网冀北电力有限公司信息通信分公司 | Huffman compression method and device |
CN106357275A (en) * | 2016-08-30 | 2017-01-25 | 国网冀北电力有限公司信息通信分公司 | Huffman compression method and device |
CN106405538A (en) * | 2016-09-13 | 2017-02-15 | 深圳大学 | Multi-target tracking method and tracking system suitable for clutter environment |
CN106405538B (en) * | 2016-09-13 | 2018-11-20 | 深圳大学 | A kind of multi-object tracking method and tracking system suitable for clutter environment |
WO2018049602A1 (en) * | 2016-09-14 | 2018-03-22 | 深圳大学 | Multi-target tracking method and tracking system applicable to cluttered environment |
US10935653B2 (en) | 2016-09-14 | 2021-03-02 | Shenzhen University | Multi-target tracking method and tracking system applicable to clutter environment |
CN106772353A (en) * | 2016-11-29 | 2017-05-31 | 深圳大学 | A kind of multi-object tracking method and system suitable for flicker noise |
CN106952290A (en) * | 2017-04-07 | 2017-07-14 | 深圳大学 | A kind of method and system that turning maneuvering target is tracked for three dimensions |
CN106952290B (en) * | 2017-04-07 | 2019-05-10 | 深圳大学 | A kind of method and system tracking turning maneuvering target for three-dimensional space |
CN110501671A (en) * | 2019-08-30 | 2019-11-26 | 深圳大学 | A kind of method for tracking target and device based on measurement distribution |
WO2021036367A1 (en) * | 2019-08-30 | 2021-03-04 | 深圳大学 | Target tracking method and apparatus based on measurement allocation |
CN110501671B (en) * | 2019-08-30 | 2021-04-13 | 深圳大学 | Target tracking method and device based on measurement distribution |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104867163A (en) | Marginal distribution passing measurement-driven target tracking method and tracking system thereof | |
CN105719312A (en) | Multi-target tracking method and tracking system based on sequential Bayes filtering | |
WO2017124299A1 (en) | Multi-target tracking method and tracking system based on sequential bayesian filtering | |
Wheeler et al. | Deep stochastic radar models | |
CN104297748B (en) | One kind is based on tracking before the enhanced Radar Targets'Detection in track | |
CN101975575B (en) | Multi-target tracking method for passive sensor based on particle filtering | |
CN103902829A (en) | Target tracking method and system transmitting edge distribution and existence probability | |
CN104730511A (en) | Tracking method for multiple potential probability hypothesis density expansion targets under star convex model | |
CN106772353B (en) | A kind of multi-object tracking method and system suitable for flicker noise | |
CN104318059A (en) | Target tracking method applied to nonlinear gauss system and tracking system | |
CN105354860A (en) | Box particle filtering based extension target CBMeMBer tracking method | |
CN105093198A (en) | Flight path fusion method for networking detection of distributed external radiation source radars | |
CN103324835A (en) | Probability hypothesis density filter target information maintaining method and information maintaining system | |
WO2019047455A1 (en) | Multi-maneuver target tracking method and system applicable to flicker noise | |
CN104794735A (en) | Extended target tracking method based on variational Bayesian expectation maximization | |
WO2018010099A1 (en) | Target tracking method for turn maneuver, and system for same | |
CN103424742B (en) | Target tracking method and target tracking system of sequential processing measurement data | |
CN104459661A (en) | Method for detecting rapid artillery type dim target | |
CN106168943A (en) | A kind of method for following the tracks of turning machine moving-target and system thereof | |
De Melo et al. | Stochastic particle flow for nonlinear high-dimensional filtering problems | |
Xia et al. | Extended object tracking with automotive radar using learned structural measurement model | |
Lindenmaier et al. | GM-PHD filter based sensor data fusion for automotive frontal perception system | |
CN103390107A (en) | Target tracking method based on Dirac weighted sum and target tracking system | |
CN104063615A (en) | Target tracking method and tracking system based on variable coefficient alpha-beta filter | |
Wan et al. | Variational Bayesian learning for robust AR modeling with the presence of sparse impulse noise |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
EXSB | Decision made by sipo to initiate substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150826 |
|
RJ01 | Rejection of invention patent application after publication |