CN111257865B - Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model - Google Patents

Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model Download PDF

Info

Publication number
CN111257865B
CN111257865B CN202010082192.0A CN202010082192A CN111257865B CN 111257865 B CN111257865 B CN 111257865B CN 202010082192 A CN202010082192 A CN 202010082192A CN 111257865 B CN111257865 B CN 111257865B
Authority
CN
China
Prior art keywords
target
frame
representing
tracking
detection
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.)
Active
Application number
CN202010082192.0A
Other languages
Chinese (zh)
Other versions
CN111257865A (en
Inventor
易伟
李武军
杨琪
张鹏辉
刘克柱
彭琪芸
曾凯
文耀毅
孔令讲
杨晓波
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
University of Electronic Science and Technology of China
Original Assignee
University of Electronic Science and Technology of China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by University of Electronic Science and Technology of China filed Critical University of Electronic Science and Technology of China
Priority to CN202010082192.0A priority Critical patent/CN111257865B/en
Publication of CN111257865A publication Critical patent/CN111257865A/en
Application granted granted Critical
Publication of CN111257865B publication Critical patent/CN111257865B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/66—Radar-tracking systems; Analogous systems
    • G01S13/72—Radar-tracking systems; Analogous systems for two-dimensional tracking, e.g. combination of angle and range tracking, track-while-scan radar
    • G01S13/723—Radar-tracking systems; Analogous systems for two-dimensional tracking, e.g. combination of angle and range tracking, track-while-scan radar by using numerical data
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Mathematics (AREA)
  • Algebra (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses a maneuvering target multi-frame detection and tracking method based on a linear pseudo-measurement model, which is applied to the technical field of radar target detection and tracking, in order to solve the problems of low tracking precision and poor detection performance of the traditional maneuvering target multi-frame detection and tracking method based on the movement limit constraint, and the prior art realizes the huge calculation cost problem of maneuvering target tracking by increasing the dimensionality of a search state space, the invention utilizes historical information as a prior multistage joint optimization idea, firstly establishes a linear pseudo-measurement model according to the historical information to estimate the high-order target motion information quantity, and then introducing a maneuvering target state transition model to model the maneuvering characteristics of the target, then adaptively adjusting the size of a multi-frame accumulated possible transition interval according to the established target maneuvering model, and finally outputting a complete target detection tracking track.

Description

Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model
Technical Field
The invention belongs to the technical field of radar target detection and tracking, and particularly relates to a multi-frame combined target detection and tracking technology under the condition of a high maneuvering target.
Background
Active radar systems often face the need for effective detection of weak targets and highly mobile targets in complex environments. According to the traditional maneuvering target single-frame detection and associated tracking algorithm, information loss exists in the single-frame threshold detection process, so that the problems of serious target loss, poor track continuity, more false tracks and the like caused by weak follow-up tracking processing results are caused, and the detection performance of a radar system is seriously restricted.
As an effective weak target detection and tracking technology, a pre-detection tracking Technology (TBD) has become a key research object in the current target detection and tracking field. Different from the traditional detection-first tracking algorithm, the TBD technology directly carries out multi-frame joint detection tracking processing on multi-frame original data planes which are not subjected to threshold detection, and improves the weak target detection performance by utilizing the correlation of the space-time dimension of a target. However, the existing research on the TBD algorithm mainly assumes that the target makes a uniform linear motion (CV), and the research on the TBD technology in the case of maneuvering target is still very rare. The multi-frame detection tracking problem under a radar scene is considered in The document "The use of track-before-detect in pulse-Doppler radar, IET Conference Proceedings,2002, pp.315-319", and effective tracking of a maneuvering target can be realized by expanding The transition interval range of search for The case that The maneuvering target exists. However, this method introduces a large amount of noise measurement, resulting in poor detection and tracking performance of the moving target. A multi-frame track-before-detect algorithm for a maneuvering target in a Radar system, in 2016IEEE Radar Conference (RadarConf),2016, pp.1-6, is disclosed, and a current statistical model (CS) is introduced to realize the maneuvering target tracking. However, a heuristic acceleration calculation method is only provided, and a complete theoretical model is not established, so that the target maneuvering characteristic estimation accuracy is low, the algorithm robustness is poor, and the like; in addition, no specific formula derivation is available for calculating the possible state transition region after the CS model is introduced, and the problem of algorithm calculation complexity is not considered. The patent "maneuvering target multiframe track-before-detect method suitable for pulse Doppler radar", CN105974402B "provides a multiframe detection track algorithm based on CS model; however, only one empirical value is given for the acceleration estimation, and an estimation model of historical information is not established, so that the algorithm estimation precision is low, and the problem of model adaptation easily exists. Therefore, the method can not really and effectively solve the maneuvering target problem in the multi-frame joint detection and tracking algorithm.
Disclosure of Invention
In order to solve the technical problem, the invention provides a maneuvering target multi-frame detection and tracking method based on a linear pseudo-measurement model, which effectively improves the detection performance of a multi-frame detection and tracking algorithm under the condition of a maneuvering target.
The technical scheme adopted by the invention is as follows: a multiple-frame detection and tracking method for maneuvering targets based on a linear pseudo-metric model, as shown in fig. 1, includes:
s1, roughly calculating a possible target state transition region of the previous frame;
s2, tracing each state in the possible target state transition area of the previous frame back to obtain a corresponding historical state sequence;
s3, performing one-step augmented target state vector estimation according to the historical state sequence;
s4, according to the constructed augmented target state vector, the augmented state vector of the current frame is predicted and estimated by introducing a current statistical model;
s5, correcting the possible target state transition region of the previous frame obtained in the step S1 according to the estimated augmented state vector at the previous moment to obtain a corrected possible target state transition region;
and S6, according to the corrected target possible state transition area, performing track recovery.
The invention has the beneficial effects that: the invention further solves the problem of the mismatching of the maneuvering target model in the actual tracking scene based on the processing framework of the multi-frame joint detection tracking algorithm, provides the target maneuvering characteristic self-adaptive estimation multi-frame detection tracking algorithm based on the historical information, and effectively improves the detection performance of the multi-frame detection tracking algorithm under the maneuvering target condition. On the other hand, in order to avoid introducing too high computational complexity, the method provides a one-step augmentation state estimation model and a target maneuvering state constraint model, avoids the huge computational cost of the traditional dimension-expanding search, and can realize higher maneuvering target detection tracking performance and simultaneously ensure that the algorithm complexity is maintained at a certain magnitude.
Drawings
FIG. 1 is a flow chart of a protocol of the present invention;
fig. 2 is a flow chart of implementing the scheme of the present invention provided by the embodiment of the present invention;
FIG. 3 is a schematic diagram of a motion trajectory of a maneuvering target provided by an embodiment of the invention;
FIG. 4 is a simulation result provided by an embodiment of the present invention;
where, fig. 4(a) shows the target detection probability P of different algorithms when the SNR is 10dBdAnd the motion time relationship diagram, and FIG. 4(b) shows the target detection probability P of different algorithms when the SNR is 12dBdAnd the motion time relationship diagram, and the target accurate tracking probability P of different algorithms is shown in the graph of FIG. 4(c) when the signal-to-noise ratio SNR is 10dBd-trackAnd the motion time relationship diagram, and the target accurate tracking probability P of different algorithms is shown in the graph of FIG. 4(d) when the signal-to-noise ratio SNR is 12dBd-trackAnd a motion time relationship diagram.
Detailed Description
For the convenience of describing the contents of the present invention, the following terms are first explained:
the term 1: iterative accumulation
And repeating the multi-frequency multi-period accumulation process, wherein the result of each accumulation is used as the initial value of the next accumulation.
The term 2: maneuvering target dimension expanding search
For the maneuvering target problem, approximate description of the maneuvering target motion can be realized by increasing the dimension of the target state space, wherein the higher the dimension is, the more accurate the described target motion is.
The term 3: augmented state vector
On the basis of the existing low-dimensional state vector, the formed new state vector becomes an augmented target state vector by introducing target state information with higher dimension (such as speed, acceleration or higher-order variable).
The term 4: probability of target detection (P)d)
Under the condition of a certain constant false alarm rate, the accumulated value function of the last frame exceeds a detection threshold gamma in the process of one batch processing, and the probability that the error between the estimated target position of the last frame and the real target position is in epsilon cells;
the term 5: probability of accurate tracking of target (P)d-track)
The last frame accumulation function exceeds the detection threshold gamma and the probability that the error between the estimated target position and the true target position of each frame is within epsilon cells.
Probability of target detection (P)d) Probability of accurate tracking with target (P)d-track) Are performance evaluation parameters.
The invention mainly adopts a simulation experiment method for verification, and all steps and conclusions are verified to be correct on Matlab2016 b. The invention will be explained in more detail below with reference to the accompanying fig. 1-4.
As shown in fig. 1, the implementation process of the present invention includes the following steps:
step 1, initializing system parameters,
in order to verify the beneficial effect of the method on the detection of the maneuvering target, the embodiment simulates a maneuvering target scene for making an S-turn in the monitoring area. Then initializing system parameters: maximum detection distance R of radarmax10km, maximum probe velocity vmax340m/s, the maximum detected acceleration is amax＝30m/s2. The total number of observation frames is 35 frames, and the inter-frame sampling interval T is 1 s. Distance resolution Δx15m, azimuth resolution Δy1.5 °, corresponding to the number of cells Nx×Ny＝100×100。
Step 2, initializing algorithm model parameters:
initializing parameters of a linear pseudo-measurement model: process noise vector vi,k-1Has a covariance matrix of
Figure GDA0003197079060000031
Initializing parameters of the CS maneuvering model: target maneuvering frequency α is 0.1, coefficient δx＝δyCorresponding to the initialization matrix q 1CS。
Step 3, radarThe k-th frame received has a measurement plane zk：
Figure GDA0003197079060000041
Where z isk(i, j) represents the measured amplitude value of the (i, j) th resolution cell in the k-th frame,
Figure GDA00031970790600000411
representing the k-th frame measurement space, NxRepresenting the number of cells resolved in the distance dimension, NyRepresenting the number of cells resolved in the azimuthal dimension, corresponding to a resolution cell size of Δx×Δy. Hypothesis measure zk(i, j) are independently and equally distributed among different cells, and the measurement distribution satisfies:
Figure GDA0003197079060000042
where n iskIs white Gaussian noise with 0 mean and the noise variance is sigman＝1。AkRepresents a constant target amplitude value, which can be calculated according to a target signal-to-noise ratio (SNR), and the corresponding relation is
Figure GDA0003197079060000043
H1Representing the assumption that the target exists, the corresponding distribution is denoted as f1(zk(xk|H1) On the contrary, the assumption is H0(i.e., the target is not present) and the corresponding distribution is f0(zk(H0) Here) of
Figure GDA0003197079060000044
Representing a target state vector, x, in an s-dimensional state spacek，ykRespectively represent the distance and the orientation information,
Figure GDA0003197079060000045
representing the corresponding speed information. Defining detection statistics
Figure GDA0003197079060000046
Figure GDA0003197079060000047
To update the accumulated value function.
And 4, performing maneuvering characteristic self-adaptive multi-frame accumulation:
step 4.1, if k is 1, directly using each quantization state x in the first frame1Initialization of the value function for the corresponding echo data detection statistic, i.e. I1(x1|z1)＝λ1(x1)，ψ1(x1)＝0，ψ1(x1) Representing a track backtracking function, storing the accumulated historical state sequence information, and executing the step 4.8; otherwise, step 4.2 is performed.
Step 4.2, roughly calculate the possible state transition region based on the motion bound constraint, where the velocity constraint (KBC) is based2) The possible state transition regions of (a) are represented as:
Figure GDA0003197079060000048
|xk-xk-1|≤B1Δx，
|yk-yk-1|≤B2Δy}
distance resolution Δx15m, azimuth resolution Δy＝1.5°。
Here, the
Figure GDA0003197079060000049
Respectively representing constraint boundaries of different dimensions; v. ofmaxRepresenting the maximum velocity of the object motion; based on maximum acceleration constraint (KBC)4) The possible state transition regions of (a) are represented as:
Figure GDA00031970790600000410
Figure GDA0003197079060000051
Figure GDA0003197079060000052
Figure GDA0003197079060000053
Figure GDA0003197079060000054
here, the
Figure GDA0003197079060000055
Figure GDA0003197079060000056
Wherein, amaxRepresenting the maximum acceleration of the object motion. Here is selected
Figure GDA0003197079060000057
The following steps are continued for example.
Step 4.3, if k is less than or equal to 3, executing step 4.7; otherwise, for each state in the selected possible state transition region
Figure GDA0003197079060000058
And obtaining a corresponding historical state sequence by utilizing a backtracking function:
χ1:k-1(xk-1)＝[χ1(xk-1)T，…，χk-1(xk-1)T]T，
here, the
Figure GDA0003197079060000059
The initial measurement information needs three frames to estimate the acceleration information, so that k is determined to be less than or equal to 3 in this step.
In this embodiment, for simplifying the notation, χ is used directlyiIndicating the ith frame history status.
Step 4.4, carrying out one-step augmented state estimation on the state sequence:
here augmented state
Figure GDA00031970790600000510
Is defined as in the original state
Figure GDA00031970790600000511
Figure GDA00031970790600000512
Based on the historical state sequence x1:k-1(xk-1) Information, estimating the amount of higher order object motion information, such as: the speed, the acceleration, etc. of the vehicle,
Figure GDA00031970790600000513
representing the target state space. Here, first, a linear pseudo-metric model based on historical information, corresponding to an augmented state yk-1And history state xiThe conversion relationship between the two is as follows:
χi＝Hi，k-1yk-1+vi，k-1，i＝1，...，k-1,
here Hi，k-1Is a 2 x 6 matrix, represented as
Figure GDA00031970790600000514
Where T is the inter-frame time interval, I2Is an identity matrix. Process noise vector vi，k-1Assuming zero mean white Gaussian noise, its covariance matrix is Ri，k-1. Then use
Figure GDA00031970790600000515
Representing a multi-frame state transition matrix, v1:k-1＝[v1，k-1，…，vk-1，k-1]TRepresenting a multi-frame process noise vector. According to the above relationship, can be selected from chi1:k-1(xk-1) One-step estimation of augmented target state vector yk-1。
Figure GDA0003197079060000061
According to the established linear pseudo-measurement model and by combining the formula, the one-step estimation of the augmented state vector can be obtained
Figure GDA0003197079060000062
Pk-1＝[(H1:k-1)T(R1:k-1)-1H1:k-1]-1
Wherein M isk-1＝Pk-1(H1:k-1)T(R1:k-1)-1。
For the above calculations, the following are specifically demonstrated:
and (3) proving that:
it is known that
Figure GDA0003197079060000063
Both sides of the above formula are multiplied simultaneously
Pk＝[(H1:k)T(R1:k)-1(H1:k)]-1(H1:k)T(R1:k)-1.
Then obtain
yk＝Pk(χ1:k-v1:k)＝Pkχ1:k-Pkv1:k
For a given χ1:kThe above formula explicitly gives ykIs a multivariate Gaussian random sample with a corresponding mean value of Pkχ1:kThe covariance matrix is PkR1:k(Pk)T. In addition, due to the fact that
Figure GDA0003197079060000066
Is a symmetric matrix, and can further obtain
Figure GDA0003197079060000067
And 4.5, predicting the target state based on the CS model:
introducing a CS model to model the randomness of the acceleration in the moving process of the maneuvering target, wherein the corresponding transition density of the augmentation state is expressed as:
Figure GDA0003197079060000064
here, the
Figure GDA0003197079060000065
Process noise covariance matrix of
Figure GDA0003197079060000071
α represents the target maneuver frequency, calculated as follows for the process noise levels in the different x and y directions:
Figure GDA0003197079060000072
Figure GDA0003197079060000073
represents the mean value of the acceleration of the (k-1) th frame, a-max＝-30m/s2Representing the minimum acceleration constraint value. Matrix qCSIs composed of
Figure GDA0003197079060000074
The values of the corresponding elements can be found in the references "Estimation with applications to tracking and navigation" the organic algorithms and software [ M],2004". Note that: one potential assumption in the CS model
Figure GDA0003197079060000075
(or
Figure GDA0003197079060000076
) It establishes the target acceleration estimated value and the process noise level of the k-1 frame
Figure GDA0003197079060000077
The relationship between, and therefore the process noise level, once the estimate of the acceleration of the previous frame is established
Figure GDA0003197079060000078
Can be adaptively adjusted.
According to Bayes theory, the distribution of the current frame prediction state can be further deduced as
Figure GDA0003197079060000079
The augmented state vector prediction estimate for the current frame is:
Figure GDA00031970790600000710
Figure GDA00031970790600000716
and 4.6, refining and correcting the possible state transition area: according to the obtained pre-Measured state probability density function
Figure GDA00031970790600000711
Introducing statistical distance, i.e. selecting all possible states with transition probability greater than a certain value to make transition interval of possible states
Figure GDA00031970790600000712
And correcting, eliminating a large number of false states as far as possible, and keeping a real target state. The corrected possible state transition interval is represented as
Figure GDA00031970790600000713
Figure GDA00031970790600000714
Figure GDA00031970790600000715
ΓxRepresenting the statistical distance, Γ, in the x-directionyRepresenting the statistical distance in the y-direction, whose value depends on the model noise variance.
Here, the
Figure GDA0003197079060000081
Representing a prediction state vector
Figure GDA0003197079060000082
A position element of (A), and
Figure GDA0003197079060000083
here [ A ]]i，jThe (i, j) th element (i row, j column) of the matrix A, i.e. [ omega ]k]1，1Represents the matrix omegakMiddle (1, 1) th element, [ omega ]k]4，4Represents the matrix omegakThe (4, 4) th element.
Step 4.7, updating the current frame state transition value function:
Figure GDA0003197079060000084
Figure GDA0003197079060000085
step 4.8, if K is equal to or less than K, returning to the step 3; otherwise, step 5 is executed.
Step 5, threshold detection:
Figure GDA0003197079060000086
here, the
Figure GDA0003197079060000087
Represents a set of possible target state sequences, and gamma is a certain false alarm rate P according to the Neyman-Pearson (NP) criterionfaDetection threshold under conditions. If the value function does not exceed the threshold, the target is declared to be absent.
And 6, recovering the flight path, and outputting an estimated target track sequence:
Figure GDA0003197079060000088
through the steps, the self-adaptive multi-frame joint detection tracking process of the maneuvering target is completed.
FIG. 3 is a schematic view of a maneuvering target simulation scenario. Considering a maneuvering target making an 'S' turn, the total movement time is 35S, the target just starts to make constant-speed linear motion (CV) within 1-7S, then makes a sudden acceleration turn (CA) within 8-17S, then makes a short-time CV motion within 18-20S, and finally makes a long-time cooperative turning motion (CT) within 21-35S.
Fig. 4 shows the result of the treatment according to the method of the invention. The tracking performance of 6 different algorithms is analyzed in fig. 4:
(1) MF-TBD (CV): a multi-frame detection tracking algorithm under a traditional CV model;
(2)MF-TBD(KBC2): the traditional speed constraint multi-frame detection tracking algorithm;
(3)
Figure GDA0003197079060000089
: the invention provides a 2-dimensional state-transition space multi-frame detection tracking algorithm based on a historical measurement guidance strategy;
(4)
Figure GDA00031970790600000810
: the traditional acceleration constraint multi-frame detection tracking algorithm;
(5)
Figure GDA00031970790600000811
: the invention provides a multi-frame detection tracking algorithm of a 4-dimensional state space based on a historical measurement guidance strategy;
(6) SFD (IMM-KF): in the traditional method, single-frame detection is firstly carried out, and then an interactive multi-model Kalman filtering tracking algorithm is carried out.
Fig. 4 analyzes the detection tracking performance of different algorithms under the condition of SNR of 10dB and 12 dB. FIG. 4(a) shows the target detection probability P of different algorithms when the SNR is 10dBdAnd the motion time relationship diagram, and FIG. 4(b) shows the target detection probability P of different algorithms when the SNR is 12dBdAnd the motion time relationship diagram, and the target accurate tracking probability P of different algorithms is shown in the graph of FIG. 4(c) when the signal-to-noise ratio SNR is 10dBd-trackAnd the motion time relationship diagram, and the target accurate tracking probability P of different algorithms is shown in the graph of FIG. 4(d) when the signal-to-noise ratio SNR is 12dBd-trackAnd a motion time relationship diagram. It can be seen that all the multi-frame detection tracking algorithms
Figure GDA0003197079060000091
Figure GDA0003197079060000092
The performance of the method is higher than that of the traditional SFD (detection before tracking algorithm).
The traditional detection tracking algorithm based on CV only uses a CV motion model, model mismatch can be caused to target maneuvering conditions, and the detection tracking performance of the algorithm is reduced. Conventional KBC-based2And KBC4The multi-frame detection tracking algorithm can realize limited detection tracking of the maneuvering target, however, both algorithms realize maneuvering target tracking by expanding a target motion constraint boundary, a large amount of noise or false measurement is correspondingly introduced, and the detection tracking performance of the algorithm is severely restricted.
As shown in fig. 4, the multi-frame detection and tracking algorithm based on measurement guidance proposed by the present invention (
Figure GDA0003197079060000093
And
Figure GDA0003197079060000094
) Effectively avoids the traditional KBC2And KBC4The algorithm introduces the defect of a large amount of noise or false measurement, improves the accuracy of target state search, and further improves the detection and tracking performance of the algorithm. It can also be seen that KBC increases with the search dimension (SNR from 10dB to 12dB)4The performance of (A) is always better than that of KBC2And is and
Figure GDA0003197079060000095
also always has better performance than
Figure GDA0003197079060000096
This is consistent with algorithmic theory. However, the search dimension cannot be increased in the actual algorithm application process, and the corresponding computational complexity is also multiplied, so that the selection can be performed in combination with the actual system requirements.
It will be appreciated by those of ordinary skill in the art that the embodiments described herein are intended to assist the reader in understanding the principles of the invention and are to be construed as being without limitation to such specifically recited embodiments and examples. Various modifications and alterations to this invention will become apparent to those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.

Claims (7)

1. A maneuvering target multi-frame detection tracking method based on a linear pseudo-measurement model is characterized by comprising the following steps:
s1, roughly calculating a possible target state transition region of the previous frame;
s2, tracing each state in the possible target state transition area of the previous frame back to obtain a corresponding historical measurement state sequence;
s3, performing one-step augmented target state vector estimation according to the historical measurement state sequence;
s4, according to the constructed augmented target state vector, the augmented state vector of the current frame is predicted and estimated by introducing a current statistical model;
s5, constructing a statistical model according to the current frame augmented state vector of the prediction estimation and the current frame measurement to correct the possible target state transition region of the previous frame obtained in the step S1, and eliminating the possible target state vector with the confidence coefficient lower than a certain threshold value to obtain the corrected possible target state transition region;
and S6, performing multi-frame iteration energy accumulation according to the corrected possible state transition area of the target, and finally outputting a complete target track.
2. The method as claimed in claim 1, wherein the step S1 is specifically to obtain the possible target state transition region of the previous frame by rough calculation based on the motion limit constraint.
3. The method according to claim 2, wherein the motion limit constraint is a maximum velocity constraint or a maximum acceleration constraint.
4. The method as claimed in claim 3, wherein the step S3 is calculated by:
Figure FDA0003197079050000011
Figure FDA0003197079050000015
wherein the content of the first and second substances,
Figure FDA0003197079050000012
representing an augmented state vector estimate, Pk-1Represents the estimated error covariance matrix, R, corresponding to the k-1 th frame1:k-1Representing process noise vector v1，k-1Covariance matrix of H1:k-1Representing a multi-frame state transition matrix.
5. The method for multi-frame detection and tracking of maneuvering targets based on linear pseudo-metric model as claimed in claim 4, characterized in that step S4 is calculated as:
Figure FDA0003197079050000013
Figure FDA0003197079050000016
wherein the content of the first and second substances,
Figure FDA0003197079050000014
one-step predictive estimate, Ω, representing the state of augmentationkRepresenting the corresponding estimation error covariance matrix, F representing the state transition matrix, PkEstimate representing the correspondence of the k-th frameError covariance matrix, QkRepresenting the process noise covariance matrix.
6. The method as claimed in claim 5, wherein the modified target possible state transition region of step S5 is obtained by performing multi-frame detection and tracking on the maneuvering target based on the linear pseudo-metric model
Figure FDA0003197079050000021
The expression is as follows:
Figure FDA0003197079050000022
Figure FDA0003197079050000023
Figure FDA0003197079050000024
wherein the content of the first and second substances,
Figure FDA0003197079050000025
representing a prediction state vector
Figure FDA0003197079050000026
The x-axis coordinate of (a) is,
Figure FDA0003197079050000027
representing a prediction state vector
Figure FDA0003197079050000028
Y-axis coordinate of (1), τKBC(xk) Representing the possible target state transition region, Γ, of the previous frame calculated in step S1xRepresenting the statistical distance, Γ, in the x-directionyRepresenting statistical distances in the y-direction whose values depend on model noiseVariance of sound, xk-1Distance information, x, representing the k-1 th framekIndicating distance information of the k-th frame, ykIndicating the orientation information of the k-th frame.
7. The method as claimed in claim 6, wherein the step S6 is specifically as follows: and updating the current frame state value function according to the corrected possible state transition area of the target, accumulating according to the value functions of all frames, and obtaining the target tracking track through threshold detection.
CN202010082192.0A 2020-02-07 2020-02-07 Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model Active CN111257865B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010082192.0A CN111257865B (en) 2020-02-07 2020-02-07 Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010082192.0A CN111257865B (en) 2020-02-07 2020-02-07 Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model

Publications (2)

Publication Number Publication Date
CN111257865A CN111257865A (en) 2020-06-09
CN111257865B true CN111257865B (en) 2021-09-24

Family

ID=70949242

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010082192.0A Active CN111257865B (en) 2020-02-07 2020-02-07 Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model

Country Status (1)

Country Link
CN (1) CN111257865B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113126086B (en) * 2020-12-30 2022-09-02 西安电子科技大学 Life detection radar weak target detection method based on state prediction accumulation
CN113156418B (en) * 2021-05-08 2023-02-24 中国科学院空天信息创新研究院 Monte Carlo simulation-based radar target tracking precision prediction method
CN113408422B (en) * 2021-06-21 2022-09-09 电子科技大学 Multi-frame joint detection tracking and classification method suitable for weak targets
CN113742845B (en) * 2021-09-08 2023-06-09 中国空气动力研究与发展中心超高速空气动力研究所 Method for correcting transition model by hypersonic flow field temperature

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2010064283A1 (en) * 2008-12-05 2010-06-10 トヨタ自動車株式会社 Method for determining reliability of traveling direction vector and reliability determining device
CN103324835B (en) * 2013-05-30 2016-09-28 深圳大学 The keeping method of probability hypothesis density wave filter target information and information keep system
US9530400B2 (en) * 2014-09-29 2016-12-27 Nuance Communications, Inc. System and method for compressed domain language identification
CN104965199B (en) * 2015-07-28 2017-11-17 中国人民解放军海军航空工程学院 Radar video moving target Fusion Features decision method
CN106199584B (en) * 2016-07-05 2018-12-18 电子科技大学 A kind of track initiation method based on measurement storage
CN106226751B (en) * 2016-07-08 2018-11-16 西安电子科技大学 Maneu-vering target detection and tracking based on DP-TBD
CN107656265B (en) * 2017-09-19 2021-03-30 电子科技大学 Particle filter fusion method for tracking short flight path before multi-frame detection
CN108646237A (en) * 2018-05-09 2018-10-12 西安电子科技大学 Radar maneuvering target tracking optimization method based on current statistical model
CN108828584B (en) * 2018-06-19 2020-11-24 电子科技大学 Multi-frequency target tracking-before-detection method based on track folding factor ambiguity resolution
CN109946671B (en) * 2019-04-12 2020-11-03 哈尔滨工程大学 Underwater maneuvering weak target detection and tracking method based on double-threshold judgment
CN110084197A (en) * 2019-04-28 2019-08-02 苏州清研微视电子科技有限公司 Bus passenger flow volume statistical method and system based on deep learning

Also Published As

Publication number Publication date
CN111257865A (en) 2020-06-09

Similar Documents

Publication Publication Date Title
CN111257865B (en) Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model
Wang et al. Fuzzy-control-based particle filter for maneuvering target tracking
CN107688179A (en) Combined chance data interconnection method based on doppler information auxiliary
CN113673565B (en) Multi-sensor GM-PHD self-adaptive sequential fusion multi-target tracking method
CN112328959B (en) Multi-target tracking method based on adaptive extended Kalman probability hypothesis density filter
CN111259332B (en) Fuzzy data association method and multi-target tracking method in clutter environment
CN116047494A (en) Tracking method of multi-expansion target based on multi-distance division under general sense integrated background
CN116047495B (en) State transformation fusion filtering tracking method for three-coordinate radar
CN107561506B (en) Multistage false alarm probability joint recursion target detection method
CN116520311A (en) GLMB-based adaptive track initiation method
CN113484866B (en) Multi-target detection tracking method based on passive sonar azimuth history map
CN115619825A (en) Ground multi-target tracking state and track determining method
CN111856442A (en) Multi-target tracking method for self-adaptively estimating strength of newborn target based on measured value driving
CN113190960A (en) Parallel IMM maneuvering target tracking method based on non-equal-dimension state hybrid estimation
CN111796267A (en) Maneuvering turning target tracking-before-detection method based on pseudo-spectrum matched filtering
Nussbaum et al. EM-based Radar Signal Processing and Tracking
CN116794646B (en) Mixed high-frequency radar target tracking method based on variable decibel leaf reasoning
CN113514823B (en) Multi-model maneuvering target tracking-before-detection method based on pseudo-spectrum
CN113917434B (en) Pre-detection tracking method based on cost reference particle filter bank
Li et al. Adaptive interacting multiple model for underwater maneuvering target tracking with one-step randomly delayed measurements
CN113408422B (en) Multi-frame joint detection tracking and classification method suitable for weak targets
CN116736287B (en) Ground wave multipath target tracking method based on multiple detection modes
Sun et al. H-PMHT track-before-detect method using Markov chain Monte Carlo particle filter
CN116500574B (en) Nonlinear extended target tracking filtering method and device under bias distribution condition
CN112816973B (en) Tracking information-assisted target detection method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant