CN113408422A - Multi-frame joint detection tracking and classification method suitable for weak targets - Google Patents

Multi-frame joint detection tracking and classification method suitable for weak targets Download PDF

Info

Publication number
CN113408422A
CN113408422A CN202110685810.5A CN202110685810A CN113408422A CN 113408422 A CN113408422 A CN 113408422A CN 202110685810 A CN202110685810 A CN 202110685810A CN 113408422 A CN113408422 A CN 113408422A
Authority
CN
China
Prior art keywords
target
tracking
representing
type
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.)
Granted
Application number
CN202110685810.5A
Other languages
Chinese (zh)
Other versions
CN113408422B (en
Inventor
易伟
曾楷
邓杰
文耀毅
秦雯
李武军
孔令讲
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 CN202110685810.5A priority Critical patent/CN113408422B/en
Publication of CN113408422A publication Critical patent/CN113408422A/en
Application granted granted Critical
Publication of CN113408422B publication Critical patent/CN113408422B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/246—Analysis of motion using feature-based methods, e.g. the tracking of corners or segments
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing
    • G06T2207/10044—Radar image

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Probability & Statistics with Applications (AREA)
  • Multimedia (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

The invention discloses a multi-frame joint detection tracking and classifying method suitable for a weak target, which is applied to the field of radar target detection tracking and classification; aiming at the problems of poor tracking performance and low classification accuracy of target joint tracking and classification algorithm detection after traditional single-frame threshold detection, firstly, multi-dimensional analysis is carried out on target characteristics, and then a high-dimensional characteristic space based on target types is established; secondly, establishing a multidimensional conditional type probability function and a measuring envelope likelihood function by using the selected target state space so as to establish a weighted likelihood ratio based on the conditional type; then, the optimal estimation of the target state is found out through multi-stage joint iterative optimization, and multi-frame joint detection tracking based on the multi-dimensional characteristics of the target is realized; finally, the target type under the condition that the target estimation state is known is subjected to joint optimization, so that the target type is effectively estimated; the method can effectively solve the problems of detection, tracking and classification based on the target multi-dimensional characteristic space in the complex environment.

Description

Multi-frame joint detection tracking and classification method suitable for weak targets
Technical Field
The invention belongs to the field of radar target detection tracking and classification, and particularly relates to a multi-frame joint detection tracking and classification technology for weak targets in a complex environment.
Background
With the development of radar detection technology and the current situation of international military change at many ends, active radar systems face more challenging tracking and classification requirements. However, in the conventional joint tracking and classifying algorithm, a binarization trace is formed after single-frame threshold detection, and then the binarization trace is used in the subsequent joint tracking and classifying algorithm. Aiming at a weak target in a complex environment, in the process of single-frame threshold detection, due to the fact that target energy is too low, a target point track cannot pass through a threshold, the target is lost, track association and target type estimation cannot be carried out, and the disadvantages that track continuity is poor, the target is seriously lost, the target type cannot be estimated and the like are formed. The disadvantage seriously restricts the active radar from meeting the requirement of joint tracking and target classification.
The multi-frame joint detection technology is an energy accumulation method for weak targets, which skips single-frame threshold detection, directly performs multi-frame joint processing on original radar measurement, and achieves the purpose of target energy accumulation through the difference of motion correlation between targets and noise (or clutter) between frames. However, the energy accumulation method ignores the difference between different types of targets, and cannot achieve the purpose of target differentiation. In the traditional joint tracking and classifying algorithm, the binaryzation trace point data after the single-frame threshold is generally adopted for relevant filtering, and meanwhile, model matching, a target characteristic likelihood function and the like are introduced to assist target judgment so as to realize effective target distinguishing. For example, in the document "Joint target tracking and classification with particulate filtering and mixture Kalman filtering using kinematic radar information," Digital Signal Processing,2006, pp,180-204 ", Joint target tracking and classification based on particle filtering and hybrid Kalman filtering is realized by using radar motion measurement, and the method directly performs target modeling on a point trace level, and simultaneously considers model constraints and speed constraints of different types of targets to improve classification performance; the document 'model type matching based multi-target joint tracking classification of PHD filter and TBM, system engineering and electronic technology, 2016, pp, 2235-2243' proposes a model type matching probability hypothesis density filter, and combines the model type matching probability hypothesis density filter with a transferable reliability model of a multi-sensor, and realizes the joint tracking and classification of multiple targets by using a plurality of kinematic radars and particle filter recursion. The methods in the two documents firstly need to detect through a single-frame threshold, and can inhibit the trace point formation of a weak target to a certain extent; in CN202010082192.0, a maneuvering target state transition model is used to model the maneuvering characteristics of a target, and then the size of a multi-frame accumulated possible state transition interval is adaptively adjusted according to the established target maneuvering model, thereby realizing the state estimation of a weak maneuvering target. However, the influence of different kinds of targets on the state transition interval and the accumulation value function is not considered, so that target type estimation cannot be realized. Therefore, the above methods are not suitable for detection, tracking and classification of weak targets.
Disclosure of Invention
In order to solve the technical problems, the invention provides a multi-frame joint detection tracking and classifying method suitable for weak targets, and particularly suitable for joint tracking and classifying after the estimation performance under the environment with low signal-to-noise ratio is superior to that of the conventional single-frame detection.
The technical scheme adopted by the invention is as follows: a multi-frame joint detection tracking and classification method suitable for weak targets comprises the following steps:
s1, establishing a high-dimensional characteristic space based on the target type; after the high-dimensional characteristic space is determined, a high-dimensional conditional type probability function can be established according to the high-dimensional conditional type probability function, and then the high-dimensional conditional type probability function is used for weighting likelihood ratios in value function accumulation;
s2, obtaining a weighted likelihood ratio based on the condition type according to the measurement envelope likelihood function and the condition type probability of different types of targets;
s3, using the weighted likelihood ratio for value function accumulation, and finding out the optimal estimation of the target state through multi-stage joint iterative optimization;
and S4, carrying out joint optimization on the target types under the condition that the target estimation states are known, and finding out the estimation value of the target type.
In step S2, the weighted likelihood ratio is calculated by:
Figure BDA0003124562790000021
wherein, ykRepresenting the target state, zkRepresenting the k frame radar echo data, c representing the target type, H0Indicates no target, Mr (z)k|yk) A weighted likelihood ratio is represented that represents the ratio of the likelihoods,
Figure BDA0003124562790000022
indicating that class c object is in state ykLower corresponding measuring unit
Figure BDA0003124562790000023
The envelope likelihood function of (a);
Figure BDA0003124562790000024
indicating that there is no target lower metrology unit
Figure BDA0003124562790000025
The envelope likelihood function of (a), ln (-) represents logarithm, p (c | y)k) Representing the target multi-dimensional conditional type probability.
Step S3 specifically includes the following substeps:
s31, value function Ik(yk) And (3) performing iterative update, wherein the expression of the iterative update is as follows:
Figure BDA0003124562790000031
Figure BDA0003124562790000032
wherein, Cr (y)k|yk-1) Representing a weighted transfer cost based on a condition type; Ψ (y)k) Representing a backtracking function for state transition relationships between frames, Ik-1(yk-1) A value function representing the k-1 th iteration;
s32, judging whether the combined processing frame number is reached, if yes, executing a step S32; otherwise, returning to the step S31;
s33, judging whether the current function exceeds the detection threshold, if so, declaring a target and executing the step S34, otherwise, ending, and indicating that no target exists;
and S34, according to the declared target, utilizing a backtracking function to backtrack the target state to obtain a target state estimation sequence.
Cr (y) in step S31k|yk-1) The calculation formula of (A) is as follows:
Figure BDA0003124562790000033
wherein, p (y)k|c,yk-1) Representing class c object slave status yk-1Transition to State ykThe corresponding cost.
The effective estimation specific expression of the target type in step S4 is as follows:
Figure BDA0003124562790000034
wherein the content of the first and second substances,
Figure BDA0003124562790000035
indicating the target state estimate obtained in step S3,
Figure BDA0003124562790000036
the log sums of the likelihood functions of the metrology units representing different types of targets,
Figure BDA0003124562790000037
representing the state transition probability logarithmic sums for different types of targets.
The invention has the beneficial effects that: the method of the invention establishes a high-dimensional characteristic space based on the target type by analyzing the multi-dimensional characteristics of the target, and ensures that the selected characteristic space can match the requirements of target classification; secondly, establishing a multidimensional conditional type probability function and a measuring envelope likelihood function to further establish a weighted likelihood ratio based on the conditional type, and realizing the quantitative distribution of the likelihood ratio according to the possibility of the type under a given target state; then, finding out the optimal estimation of the target state through multi-stage joint iterative optimization; finally, carrying out joint optimization on the target type under the condition that the target estimation state is known, and finding out the estimation value of the target type; the method effectively solves the problems of detection tracking and classification based on the target multidimensional characteristic space in a complex environment, and particularly has the estimation performance superior to that of the conventional method based on single-frame detection and then combined tracking and classification in a low signal-to-noise ratio environment.
Drawings
FIG. 1 is a block flow diagram of the present invention;
FIG. 2 is an iterative flow diagram of the present invention;
FIG. 3 is a schematic diagram of a simulation provided in an embodiment of the present invention;
FIG. 4 is a comparison graph of simulation results of a joint tracking and classification algorithm after single-frame threshold detection according to an embodiment of the present invention;
wherein, fig. 4(a) is a comparison graph of the target detection probability Pd and the observation time of the method and the conventional algorithm of the present invention; FIG. 4(b) is a comparison of the target detection and correct classification probability Pdc versus observation time for the method and conventional algorithm of the present invention; FIG. 4(c) is a comparison graph of the target detection type number probability Pnc versus observation time for the method of the present invention and the conventional algorithm.
Detailed Description
For the convenience of describing the contents of the present invention, the following terms are first explained:
the term 1: iterative updating
In the process of one-time multi-frame combined processing, the value function is updated once every time a new frame of measurement data is input.
The term 2: probability of target detection Pd
In the process of multi-frame joint processing once, the value function of the last frame exceeds the detection threshold Vt, and the probability that the error between the target position estimation of the last frame and the theoretical target real position is in err cells is obtained.
The term 3: probability of target detection and correct classification Pdc
In the process of one-time multi-frame joint processing, the value function of the last frame exceeds a detection threshold Vt, the error between the target position estimation of the last frame and the theoretical target real position is in err cells, and the probability that the target classification result is consistent with the theoretical target type is obtained.
The term 4: probability of number of types of object detection Pnc
In the process of multi-frame joint processing once, the value function of the last frame exceeds a detection threshold Vt, and when the errors of the target position estimation of the last frame and the theoretical real position of the target are all in err cells, the probability of the same type target estimation number/the same type target theoretical total number is obtained.
The target detection probability Pd, the target detection and correct classification probability Pdc and the target detection type number probability Pnc are used as the performance evaluation indexes of the invention.
The invention mainly adopts a simulation experiment method for verification, and all the steps and conclusions are verified to be correct on Matlab 2019 b. The invention will be explained in more detail below with reference to the accompanying fig. 1-3.
As shown in fig. 1, the implementation process of the present invention includes the following steps:
s1, establishing a high-dimensional characteristic space based on the target type;
s2, obtaining a weighted likelihood ratio based on the condition type according to the measurement envelope likelihood function and the condition type probability of different types of targets;
s3, using the weighted likelihood ratio for value function accumulation, and finding out the optimal estimation of the target state through multi-stage joint iterative optimization;
and S4, carrying out joint optimization on the target types under the condition that the target estimation states are known, and finding out the estimation value of the target type.
As shown in fig. 2, the specific iterative process of the present invention is:
step 1, initializing system parameters, and then initializing the system parameters,
in order to verify the beneficial effect of the method on the detection and classification of the weak targets, the present embodiment simulates a scenario in which a certain radar detects five moving targets (specifically including 2 first-type targets, 1 second-type target, and 2 third-type targets) in a three-dimensional space, as shown in fig. 3. The initialized system parameters are as follows: the total number of target types M is 3, and the average RCS corresponding to the three types of targets is [7.08,25.12,7.94]m2The target multidimensional characteristic space dimension s is 6, the joint processing frame number K is 6, and the state transition matrix set Q is { L ═ L1(q),L2(q),...,LM(q) }, detection threshold Vt-27.8061;
step 2, initializing a variable k to be 1;
step 3, obtaining kth frame radar echo data z from a radar receiverk，zkThe specific form of (A) is expressed as:
Figure BDA0003124562790000051
wherein the measured value
Figure BDA0003124562790000052
The envelope value of the k frame radar echo in a measuring unit (i, j, m, N) is shown, i represents the ith distance unit, j represents the jth azimuth unit, m represents the mth elevation unit, N represents the nth Doppler unit, Nr＝185,Nθ＝15,
Figure BDA0003124562790000053
Respectively quantifying the total number of units of each axis; the axes here refer in particular to: distance axis, azimuth axis, pitch axis, doppler axis.
Step 4, according toCarrying out pseudo measurement updating on the k frame radar echo data by using the measurement envelope likelihood functions and the conditional type probabilities of different types of targets to obtain a weighted likelihood ratio Mr (z) based on the conditional typek|yk) The method is used for value function iterative accumulation, and comprises the following steps:
Figure BDA0003124562790000061
Figure BDA0003124562790000062
Figure BDA0003124562790000063
Figure BDA0003124562790000064
wherein, ykRepresenting the target status, c representing the target type, H0Denotes no object, p (c | y)k) Representing the probability of the target multi-dimensional conditional type and describing the target state ykThe likelihood of target type c is known.
Figure BDA0003124562790000065
Indicating that class c object is in state ykLower corresponding measuring unit
Figure BDA0003124562790000066
The envelope likelihood function of (a);
Figure BDA0003124562790000067
indicating that there is no target lower metrology unit
Figure BDA0003124562790000068
The envelope likelihood function of (a), ln (-) represents taking the logarithm.
Figure BDA0003124562790000069
σcMean RCS (Radar Cross-Section), σ, for class c targetsvThe average power of the background noise is 1. p (v)k,hkAnd | c) represents the joint speed and the height likelihood function of the class c target, in the simulation, two-dimensional Gaussian distribution is assumed to obey, and the mean value and the covariance are respectively as follows:
the first class of targets:
Figure BDA00031245627900000610
second class of targets:
Figure BDA0003124562790000071
the third class of targets:
Figure BDA0003124562790000072
step 5, performing one-step state prediction based on the target type, and fusing all possible target state transition regions to obtain a set k (y)k) The method is used for the state transition of the target between frames, and specifically comprises the following steps:
κ(yk)＝κ(yk,c＝1)∪κ(yk,c＝2)∪...∪κ(yk,c＝M)
κ(yk,c)＝{yk-1|d(yk,fc(yk-1))≤Lc(q)}
wherein the target state ykAll values of (a) constitute a high-dimensional property space, k (y)kC) represents the state transition range of class c targets, fc(. h) is the motion transfer function of class c targets, d (· h- · is the distance between two states, Lc(q)＝[q1,q2,...,qs]T(quE {1,4,9, · is, u ═ 1,2,. and s) is a state transition range matrix of the class c target.
And 6, performing interframe transfer cost updating on the state of the kth frame target by using state transfer probability functions of different types of targetsObtaining a weighted transfer cost Cr (y) based on the condition typek|yk-1) The method comprises the following steps:
Figure BDA0003124562790000073
Figure BDA0003124562790000074
wherein, p (y)k|c,yk-1) Representing class c object slave status yk-1Transition to State ykThe corresponding cost;
Figure BDA0003124562790000075
representing a gaussian distribution with mean μ and covariance matrix Q.
Step 7, value function Ik(yk) And (4) performing iterative updating, wherein the specific operation is as follows:
Figure BDA0003124562790000076
Figure BDA0003124562790000077
wherein Mr (z)k|yk) Representing a weighted likelihood ratio based on the type of condition; y isk-1∈κ(yk) Representing a set of state transitions of the target between frames; cr (y)k|yk-1) Representing a weighted transfer cost based on a condition type; Ψ (y)k) A backtracking function is represented for the state transition relationship between frames.
Step 8, if k<K, then K equals K +1, andk(yk) Storing the value into a value function buffer as an iteration value for updating the value function of the (k + 1) th frame, and returning to the step 3; if K is equal to K, performing step 9;
step 9, judging a threshold of a value function domain, if the value function exceeds a detection threshold Vt, declaring a target and executing step 10; otherwise, the algorithm is ended, no target is declared, and the step 12 is executed;
step 10, according to the declared target, utilizing a backtracking function to backtrack the target state to obtain a target state estimation sequence
Figure BDA0003124562790000081
Step 11, carrying out joint optimization on the target types under the condition that the target estimation states are known, and obtaining effective estimation of the target types by calculating the likelihood functions of the measurement units of different types of targets and calculating the state sequence transition probability
Figure BDA0003124562790000082
The specific expression is as follows:
Figure BDA0003124562790000083
wherein the content of the first and second substances,
Figure BDA0003124562790000084
the log sums of the likelihood functions of the metrology units representing different types of targets,
Figure BDA0003124562790000085
representing the state transition probability logarithmic sums for different types of targets.
Through the steps, the process of multi-frame joint detection tracking and classification under the complex environment is completed.
FIG. 4(a) is a comparison graph of the target detection probability Pd of the Multi-frame joint detection and classification (MJDTC) algorithm and the Single-frame detection and classification (SDJTC) algorithm of the present invention with the observation time; FIG. 4(b) is a comparison graph of the target detection and correct classification probability Pdc of the proposed algorithm and the comparison algorithm with the result of the observation time; fig. 4(c) is a result comparison graph of the target detection type number probability Pnc and the observation time of the proposed algorithm and the comparison algorithm. The err parameters for the above simulations are all 2. As can be seen from FIG. 4, for weak targets, the detection tracking and classification performance of the method of the present invention is superior to that of the traditional joint tracking and classification algorithm based on single-frame threshold detection.
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 (5)

1. A multi-frame joint detection tracking and classification method suitable for weak targets is characterized by comprising the following steps:
s1, establishing a high-dimensional characteristic space based on the target type;
s2, based on the high-dimensional characteristic space, obtaining a weighted likelihood ratio based on the condition type according to the measured envelope likelihood function and the condition type probability of different types of targets;
s3, using the weighted likelihood ratio for value function accumulation, and finding out the optimal estimation of the target state through multi-stage joint iterative optimization;
and S4, carrying out joint optimization on the target types under the condition that the target estimation states are known, and finding out the estimation value of the target type.
2. The method for multi-frame joint detection, tracking and classification of weak targets according to claim 1, wherein the weighted likelihood ratio of step S2 is calculated by:
Figure FDA0003124562780000011
wherein, ykRepresenting the target state, zkRepresenting the k frame radar echo data, c representing the target type, H0Indicates no purposeLabel Mr (z)k|yk) A weighted likelihood ratio is represented that represents the ratio of the likelihoods,
Figure FDA0003124562780000012
Figure FDA0003124562780000013
indicating that class c object is in state ykLower corresponding measuring unit
Figure FDA0003124562780000014
The envelope likelihood function of (a);
Figure FDA0003124562780000015
indicating that there is no target lower metrology unit
Figure FDA0003124562780000016
The envelope likelihood function of (a), ln (-) represents logarithm, p (c | y)k) Representing the target multi-dimensional conditional type probability.
3. The multi-frame joint detection tracking and classification method suitable for the weak target according to claim 2, wherein the step S3 specifically includes the following sub-steps:
s31, value function Ik(yk) And (3) performing iterative update, wherein the expression of the iterative update is as follows:
Figure FDA0003124562780000017
Figure FDA0003124562780000018
wherein, Cr (y)k|yk-1) Representing a weighted transfer cost based on a condition type; Ψ (y)k) Representing a backtracking function for state transition relationships between frames, Ik-1(yk-1) Representing the (k-1) th iterationA value function;
s32, judging whether the combined processing frame number is reached, if yes, executing a step S32; otherwise, returning to the step S31;
s33, judging whether the current function exceeds the detection threshold, if so, declaring a target and executing the step S34, otherwise, ending, and indicating that no target exists;
and S34, according to the declared target, utilizing a backtracking function to backtrack the target state to obtain a target state estimation sequence.
4. The multi-frame joint detection tracking and classification method for weak targets according to claim 3, wherein Cr (y) in step S31k|yk-1) The calculation formula of (A) is as follows:
Figure FDA0003124562780000021
wherein, p (y)k|c,yk-1) Representing class c object slave status yk-1Transition to State ykThe corresponding cost.
5. The method for multi-frame joint detection, tracking and classification as claimed in claim 4, wherein the effective estimation concrete expression of the target type in step S4 is as follows:
Figure FDA0003124562780000022
wherein the content of the first and second substances,
Figure FDA0003124562780000023
indicating the target state estimate obtained in step S3,
Figure FDA0003124562780000024
the log sums of the likelihood functions of the metrology units representing different types of targets,
Figure FDA0003124562780000025
representing the state transition probability logarithmic sums for different types of targets.
CN202110685810.5A 2021-06-21 2021-06-21 Multi-frame joint detection tracking and classification method suitable for weak targets Active CN113408422B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110685810.5A CN113408422B (en) 2021-06-21 2021-06-21 Multi-frame joint detection tracking and classification method suitable for weak targets

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110685810.5A CN113408422B (en) 2021-06-21 2021-06-21 Multi-frame joint detection tracking and classification method suitable for weak targets

Publications (2)

Publication Number Publication Date
CN113408422A true CN113408422A (en) 2021-09-17
CN113408422B CN113408422B (en) 2022-09-09

Family

ID=77681943

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110685810.5A Active CN113408422B (en) 2021-06-21 2021-06-21 Multi-frame joint detection tracking and classification method suitable for weak targets

Country Status (1)

Country Link
CN (1) CN113408422B (en)

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5960097A (en) * 1997-01-21 1999-09-28 Raytheon Company Background adaptive target detection and tracking with multiple observation and processing stages
US20120280853A1 (en) * 2009-11-06 2012-11-08 Saab Ab Radar system and method for detecting and tracking a target
CN103247057A (en) * 2013-05-08 2013-08-14 西北工业大学 Road target multi-hypothesis tracking algorithm under target-echo-road network data association
CN104062651A (en) * 2014-06-30 2014-09-24 电子科技大学 Track-before-detect method based on G0 clutter background and constant target amplitude
WO2015040121A1 (en) * 2013-09-20 2015-03-26 Thales Method for detecting targets and associated multifunction radar
CN107340517A (en) * 2017-07-04 2017-11-10 电子科技大学 Tracking before a kind of multisensor multi frame detection
CN108333571A (en) * 2018-02-07 2018-07-27 电子科技大学 A kind of multisensor multi-frame joint detecting and tracking method merged based on mark sequence
CN109991597A (en) * 2019-04-04 2019-07-09 中国人民解放军国防科技大学 Weak-expansion-target-oriented tracking-before-detection method
CN111257865A (en) * 2020-02-07 2020-06-09 电子科技大学 Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5960097A (en) * 1997-01-21 1999-09-28 Raytheon Company Background adaptive target detection and tracking with multiple observation and processing stages
US20120280853A1 (en) * 2009-11-06 2012-11-08 Saab Ab Radar system and method for detecting and tracking a target
CN103247057A (en) * 2013-05-08 2013-08-14 西北工业大学 Road target multi-hypothesis tracking algorithm under target-echo-road network data association
WO2015040121A1 (en) * 2013-09-20 2015-03-26 Thales Method for detecting targets and associated multifunction radar
CN104062651A (en) * 2014-06-30 2014-09-24 电子科技大学 Track-before-detect method based on G0 clutter background and constant target amplitude
CN107340517A (en) * 2017-07-04 2017-11-10 电子科技大学 Tracking before a kind of multisensor multi frame detection
CN108333571A (en) * 2018-02-07 2018-07-27 电子科技大学 A kind of multisensor multi-frame joint detecting and tracking method merged based on mark sequence
CN109991597A (en) * 2019-04-04 2019-07-09 中国人民解放军国防科技大学 Weak-expansion-target-oriented tracking-before-detection method
CN111257865A (en) * 2020-02-07 2020-06-09 电子科技大学 Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
AUBREY B. POORE 等: "Some assignment problems arising from multiple target tracking", 《MATHEMATICAL AND COMPUTER MODELLING》 *
JINGHE WANG 等: "An Efficient Recursive Multiframe Track-Before-Detect Algorithm", 《IEEE TRANSACTIONS ON AEROSPACE AND ELECTRONIC SYSTEMS》 *
KEZHU LIU 等: "A Data Association Approach for Plot-Sequences Outputted by Multi-Frame Track-Before-Detect", 《2020 IEEE RADAR CONFERENCE (RADARCONF20)》 *
占荣辉 等: "基于序贯蒙特卡罗概率假设密度滤波的多目标检测前跟踪改进算法", 《电子与信息学报》 *
崔威威 等: "基于粒子滤波改进的VTS弱目标检测前跟踪算法", 《火力与指挥控制》 *

Also Published As

Publication number Publication date
CN113408422B (en) 2022-09-09

Similar Documents

Publication Publication Date Title
Eltrass et al. Automotive radar system for multiple‐vehicle detection and tracking in urban environments
CN107346020B (en) Distributed batch estimation fusion method for asynchronous multi-base radar system
CN108363054A (en) Passive radar multi-object tracking method for Single Frequency Network and multipath propagation
CN110376581B (en) Explicit multi-target tracking method based on Gaussian mixture probability hypothesis density filter
CN111259332B (en) Fuzzy data association method and multi-target tracking method in clutter environment
Lan et al. Joint target detection and tracking in multipath environment: A variational Bayesian approach
CN109214432B (en) Multi-sensor multi-target joint detection, tracking and classification method
CN111398948A (en) Maneuvering small target track association method under strong clutter background
CN116148837A (en) Robust multi-sensor multi-target tracking method under heterogeneous clutter background
CN110849372A (en) Underwater multi-target track association method based on EM clustering
CN111257865A (en) Maneuvering target multi-frame detection tracking method based on linear pseudo-measurement model
CN115508824A (en) Multi-target big data association fusion tracking method and system
CN112213697B (en) Feature fusion method for radar deception jamming recognition based on Bayesian decision theory
CN117433538A (en) Multi-source heterogeneous sensor track fusion method
CN113408422B (en) Multi-frame joint detection tracking and classification method suitable for weak targets
CN115220002B (en) Multi-target data association tracking method and related device for fixed single station
CN114488104B (en) Sky wave beyond-view range radar target tracking method based on interaction consistency
CN113126086B (en) Life detection radar weak target detection method based on state prediction accumulation
CN115619825A (en) Ground multi-target tracking state and track determining method
CN110031797B (en) Detection and tracking method for passive sensing system on target with discontinuous characteristic
CN113933806A (en) Sky wave radar multi-path measurement clustering method
CN113484866A (en) Multi-target detection tracking method based on passive sonar azimuth history map
Hanusa et al. Contact clustering and fusion for preprocessing multistatic active sonar data
Coraluppi Seeing the forest through the trees: Multiple-hypothesis tracking and graph-based tracking extensions
CN117930142B (en) Radar waveform design method for coping with high sea state sea surface maneuvering target tracking

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