CN117075097B - Maritime radar target tracking method and system based on expanded target cluster division - Google Patents
Maritime radar target tracking method and system based on expanded target cluster division Download PDFInfo
- Publication number
- CN117075097B CN117075097B CN202311318046.3A CN202311318046A CN117075097B CN 117075097 B CN117075097 B CN 117075097B CN 202311318046 A CN202311318046 A CN 202311318046A CN 117075097 B CN117075097 B CN 117075097B
- Authority
- CN
- China
- Prior art keywords
- target
- measurement
- cluster
- measuring
- targets
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 93
- 238000005259 measurement Methods 0.000 claims abstract description 213
- 239000011159 matrix material Substances 0.000 claims abstract description 72
- 238000001514 detection method Methods 0.000 claims description 38
- 230000008569 process Effects 0.000 claims description 26
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000004891 communication Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 7
- 238000004422 calculation algorithm Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 5
- 239000000203 mixture Substances 0.000 claims description 5
- 230000008033 biological extinction Effects 0.000 claims description 3
- 230000000977 initiatory effect Effects 0.000 claims description 3
- 230000009707 neogenesis Effects 0.000 claims description 3
- 230000007704 transition Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 5
- 238000012360 testing method Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 4
- 238000001914 filtration Methods 0.000 description 3
- 230000000877 morphologic effect Effects 0.000 description 3
- 238000000638 solvent extraction Methods 0.000 description 3
- 238000003491 array Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 230000008092 positive effect Effects 0.000 description 2
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
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
- G01S13/726—Multiple target tracking
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention belongs to the technical field of maritime radar target tracking, and discloses a maritime radar target tracking method and system based on expanded target cluster division, wherein the method comprises the following steps: constructing a radar expansion target model based on a random matrix method, firstly converting detected measuring points into expansion target clusters, and initializing the state of a target according to an initial moment measuring result; aiming at the problem that tracking errors are caused by difficult distinction of radar targets adhered to each other in a dense scene, the relation between the target expansion state and the probability density distribution of the corresponding measurement clusters is quantized, the Gaussian probability density distribution is utilized to fit the predicted measurement clusters of the corresponding targets, the Gaussian hybrid clustering method is adopted to measure and divide the adhered target measurement clusters, accurate matching of the targets and measurement points is achieved, and finally the target expansion state is estimated. The invention reduces the influence of the target form on the position estimation, in particular to the tracking error when the extended target measurement has adhesion, and realizes the stable tracking of the target.
Description
Technical Field
The invention belongs to the technical field of maritime radar target tracking, and particularly relates to a maritime radar target tracking method and system based on expanded target cluster division.
Background
Target tracking generally refers to a process of predicting and updating the state of a target to be tracked by using measurement data obtained by a sensor to finally obtain a continuous track of the target. The traditional radar target tracking method generally models targets as point models, namely, each target is considered to correspond to one measuring point, along with the development of technology, the resolution capability of a modern radar system is improved, the radar target has more details, a single target comprises a plurality of measuring units, an 'extended target' is formed, the extended target has more abundant information compared with the point target, and the effect of target tracking can be improved by utilizing the information.
However, unlike point targets, tracking extended targets presents new problems. In areas where targets are dense, measurement of targets in radar is stuck together due to proximity parallelism of extended targets, in which case a simple point target tracking method may determine multiple target measurements as one target, which may cause errors in correlation of targets to be tracked and measurements, and thus tracking errors. It is a matter to be solved to design radar extended target tracking schemes that consider how to partition the combined target measurements.
Through the above analysis, the problems and defects existing in the prior art are as follows:
1) The point target tracking method is not suitable for radar multi-expansion target tracking scenes: point target tracking is generally modeled as a filtering and updating process of target states, converting multi-target tracking into a problem of correlation of metrology and targets to be tracked. Under a dense multi-target scene, a plurality of target measurement results are judged to be one target, and measurement cannot be correctly matched, so that the problems of track merging, target track loss and the like occur.
2) The object measurement and division are difficult: the measurement division is a difficulty in tracking an extended target, and currently, a method is available for carrying out related calculation on all measurement points in a scene and the target to be tracked, and measuring division is carried out by using indexes such as distance, so that the calculated amount is huge, the method is not applicable to a real radar target tracking scene, and the result is not necessarily accurate because the form and size change factors of the extended target are not considered in the division process.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a maritime radar target tracking method and a maritime radar target tracking system based on expanded target cluster division, which improve the tracking effect on adhesion targets. An extended target model is constructed by considering target motion characteristics and morphological characteristics, a target combination of measurement combination is extracted by adopting a correlation algorithm, probability density distribution of target measurement is predicted through a target extended state, and finally, a plurality of adhesion target measurement divisions are realized by utilizing Gaussian mixture clustering, so that target tracking is realized.
The method is realized by converting measurement points adjacent to each other into measurement clusters, constructing a radar expansion target model, initializing the states of the targets according to initial measurement cluster results, calculating a correlation matrix according to the matching degree between the targets to be tracked and the measurement clusters, marking the tracking states of the targets, and finally dividing the measurement cluster points by utilizing the expansion state information of the matched targets for the targets and the measurement clusters in the presence of adhesion, and taking the measurement cluster division results as final measurement to track the targets.
Further, the maritime radar target tracking method based on the expanded target cluster division comprises the following steps:
s1, for measuring points detected by a radar, converting the measuring points adjacent to each other into a measuring cluster before tracking;
s2, when the moment k=1, namely at the initial moment of tracking, constructing a radar expansion target model based on a random matrix method, and initializing the state of a target to be tracked according to an initial cluster measurement result;
s3, when the moment k is greater than 1, predicting the target state of the previous moment in one step, and calculating an incidence matrix according to the relationship between the prediction result and the position, shape, distance and the like of the measuring cluster at the current moment;
S4, judging the matching relation between the target and the measurement according to the incidence matrix, and distinguishing the tracking state of the target according to the presence or absence of the corresponding matching measurement clusters, the target and the corresponding matching quantity, wherein the tracking state comprises target neogenesis, target extinction and target adhesion;
s5, for the targets and the measurement clusters with adhesion, quantifying the target prediction state into probability density distribution prediction of the measurement points, dividing the measurement points in the measurement clusters according to the distribution, and enabling the measurement points to form measurement clusters with corresponding target numbers and to be matched with the targets;
s6, starting/stopping a target track for the measurement clusters/targets which are not associated; and updating the state of the target by using the information of the measuring points in the measuring cluster for the target and the measuring cluster which are matched.
Further, the step S1 specifically includes the following steps:
s11, analyzing radar data, performing linear interpolation processing on non-uniform azimuth-distance echo data, and forming a radar image of a uniform X-Y coordinate system by utilizing coordinate conversion;
s12, performing target detection on the radar image, performing image binarization according to the existence of a target point, then performing image communication region statistics on a region where the target exists, and dividing the target region by adopting a judgment mode of neighborhood 4 communication, wherein one target region is an extended target cluster (or called a measuring cluster).
Further, the step S2 specifically includes the following steps:
s21, modeling a target expansion state by adopting a random matrix method, fitting the target by utilizing an ellipse, and if the lengths of a long half shaft and a short half shaft of the ellipse are a and b respectively and the rotation angle is theta, then a matrix variable X describing the ellipse expansion state can be expressed by the following formula:
;
in addition, a measurement rate variable is also usedTo represent the number of measurement points involved;
s22, modeling radar target motion by adopting a uniform-speed Kalman model, wherein a target motion state model and a measurement model can be expressed as the following formula:
;
wherein,representing the state of movement of the object->For state transition matrix>And->Represents process noise and measurement noise and assumes that they obey mean zero covariance +.>And->H is the observation matrix;
s23, regarding each measuring cluster in the step S1 as a single target at the initial tracking time, and assuming that the unit occupied by the corresponding radar echo image area forms a measuring cluster Z= { Z 1 ,z 2 ,…,z n },z i ={x i ,y i X, where x i 、y i Respectively the abscissa and ordinate of the measuring points, n is the number of measurements in the measuring cluster, and the initial parameter a of the variable X is calculated according to the point information of the measuring cluster 0 、b 0 And theta 0 The ellipse mode of the graph is calculated:
;
In addition, the initial time measurement rate variableWherein m is the central moment of the graphic region, and the related graphic moment parameter calculation is given by the following formula:
;
;
;
s24, initial variable of target motion stateThe relevant initial values are set as follows:
wherein the method comprises the steps ofAnd->The abscissa of the center point of the fitting ellipse for the object, respectively, has been given in step S23, velocity variable +.>State covariance matrix->For the adjustable parameter, a recommended quantity is given here, on the basis of which an adjustment can be made:
,/>。
further, the step S3 specifically includes the following steps:
s31, assuming m targets at the moment k-1, when the tracking process reaches the moment k, firstly predicting the motion and the expansion state of all targets at the previous moment in one step, and updating the process formula as follows:
;
;
;
;
according to the updated state variables, for the ith target, a target prediction center and a minimum circumscribed rectangle tBOX are obtained i ;
S32, for the measuring point set detected at the moment k, dividing the measuring point set into n measuring clusters by the method in the step S1, and extracting the center of each measuring clusterAnd circumscribe rectangle->Consider the correlation of the ith target with the jth cluster of measurements:
(1) The distance between the target and the cluster of measurements is calculated and can be expressed as:
;
Wherein,the residual error of the Kalman filter between the target and the center of the measurement cluster,covariance of the residual;
(2) Calculating the intersection ratio between the target prediction result and the measurement cluster can be expressed as:
;
s33, calculating an association matrix A according to the relation between the target and the measurement. Can be expressed in the following form:
;
the cluster of measurements that can be associated with a target should satisfy the following conditions: the distance from the intersection is smaller than the threshold value, and the intersection condition of the circumscribed rectangle is larger than a certain intersection ratio threshold value. The matrix elements are calculated in the following manner:
wherein,and->For the correlation of the target and the cluster, the comparison and distance discrimination threshold value, +.>Can be adjusted according to the actual situation, and the recommended value +.>,/>The following formula is adopted for calculation:
。
further, the step S4 specifically includes the following steps:
s41, analyzing the association matrix A and the association condition of the target and the measurement cluster, wherein the association condition of the expanded target data is divided into the following categories:
a. the target prediction result and the detection result can be well matched, and the center distance of the target prediction result and the detection result is relatively close, so that the overlapping rate of the external rectangular frame is high;
b. the target prediction result has no detection result which can be matched with the target prediction result within a certain range, and the detection result is represented as missed detection;
c. The detection result has no target prediction result which can be matched with the detection result in a certain range, and the detection result is expressed as a new target or clutter;
d. the multiple target prediction results and the same detection result have certain matching degree, and the fact that the multiple targets are mistakenly detected as one target in the detection step is the target association condition of the tracking problem aimed at in the text;
s42, analyzing each target, namely traversing the rows of the incidence matrix; when an element is=1 and corresponding toWhen the row minimum value is the row minimum value, recording the association condition of the jth measuring cluster and the ith target, and setting all other elements of the row of the matrix A to be 0; when the row elements are all 0, recording the association classification of (b), and recording the corresponding target sequence number;
s43, analyzing each measurement cluster, namely traversing the columns of the incidence matrix; and accumulating the values of each column of the correlation matrix by taking the number of targets associated with a single measurement cluster as a reference, when the number of targets is more than 2, considering the situation of adhering to the target (d), when the number of targets is 1, classifying the targets into the situation (a), when the number of targets is 0, classifying the targets into the situation (c), and finally recording the corresponding measurement clusters and the target serial numbers.
Further, the step S5 specifically includes the following steps:
S51, for targets and measurement clusters with adhesion, the target prediction state is quantized into probability density distribution prediction of measurement points, and for a plurality of targets in adhesion, the following Gaussian mixture representation (GMM model) can be adopted:
;
wherein the method comprises the steps ofFor point set data in a measurement cluster to be classified, n k The number of Gaussian distribution or the number of targets which are mixed corresponds to the number of measurement classification categories; />For the mixing coefficient, represent the probability of selecting the measurement division i-th class, corresponding to the target measurement point number +.>A predicted value; />Respectively corresponding to the central position and the expansion state in each target state x>Is a one-step predictor of (a);
s52, according to the model in the step S51, calculating the probability that each measuring point in the measurement cluster corresponding to the adhesion target belongs to a certain target, wherein the probability can be calculated by the following formula:
;
wherein,(/>{1,2,…,n k }) is a variable representing the division of the measurement into a certain target distribution,/>Representing individual measurement points in a measurement cluster;
solving the model by using an EM algorithm, and usingIndicating measurement->Probability +.>In the parameter iterative solving process, the +.>、/>And->The updated formula and log likelihood function of (2) are as follows:
;
;
;
;
wherein,for measuring set- >The number of medium measurement points;
s53, calculating and iterating the parameter updating process in the step S52 until the convergence of the log-likelihood function reaches a certain precision; finally, the probability of dividing each measurement sample into a certain distribution or a conditional probability matrix Div belonging to a certain target can be obtained, wherein the probability is shown as the following formula:
;
for each row of the conditional probability matrix Div, a maximum term for each row is calculated, which corresponds to the final division of the measurement, as shown in the following equation:
;
after the measuring points in the measuring clusters corresponding to all the merging targets are divided, new measuring clusters with the same number as the targets are formed according to the categories, and the new measuring clusters are matched with the corresponding targets.
Further, the step S6 specifically includes the following steps:
s61, for the targets which are not associated with the measurement cluster, determining whether the track is terminated according to the total number of times which is not associated with the tracking process, wherein the number of times is designed to be 3, and the adjustment can be performed on the basis; initializing the measurement clusters which are not associated with the target in the step S2 to serve as candidate tracking targets, and performing track initiation after the target can be associated with the measurement cluster at the next moment, otherwise, deleting the candidate targets;
s62, for the matched target and the measuring cluster, updating the target state by using the information of the measuring points in the measuring cluster, wherein the updating formula is as follows:
;
;
;
;
;
;
;
;
;
;
。
Other unexplained symbols are mainly some intermediate variables and adjustment parameters, suggested ƞ =0.25.
Another object of the present invention is to provide a maritime radar target tracking system based on extended target cluster division, which applies the maritime radar target tracking method based on extended target cluster division, the maritime radar target tracking system based on extended target cluster division comprising:
the measuring cluster conversion module is used for converting measuring points which are adjacent to each other into measuring clusters by utilizing a 4-neighborhood communication analysis mode for the measuring points detected by the radar before tracking;
the state initialization module is used for constructing a radar expansion target model based on a random matrix method when the moment k=1, and initializing the state of the target measurement set according to an initial measurement cluster result;
the correlation matrix calculation module is used for carrying out one-step prediction on the target state of the previous moment when the moment k is more than 1, and calculating a correlation matrix according to the relationship between the prediction result and the position, the shape, the distance and the like of the measuring cluster at the current moment;
the tracking state distinguishing module is used for judging the matching relation between the target and the measurement according to the incidence matrix and distinguishing the tracking state of the target according to the existence of corresponding matching measurement clusters, the target and the corresponding matching quantity;
The target matching module is used for quantifying the target prediction state into probability density distribution prediction of the measuring points, dividing the measuring points in the measuring clusters according to the probability density distribution prediction of the measuring points, and enabling the measuring points to form measuring clusters with corresponding target numbers and to be matched with the targets;
the track updating module is used for starting/stopping the target track for the measurement clusters/targets which are not associated;
and the state updating module is used for updating the state of the target by using the information of the measuring points in the measuring cluster for the matched target and the measuring cluster.
Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and the computer program when executed by the processor causes the processor to execute the steps of the maritime radar target tracking method based on the extended target cluster division.
Another object of the present invention is to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the maritime radar target tracking method based on the extended target cluster division.
The invention further aims to provide an information data processing terminal which is used for realizing the maritime radar target tracking system based on the expanded target cluster division.
In combination with the technical scheme and the technical problems to be solved, the technical scheme to be protected has the following advantages and positive effects:
firstly, an extended target model is constructed by utilizing the target motion state and the morphological characteristics, the relation between the target extended state and the measurement probability density distribution is quantized on the basis, the measurement division is performed by utilizing a Gaussian mixture model clustering method, the influence of the target morphology on position estimation is reduced, and the accurate matching of the target and measurement is realized; aiming at the problem of large calculation amount of division of all measuring points, the invention firstly uses the incidence matrix to judge the matching relation between the targets and the measurement, and distinguishes the tracking state of the targets according to the existence of corresponding matching measuring clusters, the targets and the corresponding matching quantity, and only carries out measurement division on the combined and adhered target measurement, thereby greatly reducing the calculation amount and improving the efficiency of the target tracking method in dense scenes.
Secondly, aiming at the problems existing in the prior art, the invention provides a maritime radar target tracking method based on the division of the extended target cluster, which realizes the tracking task of multiple extended targets in a maritime radar dense scene; an extended target model is constructed by considering the motion characteristics and the morphological characteristics of the radar extended target, and the filtering and updating of the extended target state are realized in the form of an extended target measurement cluster; an expanded target association mode is designed, a target combination of measurement combination is extracted according to the matching relation between the measurement cluster and the target, and measurement division of adhesion targets can be performed pertinently; and predicting probability density distribution of target measurement through a target expansion state, and utilizing mixed Gaussian clustering to realize the division of a plurality of adhesion target measurements so as to realize target tracking.
Third, the following are significant technological advances and positive effects of each step of the present invention:
after the radar data are analyzed, the target measurement is reduced by using the connected region statistics mode and converted into the extended target cluster, and a data basis is provided for the subsequent extended target modeling.
In the second step provided by the invention, in the process of carrying out state initialization on the target measurement set according to the initial measurement cluster result, the initial expansion state of the target measurement is fitted by adopting a form of calculating the image moment, so that the filter updating process of the target state can be stably and rapidly converged, and the robustness of the method is improved.
According to the method, the matching relation between the target and the measurement is judged by calculating the incidence matrix, the tracking state of the target is distinguished according to whether the corresponding matching measurement cluster, the target and the corresponding matching quantity exist, the adhesion target can be accurately distinguished in the process, the target can be appointed for subsequent measurement division, and the efficiency of the method is improved.
The fifth step provided by the invention is to quantize the target prediction state into the probability density distribution prediction of the measuring points, divide the measuring points in the measuring cluster according to the probability density distribution prediction, reduce the influence of the size change of the target form on the division, improve the accuracy of the measurement division and have good technical effects.
The sixth step provided by the invention is to perform target track starting/stopping and state updating on the measurement clusters/targets which are not associated and have been formed with association respectively, and the mode of processing respectively can be realized in parallel processing in practice, so that the calculation efficiency of the system is improved.
Drawings
FIG. 1 is a flowchart of a maritime radar target tracking method based on extended target cluster partitioning provided by an embodiment of the invention;
FIG. 2 is a schematic diagram illustrating the conversion of measurement points into measurement clusters according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of initializing a target state based on a random matrix method according to an embodiment of the present invention; wherein, (a) is a relation schematic of a target expansion ellipse and a measurement cluster, and (b) is a target expansion state parameter schematic;
FIG. 4 is a schematic diagram of a process for associating with a cluster of measurements provided by an embodiment of the present invention;
FIG. 5 is a schematic diagram of an association matrix and a target tracking status class according to an embodiment of the present invention;
FIG. 6 is a diagram illustrating merging target measurement partitioning according to an embodiment of the present invention; wherein, (a) is the relation between two combined target measurements and the predicted extended ellipse position in the example, and (b) is a measurement division schematic formed by utilizing the relation;
FIG. 7 is an effect diagram of a radar target tracking system provided by an embodiment of the present invention;
FIG. 8 is a schematic view of a test data area provided by an embodiment of the present invention;
FIG. 9 is a schematic illustration of a scenario of experimental data provided by an embodiment of the present invention; (a) frame 1, (b) frame 55, and (c) frame 150;
FIG. 10 is a diagram of a center positioning error of each target tracking result provided by an embodiment of the present invention; (a) target 1, (b) target 2, (c) target 3;
fig. 11 shows a target tracking result overlapping ratio chart (a) target 1, (b) target 2, and (c) target 3 according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Aiming at the problems existing in the prior art, the invention provides a maritime radar target tracking method and system based on expanded target cluster division.
The maritime radar target tracking method based on the expanded target cluster division provided by the embodiment of the invention comprises the following steps:
s1, for measuring points detected by a radar, converting the measuring points adjacent to each other into a measuring cluster before tracking;
S2, when the moment k=1, namely when the initial moment is tracked, constructing a radar expansion target model based on a random matrix method, and initializing the state of a target measurement set according to an initial measurement cluster result;
s3, when the moment k is greater than 1, predicting the target state of the previous moment in one step, and calculating an incidence matrix according to the relationship between the prediction result and the position, shape, distance and the like of the measuring cluster at the current moment;
s4, judging the matching relation between the target and the measurement according to the incidence matrix, and distinguishing the tracking state of the target according to the presence or absence of the corresponding matching measurement clusters, the target and the corresponding matching quantity, wherein the tracking state comprises target neogenesis, target extinction and target adhesion;
s5, for the targets and the measurement clusters with adhesion, quantifying the target prediction state into probability density distribution prediction of the measurement points, dividing the measurement points in the measurement clusters according to the distribution, and enabling the measurement points to form measurement clusters with corresponding target numbers and to be matched with the targets;
s6, starting/stopping a target track for the measurement clusters/targets which are not associated; and updating the state of the target by using the information of the measuring points in the measuring cluster for the target and the measuring cluster which are matched.
Further, the step S1 specifically includes the following steps:
S11, analyzing radar data, performing linear interpolation processing on non-uniform azimuth-distance echo data, and forming a radar image of a uniform X-Y coordinate system by utilizing coordinate conversion;
s12, performing target detection on the radar image, performing image binarization according to the existence of a target point, then performing image communication region statistics on a region where the target exists, and dividing the target region by adopting a judgment mode of neighborhood 4 communication, wherein one target region is an extended target cluster (or called a measuring cluster).
The step S2 specifically comprises the following steps:
s21, modeling a target expansion state by adopting a random matrix method, fitting the target by utilizing an ellipse, and if the lengths of a long half shaft and a short half shaft of the ellipse are a and b respectively and the rotation angle is theta, then a matrix variable X describing the ellipse expansion state can be expressed by the following formula:
;
in addition, a measurement rate variable is also usedTo represent the number of measurement points involved;
s22, modeling radar target motion by adopting a uniform-speed Kalman model, wherein a target motion state model and a measurement model can be expressed as the following formula:
;
wherein,representing the state of movement of the object->For state transition matrix>And->Represents process noise and measurement noise and assumes that they obey mean zero covariance +. >And->H is the observation matrix;
s23, regarding each measuring cluster in the step S1 as a single target at the initial tracking time, and assuming that the unit occupied by the corresponding radar echo image area forms a measuring cluster Z= { Z 1 ,z 2 ,…,z n },z i ={x i ,y i X, where x i 、y i Respectively the abscissa and ordinate of the measuring points, n is the number of measurements in the measuring cluster, and the initial parameter a of the variable X is calculated according to the point information of the measuring cluster 0 、b 0 And theta 0 The ellipse mode of the graph is calculated:
;
in addition, the initial time measurement rate variableWhere m is the central moment of the graphic region, the relevant graphic moment parameter calculation is given by:
;
;
;
wherein m is 00 For the zero-order moment of the graphic region, the area of the region is represented, m 10 And m 01 First moments of the graphic region about the x-axis and the y-axis, respectively, are used to determine centroid coordinates of the region、/>Second moment m 20 And m 02 Representing the stretching degree of the target area in the horizontal and vertical directions, m 22 Indicating the inclination of the region.
S24, initial variable of target motion stateThe relevant initial values are set as follows:
wherein,and->The abscissa of the center point of the fitting ellipse to the target, the centroid coordinates of the region, are given in step S23, v x 、v y For the speed of the object in the x-axis and y-axis directions, the speed variable +. >State covariance matrix->For the adjustable parameter, a recommended quantity is given here, on the basis of which an adjustment can be made:
,/>。
the step S3 specifically comprises the following steps:
s31, assuming m targets at the moment k-1, when the tracking process reaches the moment k, firstly, predicting the motion and the expansion states of all targets at the previous moment in one step, wherein the process formula is as follows:
;
;
;
;
wherein the method comprises the steps of、/>、/>And->Respectively represent the targets according to the k-1 timeState, prediction of motion state variable, covariance variable, extended state variable and measurement rate state variable at time k. According to the updated state variables, for the ith target, a target prediction center and a minimum circumscribed rectangle tBOX are obtained i ;
S32, for the measuring point set detected at the moment k, dividing the measuring point set into n measuring clusters by the method in the step S1, and extracting the center of each measuring clusterAnd circumscribe rectangle->Consider the correlation of the ith target with the jth cluster of measurements:
(1) The weighted distance between the target and the cluster of measurements is calculated and can be expressed as:
;
wherein,the residual error of the Kalman filter between the target and the center of the measurement cluster,covariance of the residual;
(2) Calculating the intersection ratio between the target prediction result and the measurement cluster can be expressed as:
;
The formula is used for measuring the coincidence degree of the prediction result and the measurement result, wherein the formula respectively represents the coincidence area and the coincidence area between the target circumscribed rectangles.
S33, calculating an association matrix A according to the relation between the target and the measurement. Can be expressed in the following form:
;
the cluster of measurements that can be associated with a target should satisfy the following conditions: the distance from the intersection is smaller than the threshold value, and the intersection condition of the circumscribed rectangle is larger than a certain intersection ratio threshold value. The matrix elements are calculated in the following manner:
wherein,and->For the correlation of the target and the cluster, the comparison and distance discrimination threshold value, +.>Can be adjusted according to the actual situation, and the recommended value +.>,/>The following formula is adopted for calculation: />
。
The step S4 specifically comprises the following steps:
s41, analyzing the association matrix A and the association condition of the target and the measurement cluster, wherein the association condition of the expanded target data is divided into the following categories:
a. the target prediction result and the detection result can be well matched, and the center distance of the target prediction result and the detection result is relatively close, so that the overlapping rate of the external rectangular frame is high;
b. the target prediction result has no detection result which can be matched with the target prediction result within a certain range, and the detection result is represented as missed detection;
c. the detection result has no target prediction result which can be matched with the detection result in a certain range, and the detection result is expressed as a new target or clutter;
d. The multiple target prediction results and the same detection result have certain matching degree, and the fact that the multiple targets are mistakenly detected as one target in the detection step is the target association condition of the tracking problem aimed at in the text;
s42, analyzing each target, namely traversing the rows of the incidence matrix; when an element is=1 and corresponding toWhen the row minimum value is the row minimum value, recording the association condition of the jth measuring cluster and the ith target, and setting all other elements of the row of the matrix A to be 0; when the row elements are all 0, recording the association classification of (b), and recording the corresponding target sequence number;
s43, analyzing each measurement cluster, namely traversing the columns of the incidence matrix; and accumulating the values of each column of the correlation matrix by taking the number of targets associated with a single measurement cluster as a reference, when the number of targets is more than 2, considering the situation of adhering to the target (d), when the number of targets is 1, classifying the targets into the situation (a), when the number of targets is 0, classifying the targets into the situation (c), and finally recording the corresponding measurement clusters and the target serial numbers.
The step S5 specifically comprises the following steps:
s51, for targets and measurement clusters with adhesion, the target prediction state is quantized into probability density distribution prediction of measurement points, and for a plurality of targets in adhesion, the following Gaussian mixture representation (GMM model) can be adopted:
;
Wherein the method comprises the steps ofFor point set data in a measurement cluster to be classified, n k For mixed Gaussian distribution quantity or target quantity, pairThe number of categories is divided according to the measurement; />For the mixing coefficient, represent the probability of selecting the measurement division i-th class, corresponding to the target measurement point number +.>A predicted value; />Respectively corresponding to the central position and the expansion state in each target state x>Is a one-step predictor of (a);
s52, according to the model in the step S51, calculating the probability that each measuring point in the measurement cluster corresponding to the adhesion target belongs to a certain target, wherein the probability can be calculated by the following formula:
;
wherein,(/>{1,2,…,n k }) is a variable representing the division of the measurement into a certain target distribution,/>Representing individual measurement points in a measurement cluster; />
Solving the model by using an EM algorithm, and usingIndicating measurement->Probability +.>In the parameter iterative solving process, the +.>、/>And->The updated formula and log likelihood function of (2) are as follows:
;
;
;
;
wherein,for measuring set->The number of medium measurement points;
s53, calculating and iterating the parameter updating process in the step S52 until the convergence of the log-likelihood function reaches a certain precision; finally, the probability of dividing each measurement sample into a certain distribution or a conditional probability matrix Div belonging to a certain target can be obtained, wherein the probability is shown as the following formula:
;
For each row of the conditional probability matrix Div, a maximum term for each row is calculated, which corresponds to the final division of the measurement, as shown in the following equation:
;
after the measuring points in the measuring clusters corresponding to all the merging targets are divided, new measuring clusters with the same number as the targets are formed according to the categories, and the new measuring clusters are matched with the corresponding targets.
The step S6 specifically comprises the following steps:
s61, for the targets which are not associated with the measurement cluster, determining whether the track is terminated according to the total number of times which is not associated with the tracking process, wherein the number of times is designed to be 3, and the adjustment can be performed on the basis; initializing the measurement clusters which are not associated with the target in the step S2 to serve as candidate tracking targets, and performing track initiation after the target can be associated with the measurement cluster at the next moment, otherwise, deleting the candidate targets;
s62, for the matched target and the measuring cluster, updating the target state by using the information of the measuring points in the measuring cluster, wherein the updating formula is as follows:
;
;/>
;
;
;
;
;
;
;
;
。
other unexplained symbols are mainly some intermediate variables and adjustment parameters, suggested ƞ =0.25.
As shown in fig. 7, preferably, the radar target tracking system effect diagram provided by the embodiment of the present invention, a maritime radar target tracking system based on extended target cluster division provided by the embodiment of the present invention includes:
The measuring cluster conversion module is used for converting measuring points which are adjacent to each other into measuring clusters by utilizing a 4-neighborhood communication analysis mode for the measuring points detected by the radar before tracking;
the state initialization module is used for constructing a radar expansion target model based on a random matrix method when the moment k=1, and initializing the state of the target measurement set according to an initial measurement cluster result;
the correlation matrix calculation module is used for carrying out one-step prediction on the target state of the previous moment when the moment k is more than 1, and calculating a correlation matrix according to the relationship between the prediction result and the position, the shape, the distance and the like of the measuring cluster at the current moment;
the tracking state distinguishing module is used for judging the matching relation between the target and the measurement according to the incidence matrix and distinguishing the tracking state of the target according to the existence of corresponding matching measurement clusters, the target and the corresponding matching quantity;
the target matching module is used for quantifying the target prediction state into probability density distribution prediction of the measuring points, dividing the measuring points in the measuring clusters according to the probability density distribution prediction of the measuring points, and enabling the measuring points to form measuring clusters with corresponding target numbers and to be matched with the targets;
the track updating module is used for starting/stopping the target track for the measurement clusters/targets which are not associated;
And the state updating module is used for updating the state of the target by using the information of the measuring points in the measuring cluster for the matched target and the measuring cluster.
As shown in fig. 1, a flowchart provided by an embodiment of the present invention includes the following specific procedures:
s101, at the initial tracking moment, as shown in FIG. 2, measuring points are converted into measuring clusters, and each measuring cluster is regarded as a target;
s102, as shown in FIG. 3, calculating a graph ellipse of the measurement cluster area to fit a target, and simultaneously obtaining parameters of the ellipse for initializing an expansion state of the target;
s103, in the tracking process, as shown in FIG. 4, the state information of the target tracked at the previous moment is utilized to conduct one-step prediction, so as to obtain a prediction center and a prediction circumscribed rectangle of the target at the current moment;
for a measuring point at the current moment, after the measuring point is converted into a measuring cluster by utilizing 4-communication area analysis, the center of the measuring cluster and an external rectangle are obtained;
calculating the center distance and the intersection ratio between the target prediction state and the measurement cluster state according to the target prediction state and the measurement cluster state, and comparing the calculated center distance and the intersection ratio with a designed threshold value to generate an association matrix;
s104, for the association case in FIG. 4, a specific association matrix is shown in FIG. 5, where T represents a target and B represents a measurement cluster;
Traversing the correlation matrix, the correlation between the target and the cluster can be categorized. According to the correlation result, the normal matching of the target T1 and the measurement cluster B1 can be judged, the targets T2 and T3 are simultaneously matched with the measurement cluster B2, namely, the situation that the measurement adhesion combination exists in the targets, the target T4 is in missed detection, and the measurement clusters B3 and B4 are in clutter or new targets;
s105, for T2 and T3 of the target measurement adhesion condition and a corresponding measurement cluster B2, as shown in FIG. 6, converting an ellipse corresponding to each target prediction state into a corresponding two-dimensional Gaussian distribution, generating a mixed Gaussian distribution, namely a GMM model, for each point in the measurement cluster, calculating probability of each Gaussian distribution, solving by using an EM algorithm, and finally dividing the probability into two corresponding measurement clusters according to the probability and matching with the corresponding targets;
s106, according to the judgment of the situation, for the target T4, according to the times of missed detection (not related to the measurement cluster), performing track termination when the number of missed detection is larger than a designed threshold value; initializing a target state for the measurement clusters B3 and B4, and determining whether to start a track according to whether correlation is continued at the subsequent moment; and updating the state of the target by using the corresponding measuring points in the measuring cluster for the matched target and the measuring cluster.
Example 1: in the water traffic management system, under the condition that marine radars are used as target detection and tracking equipment, the detection capability and the tracking capability of the radars in dense target scenes can be enhanced by the target tracking method based on the expanded target cluster division.
An application embodiment of the present invention provides a computer device including a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of a maritime radar target tracking method based on expanded target cluster partitioning.
An application embodiment of the present invention provides a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to execute steps of a maritime radar target tracking method based on expanded target cluster division.
The application embodiment of the invention provides an information data processing terminal which is used for realizing a maritime radar target tracking system based on expanded target cluster division.
(1) The area where the test data is located is the area of the anchor land of the sauvignon of south Beijing as shown in FIG. 8;
specific test a target parallel-chase scene as shown in fig. 9, where the radar data total 150 frames: mainly for 3 targets shown in fig. 9 (a) (white solid line boxes are marked and serial numbers are marked by numbers), wherein the target No. 2 is in parallel motion state with the target No. 3 at the beginning, and lasts for about 50 frames, and the target No. 2 exceeds the target No. 3; target number 1 is in an independent state from the beginning to about 55 frames, then starts to cross target number 3, and keeps a parallel state in the middle until the end.
(2) Test results. Tracking experiments are carried out on 3 selected targets, wherein the tracking method (EOT), the Kalman filtering method (KF), the KCF and the MKCF algorithms provided by the invention are all used for tracking under the condition of given target positions in a first frame.
Fig. 10, 11, table 1, table 2 show the centering errors and the overlap ratio of the methods used during the test. For the target 1, the KF can obtain a better effect when being in an independent state, and meanwhile, as the result of the proximity correlation of the target 3 is transferred to the target 2, the KF method has a better effect on the target 2, but the tracking of the target 3 is transferred, and the effect is poor; since the appearance of the target changes greatly, the effect of the KCF method is poor, tracking will drift in the previous tens of frames, and similarly, the MKCF method also has the problem that a model can be built by utilizing the new state of the latest target in a data association mode, so that the method has a certain effect on the target 1, and the problem that the method can find the wrong target to build the target model and finally cause tracking drift due to the extremely long adhesion time of the target 2 and the target 3. From the end result, the method provided by the invention is effective for the goal of long-time adhesion in dense scenes.
TABLE 1 statistical results of target tracking center positioning errors
TABLE 2 statistics of target tracking overlap ratio
It should be noted that the embodiments of the present invention can be realized in hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or special purpose design hardware. Those of ordinary skill in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such as provided on a carrier medium such as a magnetic disk, CD or DVD-ROM, a programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The device of the present invention and its modules may be implemented by hardware circuitry, such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., as well as software executed by various types of processors, or by a combination of the above hardware circuitry and software, such as firmware.
The foregoing is merely illustrative of specific embodiments of the present invention, and the scope of the invention is not limited thereto, but any modifications, equivalents, improvements and alternatives falling within the spirit and principles of the present invention will be apparent to those skilled in the art within the scope of the present invention.
Claims (7)
1. A maritime radar target tracking method based on expanded target cluster division is characterized in that firstly, measuring points which are adjacent to each other are converted into measuring clusters, then a radar expanded target model is constructed, the targets are initialized according to initial measuring cluster results, then an association matrix is calculated according to the matching degree between the targets to be tracked and the measuring clusters, the tracking states of the targets are marked, finally, the targets and the measuring clusters with adhesion conditions are divided by utilizing expanded state information of the matched targets, and the measuring cluster division results are used as final measurement to track the targets;
the method specifically comprises the following steps:
s1, for measuring points detected by a radar, converting the measuring points adjacent to each other into a measuring cluster before tracking;
s2, when the moment k=1, namely when the initial moment is tracked, constructing a radar expansion target model based on a random matrix method, and initializing the state of a target measurement set according to an initial measurement cluster result;
S3, when the moment k is more than 1, further predicting the target state of the previous moment, and calculating an incidence matrix according to the predicting result and the relation between the current moment measuring cluster position, the current moment measuring cluster shape and the distance;
s4, judging the matching relation between the target and the measurement according to the incidence matrix, and distinguishing the tracking state of the target according to the presence or absence of the corresponding matching measurement clusters, the target and the corresponding matching quantity, wherein the tracking state comprises target neogenesis, target extinction and target adhesion;
s5, for the targets and the measurement clusters with adhesion, quantifying the target prediction state into probability density distribution prediction of the measurement points, dividing the measurement points in the measurement clusters according to the distribution, and enabling the measurement points to form measurement clusters with corresponding target numbers and to be matched with the targets;
s6, starting or stopping a target track for the measurement clusters or targets which are not associated; for the matched target and the measuring cluster, updating the target state by using the information of the measuring points in the measuring cluster;
the step S1 specifically comprises the following steps:
s11, analyzing radar data, performing linear interpolation processing on non-uniform azimuth-distance echo data, and forming a radar image of a uniform X-Y coordinate system by utilizing coordinate conversion;
S12, performing target detection on a radar image, performing image binarization according to the existence of a target point, then performing image communication region statistics on a region where the target exists, and dividing the target region by adopting a judgment mode of neighborhood 4 communication, wherein one target region is an extended target cluster or a measurement cluster;
the step S2 specifically comprises the following steps:
s21, modeling a target expansion state by adopting a random matrix method, and fitting the target by utilizing an ellipse, wherein the lengths of a long half shaft and a short half shaft of the ellipse are respectively a and b, the rotation angle is theta, and a matrix variable X of the ellipse expansion state is expressed by the following formula:
;
using a variable of measurement rateTo represent the number of measurement points involved;
s22, modeling radar target motion by adopting a uniform-speed Kalman model, wherein a target motion state model and a measurement model are expressed as follows:
;
;
wherein,representing the state of movement of the object->For state transition matrix>And->Representing process noise and metrology noise and assuming that they obey a mean value of zero covariance fractionLet alone->And->H is the observation matrix;
s23, regarding each measuring cluster in the step S1 as a single target at the initial tracking time, and assuming that the unit occupied by the corresponding radar echo image area forms a measuring cluster Z= { Z 1 ,z 2 ,…,z n },z i ={x i ,y i X, where x i 、y i Respectively the abscissa and ordinate of the measuring points, n is the number of measurements in the measuring cluster, and the initial parameter a of the variable X is calculated according to the point information of the measuring cluster 0 、b 0 And theta 0 The ellipse mode of the graph is calculated:
;
in addition, the initial time measurement rate variableWhere m is the central moment of the graphic region, the relevant graphic moment parameter calculation is given by:
;
;
;
s24, initial variable of target motion stateThe relevant initial values are set as follows:
wherein the method comprises the steps ofAnd->The abscissa of the center point of the fitting ellipse for the object, respectively, has been given in step S23, velocity variable +.>State covariance matrix->For adjustable parameters, here +>The adjustment can be performed on the basis:
,/>。
2. the maritime radar target tracking method based on the expanded target cluster division as claimed in claim 1, wherein the step S3 specifically includes the steps of:
s31, assuming m targets at the moment k-1, when the tracking process reaches the moment k, firstly predicting the motion and the expansion state of all targets at the previous moment in one step, and updating the process formula as follows:
;
;
;
;
according to the updated state variables, for the ith target, a target prediction center and a minimum circumscribed rectangle tBOX are obtained i ;
S32, for the measuring point set detected at the moment k, dividing the measuring point set into n measuring clusters by the method in the step S1, and extracting the center of each measuring cluster And circumscribe rectangle->Consider the correlation of the ith target with the jth cluster of measurements:
(1) The distance between the target and the cluster of measurements is calculated and can be expressed as:
;
wherein,the residual error of the Kalman filter between the target and the center of the measurement cluster,covariance of the residual;
(2) Calculating the intersection ratio between the target prediction result and the measurement cluster can be expressed as:
;
s33, calculating an association matrix A according to the relation between the target and the measurement, wherein the association matrix A can be expressed as the following form:
;
the cluster of measurements that can be associated with a target should satisfy the following conditions: the distance from the matrix element is smaller than a threshold value, and the intersection condition of the circumscribed rectangle is larger than the intersection ratio threshold value, the matrix element is calculated by the following way:
;
wherein,and->For the correlation of the target and the cluster, the comparison and distance discrimination threshold value, +.>,/>The following formula is adopted for calculation:
。
3. the maritime radar target tracking method based on the expanded target cluster division as claimed in claim 2, wherein the step S4 specifically includes the steps of:
s41, analyzing the association matrix A and the association condition of the target and the measurement cluster, wherein the association condition of the expanded target data is divided into the following categories:
a. the target prediction result and the detection result can be well matched, and the center distance of the target prediction result and the detection result is relatively close, so that the overlapping rate of the external rectangular frame is high;
b. The target prediction result has no detection result which can be matched with the target prediction result within a certain range, and the detection result is represented as missed detection;
c. the detection result has no target prediction result which can be matched with the detection result in a certain range, and the detection result is expressed as a new target or clutter;
d. the multiple target prediction results and the same detection result have certain matching degree, and the fact that the multiple targets are mistakenly detected as one target in the detection step is the target association condition of the tracking problem aimed at in the text;
s42, analyzing each target, namely traversing the rows of the incidence matrix; when an element is=1 and corresponding ++>When the row minimum value is the row minimum value, recording the association condition of the jth measuring cluster and the ith target, and setting all other elements of the row of the matrix A to be 0; when the row elements are all 0, recording the association classification b, and recording the corresponding target serial numbers;
s43, analyzing each measurement cluster, namely traversing the columns of the incidence matrix; accumulating the values of each column of the correlation matrix by taking the number of targets associated with a single measurement cluster as a reference, wherein the number of targets is more than 2 and can be regarded as the situation of adhesion to the target d, and the number of targets is 1 and is classified as the situation a , And when the target number is 0, classifying the target number into the case c, and finally recording the corresponding measuring cluster and the target sequence number.
4. The maritime radar target tracking method based on the expanded target cluster division as claimed in claim 1, wherein the step S5 specifically comprises the steps of:
s51, for targets and measurement clusters with adhesion, the target prediction state is quantized into probability density distribution prediction of measurement points, and for a plurality of targets in adhesion, the following Gaussian Mixture Model (GMM) can be used for representing:
;
wherein the method comprises the steps ofFor point set data in a measurement cluster to be classified, n k The number of Gaussian distribution or the number of targets which are mixed corresponds to the number of measurement classification categories; />For the mixing coefficient, represent the probability of selecting the measurement division i-th class, corresponding to the target measurement point number +.>A predicted value; />Respectively corresponding to the central position and the expansion state in each target state x>Is a one-step predictor of (a);
s52, according to the model in the step S51, calculating the probability that each measuring point in the measurement cluster corresponding to the adhesion target belongs to a certain target, wherein the probability can be calculated by the following formula:
;
wherein,(/>{1,2,…,n k }) is a variable representing the division of the measurement into a certain target distribution,/>Representing individual measurement points in a measurement cluster;
solving the model by using an EM algorithm, and usingIndicating measurement->Probability belonging to the ith gaussian distribution In the parameter iterative solving process, the +.>、/>And->The updated formula and log likelihood function of (2) are as follows:
;
;
;
;
wherein,for measuring set->The number of medium measurement points;
s53, calculating and iterating the parameter updating process in the step S52 until the convergence of the log-likelihood function reaches a certain precision; finally, the probability of dividing each measurement sample into a certain distribution or a conditional probability matrix Div belonging to a certain target can be obtained, wherein the probability is shown as the following formula:
;
for each row of the conditional probability matrix Div, a maximum term for each row is calculated, which corresponds to the final division of the measurement, as shown in the following equation:
;
after the measuring points in the measuring clusters corresponding to all the merging targets are divided, new measuring clusters with the same number as the targets are formed according to the categories, and the new measuring clusters are matched with the corresponding targets.
5. The maritime radar target tracking method based on the expanded target cluster division as claimed in claim 1, wherein the step S6 specifically includes the steps of:
s61, for the targets which are not associated with the measurement cluster, determining whether the track is terminated according to the total number of times which is not associated with the tracking process, wherein the number of times is designed to be 3, and the adjustment can be performed on the basis; initializing the measurement clusters which are not associated with the target in the step S2 to serve as candidate tracking targets, and performing track initiation after the target can be associated with the measurement cluster at the next moment, otherwise, deleting the candidate targets;
S62, for the matched target and the measuring cluster, updating the target state by using the information of the measuring points in the measuring cluster, wherein the updating formula is as follows:
;
;
;
;
;
;
;
;
;
;
。
6. an extended target cluster division-based maritime radar target tracking system applying the extended target cluster division-based maritime radar target tracking method of any one of claims 1 to 5, the extended target cluster division-based maritime radar target tracking system comprising:
the measuring cluster conversion module is used for converting measuring points which are adjacent to each other into measuring clusters by utilizing a 4-neighborhood communication analysis mode for the measuring points detected by the radar before tracking;
the state initialization module is used for constructing a radar expansion target model based on a random matrix method when the moment k=1, and initializing the state of the target measurement set according to an initial measurement cluster result;
the correlation matrix calculation module is used for further predicting the target state of the previous moment when the moment k is more than 1, and calculating a correlation matrix according to the predicting result and the relation between the current moment measuring cluster position, the shape and the distance;
the tracking state distinguishing module is used for judging the matching relation between the target and the measurement according to the incidence matrix and distinguishing the tracking state of the target according to the existence of corresponding matching measurement clusters, the target and the corresponding matching quantity;
The target matching module is used for quantifying the target prediction state into probability density distribution prediction of the measuring points, dividing the measuring points in the measuring clusters according to the probability density distribution prediction of the measuring points, and enabling the measuring points to form measuring clusters with corresponding target numbers and to be matched with the targets;
the track updating module is used for starting or stopping the target track for the measurement clusters or targets which are not associated;
and the state updating module is used for updating the state of the target by using the information of the measuring points in the measuring cluster for the matched target and the measuring cluster.
7. An information data processing terminal for implementing the maritime radar target tracking system based on the expanded target cluster division according to claim 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311318046.3A CN117075097B (en) | 2023-10-12 | 2023-10-12 | Maritime radar target tracking method and system based on expanded target cluster division |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202311318046.3A CN117075097B (en) | 2023-10-12 | 2023-10-12 | Maritime radar target tracking method and system based on expanded target cluster division |
Publications (2)
Publication Number | Publication Date |
---|---|
CN117075097A CN117075097A (en) | 2023-11-17 |
CN117075097B true CN117075097B (en) | 2023-12-22 |
Family
ID=88717342
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202311318046.3A Active CN117075097B (en) | 2023-10-12 | 2023-10-12 | Maritime radar target tracking method and system based on expanded target cluster division |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN117075097B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117665795A (en) * | 2023-12-07 | 2024-03-08 | 兰州理工大学 | Poisson Bernoulli extended target tracking method based on track set |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107831490A (en) * | 2017-12-01 | 2018-03-23 | 南京理工大学 | A kind of improved more extension method for tracking target |
CN108427112A (en) * | 2018-01-22 | 2018-08-21 | 南京理工大学 | A kind of improved more extension method for tracking target |
CN114002667A (en) * | 2021-10-29 | 2022-02-01 | 西安交通大学 | Multi-neighbor extended target tracking algorithm based on random matrix method |
CN115575935A (en) * | 2022-10-18 | 2023-01-06 | 西安交通大学 | Partial shielding extended target tracking method and device based on random matrix method |
CN115792895A (en) * | 2022-12-16 | 2023-03-14 | 浙江飞碟汽车制造有限公司 | Millimeter wave radar target tracking method combining UKF and random matrix |
CN116047494A (en) * | 2022-12-29 | 2023-05-02 | 西安电子科技大学 | Tracking method of multi-expansion target based on multi-distance division under general sense integrated background |
CN116500575A (en) * | 2023-05-11 | 2023-07-28 | 兰州理工大学 | Extended target tracking method and device based on variable decibel leaf theory |
-
2023
- 2023-10-12 CN CN202311318046.3A patent/CN117075097B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107831490A (en) * | 2017-12-01 | 2018-03-23 | 南京理工大学 | A kind of improved more extension method for tracking target |
CN108427112A (en) * | 2018-01-22 | 2018-08-21 | 南京理工大学 | A kind of improved more extension method for tracking target |
CN114002667A (en) * | 2021-10-29 | 2022-02-01 | 西安交通大学 | Multi-neighbor extended target tracking algorithm based on random matrix method |
CN115575935A (en) * | 2022-10-18 | 2023-01-06 | 西安交通大学 | Partial shielding extended target tracking method and device based on random matrix method |
CN115792895A (en) * | 2022-12-16 | 2023-03-14 | 浙江飞碟汽车制造有限公司 | Millimeter wave radar target tracking method combining UKF and random matrix |
CN116047494A (en) * | 2022-12-29 | 2023-05-02 | 西安电子科技大学 | Tracking method of multi-expansion target based on multi-distance division under general sense integrated background |
CN116500575A (en) * | 2023-05-11 | 2023-07-28 | 兰州理工大学 | Extended target tracking method and device based on variable decibel leaf theory |
Also Published As
Publication number | Publication date |
---|---|
CN117075097A (en) | 2023-11-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113866742B (en) | Method for point cloud processing and target classification of 4D millimeter wave radar | |
CN117075097B (en) | Maritime radar target tracking method and system based on expanded target cluster division | |
US11530915B2 (en) | Dimension measuring device, dimension measuring method, and semiconductor manufacturing system | |
CN108921873A (en) | The online multi-object tracking method of Markovian decision of filtering optimization is closed based on nuclear phase | |
CN104199022B (en) | Target modal estimation based near-space hypersonic velocity target tracking method | |
CN111476827A (en) | Target tracking method, system, electronic device and storage medium | |
CN110940971B (en) | Radar target point trace recording method and device and storage medium | |
CN110349188B (en) | Multi-target tracking method, device and storage medium based on TSK fuzzy model | |
CN113486961A (en) | Radar RD image target detection method and system based on deep learning under low signal-to-noise ratio and computer equipment | |
CN113012200B (en) | Method and device for positioning moving object, electronic equipment and storage medium | |
CN112634368A (en) | Method and device for generating space and OR graph model of scene target and electronic equipment | |
CN112130142A (en) | Micro Doppler feature extraction method and system for complex moving target | |
Liu et al. | Plant cell tracking using Kalman filter based local graph matching | |
Liu et al. | High-precision detection method for structure parameters of catenary cantilever devices using 3-D point cloud data | |
CN115082444B (en) | Copper pipe weld defect detection method and system based on image processing | |
CN111259332B (en) | Fuzzy data association method and multi-target tracking method in clutter environment | |
CN112966553A (en) | Strong coupling target tracking method, device, medium and equipment based on twin network | |
CN115546705A (en) | Target identification method, terminal device and storage medium | |
CN111738319A (en) | Clustering result evaluation method and device based on large-scale samples | |
CN112748735B (en) | Extended target tracking method introducing color features | |
CN111274529B (en) | Robust Gao Sini Weisal PHD multi-expansion target tracking algorithm | |
CN110059591B (en) | Method for identifying moving target area | |
CN112614158B (en) | Sampling frame self-adaptive multi-feature fusion online target tracking method | |
CN112861708B (en) | Semantic segmentation method and device for radar image and storage medium | |
CN115220002A (en) | Multi-target data association tracking method and related device for fixed single station |
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 |