CN-121978675-A - Multi-target tracking method, system, equipment and medium based on compressed sensing and particle filtering

CN121978675ACN 121978675 ACN121978675 ACN 121978675ACN-121978675-A

Abstract

The invention discloses a multi-target tracking method, a system and a medium based on compressed sensing and particle filtering, belonging to the field of signal monitoring and tracking; the method comprises the steps of initializing a group of independent particle swarms for each target to be tracked, executing time recursion tracking on the particle swarm of each target to be tracked according to the low-dimensional compressed observation signals, jointly calculating the weight of each particle in each particle swarm to obtain a weighted particle set of each target to be tracked, carrying out state estimation according to the weighted particle set of each target to be tracked to obtain an angle estimated value and an angular velocity estimated value of each target to be tracked, and outputting a tracking result of each target to be tracked.

Inventors

GAO QIAN
HUANG JIALIN
CUI XINYUE

Assignees

海南热带海洋学院

Dates

Publication Date: 20260505
Application Date: 20260202

Claims (10)

1. A multi-target tracking method based on compressed sensing and particle filtering, comprising: Performing compression observation on the received original array signals to obtain low-dimensional compression observation signals; initializing a group of independent particle swarms for each target to be tracked, wherein each particle in the particle swarm comprises an angle state variable and an angular velocity state variable of the target to be tracked; According to the low-dimensional compressed observation signals, respectively performing time recursion tracking on particle swarms of each target to be tracked, and jointly calculating the weight of each particle in each particle swarm to obtain a weighted particle set of each target to be tracked; and carrying out state estimation according to the weighted particle set of each target to be tracked, obtaining an angle estimation value and an angular velocity estimation value of each target to be tracked, and outputting a tracking result of each target to be tracked.
2. The multi-target tracking method based on compressed sensing and particle filtering according to claim 1, wherein the performing time recursive tracking on the particle swarm of each target to be tracked based on the low-dimensional compressed observed signal, respectively, comprises: dynamically adjusting an initial process noise variance in a state prediction process according to errors between a historical state estimated value and a priori value of each target to be tracked, and obtaining a first process noise variance; and respectively carrying out state prediction on each particle in the particle swarm of each target to be tracked according to the first process noise variance, and applying physical range constraint to obtain a constrained predicted particle state.
3. The multi-target tracking method based on compressed sensing and particle filtering according to claim 2, wherein the step of jointly calculating the weight of each particle in each particle group to obtain a weighted particle set of each target to be tracked is specifically as follows: According to the predicted particle state, particles with the same serial number are respectively extracted from each particle swarm of the target to be tracked, an extracted particle swarm is obtained, and a joint guide matrix is constructed according to the angular velocity state variables of the particles in the extracted particle swarm; estimating signal amplitude according to the low-dimensional compressed observation signal, the joint guide matrix and a preset compressed observation matrix; And calculating an observation residual according to the signal amplitude, calculating a combined basic weight of the extracted particle group by combining the predicted particle state with a signal energy penalty term, and determining a corresponding particle weight of each target to be tracked according to the combined basic weight to obtain a weighted particle set of each target to be tracked.
4. The method for multi-target tracking based on compressed sensing and particle filtering of claim 3, wherein determining the weight of the corresponding particle in each target particle group to be tracked according to the joint base weight further comprises: calculating a speed penalty factor according to the predicted particle state through a speed penalty term formula; Correcting the combined basic weight through the speed penalty factor to determine the corresponding particle weight of each target to be tracked; Wherein the speed penalty term formula is as follows: Wherein, the For the speed penalty factor, Is a priori angular velocity range; The angular velocity state variable of the ith particle of the kth target to be tracked in the predicted particle state at the moment t is used as K, wherein K is the number of the targets to be tracked; is a speed penalty factor.
5. The method for multi-target tracking based on compressed sensing and particle filtering according to claim 1, wherein the state estimation is performed according to a weighted particle set of each target to be tracked, an angle estimation value and an angular velocity estimation value of each target to be tracked are obtained, and a tracking result of each target to be tracked is output, specifically: according to the weighted particles, mapping the value of the angle state variable in the weighted particles to a complex unit circle for weighted average by adopting a complex average method to obtain a complex result, and reversely calculating the complex result as an angle value to obtain an angle estimated value of a target to be tracked; according to the weighted particles, carrying out weighted average on the values of the angular velocity state variables in the weighted particles to obtain a weighted average result, and carrying out fine adjustment on the weighted average result through a compensation mechanism to obtain an angular velocity estimated value of a target to be tracked; And outputting a tracking result of each target to be tracked according to the angle estimation value and the angular speed estimation value.
6. The compressed sensing and particle filtering based multi-target tracking method of claim 1, further comprising intelligent resampling after obtaining a weighted particle set for each target to be tracked: according to the weighted particle set of each target to be tracked, a system resampling method is adopted to generate a new particle set of each target to be tracked; and randomly selecting part of new particles in the new particle set, and applying differential disturbance for maintaining particle diversity to state variables of the new particles.
7. The multi-target tracking method based on compressed sensing and particle filtering according to claim 1, wherein the compressed observing is performed on the received original array signal to obtain a low-dimensional compressed observed signal, specifically: Constructing an original received signal vector according to the received original array signal, wherein the original received signal vector represents superposition of signals of a plurality of targets to be tracked and noise; And carrying out linear dimension reduction projection on the original received signal vector through a preset compressed observation matrix to obtain a low-dimension compressed observation signal, wherein the low-dimension compressed observation signal comprises a compressed observation signal vector, and the number of vector lines of the compressed observation signal is far smaller than the number of array elements of the array.
8. The multi-target tracking system based on compressed sensing and particle filtering is characterized by comprising a compressed observation module, an initialized particle module, a calculation weight module and a state estimation module; The compression observation module is used for carrying out compression observation on the received original array signals to obtain low-dimensional compression observation signals; The particle initializing module is used for initializing a group of independent particle groups for each target to be tracked respectively, wherein each particle in the particle groups comprises an angle state variable and an angular speed state variable of the target to be tracked; the calculation weight module is used for respectively executing time recursion tracking on the particle swarm of each target to be tracked according to the low-dimensional compression observation signal, and jointly calculating the weight of each particle in each particle swarm to obtain a weighted particle set of each target to be tracked; The state estimation module is used for carrying out state estimation according to the weighted particle set of each target to be tracked, obtaining an angle estimation value and an angular velocity estimation value of each target to be tracked, and outputting a tracking result of each target to be tracked.
9. A terminal device comprising a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the compressed sensing and particle filtering based multi-objective tracking method according to any of claims 1-7 when the computer program is executed.
10. A computer readable storage medium comprising a stored computer program, wherein the computer program, when run, controls a device in which the computer readable storage medium is located to perform a compressed sensing and particle filtering based multi-objective tracking method according to any one of claims 1-7.

Description

Multi-target tracking method, system, equipment and medium based on compressed sensing and particle filtering Technical Field The invention relates to the field of signal monitoring and tracking, in particular to a multi-target tracking method, a system, equipment and a medium based on compressed sensing and particle filtering. Background Currently, the multi-target angle tracking technology mainly depends on methods based on array signal processing such as traditional beam forming, MUSIC algorithm, ESPRIT algorithm and the like. These methods typically require a complete array to receive the data, have high computational complexity, and require high signal-to-noise ratios and target separation angles. In recent years, the compressed sensing technology plays an important role in angle estimation, the compressed sensing can alleviate the requirements on high signal-to-noise ratio and multiple snapshots, and the hardware cost and the calculation burden are reduced by reducing the sampling dimension, but the problems of insufficient tracking capability of a dynamic target, inaccurate speed estimation, non-robust particle filter initialization and the like still exist. The existing multi-target tracking method has the following defects that firstly, high-dimensional data dependence is needed, a signal covariance matrix is built on the basis of complete array received data in the traditional method, hardware cost is high, calculation complexity is high, secondly, dynamic tracking capacity is weak, the existing compressed sensing method is mainly based on a static sparse signal model to design an observation matrix, dynamic constraint conditions of a target motion state are not integrated, the estimation of an angle change rate (speed) is inaccurate, thirdly, particle filtering initialization is inaccurate, generation of initial particle sets is mainly dependent on empirical prior distribution, gaussian distribution and the like, target motion prior information is not combined, convergence is slow, tracking deviation is large, four targets interfere with each other, angle confusion or tracking loss is easily caused, and fifthly, a joint estimation mechanism is lacked, the prior art generally splits angle estimation and speed estimation into independent processing modules, and collaborative optimization is lacked. Disclosure of Invention The invention provides a multi-target tracking method, a system, equipment and a medium based on compressed sensing and particle filtering, which reduce data dimension through compressed observation and realize real-time joint tracking of angles and speeds of a plurality of targets to be tracked. The invention provides a multi-target tracking method based on compressed sensing and particle filtering, which comprises the following steps: Performing compression observation on the received original array signals to obtain low-dimensional compression observation signals; initializing a group of independent particle swarms for each target to be tracked, wherein each particle in the particle swarm comprises an angle state variable and an angular velocity state variable of the target to be tracked; According to the low-dimensional compressed observation signals, respectively performing time recursion tracking on particle swarms of each target to be tracked, and jointly calculating the weight of each particle in each particle swarm to obtain a weighted particle set of each target to be tracked; and carrying out state estimation according to the weighted particle set of each target to be tracked, obtaining an angle estimation value and an angular velocity estimation value of each target to be tracked, and outputting a tracking result of each target to be tracked. The method comprises the steps of obtaining low-dimensional compression observation signals through compression observation on received original array signals, completing low-dimensional compression of data quantity in a signal acquisition stage, providing a low-dimensional data base for subsequent data real-time processing, respectively initializing a group of independent particle swarms for each target to be tracked, establishing a multi-target-oriented distributed state representation framework by means of angle and angular velocity state variables of each particle, reducing calculation complexity from an exponential level to a linear level, providing a state carrier for subsequent joint estimation of angles and velocities, performing time recursive tracking on the particle swarms according to the low-dimensional compression observation signals, adopting a joint processing mechanism in a weight calculation link, remarkably improving accuracy of state estimation, obtaining an angle estimated value and an angular velocity estimated value of each target to be tracked through state estimation, and outputting a tracking result of each target to be tracked, and reducing the dimensional burden of data processing through compression ob