A motorized target tracking method, system, electronic device and storage medium
By using the NVSIMM algorithm to compute the likelihood function and probability of multiple model subsets in parallel, and selecting the subset with the highest probability as the tracking result, the problem of low tracking accuracy of maneuvering targets in existing technologies is solved, and higher tracking accuracy is achieved.
Patent Information
- Application Number
- CN202211656412.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-22
- Publication Date
- 2026-01-02
- Estimated Expiration
- 2042-12-22
AI Technical Summary
Existing maneuvering target tracking methods suffer from low accuracy when the model is mismatched, especially the variable structure IMM algorithm, which suffers from reduced tracking accuracy when calculating the likelihood function and model subset probabilities.
A novel Variable Structure Interactive Multiple Model (NVSIMM) algorithm is designed. By running multiple model subsets in parallel, calculating the likelihood function and probability, and selecting the model subset with the highest probability as the final tracking result, the tracking accuracy is improved.
Without increasing the number of models, the tracking accuracy of maneuvering targets is significantly improved, especially maintaining high accuracy when the maneuvering mode changes.
Smart Images
Figure CN115880335B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the field of maneuvering target tracking, and more particularly, relates to a maneuvering target tracking method, system, electronic device and storage medium. BACKGROUND
[0002] At present, Kalman filter and its improved forms have been widely applied in maneuvering target tracking problems. However, the maneuvering target will change the maneuvering model in the movement process, and when the filter model does not match the target maneuvering mode, the tracking accuracy will be greatly reduced. In view of this problem, the interacting multiple model (IMM) shows more superior performance. Therefore, the IMM algorithm has attracted the attention of many scholars, and the IMM algorithm has been improved in structure and parameters.
[0003] The prior art provides an IMM algorithm based on a uniform speed model and a current statistical model, wherein the average speed of the current statistical model is first estimated by the least square method, and then used in the IMM algorithm, which can improve the model accuracy, thereby improving the maneuvering target tracking accuracy. The prior art improves the IMM algorithm in many aspects for tracking of maneuvering targets, including using an improved Kalman filter as a sub-filter, asymmetric state estimation between different models, and a model probability formula based on entropy. The prior art proposes an alternative method of IMM, wherein the models in the model set are all composed of uniform acceleration models, which can reduce the complexity of the model set. On this basis, the prior art proposes an adaptive IMM algorithm, which first estimates the acceleration of the maneuvering target using a filter, then constructs a model set according to the estimated acceleration, and finally estimates using the IMM algorithm, which reduces the number of models in the model set. The prior art proposes a second-order IMM algorithm based on a second-order Markov chain, which further improves the maneuvering target tracking accuracy due to more use of prior information. The above IMM is also called fixed structure IMM (FSIMM) because the model set model used is unchanged.
[0004] In order to avoid the reduction of the tracking precision of the maneuvering target due to the model mismatch, when the IMM algorithm is used, as many models as possible should be used to cover the maneuvering modes of the target. However, it is worth noting that too many models in a single model set will also reduce the tracking precision. In this context, the variable structure IMM (VSIMM) algorithm emerges as the times require. The VSIMM is also various, and can be roughly divided into four types: model group switching (MGS), likely mode set (LMS), expected mode augmentation (EMA) and adaptive grid (AG). Among them, the MGS divides the model set into a model subset, and only one model subset is selected for the maneuvering target trajectory estimation at each moment in the process of the maneuvering target tracking, so that the model subset will be switched constantly, and the switching between the model subsets is selected according to the model subset transition probability.
[0005] The LMS finds a model subset from a large model set for the maneuvering target tracking, and all the models are divided into three types at each moment: impossible, important and main, and the model subset for the trajectory estimation at each moment is composed of the main and the near-main models. Similar to the MGS, the EMA divides a large model set into small model subsets, and then calculates the probability of all the model subsets at the next moment, and selects the model subset with the maximum probability for the maneuvering target tracking. The AG combines the graph theory, and all the models constitute a grid, a local refined grid is obtained by using the prior information and the current data, the models in the refined grid constitute the candidate model subset, and then the model subset for the maneuvering target tracking at the next moment is selected according to some rules. It is worth noting that the methods mentioned in the prior art approximate the formula when calculating the likelihood sum and the model subset probability, so that the tracking precision of the VSIMM for the maneuvering target is reduced. SUMMARY
[0006] In view of the defects of the prior art, the purpose of the present application is to provide a maneuvering target tracking method, system, electronic equipment and storage medium, which aims to solve the problem of low precision of the existing maneuvering target tracking method.
[0007] To achieve the above purpose, in a first aspect, the present application provides a maneuvering target tracking method, comprising the following steps:
[0008] determining a motion model set of the maneuvering target; the motion model set comprises a plurality of model subsets, each model subset comprises a plurality of motion models, and each motion model represents a maneuvering mode of the maneuvering target;
[0009] An interactive multi-model (IMM) algorithm is run in parallel for each model subset to obtain a target tracking result of each model subset; the IMM algorithm is based on a motion model to denoise measurement noise, so that the tracking result of the maneuvering target is close to the actual motion state of the maneuvering target; due to the difference between the motion models in the model subset, the tracking accuracy of different model subsets is different;
[0010] A likelihood function of each model subset is calculated based on the target tracking result of each model subset; and a probability of each model subset is calculated based on the likelihood function of each model subset; the likelihood function reflects the tracking accuracy of the model subset;
[0011] The target tracking result of the model subset with the highest probability is selected as the tracking result of the maneuvering target.
[0012] In an optional example, the likelihood function of each model subset is calculated based on the target tracking result of each model subset, specifically as follows:
[0013]
[0014] wherein C k (n) represents the likelihood function of the nth model subset, represents the probability density of the nth model subset, and represents the prediction probability of the nth model subset.
[0015] In an optional example, the probability κ k (n) of each model subset is calculated based on the likelihood function of each model subset, specifically as follows:
[0016]
[0017] wherein the normalization factor p(z k |Z k - 1 ) and the model subset prediction probability p(Π k (n)|Z k-1 ) are respectively:
[0018]
[0019]
[0020] wherein κ k-1 (m)=p(Π k-1 (m)|Z k-1 ) is the probability of the mth model subset at the previous moment, p ij is the transition probability from the ith model subset to the jth model subset.
[0021] In a second aspect, the present application provides a motorized target tracking system, comprising:
[0022] a motion model set determining unit configured to determine a motion model set of the motorized target, wherein the motion model set comprises a plurality of model subsets, each model subset comprises a plurality of motion models, and each motion model represents a motorized mode of the motorized target;
[0023] a model subset parallel IMM unit configured to run an interactive multiple model (IMM) algorithm in parallel for each model subset to obtain a target tracking result of each model subset, wherein the IMM algorithm is based on the motion models to denoise measurement noise, so that the tracking result of the motorized target is close to the actual motion state of the motorized target, and the tracking accuracy of different model subsets is different due to the difference between the motion models in the model subsets;
[0024] a tracking accuracy determining unit configured to calculate a likelihood function of each model subset based on the target tracking result of each model subset, and calculate a probability of each model subset based on the likelihood function of each model subset, wherein the likelihood function reflects the tracking accuracy of the model subset;
[0025] a tracking result determining unit configured to select the target tracking result of the model subset with the highest probability as the tracking result of the motorized target.
[0026] In an optional example, the tracking accuracy determining unit calculates the likelihood function of each model subset based on the target tracking result of each model subset, specifically as follows:
[0027]
[0028] wherein C k (n) represents the likelihood function of the nth model subset, represents the probability density of the nth model subset, and represents the prediction probability of the nth model subset.
[0029] In an optional example, the tracking accuracy determining unit calculates the probability of each model subset based on the likelihood function of each model subset, specifically as follows: k
[0030]
[0031] wherein the normalization factor p(z k |Z k - 1 ) and the model subset prediction probability p(Π k (n)|Z k-1 ) are respectively:
[0032]
[0033]
[0034] wherein, κ k-1 (m) = p (Π k-1 (m) | Z k-1 ) is the probability of the mth model subset at the last time, p ij is the transition probability from the ith model subset to the jth model subset.
[0035] In a third aspect, the present application provides an electronic device, comprising: a memory and a processor; the memory is used for storing a computer program; the processor is used for realizing the motor target tracking method provided by the first aspect when the computer program is executed.
[0036] In a fourth aspect, the present application provides a computer readable storage medium, and the storage medium stores a computer program, and when the computer program is executed by a processor, the motor target tracking method provided by the first aspect is realized.
[0037] Overall, compared with the prior art, the above technical solutions conceived by the present application have the following beneficial effects:
[0038] The present application provides a motor target tracking method, system, electronic device and storage medium, and a motor target tracking method based on a new variable structure interactive multiple model (NVSIMM) is designed. Compared with the existing EMA method and FSIMM method, the present application provides a more accurate calculation model, and the tracking accuracy of the motor target is higher. BRIEF DESCRIPTION OF DRAWINGS
[0039] Figure 1 is a motor target tracking method flowchart provided by the embodiment of the present application;
[0040] Figure 2 is an IMM algorithm block diagram provided by the embodiment of the present application;
[0041] Figure 3 is an NVSIMM algorithm block diagram provided by the embodiment of the present application;
[0042] Figure 4 is a motor target motion trajectory schematic diagram provided by the embodiment of the present application;
[0043] Figure 5 is a motor target tracking result schematic diagram provided by the embodiment of the present application;
[0044] Figure 6 is a motor target tracking system architecture diagram provided by the embodiment of the present application. DETAILED DESCRIPTION
[0045] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the following will be combined with the accompanying drawings for the embodiments of the present application to make a clear and complete description of the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only a part but not all of the embodiments of the present application. The following description of at least one exemplary embodiment is merely illustrative in nature and not intended to be limiting on the present application and its applications or uses. Based upon the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort are within the scope of protection of the present application.
[0046] In the description of the present application, the description of the terms “one embodiment”, “some embodiments”, “exemplary embodiment”, “example”, “specific example” or “some examples” means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the exemplary description of the above terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.
[0047] Firstly, the present application describes the problem of maneuvering target tracking and gives an interactive multiple model filtering method; then, on this basis, a new variable structure interactive multiple model method is designed, which can make the actual maneuvering model of the maneuvering target in the model subset most matched without increasing the number of models in the model subset, thereby improving the tracking precision of the maneuvering target.
[0048] Figure 1 is a flowchart of the maneuvering target tracking method provided by the embodiments of the present application; as shown in Figure 1 the figure, the method comprises the following steps:
[0049] S101, determining a set of motion models of a maneuvering target; the set of motion models comprises a plurality of model subsets, each model subset comprises a plurality of motion models, and one motion model represents one maneuvering mode of the maneuvering target;
[0050] S102, performing an interactive multiple model (IMM) algorithm on each model subset in parallel to obtain a target tracking result of each model subset; the IMM algorithm is based on the motion model to denoise the measurement noise, so that the tracking result of the maneuvering target is close to the actual motion state of the maneuvering target; due to the differences between the motion models in the model subset, the tracking precisions of different model subsets are different;
[0051] S103, calculate the likelihood function of each model subset based on the target tracking result of each model subset; and calculate the probability of each model subset based on the likelihood function of each model subset; the likelihood function reflects the tracking accuracy of the model subset;
[0052] S104, select the target tracking result of the model subset with the highest probability as the tracking result of the maneuvering target.
[0053] Specifically, the interactive multiple model algorithm designed by the present application is as follows:
[0054] 1.1 Interactive multiple model algorithm
[0055] 1.1.1 Problem description
[0056] Suppose that the maneuvering target may have r motion models Ω=M 1 ,...,M r , and the transition matrix between the models is:
[0057]
[0058] Where p ij (1≤i≤r,1≤j≤r) is the transition probability from model M i to model M j .
[0059] The state equation of the system discretization is as follows:
[0060]
[0061] Where x is the state vector of the system, F j is the state transition matrix of model M j , is a Gaussian white noise with mean 0, and the covariance matrix is Q j .
[0062] The observation equation of model M j is:
[0063]
[0064] Where z k is the measurement vector, H j is the measurement matrix of model M j , is a Gaussian white noise with mean 0, and the covariance matrix is R j .
[0065] The IMM algorithm is then based on the observation measurement set Z k ={z1,...,z kand model set Ω, the state xk at time k is estimated based on Bayesian theory.
[0066] 1.1.2 IMM algorithm
[0067] The IMM algorithm block diagram is shown in Figure 2 , which can be generally divided into the following four steps:
[0068] Step 1: input interaction (solving )
[0069]
[0070]
[0071] In the above formula, and are the mixed maneuvering target tracking state estimation and the corresponding covariance matrix of model M j , and are the Kalman estimation and the corresponding covariance matrix of model M i , is the mixed probability, and its calculation formula is:
[0072]
[0073] Wherein, is the probability of model M i , the prediction probability is the normalization factor.
[0074] Step 2: filtering (solving )
[0075] Kalman filtering is performed on each model respectively:
[0076] 1) Predicting state
[0077]
[0078] 2) Predicting covariance matrix
[0079]
[0080] 3) Kalman gain
[0081]
[0082] 4) Filtering (weighting value)
[0083]
[0084] 5) Covariance of filtering value
[0085]
[0086] Step3: Model probability update
[0087] The model probability update formula is:
[0088]
[0089] Wherein, the denominator p(z k |Z k-1 ) is a normalization factor, and the integral of the numerator can be obtained, In Step 1, it has been calculated that obeys a Gaussian distribution, and the specific form is:
[0090]
[0091] Wherein, is the mean, is the variance, The probability density of is:
[0092]
[0093] Wherein,
[0094] Step4: Output interaction
[0095] The final maneuvering target tracking state estimation and covariance estimation:
[0096]
[0097]
[0098] 1.2 Variable structure interactive multiple model algorithm
[0099] In order to avoid too many models in the model set and improve the maneuvering target tracking precision, the NVSIMM algorithm is designed. The NVSIMM algorithm includes multiple model subsets, each model subset is calculated independently in parallel, and the estimation result of the model subset with the highest probability is selected as the final maneuvering target tracking state estimation result. The block diagram of the NVSIMM algorithm is as shown in Figure 3 The following steps are included:
[0100] 1.2.1: Parallel independent IMM state estimation
[0101] For different model subsets, the IMM algorithm is used respectively to obtain the prediction probability of each model Target tracking state estimation value The corresponding covariance P k|k(n), and the probability of the model in each model subset and the likelihood function Here n is the index of the model subset, 1≤n≤N, and i is the index of the model in the model subset.
[0102] 1.2.2: Calculate the likelihood function of each model subset
[0103] The likelihood function of the model subset is:
[0104]
[0105] where Π k (n) represents the nth model subset, calculated by Step 3 of the above IMM, calculated by Step 1 of the IMM. Let p(z k |Z k-1 ,Π k (n)) be C k (n), then:
[0106]
[0107] 1.2.3: Calculate the model subset probability
[0108] Let the model subset probability κ k (n) be:
[0109]
[0110] where p(z k |Z k-1 ,Π k (n)) has been calculated in 1.2.2, and the denominator is the normalization factor, and p(Π k (n)|Z k-1 ) has the following formula according to the total probability formula:
[0111]
[0112] where κ k-1 (m) = p(Π k-1 (m)|Z k-1 ) is the model subset probability at the previous time, and the calculation formula of p(Π k (n)|Z k-1 ) is:
[0113]
[0114] 1.2.4: State estimation and corresponding covariance of maneuvering target tracking
[0115] The IMM corresponding to the highest model subset probability is selected as the final output of the maneuvering target tracking. First, the number of the highest model subset is calculated:
[0116]
[0117] Thus, the final maneuvering target tracking state estimation and the corresponding covariance of the filter can be obtained as:
[0118]
[0119]
[0120] Specifically, the target tracking can be any state data of a target, including position, velocity, acceleration, etc. The data measured by a sensor often contains a large amount of noise, and the purpose of designing a tracking algorithm is to reduce the noise and improve the accuracy of the data. That is, the input of the algorithm is the measurement data with noise (which can be position data, velocity data or other data, but the accuracy of the measurement noise is low), and the output of the algorithm is data with high accuracy.
[0121] To verify the effectiveness of the NVSIMM for maneuvering target tracking, simulation verification is carried out as follows. It is assumed that the maneuvering target moves in a two-dimensional plane, and the initial position, velocity and acceleration are (10 km, 40 km), (300 m / s, 0) and (0, 0), respectively.
[0122] The maneuvering target first moves at a constant speed in a straight line for 30 seconds, then moves at an acceleration of (-10 m / s 2 , -10 m / s 2 ) for 30 seconds, then moves at a constant speed for 38 seconds, and the angular velocity rate is 0.2, and finally moves at a constant speed in a straight line for 30 seconds. The real motion trajectory of the maneuvering target is shown in Figure 4 .
[0123] It is assumed that the standard deviation of the system process noise is 1 m / s 2 , and the standard deviation of the measurement noise is 50 m. The NVSIMM and the FSIMM are used to estimate the trajectory of the maneuvering target, respectively. The available models include constant velocity (CV), constant acceleration (CA) and constant turn (CT), and the transition probabilities of the three models are:
[0124]
[0125] In the FSIMM algorithm, the model subsets include CV, VA and CT models. The NVSIMM includes two model subsets, the first subset includes CV and VA models, and the second subset includes CV and CT models. The simulation results are shown in Figure 5 .
[0126] Figure 5 The true value, the measured value, and the estimated values of the FSIMM and the NVSIMM of the trajectory of the maneuvering target are shown in (a) in the drawings. Figure 5 It can be seen from (a) that the FSIMM and the NVSIMM can better track the trajectory of the maneuvering target. Figure 5 The model probability of the FSIMM is shown in (b) in the drawings, and the model with the maximum probability is consistent with the actual maneuvering model of the target. Figure 5 The probability changes of the model subsets in the NVSIMM are shown in (c) in the drawings, and the models covered in the model subsets can be consistent with the actual maneuvering model of the target. Figure 5 The mean square errors of the position estimation of the maneuvering target by the FSIMM and the NVSIMM are shown in (d) in the drawings. In most time, the mean square error of the NVSIMM is smaller than that of the FSIMM. Only when the maneuvering mode of the target changes, the mean square error of the FSIMM is temporarily smaller than that of the NVSIMM, but it can be recovered immediately. Overall, the tracking accuracy of the NVSIMM for the maneuvering target is higher.
[0127] Figure 6 The figure is a schematic diagram of a maneuvering target tracking system provided by an embodiment of the present application, as shown in the figure, and includes: Figure 6
[0128] The motion model set determination unit 610 is configured to determine a motion model set of the maneuvering target. The motion model set includes a plurality of model subsets, and each model subset includes a plurality of motion models. One motion model represents one maneuvering mode of the maneuvering target.
[0129] The model subset parallel IMM unit 620 is configured to run an interactive multiple model (IMM) algorithm in parallel for each model subset to obtain a target tracking result of each model subset. The IMM algorithm is based on the motion models to denoise the measurement noise, so that the tracking result of the maneuvering target is close to the actual motion state of the maneuvering target. Due to the differences between the motion models in the model subsets, the tracking accuracies of different model subsets are different.
[0130] The tracking accuracy determination unit 630 is configured to calculate a likelihood function of each model subset based on the target tracking result of each model subset, and calculate a probability of each model subset based on the likelihood function of each model subset. The likelihood function reflects the tracking accuracy of the model subset.
[0131] The tracking result determination unit 640 is configured to select the target tracking result of the model subset with the highest probability as the tracking result of the maneuvering target.
[0132] It can be understood that the detailed function implementation of each unit described above can refer to the description in the foregoing method embodiments, which will not be repeated here.
[0133] In addition, the embodiment of the present application provides an electronic device, comprising a memory and a processor.
[0134] The memory is configured to store a computer program.
[0135] The processor is configured to implement the method in the foregoing embodiments when executing the computer program.
[0136] In addition, the present application also provides a computer readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the method in the foregoing embodiments is implemented.
[0137] Based on the method in the foregoing embodiments, the embodiment of the present application provides a computer program product, which, when running on a processor, causes the processor to execute the method in the foregoing embodiments.
[0138] Based on the method in the foregoing embodiments, the embodiment of the present application also provides a chip, comprising one or more processors and an interface circuit. Optionally, the chip can also include a bus. Wherein:
[0139] The processor can be an integrated circuit chip having a processing capability of signals. In the implementation process, each step of the foregoing method can be completed by an integrated logic circuit or an instruction in the form of software in the processor. The processor described above can be a general processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. Each method, step disclosed in the embodiment of the present application can be implemented or executed. The general processor can be a microprocessor or any conventional processor.
[0140] The interface circuit can be used for sending or receiving data, instructions or information. The processor can process the data, instructions or other information received by the interface circuit, and can send the processed information out through the interface circuit.
[0141] Optionally, the chip further comprises a memory, which can include a read-only memory and a random access memory, and provides operation instructions and data for the processor. A part of the memory can also include a non-volatile random access memory (NVRAM).
[0142] Optionally, the memory stores executable software modules or data structures, and the processor can execute corresponding operations by calling operation instructions stored in the memory (the operation instructions can be stored in an operating system).
[0143] Optionally, the interface circuit can be used to output the execution result of the processor.
[0144] It should be noted that the functions of the processor and the interface circuit can be realized by hardware design, software design, or a combination of hardware and software, and the present application is not limited in this regard.
[0145] It should be understood that each step of the above method embodiments can be completed by a logic circuit in the form of hardware or instructions in the form of software in the processor.
[0146] It should be understood that the size of the serial number of each step in the above embodiments does not mean the execution order, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. In addition, in some possible implementations, each step in the above embodiments can be selectively executed, partially executed, or fully executed according to actual conditions, and the present application is not limited in this regard.
[0147] It should be understood that the processor in the embodiments of the present application can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor can be a microprocessor or any conventional processor.
[0148] The method steps in the embodiments of the present application can be implemented by hardware, or by a combination of software and hardware executed by a processor. The software instructions can be composed of a corresponding software module, which can be stored in a random access memory (RAM), a flash memory, a read-only memory (ROM), a programmable read-only memory (PROM), an erasable PROM (EPROM), an electrically EPROM (EEPROM), a register, a hard disk, a mobile hard disk, a CD-ROM, or any other form of storage medium well known in the art. An exemplary storage medium is coupled to the processor, so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and the storage medium can be located in an ASIC.
[0149] In the above embodiments, all or part of the embodiments can be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part of the embodiments can be implemented in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted by the computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wire (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that includes one or more available media sets. The available media can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk (SSD)), etc.
[0150] Those skilled in the art will easily understand that the above description is only a preferred embodiment of the present application, and is not intended to limit the present application. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A method of motorized target tracking, characterized in that, The method comprises the following steps: determining a motion model set of the maneuvering target; the motion model set comprises a plurality of model subsets, each model subset further comprises a plurality of motion models, and each motion model represents a maneuvering mode of the maneuvering target; running an interactive multiple model (IMM) algorithm on each model subset in parallel to obtain a target tracking result of each model subset; the IMM algorithm is based on the motion models to denoise measurement noise, so that the tracking result of the maneuvering target is close to the actual motion state of the maneuvering target; due to the differences between the motion models in the model subsets, the tracking accuracies of different model subsets are different; calculating a likelihood function of each model subset based on the target tracking result of each model subset, and calculating a probability of each model subset based on the likelihood function of each model subset; the likelihood function reflects the tracking accuracy of the model subset; selecting the target tracking result of the model subset with the highest probability as the tracking result of the maneuvering target; calculating the likelihood function of each model subset based on the target tracking result of each model subset, specifically comprising: wherein, represents the likelihood function of the n th model subset, represents the probability density of the n th model subset, represents the prediction probability of the n th model subset; calculating a probability of each model subset based on a likelihood function of each model subset , specifically: where the normalization factor and model subset prediction probabilities are respectively: wherein, is the probability of the last model subset, m , is the transition probability from the i j model subset to themodel subset. 2. A motorized target tracking system characterized by, including: a motion model set determining unit configured to determine a motion model set of the maneuvering target; the motion model set comprises a plurality of model subsets, each model subset further comprises a plurality of motion models, and each motion model represents a maneuvering mode of the maneuvering target; a model subset parallel IMM unit configured to run an interactive multiple model (IMM) algorithm on each model subset in parallel to obtain a target tracking result of each model subset; the IMM algorithm is based on the motion models to denoise measurement noise, so that the tracking result of the maneuvering target is close to the actual motion state of the maneuvering target; due to the differences between the motion models in the model subsets, the tracking accuracies of different model subsets are different; a tracking accuracy determining unit configured to calculate a likelihood function of each model subset based on the target tracking result of each model subset, and calculate a probability of each model subset based on the likelihood function of each model subset; the likelihood function reflects the tracking accuracy of the model subset; a tracking result determining unit configured to select the target tracking result of the model subset with the highest probability as the tracking result of the maneuvering target; the tracking accuracy determining unit calculates the likelihood function of each model subset based on the target tracking result of each model subset, specifically comprising: wherein, represents a likelihood function of the n th model subset, represents a probability density of the n th model subset, represents a prediction probability of the n th model subset. The tracking accuracy determination unit calculates the probability of each model subset based on the likelihood function of each model subset , and specifically: where the normalization factor and model subset prediction probabilities are respectively: wherein, is the probability of the last time step, m , is the transition probability from the i j model subset to themodel subset. 3. An electronic device, comprising: including: a memory and a processor; the memory is configured to store a computer program; the processor is configured to implement the maneuvering target tracking method of claim 1 when executing the computer program.
4. A computer-readable storage medium, characterized in that, The storage medium has a computer program stored thereon, and the computer program, when executed by a processor, implements the maneuvering target tracking method of claim 1.