Video synthetic aperture radar imaging and tracking integrated method

Abstract

The application discloses a video synthetic aperture radar imaging and tracking integrated method, and mainly solves the problems of low imaging automation degree and large calculation amount in the prior art.The implementation scheme is as follows: reading SAR echo data containing N segments of synthetic apertures and setting an imaging grid; projecting and imaging the first two segments of echo data; preliminarily detecting on the image to obtain a target initial state set; performing inter-frame state transition on the targets in the initial state set to obtain a total target state set; setting an imaging region of interest according to the total target state set; performing value function accumulation and backtracking function updating on the second frame of image; performing target value function accumulation and backtracking function updating on the images of the remaining frame numbers according to the region of interest; and performing target track backtracking.The application reduces the calculation amount of video synthetic aperture radar imaging, improves the overall efficiency of video synthetic aperture radar target detection, and can be used for detecting ground moving targets by various airborne radars.

Description

Technical Field

[0001] This invention belongs to the field of radar technology, and specifically relates to an integrated video synthetic aperture radar imaging and tracking method, which can be used by various types of airborne radars to detect moving targets on the ground. Background Technology

[0002] Video Synthetic Aperture Radar (SAR) is a high-resolution, high-frame-rate radar imaging system capable of acquiring dynamic SAR images of an observed scene and sensing moving targets or dynamic changes within it. Compared to traditional optical and infrared sensors, SAR has stronger penetration capabilities and can maintain uninterrupted imaging of the observed scene even under degraded environmental conditions such as cloud cover or smoke obstruction. SAR typically operates in spotlight mode, and its continuous imaging characteristic places high demands on the robustness of the imaging algorithm. Backprojection is a classic SAR imaging algorithm, characterized by its simplicity, robustness, and flexible imaging area settings, making it suitable for SAR imaging. However, the high computational complexity of backprojection limits its application in real-time imaging systems.

[0003] Compared to traditional SAR systems, video synthetic aperture radar (SAR) systems typically operate at higher frequencies and have shorter synthetic apertures. Due to the occlusion effect of moving targets on the ground within a single synthetic aperture time, the moving target leaves a shadow on the SAR image reflecting its true location. Utilizing the dynamic shadows in the video SAR image sequence, moving targets can be detected and tracked. Pre-detection tracking is a detection and tracking technique for weak targets. Compared to traditional detection-before-tracking techniques, it improves detection and tracking performance by accumulating target energy across multiple frames.

[0004] In the existing video synthetic aperture radar (SAR) signal and data processing workflow, SAR imaging and target tracking are two separate steps. SAR imaging must be completed first, and then target tracking must be performed based on these multiple frames. This independent processing method results in a waste of time resources.

[0005] In their paper "Processing video-SAR data with the fastbackprojection method," X. Song et al. proposed a video synthetic aperture radar (SAR) imaging processing workflow. This workflow employs a fast backprojection algorithm for efficient imaging and divides the imaging scene into ordinary regions and regions of interest (ROIs). While maintaining full resolution in the ROIs, the resolution of the ordinary regions is reduced to decrease computational complexity, further improving imaging efficiency. However, this workflow requires manual annotation of the ROIs, resulting in low automation. Furthermore, the ROIs in this method are for stationary targets, making it impractical for applications involving focused observation of moving targets.

[0006] In their paper "Simultaneous Detection and Tracking of Moving-Target Shadows in ViSAR Imagery," X. Tian et al. proposed a video synthetic aperture radar (VSAR) moving target tracking method based on pre-detection tracking technology. This method accumulates shadow energy by combining multiple frames of images to achieve robust tracking of moving targets. However, the tracking and imaging processes in this method are independent, and the tracking process cannot provide prior information to the imaging process to improve the overall efficiency of target detection. When using time-domain algorithms for SAR imaging, there are problems with high computational load and difficulty in real-time processing. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of the prior art by proposing an integrated video synthetic aperture radar (SSAR) imaging and tracking method. This method reduces the computational load of SSAR imaging and improves the overall efficiency of SSAR target detection by fusing the imaging and tracking processes.

[0008] To achieve the above objectives, the implementation steps of the technical solution of the present invention include the following:

[0009] 1. A video synthetic aperture radar imaging and tracking integrated method, characterized by comprising the following:

[0010] (1) Read SAR echo data containing N synthetic aperture segments and set the imaging grid;

[0011] (2) Perform fast back-projection imaging on the echo data corresponding to the first two synthetic aperture segments to obtain the first high-resolution image I1 and the second high-resolution image I2.

[0012] (3) Perform preliminary detection on the two frames I1 and I2 respectively to obtain the initial state set of the target. in Let P be the initial state of the i-th target in the initial state set C, and let P be the number of targets in the initial state set C.

[0013] (4) Perform inter-frame state transitions on the target in the initial state set C to obtain the target's state candidate regions on each frame image, thus forming the total target state set Ω.

[0014] (5) Based on the total set of target states Ω, set the region of interest for imaging Ε;

[0015] (6) On the second high-resolution image I2, the target is accumulated using a value function and updated using a backtracking function;

[0016] (7) Generate SAR images from the third to the Nth frame based on the imaging region of interest E, and accumulate the target's value function and update the backtracking function:

[0017] (7a) Suppose that the current processing segment of synthetic aperture data is the m-th segment, 3≤m≤N. Back-project the m-th segment of synthetic aperture data onto the imaging region of interest E to obtain the m-th frame image I. m ;

[0018] (7b) In image I m Above, accumulate the target's value function and update the backtracking function;

[0019] (7c) Repeat steps (7a) to (7b) until the synthetic aperture from the third segment to the Nth segment is traversed, and the SAR image generation from the third frame to the Nth frame is completed, as well as the accumulation of the target value function and the update of the backtracking function.

[0020] (8) Perform target trajectory retracing:

[0021] (8a) Set the value function threshold to T N ;

[0022] (8b) Suppose that the current target being processed is the i-th target, and the maximum value V of the value function of the i-th target in the N-th frame image is... N With T N Comparison: If V N >T N If the target exists, the optimal state of the i-th target in each frame is extracted according to the backtracking function to obtain the trajectory of the i-th target; otherwise, the target does not exist.

[0023] (8c) Repeat step (8b) until all targets in the initial state set C are traversed, and target track backtracking is completed to achieve target tracking.

[0024] Compared with the prior art, the present invention has the following advantages:

[0025] 1) This invention integrates the imaging and tracking processes, enabling simultaneous imaging and moving target tracking of video synthetic aperture radar.

[0026] 2) By performing inter-frame state transitions on targets in the initial state set and using the obtained target state candidate regions as prior information to guide the SAR imaging process, this invention can not only achieve high frame rate imaging of local regions of interest, but also reduce the total computational load of the video synthetic aperture radar target detection process.

[0027] 3) This invention improves the automation of the imaging and tracking process by extracting the coordinate parameters of each state in the total set of states to achieve automatic selection of the region of interest during the tracking process. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating the implementation of the present invention. Detailed Implementation

[0029] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0030] Reference Figure 1 The implementation steps for this example are as follows:

[0031] Step 1: Read the SAR echo data and set the imaging grid.

[0032] (1.1) Read the radar echo signal echo data after range pulse compression processing. It contains N synthetic aperture segments. The echo data corresponding to each synthetic aperture segment is used to generate a frame of high-resolution SAR image. The m-th synthetic aperture segment data s m (τ,t) is represented as:

[0033]

[0034] Where τ is the fast time, t is the slow time, B is the signal bandwidth, R(t) is the instantaneous slant range of the radar reaching the point target, c is the speed of light, λ is the wavelength, j is the imaginary unit, and 1≤m≤N;

[0035] (1.2) Divide the imaging scene into N a ×N r An imaging grid of a certain size, where each grid point corresponds to a pixel in the image.

[0036] Step 2: Obtain the first two high-resolution SAR images.

[0037] (2.1) Divide the first segment of synthetic aperture data s1(τ,t) into Q sub-apertures, where the p-th segment of sub-aperture data is represented as follows:

[0038] (2.2) The Q sub-aperture data are sequentially back-projected onto the imaging grid to obtain Q low-resolution sub-images, where the low-resolution sub-image corresponding to the p-th sub-aperture is denoted as F1. p (x,y), where x and y are the X-coordinate and Y-coordinate of the imaging grid point, respectively;

[0039] (2.3) Upsample the Q low-resolution sub-images to obtain Q upsampled sub-images, where the upsampled sub-image corresponding to the p-th sub-aperture is represented as follows:

[0040] (2.4) The Q upsampled sub-images are coherently superimposed to obtain the first high-resolution image I1 of the synthetic aperture;

[0041] (2.5) Repeat steps (2.1) to (2.4) on the second synthetic aperture data to obtain a high-resolution image I2 of the second synthetic aperture.

[0042] Step 3: Perform preliminary detection on the first two frames of images to obtain the initial state set of the target.

[0043] (3.1) Initialize the initial state set

[0044] (3.2) Set the distance threshold between two adjacent frames for the same target to be L. max ;

[0045] (3.3) A constant false alarm rate detector is used to perform preliminary detection on two frames of images I1 and I2 with a high false alarm rate to obtain the preliminary detection point set Ψ1 and Ψ2 on the two frames of images;

[0046] (3.4) Calculate any preliminary detection point (x) in the preliminary detection point set Ψ1. u ,y u The distance (x) between the set Ψ2 and the set Ψ2 u ,y u The nearest preliminary testing site Distance L between them:

[0047]

[0048] (3.5) Relative distance L and L max Comparison:

[0049] If L>L max Then the initial detection point (x) u ,y u Discard if necessary, otherwise retain the initial testing point;

[0050] (3.6) Calculate the initial detection point (x) u ,y u) speed (v xu ,v yu ):

[0051]

[0052] Among them, v xu and v yu These are the preliminary detection points (x) u ,y u The velocity along the X and Y directions, where Δt is the time interval between two adjacent frames;

[0053] (3.7) The initial detection point (x) u ,y u The initial state of [x] u ,v xu ,y u ,v yu ] T Add it to the initial state set C, where the symbol T represents the transpose operation;

[0054] (3.8) Repeat steps (3.3) to (3.7) until all preliminary detection points in the preliminary detection point set Ψ1 are traversed to obtain the target initial state set. in Let be the initial state of the i-th target in the initial state set C. Let X be the X-coordinate, X-velocity, Y-coordinate, and Y-velocity of the i-th target in image I1, and let P be the number of targets in the initial state set C.

[0055] Step 4: Perform inter-frame state transitions on the target to obtain the total set of target states.

[0056] (4.1) Initialize the state candidate region of the i-th target in image I1 Value function and backtracking function Where I1(x1) i ,y1 i ) is in image I1 (x1) i ,y1 i Pixel value at point )

[0057] (4.2) Suppose that the current calculation of the state candidate region of the i-th target in the m-th frame image is performed. The target's state candidate region in the (m-1)th frame image. Represented as:

[0058]

[0059] in, Let k be the state of the i-th target in the (m-1)-th frame of the image. These represent the target's X-coordinate, X-velocity, Y-coordinate, and Y-velocity in this state. Let P be the total number of states of the i-th target in the m-1-th frame of the image, where i = 1, 2, ..., P, and P is the total number of targets in the initial state set C.

[0060] (4.3) For the state candidate region any state within Perform inter-frame state transitions to obtain the state in the m-th frame image.

[0061]

[0062] Where F is the state transition matrix under the uniform motion model, δ x and δ y These are the acceleration process noises in the X and Y directions, respectively.

[0063] (4.4) Repeat step (4.3) until the candidate state region has been traversed. Given all states, obtain the state candidate region of the i-th target in the m-th frame image. The total number of states of the i-th target in the m-th frame image;

[0064] (4.5) Repeat steps (4.2) to (4.4) until the second to Nth frames of images are traversed to obtain the set of state candidate regions for the i-th target in the N frames of images.

[0065] (4.6) Repeat steps (4.1) to (4.5) until all targets in the initial state set C are traversed, obtaining the total set of target states Ω = {Γ} for P targets in N frames of images. i ,i=1,2,...,P}.

[0066] Step 5: Set the region of interest for imaging.

[0067] (5.1) Initialize the region of interest for imaging

[0068] (5.2) For any state in the total set of states Ω Extract its coordinate parameters and This coordinate parameter That is, it is a point in the region of interest E of the imaging;

[0069] (5.3) Repeat step (5.2) until all states in the total set of states Ω are traversed to obtain the imaging region of interest Ε.

[0070] Step six: Accumulate the target's value function and update the backtracking function on the second frame image.

[0071] (6.1) Take the state candidate region of the i-th target in the second frame image I2.

[0072] (6.2) Select the candidate state region any state within Calculate the value function of this state and backtracking function

[0073]

[0074] in, For the second frame image I2 Pixel value at point This represents the k-th state of the i-th target in the second frame image I2. This represents the initial state of the i-th target;

[0075] (6.3) Repeat step (6.2) until the candidate state region has been traversed. Given all states within a given state, a function to obtain the value of each state and a backtracking function;

[0076] (6.4) Repeat steps (6.1) to (6.3) until all targets in the initial state set C are traversed, and the value function accumulation and backtracking function update of the targets in image I2 are completed.

[0077] Step 7: Generate SAR images from the third to the Nth frame, and perform target value function accumulation and backtracking function update.

[0078] (7.1) Suppose that the current processing segment of synthetic aperture data is the m-th segment, 3≤m≤N. Back-project the m-th segment of synthetic aperture data onto the imaging region of interest E to obtain the m-th frame image I. m :

[0079] 7.1.1) Divide the m-th segment of the synthesized aperture into Q sub-apertures to obtain Q segment sub-aperture data, where the p-th segment sub-aperture data is represented as...

[0080] 7.1.2) The Q sub-aperture data are sequentially projected onto the imaging region of interest E to obtain Q low-resolution sub-images, where the low-resolution image corresponding to the p-th sub-aperture is represented as follows:

[0081] 7.1.3) Upsample the Q low-resolution sub-images to obtain Q upsampled sub-images, where the upsampled sub-image corresponding to the p-th sub-aperture is represented as follows:

[0082] 7.1.4) Coherently superimpose the Q upsampled sub-images to obtain the high-resolution image I. m ;

[0083] (7.2) In image I m Above, accumulate the target's value function and update the backtracking function:

[0084] 7.2.1) Take the i-th target in image I m Candidate region of states

[0085] 7.2.2) Retrieve the candidate state region any state within The value function of this state is obtained through the accumulation method. The backtracking function for this state is obtained through a search method.

[0086]

[0087] in, For image I m middle Pixel value at point Candidate region of states All states that can be transitioned to The set of states, It represents the g-th state of the i-th target in the (m-1)-th frame of the image. This represents the k-th state of the i-th target in the m-th frame of the image. For set The number of states in the middle, It is a state Value function;

[0088] 7.2.3) Repeat step 7.2.2) until the candidate state region has been traversed. Given all states within the range, obtain the value function and backtracking function for all states;

[0089] 7.2.4) Repeat steps 7.2.1) to 7.2.3) until all targets in the initial state set C have been traversed, completing image I. m Value function accumulation and backtracking function update in the context of [the system / mechanism].

[0090] (7.3) Repeat steps (7.1) to (7.2) until the synthetic aperture from the third segment to the Nth segment is traversed, and the SAR image generation from the third frame to the Nth frame is completed, as well as the accumulation of the target value function and the update of the backtracking function.

[0091] Step 8: Perform target trajectory retracement.

[0092] (8.1) Set the threshold of the value function to T N ;

[0093] (8.2) Take the i-th target in the N-th frame image I N The maximum value function V in N and V N With threshold T N Comparison:

[0094] If V N >T N If so, the target is determined to exist, and the maximum value function V is recorded. N The corresponding optimal state is Then, the optimal state of the i-th target in each frame is obtained through a backtracking function. This leads to the predicted trajectory of the i-th target.

[0095] Otherwise, the target is determined to not exist;

[0096] (8.3) Repeat step (8.2) until all targets in the initial state set C are traversed, and target track backtracking is completed to achieve target tracking.

[0097] The above description is merely a specific example of the present invention and does not constitute any limitation on the present invention. Obviously, those skilled in the art, after understanding the content and principles of the present invention, may make various modifications and changes in form and details without departing from the principles and structure of the present invention. However, these modifications and changes based on the ideas of the present invention are still within the scope of protection of the claims of the present invention.

Claims

1. A method for video synthetic aperture radar imaging and tracking integration, the method comprising: Comprise the following: (1) reading a set of SAR echo data comprising segmenting the SAR echo data, and setting an imaging grid; (2) respectively performing fast back-projection imaging on echo data corresponding to the first two paragraphs of synthetic aperture to obtain a first frame of image and a second frame of image ; (3) In the two frames of images and images Preliminary detections are performed on each of the above to obtain the initial state set of the target. ,in For the initial state set The first in The initial state of each target. For the initial state set The number of targets in the middle; (4) Inter-frame state transition is performed on the target in the initial state set to obtain a state candidate region of the target on each frame image, and a total set of target states is constructed ; the implementation is as follows: (4a) Initialize the first A target in the image Candidate region of states Value functions and backtracking function ,in For image middle Pixel value at the point; (4b) Set the current calculation of the target in the state candidate region of the frame image , , the target in the state candidate region of the frame image is expressed as: ， wherein, is the th target in the th frame image, th state, respectively the direction coordinate, direction velocity, direction coordinate and direction velocity of the target in the state, is the total number of states of the th target in the th frame image; (4c) any of the state candidate regions within the state inter-frame state transition is performed to obtain the first state in the frame image : ， wherein, is the state transition matrix under the uniform motion model, and are the acceleration process noises in the direction and direction, respectively. (4d) repeat (4c) until all states of the target are traversed the total number of states of the i-th target in the m-th frame image, the total number of states of the i-th target in the m-th frame image, , the total number of states of the i-th target in the m-th frame image, ;​ (4e) repeating (4b) to (4d) until a second frame to a last frame of the frame images are traversed, obtaining a set of candidate regions of states of the target in the frame images ​​​​ ； (4f) repeat (4a) through (4e) until all goals in the initial state set are achieved The goal state set of the goal on the frame image : ； (5) setting the imaging region of interest according to the total set of target states ;​ (6) in the second frame high-resolution image In the above, the value function accumulation and backtracking function update are performed on the target, and the following is realized: (6a) take the state candidate region of the target in the image ;​​ (6b) taking a state candidate region any of the states , computing a value function for the state and a backtracking function : ， wherein, is a second frame image in is a pixel value at a point, is a kth state of the ith object in a second frame image , is an initial state of the ith object; (6c) repeating step (6b) until all states within the state candidate region are traversed, obtaining the value function and the backtracking function for all states the value function and the backtracking function for all states (6d) repeat steps (6a) through (6c) until all goals in the initial state set are completed value function accumulation and backup function update for goals in the image (7) Based on the imaging region of interest SAR image generation of the third frame to the frame, and the value function accumulation and the backtracking function update of the target (7a) if the current processing is the first segment of the synthetic aperture data, , the first segment of the synthetic aperture data is back projected to the imaging region of interest , the first frame of images is obtained; (7b) On the image above, the value function accumulation of the target is performed, and the backtracking function is updated; (7c) repeating steps (7a) to (7b) until the third segment to the synthetic aperture is traversed, the third frame to the SAR image generation of the third frame to the target value function accumulation and backtracking function update are completed; (8) target track backtracking is carried out: (8a) setting the value function threshold to ; (8b) Let the current process be the first... The first goal will be the first The first goal in Maximum value of the value function in the frame image and Comparison: If If the target exists, then the first step is determined to be the target, and the backtracking function is used to extract the target. The optimal state of the target in each frame of the image is obtained to obtain the th target. The target's trajectory is recorded; otherwise, the target is deemed not to exist. (8c) repeat step (8b) until all goals in the initial state set are traversed, the goal path backtracking is completed, and the goal tracking is realized.

2. The method of claim 1, wherein in step (2), the echo data corresponding to the first two synthetic apertures are respectively subjected to fast back-projection imaging to achieve the following: (2a) uniformly dividing the first segment synthetic aperture into sub-apertures, obtaining segment sub-aperture data, wherein the pth segment sub-aperture data is represented as , ; (2b) the segment aperture data are sequentially back-projected to the imaging grid to obtain low-resolution sub-images, wherein the low-resolution sub-image corresponding to the pth segment aperture is denoted as , and are the directional coordinates and directional coordinates of the imaging grid point, respectively. (2c) Each low-resolution sub-image is upsampled to obtain an upsampled sub-image, where the upsampled sub-image corresponding to the p-th sub-aperture is denoted as... ; (2d) the sub-holographic images are added coherently to obtain a high-resolution image ; (2e) repeating (2a) to (2d) on the second segment of synthetic aperture data to obtain a high resolution image .

3. The method of claim 1, wherein in step (3), preliminary detection is performed on the first two images to obtain an initial state set of the target to achieve the following: (3.1) initialize the initial state set ; (3.2) set the distance threshold between the same target in two adjacent frames of images as L max ; (3.3) A constant false alarm rate detector is used in two frames of images. and The initial detection was performed with a high false alarm rate, and the initial detection point sets Ψ1 and Ψ2 on the two frames of images were obtained. (3.4) calculating the distance L between any of the preliminary detection points in the set Ψ1 and the closest preliminary detection point in the set Ψ2 the distance L between any of the preliminary detection points in the set Ψ1 and the closest preliminary detection point in the set Ψ2 the closest preliminary detection point in the set Ψ2 the closest preliminary detection point in the set Ψ2 ； (3.5) comparing the distance L with L max L2 If L > L max then the preliminary detection point is discarded, otherwise it is retained; (3.6) Calculate preliminary detection points of the speed : ， wherein, and are preliminary detection points speeds in the X and Y directions, is a time interval between two adjacent frames of images; (3.7) adding the initial state of the preliminary detection point to the initial state set C, the symbol T denotes the transpose operation; ​ (3.8) repeating steps (3.3) to (3.7) until all the preliminary detection points in the preliminary detection point set Ψ1 are traversed, to obtain a target initial state set wherein is an initial state of an i-th target in the initial state set C, is an X-direction coordinate, an X-direction velocity, a Y-direction coordinate and a Y-direction velocity of the i-th target in the image , respectively, and P is a number of targets in the initial state set C.

4. The method of claim 1, wherein in step (5), an imaging region of interest is set to achieve the following: (5a) initializing an imaging region of interest ; (5b) For the total set of states any state in ,extract coordinate parameters in and , Use it as the region of interest for imaging. One of the points; (5c) repeating (5b) until all states in the total set of states are traversed, obtaining an imaged region of interest .​ 5. The method of claim 1, wherein the third frame to the nth frame in step (7) is performed. The SAR image generation of the frame and the value function accumulation and the backtracking function update of the target are implemented as follows: (7a) if the current processing is the first segment synthetic aperture data, , the first segment synthetic aperture data is back projected to an imaging region of interest , the first frame image is obtained: 7a1) the first segment is evenly divided into subsegments, obtaining subsegment data, wherein the first subsegment data is represented as , ; 7a2) to Subaperture data are sequentially projected to the imaging region of interest , obtaining low-resolution sub-images, wherein the low-resolution image corresponding to the th subaperture is denoted as ; 7a3) Each of the low-resolution sub-images is upsampled to obtain upsampled sub-images, where the i-th... The upsampled sub-image corresponding to each sub-aperture is represented as follows: ; 7a4) to sub-pixel image coherence superposition, obtaining a high-resolution image ; (7b) On the image , the value function accumulation and back function update of the target are performed: 7b1 ) taking the first target in the image state candidate region ; 7b2) taking a state candidate any of the states the value function of the state is obtained by the cumulative approach the backtracking function of the state is obtained by the search approach : ， wherein is the pixel value at the point in the image is the pixel value at the point is the state candidate region is the state set that can be transferred from the state is the state set that can be transferred from the state is the gth state of the ith target in the m-1th image is the kth state of the ith target in the mth image is the number of states in the set is the number of states in the set is the value function of the state is the value function of the state 7b3) repeat step 7b2) until all states within the state candidate region are traversed, resulting in value functions and back-up functions for all states 7b3) repeat step 7b2) until all states within the state candidate region are traversed, resulting in value functions and back-up functions for all states 7b4) repeat steps 7b1) to 7b3) until all goals in the initial state set are traversed accumulating value functions in the graph and updating the backtracking functions;​ (7c) repeating steps (7a) to (7b) until the third segment to the segment synthetic aperture is completed, the SAR image generation of the third frame to the frame is completed, and the value function accumulation and the backtracking function update of the target are completed.

6. The method of claim 1, wherein in step (8), target track backtracking is performed to achieve the following: (8a) setting the value function threshold to ; (8b) take the maximum function of the target in the frame image and compare it to a threshold ​​​​​ If , it is determined that the target exists, and the maximum value function is recorded The corresponding optimal state is The optimal state of the first target in each frame of image is obtained through the backtracking function The predicted trajectory of the first target is obtained ;​ Otherwise, it is determined that the target does not exist. (8c) repeat step (8b) until all goals in the initial state set are traversed, the goal path backtracking is completed, and the goal tracking is achieved.

Images