Detection and tracking method before imaging based on high-dimensional light field ghost imaging system

Through the high-dimensional light field ghost imaging system and particle filtering method, the target's three-dimensional light intensity, spectrum and polarization information are used for passive detection and tracking, solving the problem of insufficient tracking accuracy of weak targets, and achieving efficient and low-cost detection and tracking effects. Suitable for harsh environments.

CN118915176APending Publication Date: 2024-11-08SHANGHAI INST OF OPTICS & FINE MECHANICS CHINESE ACAD OF SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410893704.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-04
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The prior art lacks accuracy and accuracy in real-time detection and tracking of weak targets, especially in harsh environments, and active detection has high cost and system complexity problems.

Method used

Using a ghost imaging system based on a high-dimensional light field, a high-dimensional ghost imaging system is built by obtaining the target's three-dimensional light intensity, spectrum and polarization information in real time, including a pre-imaging system, a spatial random phase modulator, an amplification imaging system and a photoelectric polarization detector , passive detection and tracking are performed using particle filtering method, and encoding sampling and filtering algorithms solve the moving state of the target in real time.

Benefits of technology

It improves the accuracy and accuracy of target tracking, reduces system complexity and cost, enhances the tracking effect of temporarily blocking targets, and shows high concealment and stability in harsh environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118915176A_ABST
    Figure CN118915176A_ABST
Patent Text Reader

Abstract

A detection and tracking method before imaging based on a high-dimensional light field ghost imaging system comprises the following steps: selecting proper information dimensions (including spatial three-dimensional position, spectrum and polarization) based on a specific detection and tracking scene, and initializing a filtering algorithm; coding and sampling high-dimensional light field information of the moving target; and processing coding information by an algorithm to obtain a target motion state in real time. According to the method, the high-dimensional light field information of the target is coded to the spatial distribution of the detection speckle field, the target is detected and tracked by using the high-dimensional information of the light field, the decoding process of traditional ghost imaging is not needed, and the detection and tracking effect of the moving target is improved through the difference of the high-dimensional information of the target and the background.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of target detection and tracking under multi-dimensional information, and in particular to a device and method for passively detecting and tracking a target by using a particle filtering method based on a high-dimensional ghost imaging system. Background Art

[0002] With the development of intelligent driving, military technology, industrial production and other fields, the precision and accuracy of real-time detection and tracking of small targets are becoming increasingly higher. Compared with the traditional detection and tracking method that only relies on the intensity of the reflected light of the target, the use of high-dimensional information of the moving target light field, including the three-dimensional light intensity, spectrum, polarization, etc. of the target light field, can improve the detection and tracking effect of the target and solve the impact of environmental changes and harsh environmental interference on the target detection and tracking performance.

[0003] Active detection requires the launch of laser radar, which is greatly affected by bad weather such as rain and snow, while passive detection is based on the natural emission and reflection of the target and is less affected by the external environment. In addition, passive detection has the advantage of high concealment in the military, and it also has the characteristics of low cost and simple system structure in the civilian field. Summary of the invention

[0004] The purpose of the present invention is to propose a pre-imaging detection and tracking method based on a high-dimensional light field ghost imaging system, which realizes passive detection and tracking of the target by acquiring the three-dimensional light intensity, spectrum, polarization and other information of the target in real time, thereby improving the accuracy of detection and the precision of tracking.

[0005] The present invention provides a pre-imaging detection and tracking method based on a high-dimensional light field ghost imaging system, which is characterized in that it comprises the following steps:

[0006] Step S1. Building a high-dimensional ghost imaging system: a pre-imaging system (1), a spatial random phase modulator (2), an amplification imaging system (3) and a photoelectric polarization detector (4) are sequentially arranged along the optical transmission direction, and the photoelectric polarization detector (4) is connected to a computer (5);

[0007] Step S2. Calibrate the optical system:

[0008] A measurement matrix A is obtained by calibrating the high-dimensional ghost imaging system;

[0009] Step S3. Initialize the system: construct a filtering algorithm based on the information of the detection and tracking scene, including the state transfer equation, measurement equation and likelihood probability density function of the target;

[0010] Step S4. Encoding samples:

[0011] The high-dimensional ghost imaging system is used to sample and encode the target signal with weak motion, obtain the measured speckle of the target, and record it;

[0012] Step S5. Detection and tracking:

[0013] The measured speckle of the target is input into the filtering algorithm as a measurement value, and the motion state and state quantities of each dimension of the target are solved in real time.

[0014] The target light field first passes through a coding module composed of an optical system, and its three-dimensional spatial information and spectral information are encoded into two-dimensional spatially distributed speckles, which are then received by the polarization detector of the receiving module and finally processed by a computer algorithm to achieve detection and tracking of the target using the high-dimensional information of the target light field.

[0015] Preferably, the spatial random phase modulator is frosted glass or a spatial light modulator, and the amplifying imaging system is any type of optical imaging system that amplifies the modulated moving target signal and images it onto a photoelectric polarization detector.

[0016] Further, step 3, system initialization: select appropriate information dimensions (including spatial three-dimensional position, spectrum, polarization) based on the specific detection and tracking scenario, and initialize the filtering algorithm, specifically including:

[0017] S3.1. For different targets, determine the dimension of the target's motion space, whether the target moves along the axis of the optical system, or whether the axial motion speed is obvious; determine whether the spectrum or polarization of the target's light field information is different from its background, and select the dimension with the larger difference as the state quantity of filtering tracking.

[0018] S3.2. Initialize the filtering algorithm, and construct the state transfer equation, measurement equation and likelihood probability density function of the target according to the selection of the target state quantity;

[0019] Step 4: Coded sampling: Use the high-dimensional ghost imaging system to code and sample the signal of the moving weak target to obtain the measured speckle and record it in the computer; specifically, it includes:

[0020] S4.1 Encode the three-dimensional spatial information and spectral information of the target through a random phase plate: By designing a random phase plate, the system speckle response has a relationship with the spatial position and spectrum of the target in terms of scaling, imaging position, etc., thereby encoding the high-dimensional information of the target. For example, for the spatial position component of the target, the thickness of the phase plate can be designed based on the spatial rotation point spread function so that the rotation angle of the system speckle is proportional to the depth distance of the target; for the spectral component, the difference in the response of the phase plate to different wavelengths is used to make the speckle of different wavelengths imaged at different positions.

[0021] S4.2 Obtaining polarization information of the target light field through a polarization detector: The polarization detector receives redundant speckles that have encoded spatial and spectral information, and realizes real-time detection of the encoded speckle polarization information. For example, through the micro-polarization array of the polarization detector, each pixel receiving the speckle is divided into four parts, and the polarization components of 0°, 45°, 90°, and 135° are received respectively, thereby representing the polarization component of each pixel of the speckle.

[0022] Step 5: Detection and tracking: Use the measured speckle as the measurement value input of the filtering algorithm to solve the target's motion state and state quantities in various dimensions in real time.

[0023] Compared with the prior art, the technical effects of the present invention are:

[0024] (1) Compared with the traditional optical tracking algorithm that only tracks the target through its position information, this algorithm additionally introduces the target’s spectrum and polarization information. By utilizing the difference between the target’s spectrum and polarization and the background, it improves the tracking precision and accuracy of the target and improves the tracking effect of the algorithm for temporarily occluded targets.

[0025] (2) Combining the ghost imaging hardware system and software tracking algorithm, this system does not require active detection devices, which reduces the complexity of the system and achieves low-cost detection and tracking. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a system device diagram of the present invention.

[0027] Figure 2 This is a complete coding detection and tracking flow chart of the present invention.

[0028] Figure 3 This is a simulation diagram of the detection and tracking effect of the method of the present invention.

[0029] Figure 1 The markings are as follows: 1. Pre-imaging system; 2. Spatial random phase modulator; 3. Amplification imaging system; 4. Photoelectric polarization detector; 5. Computer. DETAILED DESCRIPTION

[0030] The present invention is further described below in conjunction with the accompanying drawings and embodiments, but the protection scope of the present invention shall not be limited thereto.

[0031] like Figure 1 As shown, the device of the present invention comprises: a pre-imaging system (1), a spatial random phase modulator (2), an amplifying imaging system (3), a photoelectric polarization detector (4) and a computer (5).

[0032] like Figure 2 As shown, the specific implementation steps of the method are as follows:

[0033] Step 1: Build a high-dimensional ghost imaging system

[0034] A pre-imaging system (1), a spatial random phase modulator (2), an amplifying imaging system (3) and a photoelectric polarization detector (4) are sequentially arranged along the transmission direction of the optical path, and the photoelectric polarization detector (4) is connected to a computer (5);

[0035] The spatial random phase modulator is frosted glass or a spatial light modulator, and the amplified imaging system is any optical imaging system that amplifies the modulated moving target signal and images it into a photoelectric polarization detector.

[0036] Step 2: System calibration: Calibrate the optical system to obtain the measurement matrix A of the optical system, which is used to construct the measurement equation of the system. It is necessary to ensure that the cross-correlation of the measurement speckles corresponding to adjacent object plane targets is high, so as to improve the accuracy of tracking moving targets.

[0037] Step 3: System initialization: Select appropriate information dimensions (including spatial three-dimensional position, spectrum, and polarization) based on the specific detection and tracking scenario, and initialize the filtering algorithm, including:

[0038] S3.1. For different targets, determine the dimension of the target's motion space, whether the target moves along the axis of the optical system, or whether the axial motion speed is obvious; determine whether the spectrum or polarization of the target's light field information is different from its background, and select the dimension with the larger difference as the state quantity of filtering tracking.

[0039] S3.2. Initialize the filtering algorithm and construct the state transfer equation, measurement equation and likelihood probability density function of the target according to the selection of the target state quantity:

[0040] S3.2.1. Construction of state transfer equation: According to the selection of target state (including target motion state, spatial three-dimensional coordinates, spectrum, polarization, etc.), construct the state transfer function of the target. Assume that the state of the target at time k is θ k =[r k ,s k ,μ k ,ν k ], where r k ,s k ,μ k and ν k They represent the target's position (3D coordinates), speed (2D speed), polarization component, and spectral component respectively. Assuming that the target keeps moving at a constant speed and keeps the polarization component and spectral component unchanged during the movement, its state transfer equation can be obtained as follows:

[0041] θ k =F k θk-1 +V k-1 , (1)

[0042] Among them, F k The state transfer function of the target is expressed as V k-1 is a zero-mean Gaussian process noise, whose covariance matrix Q k-1 Defined as

[0043]

[0044] S3.2.2. Construction of measurement equation: Assume that in the kth measurement, the process from the target state to the target surface image can be expressed as X k (i,j)=δ(r i,j -r k ), where X k (i,j) represents the i-th and j-th regions in the object plane image, which we divide into m×n regions, representing the field of view of the system, where i∈(1,m), j∈(1,n). Assume that the speckle received by the detector is Where p represents the total number of pixels of the detector. Then, the measurement equation of the system can be expressed as

[0045] Y k =H k (θ k-1 )+W k =AX k +W k , (2)

[0046] Among them, H k is the measurement equation of the system, W k is a Gaussian measurement noise with zero mean, and its covariance matrix is ​​assumed to be the identity matrix;

[0047] S3.2.3. Assuming the number of particles is N, and the light intensity distribution of each pixel of the detector follows a Gaussian distribution, then the likelihood probability density function of each particle can be expressed as

[0048]

[0049] in is the mean square error of the light intensity at pixel u, is the predicted light intensity at pixel u in the kth frame, is the detected light intensity at pixel u in the kth frame.

[0050] Step 4: Coded sampling: Use the high-dimensional ghost imaging system to code and sample the signal of the moving weak target to obtain the measured speckle and record it in the computer, which specifically includes:

[0051] S4.1 Encode the three-dimensional spatial information and spectral information of the target through a random phase plate: By designing a random phase plate, the system speckle response has a relationship with the spatial position and spectrum of the target in terms of scaling, imaging position, etc., thereby encoding the high-dimensional information of the target. For example, for the spatial position component of the target, the thickness of the phase plate can be designed based on the spatial rotation point spread function so that the rotation angle of the system speckle is proportional to the depth distance of the target; for the spectral component, the difference in the response of the phase plate to different wavelengths is used to make the speckle of different wavelengths imaged at different positions.

[0052] S4.2 Obtaining polarization information of the target light field through a polarization detector: The polarization detector receives redundant speckles that have encoded spatial and spectral information, and realizes real-time detection of the encoded speckle polarization information. For example, through the micro-polarization array of the polarization detector, each pixel receiving the speckle is divided into four parts, and the polarization components of 0°, 45°, 90°, and 135° are received respectively, thereby representing the polarization component of each pixel of the speckle.

[0053] Step 5: Detection and tracking: Use the measured speckle as the measurement value input of the filtering algorithm to solve the target's motion state and state quantities in various dimensions in real time.

[0054] The simulation results of the method are as follows Figure 3 As shown in the figure. In the simulation, the imaging field of view is set to 64×64 pixels, the initial position of the point target is (10,10), the initial velocity is (1,1), and there is random acceleration. For the simulated spectral information, the distribution values ​​of the three spectral bands are set to (1,1,1). The signal-to-noise ratio of the target is set to 4:1, the preset number of particles is 500, and the initial state of the particle swarm is set near the initial state of the target according to the Gaussian distribution. Figure 3 The black curve and red curve in (1) are the true path and estimated path of the target point, respectively. Figure 3 The black curve and the red curve in (2) are the velocity change of the target point and the tracking error of the algorithm for the target. The simulation verifies that the ghost camera can successfully track the weak target before imaging. Figure 3 (2), we can see that when the target’s motion state changes dramatically, such as from frame 5 to frame 20, the tracking error is improved; when the motion becomes stable, such as from frame 20 to frame 35, the tracking accuracy is improved, which is in line with reality.

Claims

1. A pre-imaging detection and tracking method based on a high-dimensional light field ghost imaging system. It is characterized in that The steps include: Step S1. Building a high-dimensional ghost imaging system: a pre-imaging system (1), a spatial random phase modulator (2), an amplification imaging system (3) and a photoelectric polarization detector (4) are sequentially arranged along the optical transmission direction, and the photoelectric polarization detector (4) is connected to a computer (5); Step S2. Calibrate the optical system: A measurement matrix A is obtained by calibrating the high-dimensional ghost imaging system; Step S3. Initialize the system: construct a filtering algorithm based on the information of the detection and tracking scene, including the state transfer equation, measurement equation and likelihood probability density function of the target; Step S4. Encoding samples: The high-dimensional ghost imaging system is used to sample and encode the target signal with weak motion, obtain the measured speckle of the target, and record it; Step S5. Detection and tracking: The measured speckle of the target is input into the filtering algorithm as a measurement value, and the motion state and state quantities of each dimension of the target are solved in real time.

2. The method for pre-imaging detection and tracking based on a high-dimensional light field ghost imaging system according to claim 1, It is characterized in that The step S3. constructs a filtering algorithm based on the information of the detection and tracking scene, specifically comprising: S3.1 For different targets, determine the dimension of the target motion space, including three-dimensional spatial position, spectrum, polarization), whether the target moves along the axis of the optical system, or whether the axial motion speed is obvious; determine whether the spectrum or polarization of the target's light field information is different from its background, and select the dimension with the largest difference as the target state quantity; S3.2 According to the selection of the target state quantity, construct the target state transfer equation, measurement equation and likelihood probability density function; The state transfer equation of the target θ k =F k θ k-1 +V k-1 , Among them, F k The state transfer function of the target is expressed as V k-1 is a zero-mean Gaussian process noise, whose covariance matrix Q k-1 Defined as The measurement equation is as follows: Y k =H k (θ k-1 )+W k =AX k +W k Among them, H k is the measurement function of the high-dimensional ghost imaging system, W k is a Gaussian measurement noise with zero mean, and its covariance matrix is ​​assumed to be the identity matrix; The likelihood probability density function is as follows: in is the mean square error of the light intensity at pixel u, is the predicted light intensity at pixel u in the kth frame, is the detected light intensity at pixel u in the kth frame.

3. The method for pre-imaging detection and tracking based on a high-dimensional light field ghost imaging system according to claim 1, It is characterized in that The step 4, coding sampling: using a high-dimensional ghost imaging system to code and sample the signal of a small moving target to obtain a measured speckle, which is recorded in a computer, specifically includes: S4.1 Encode the three-dimensional spatial information and spectral information of the target through a random phase plate: By designing a random phase plate, the system speckle response has a relationship with the spatial position and spectrum of the target in terms of scaling, imaging position, etc., thereby encoding the high-dimensional information of the target. For example, for the spatial position component of the target, the thickness of the phase plate can be designed based on the spatial rotation point spread function so that the rotation angle of the system speckle is proportional to the depth distance of the target; for the spectral component, the difference in the response of the phase plate to different wavelengths is used to make the speckle of different wavelengths imaged at different positions. S4.2 Obtaining polarization information of the target light field through a polarization detector: The polarization detector receives redundant speckles that have encoded spatial and spectral information, and realizes real-time detection of the encoded speckle polarization information. For example, through the micro-polarization array of the polarization detector, each pixel receiving the speckle is divided into four parts, and the polarization components of 0°, 45°, 90°, and 135° are received respectively, thereby representing the polarization component of each pixel of the speckle.

4. The pre-imaging detection and tracking method based on a high-dimensional light field ghost imaging system according to any one of claims 1 to 4, It is characterized in that The spatial random phase modulator is frosted glass or a spatial light modulator, and the amplified imaging system is any optical imaging system that amplifies the modulated moving target signal and images it into a photoelectric polarization detector.