A short-range line-scan laser point cloud imaging simulation method in a smog interference environment

Abstract

The application discloses a short-range line-scan laser point cloud imaging simulation method in a smoke interference environment, and belongs to the field of laser detection.The application combines a laser power model, a smoke diffusion model and a geometric measurement model to establish a short-range line-scan laser point cloud imaging model in a smoke interference environment, realizes unified multi-physical coupling modeling in a three-dimensional space, and improves the numerical prediction accuracy of the received power in the smoke interference environment.The application combines a parallel ray tracing method, particle dynamics solving and iterative radiation transmission calculation to realize synchronous simulation of smoke diffusion and line-scan laser imaging, and improves the point cloud generation fidelity.The application combines physical model driving and three-dimensional virtual environment deployment, configures different parameters for simulation, and can realize generation of large-scale line-scan laser point cloud imaging simulation data under different smoke interference conditions.The line-scan laser point cloud imaging simulation data generated by the application can be applied to target recognition algorithm training in a smoke interference environment, and the accuracy and efficiency of target recognition are improved.

Description

Technical Field

[0001] This invention relates to a simulation method for short-range line-scan laser point cloud imaging under smoke interference environment, belonging to the field of laser detection technology. Background Technology

[0002] Smoke obscuring in battlefield environments is a significant means of interfering with short-range laser imaging systems. The strong scattering effect of laser light in non-uniform smoke fields generates irregular echo noise, attenuating the target signal and severely impacting image quality, leading to reduced target recognition performance. However, the quantity of real-world battlefield measurement data is limited, and the cost of equivalent experiments is high. Collecting large-scale battlefield smoke-interference-based short-range laser imaging measurement data is challenging, hindering the development of research on laser anti-smoke interference and target recognition. Therefore, using computer simulation to generate synthetic data is an important way to address data scarcity, providing a safe, convenient, and reliable data source.

[0003] 3D point clouds are an important data representation type for laser imaging, containing both 3D and intensity information, which is beneficial for subsequent target identification. Current research on laser imaging simulation under aerosol interference environments mainly falls into two technical approaches: physical modeling-based and data-driven approaches. Physical modeling-based methods use Mie scattering theory as a framework, employing the Monte Carlo method to simulate the random walk of photons in an aerosol particle field. By defining the observation geometry to determine the number of received photons, and then simulating echo characteristics based on temporal superposition, these methods typically require simulating particle trajectory tracking with millions of particles. Single-channel simulations are time-consuming and unsuitable for multi-line laser imaging point cloud simulations. Some commercial open-source 3D simulation platforms have built-in LiDAR components, enabling direct interactive output of 3D point clouds based on scene elements. While this simplifies the detection model and improves computational efficiency, it lacks a control branch for environmental interference, often using uniformly distributed spherical particles as an equivalent, leading to decreased simulation accuracy. Data-driven approaches model smoke diffusion as a concentration flow, generating equivalent point cloud noise based on the smoke field concentration distribution, and adding it to the existing laser imaging point cloud. These methods rely on existing real-world samples, and the generated content is limited by the scope of the original dataset, making it difficult to generate edge cases.

[0004] Modeling and simulation of laser imaging under smog interference can not only reveal the mechanism of distorted imaging, but also obtain high-fidelity synthetic data, which is of great practical application value for the design and testing of actual intelligent algorithms. By establishing a coupled physical model of particle smoke diffusion and laser imaging, and realizing efficient synchronous simulation of smoke diffusion and laser measurement, it is of great significance for the research on laser anti-interference and intelligent target recognition under smog environment. Summary of the Invention

[0005] To address the lack of short-range line-scan laser point cloud imaging data under smoke interference, this invention aims to provide a simulation method for short-range line-scan laser point cloud imaging under smoke interference. This method considers laser scattering and attenuation, non-uniform smoke diffusion, and spatial geometric measurement processes under smoke interference. It establishes a short-range line-scan laser point cloud imaging model under smoke interference, and based on a three-dimensional virtual environment, combines ray tracing methods, particle dynamics solutions, and iterative radiative transfer calculations to achieve simultaneous simulation of smoke diffusion and laser imaging, thus constructing a large-scale simulation dataset of short-range line-scan laser point clouds under smoke interference.

[0006] The objective of this invention is achieved through the following technical solution:

[0007] This invention discloses a simulation method for short-range line-scan laser point cloud imaging under smoke interference. It establishes a laser power model, a smoke diffusion model, and a geometric measurement model by combining laser transmission characteristics, smoke diffusion laws, and geometric measurement relationships. These models are coupled in a unified three-dimensional space using a multi-physics model to establish a short-range line-scan laser point cloud imaging model under smoke interference. Based on a three-dimensional virtual environment, and combining parallel ray tracing methods, particle dynamics solutions, and iterative radiative transfer calculations, it achieves synchronous simulation of smoke diffusion and line-scan laser imaging, obtaining a simulation dataset of short-range line-scan laser point clouds under smoke interference.

[0008] This invention discloses a short-range line-scan laser point cloud imaging simulation method under smoke interference environment, comprising the following steps:

[0009] Step 1: Combining Mie scattering theory and lidar equations, establish a laser power model of target reflection under smoke interference environment and quantify the received power;

[0010] The laser power model is constructed based on the lidar equations, and is calculated by combining the target reflection component and the smoke particle scattering component as follows:

[0011]

[0012] In the formula, P t Let P(R) represent the transmit power, P(R) represent the receive power at a distance R, η represent the LiDAR system efficiency, O(R) represent the overlap rate between the transmit beam's field of view and the receive beam's field of view, c represent the speed of light, τ represent the pulse width, and A represent the receiver aperture area. T(R), B(R), and H(R) represent the transmission term, backscattering term, and target reflection term, respectively.

[0013] ① Transmission term. T(R) describes the bidirectional attenuation along the light transmission path. According to Beer-Lambert's law, the attenuation of laser light passing through smoke aerosol is calculated as follows:

[0014]

[0015] In the formula, L R R represents the effective distance the laser travels through the smoke at a distance of R, and α represents the extinction coefficient. According to Mie scattering theory, for aerosol particles of the same type uniformly present in a unit volume, the extinction coefficient is calculated as follows:

[0016]

[0017] In the formula, r a Q represents the average particle radius, N represents the particle number concentration, and Q represents the average particle radius. ext This represents the extinction efficiency factor.

[0018] ② Backscattering term. The backscattering term B(R) describes the scattered energy of the smoke in the 180° scattering direction, and is calculated as follows:

[0019]

[0020] In the formula, β represents the backscattering coefficient, which is calculated according to Mie scattering theory as follows:

[0021]

[0022] In the formula, Q back This represents the backscattering efficiency factor.

[0023] ③ Target reflection term. H(R) describes the reflection effect of the laser beam on the target surface. The energy relationship between the incident and reflected directions is quantified by the bidirectional reflection distribution function, as shown in the following expression:

[0024]

[0025] In the formula, θ and φ represent the zenith angle and azimuth angle, respectively; the subscripts i and r represent the incident light and reflected light, respectively; δ represents the Dirac function; R0 represents the distance to the target surface; and f(θ) i , θ r ; φ i , φ r The function represents the bidirectional reflection distribution function, calculated using a six-parameter model as follows:

[0026]

[0027] In the formula, k b k r k da, b, and d are parameters to be determined, γ represents the angle between the surface normal and the target surface normal, and ξ represents the angle between the incident light direction and the surface normal.

[0028] Equations (1), (2), (4) and (6) constitute the laser power model;

[0029] Step 2: Combining the turbulence model and particle dynamics equations, establish a smoke diffusion model based on computational fluid dynamics to describe the diffusion characteristics of smoke under the influence of natural environmental wind fields;

[0030] The diffusion of smoke under the influence of wind can be regarded as a typical gas-solid two-phase flow. The spatial distribution of particle clusters affects the wind field changes, and at the same time, smoke particles are affected by air forces to carry out diffusion motion. The gas phase model and particle phase model are established as follows:

[0031] Gas-phase model. The standard k-ε two-equation Reynolds-time-averaged Navier-Stokes (RANS) model is adopted, simplifying the flow field calculation through time averaging and utilizing an efficient turbulence model with a closed equation set. Assuming that the smoke source is under constant wind and temperature fields in the initial state, the continuity and momentum equations of the RANS model are expressed as follows:

[0032]

[0033]

[0034] In the formula, the overline represents the ensemble average, u represents the velocity vector, x represents the tensor expression of the turbulence model, and i, j = 1, 2, or 3. ρ a ρ represents gas density, p represents gas pressure, and μ represents viscosity coefficient. The Reynolds stress term is expressed as follows:

[0035]

[0036] In the formula, μ t = ρ a C μ k 2 / ε, k, and ε represent the turbulent kinetic energy and turbulent energy dissipation rate, respectively. When i = j, δ ij =1, otherwise δ ij = 0. The transport equations for k and ε are calculated as follows:

[0037]

[0038]

[0039] In the formula, P k The term representing the turbulent kinetic energy production term satisfies P k = μ t S², where , The constant term is C μ =0.09, C ε1 =1.44, C ε2 =1.92, σ k =1 and σ ε =1.3.

[0040] Particle phase model. By analyzing particle motion through Lagrange dynamics, a single-particle force balance equation is established to achieve particle trajectory tracking, as shown below:

[0041]

[0042] In the formula, u p U represents particle velocity. a F represents air velocity. D ·(u a -u p ) represents the drag force per unit mass of a particle, g·(ρ) p -ρ a ) / ρ p ρ represents the resultant force of gravity and buoyancy. p F represents the density of smoke particles. E This represents other external forces. The F in the drag force term... D The calculation is as follows:

[0043]

[0044] In the formula, C D d represents the traction coefficient. p Re represents the diameter of the smoke particles, and Re represents the Reynolds number.

[0045] Equations (8), (9), (11), (12), and (13) constitute a smoke diffusion model;

[0046] Step 3: Based on the principle of line scan imaging, establish a geometric measurement model and determine the imaging boundary through three-dimensional spatial mapping; couple the laser power model and the smoke diffusion model to establish a short-range line scan laser point cloud imaging model under smoke interference environment;

[0047] Linear laser imaging systems can acquire linear point cloud data and achieve three-dimensional imaging through relative motion between a moving platform and the target, defining a global coordinate system O. g-x g y g z g The origin is fixed on the moving platform, the positive x-direction is defined as the forward motion direction, and the positive y- and z-directions are to the left and upward, respectively. A local coordinate system Ox is defined. s y s z s The origin is fixed at the optical center of the laser detector, the central beam direction is the positive x-axis, and the left and upward directions are the positive y-axis and z-axis, respectively. The homogeneous form of the transformation matrix from the local coordinate system to the global coordinate system is as follows:

[0048]

[0049] In the formula, Rot and B represent rotation and translation transformations, respectively, determined by their relative spatial positions. The 3D coordinates of a single-frame linear point cloud are calculated in the global coordinate system as follows:

[0050]

[0051] In the formula, Let ϕ represent the distance vector, and let ϕ represent the field of view angle. m The angle between the m-th line beam and the central axis is represented, ranging from (-ϕ / 2, ϕ / 2). During relative motion, three-dimensional imaging is achieved by stitching together continuous linear point clouds in the x-direction, with a fixed total number of stitched frames n. s The current frame is selected as the upper boundary of the image, and the corrected coordinates of the nth frame in the x and z directions are as follows:

[0052]

[0053] In the formula, v m f represents the speed of the mobile platform. s ψ represents the sampling rate. a The forward tilt angle is the angle between the platform's direction of motion and the horizontal plane.

[0054] Equations (16) and (17) constitute a geometric measurement model, which is coupled with the laser power model and the smoke diffusion model to establish a short-range linear scanning laser point cloud imaging model under smoke interference environment;

[0055] Step 4: Perform multi-physics simulation based on the laser point cloud imaging model in a 3D virtual environment, that is, perform simultaneous simulation of smoke diffusion and line-scan laser imaging to obtain simulated laser point cloud data.

[0056] A multiphysics simulation application is built based on a 3D virtual environment. According to the actual application scenario, scene elements such as terrain, vegetation, targets, and smoke interference sources are established in the 3D virtual environment, and optical attribute data are configured. Simulation parameters are configured according to different smoke interference conditions and measurement conditions. The geometric measurement model obtained in step three is solved using the parallel ray tracing method to obtain spatial relative coordinate parameters. At the same time, the smoke diffusion model obtained in step two is calculated using the particle dynamics solution method to obtain the wind field and particle state parameters. The laser power model obtained in step one is solved using the iterative radiative transfer calculation method to obtain simulated laser point cloud data.

[0057] Step 5: Establish evaluation indicators. Evaluate the accuracy of the simulated laser point cloud data in Step 4 using measured laser point cloud data to obtain an effective simulation model. Based on the effective simulation model, realize laser point cloud imaging simulation prediction.

[0058] The similarity between simulated laser point cloud data and measured laser point cloud data is comprehensively evaluated by chamfer distance (CD), Hausdorff distance (HD), and bulldozer distance (EMD). When the CD, HD, and EMD values ​​all meet the preset validity requirements, an effective simulation model is obtained.

[0059] It also includes step six, which uses the effective simulation model from step five to generate short-range line-scan laser point cloud imaging data under various smoke interference conditions by configuring different parameters, constructs a large-scale short-range line-scan laser point cloud imaging dataset under smoke interference conditions, and applies it to the training of target recognition algorithms under smoke interference conditions to improve the accuracy and efficiency of target recognition.

[0060] Beneficial effects:

[0061] 1. The present invention discloses a simulation method for short-range line-scan laser point cloud imaging under smoke interference environment. By combining laser power model, smoke diffusion model and geometric measurement model, a short-range line-scan laser point cloud imaging model under smoke interference environment is established, realizing multi-physics coupling modeling in unified three-dimensional space and improving the accuracy of received power numerical prediction under smoke interference environment.

[0062] 2. The present invention discloses a short-range line-scan laser point cloud imaging simulation method under smoke interference environment. By combining parallel ray tracing method, particle dynamics solution and iterative radiative transfer calculation, it realizes synchronous simulation of smoke diffusion and line-scan laser imaging, and improves the fidelity of point cloud generation.

[0063] 3. The present invention discloses a simulation method for short-range line-scan laser point cloud imaging under smoke interference environment. By combining physical model driving and three-dimensional virtual environment deployment, it can generate large-scale line-scan laser point cloud imaging simulation data under different smoke interference conditions, providing an effective data source for the design of line-scan laser imaging system anti-smoke interference and target recognition applications.

[0064] 4. This invention discloses a simulation method for short-range line-scan laser point cloud imaging under smoke interference conditions. It configures scene elements such as terrain, vegetation, targets, and smoke interference sources in a three-dimensional virtual environment, and configures optical attribute data. Simulation parameters are configured according to different smoke interference conditions and measurement conditions. By configuring different parameters for simulation, short-range line-scan laser point cloud imaging data under various smoke interference conditions are generated, constructing a large-scale short-range line-scan laser point cloud imaging dataset under smoke interference conditions. This dataset is then applied to target recognition algorithm training under smoke interference conditions, improving the accuracy and efficiency of target recognition. Attached Figure Description

[0065] Figure 1 This is a flowchart of a short-range line-scan laser point cloud imaging simulation method under smoke interference environment disclosed in this invention;

[0066] Figure 2 This is the visualization of the three-dimensional virtual environment and simulation in step four of this embodiment;

[0067] Figure 3 This is a simulation flowchart for step four in this embodiment;

[0068] Figure 4 To visualize the simulated laser point cloud data and the measured laser point cloud data from the four experiments in step five of this embodiment,

[0069] Figure (a) shows the simulated laser point cloud diagram of Test 1, and Figure (b) shows the measured laser point cloud diagram of Test 1.

[0070] Figure (c) shows the simulated laser point cloud diagram of Test 2, and Figure (d) shows the measured laser point cloud diagram of Test 2.

[0071] Figure (e) shows the simulated laser point cloud diagram for Test 3, and Figure (f) shows the measured laser point cloud diagram for Test 3.

[0072] Figure (g) shows the simulated laser point cloud diagram of Test 4, and Figure (h) shows the measured laser point cloud diagram of Test 4. Detailed Implementation

[0073] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. The technical problems solved by the present invention and its beneficial effects are also described. It should be noted that the described embodiments are only intended to facilitate understanding of the present invention and do not constitute any limitation thereof.

[0074] The present invention discloses a short-range line-scan laser point cloud imaging simulation method under smoke interference environment, which is applicable to ground target detection application scenarios of dynamic platforms with no roll flight attitude such as loitering munitions and UAVs. This embodiment uses UAV as an example for illustration.

[0075] This embodiment discloses a simulation method for short-range line-scan laser point cloud imaging under smoke interference environment, such as... Figure 1 As shown, it includes the following steps:

[0076] Step 1: Combining Mie scattering theory and lidar equations, establish a laser power model of target reflection under smoke interference environment and quantify the received power;

[0077] The laser power model is constructed based on the lidar equation. The model input parameters are configured according to the actual line-scan laser imaging system and experimental conditions. Taking the parameters in a certain experiment as an example, the key parameters are shown in Table 1.

[0078] Table 1 Input parameters for laser power model

[0079]

[0080] The bidirectional reflection distribution function of the target surface was modeled using a six-parameter model. The parameters to be determined were fitted based on measured values. The six-parameter fitting results for typical target surface materials are shown in Table 2. Based on the obtained fitting parameters, the bidirectional reflection distribution function values ​​for the incident angle range of -70° to 70° were pre-calculated at a resolution of 0.1° for real-time simulation.

[0081] Table 2 Fitting parameters of bidirectional reflection distribution function for typical materials

[0082]

[0083] Step 2: Combining the turbulence model and particle dynamics equations, establish a smoke diffusion model based on computational fluid dynamics to describe the diffusion characteristics of smoke under the influence of wind fields in the natural environment;

[0084] A smoke diffusion model based on the coupling of the RANS model and particle dynamics equations was established. The computational boundary conditions were determined. A uniform three-dimensional grid was divided with the smoke source as the center. The grid size was 1 m × 1 m × 1 m, and the spatial boundary size was 10 m × 10 m × 10 m. The initial wind field was uniform. The particle emission velocity and direction were set. The wind field state of each grid was calculated by solving the smoke diffusion model. Then, the particle dynamics equations were solved based on the wind field information to obtain the particle motion state and perform trajectory tracking, thereby realizing the numerical calculation of microscopic smoke diffusion parameters.

[0085] Step 3: Based on the principle of line scan imaging, establish a geometric measurement model and determine the imaging boundary through three-dimensional spatial mapping; couple the laser power model and the smoke diffusion model to establish a short-range line scan laser point cloud imaging model under smoke interference environment;

[0086] Based on the actual line-scan laser imaging system parameters and experimental parameters, as shown in Table 3, the geometric beam transmission direction in three-dimensional space is determined. The smoke field parameters are determined using the smoke diffusion model in step two in the transmission path. The received power is calculated using the laser power model in step one. A short-range line-scan laser point cloud imaging model under smoke interference environment is established. The point cloud coordinates are calculated based on the ranging values ​​of multiple channels.

[0087] Table 3 Parameters of Linear Laser Imaging System

[0088]

[0089] Step 4: Establish a multiphysics simulation application based on a 3D virtual environment, deploy the physical models established in Step 1, Step 2 and Step 3, perform synchronous simulation of smoke diffusion and line scan laser imaging, and obtain simulated laser point cloud data.

[0090] Based on the Unity3D platform, a 3D virtual environment is constructed that includes scene elements such as terrain, vegetation, targets, and smoke interference sources, such as... Figure 2 As shown in the figure. A particle system is used to drive the motion of smoke particles. The initial ejection velocity and shape of the smoke particles are configured according to the continuous point source smoke emission mode. Parallel ray tracing is used to simulate the geometric measurement process. Simultaneously, the particle velocity and acceleration vectors are controlled by the calculated wind field parameters and particle dynamics. Iterative radiative transfer calculations are performed based on the interaction information of scene elements to obtain the received power and calculate the point cloud coordinates. The overall simulation process is as follows. Figure 3 As shown, the summary is as follows:

[0091] (1) Iterate through all wind field grids and calculate the wind field information for the next state, including wind speed and wind direction;

[0092] (2) Iterate through all particles and calculate the resultant force of the particles based on the current wind field information;

[0093] (3) Iterate through all beams, configure the direction vector according to the beam control mode of the line-scan laser imaging system to create ray tracing, calculate the received power signal on the optical transmission path, and generate a virtual point cloud based on the ranging value;

[0094] (4) Update the wind field, smoke particle field information and the spatial position of scene elements, and execute file output, visualization and other programs.

[0095] Step 5: Establish evaluation indicators and evaluate the accuracy of the simulated laser point cloud data in Step 4 using measured laser point cloud data to obtain an effective simulation model;

[0096] The similarity between simulated and measured laser point cloud data is comprehensively evaluated using chamfer distance (CD), Hausdorff distance (HD), and bulldozing distance (EMD). Given two sets of laser point cloud data, the evaluation indicators are calculated as follows:

[0097]

[0098]

[0099]

[0100] In the formula, Q1 and Q2 represent the simulated point cloud and the measured point cloud, respectively, q1 and q2 represent the points in the two sets of point clouds, Φ:Q1→Q2 represents the double mapping, and the CD, HD and EMD values ​​are not greater than 20 to prove that the simulation model is effective.

[0101] Four experiments were conducted with different flight attitudes, using the same experimental parameters as steps one and three. The same input parameters were configured in the simulation environment, and the measured data and simulation data were compared. The point cloud visualization results are shown below. Figure 4 As shown in Table 4, the performance evaluation results are quite close to those of the measured point cloud. The CD, EMD, and HD values ​​are all below 10, with average values ​​of 0.71 and 4.77, respectively. Therefore, the effectiveness of the simulation model in terms of accuracy is demonstrated. The simulation model was exported and packaged into software to obtain an effective simulation model for subsequent engineering applications.

[0102] Table 4 Performance Evaluation Results of Simulated Point Cloud

[0103]

[0104] It also includes step six, which uses the effective simulation model from step five to generate short-range line-scan laser point cloud imaging data under various smoke interference conditions by configuring different parameters, constructs a large-scale short-range line-scan laser point cloud imaging dataset under smoke interference conditions, and applies it to the training of target recognition algorithms under smoke interference conditions.

[0105] Based on actual application scenarios, simulations were conducted under different projectile-target encounter attitudes and smoke diffusion conditions. The simulation parameters are shown in Table 5. A total of 3000 sets of simulated laser point cloud data were obtained, and a short-range line-scan laser point cloud imaging dataset under smoke interference conditions was constructed for training a deep learning-based target recognition algorithm.

[0106] Table 5 Simulation Parameters

[0107] Height (m) 10、15、20 Lean angle (°) 0、30、45 Meeting speed (m / s) 100、200 Target velocity (m / s) 15、20、25 Wind speed (m / s) 1、3、5 Wind direction (°) 0、45、90

[0108] In summary, the short-range line-scan laser point cloud imaging simulation method disclosed in this embodiment establishes a multi-physics coupling model for line-scan laser imaging under smoke interference, and combines parallel ray tracing, particle dynamics solving, and iterative radiative transfer calculation to achieve synchronous simulation of smoke diffusion and line-scan laser imaging. This generates high-fidelity line-scan laser point cloud imaging data under smoke interference, providing an effective data source for the design of line-scan laser imaging systems for smoke interference resistance and target recognition applications.

[0109] The above detailed description further illustrates the purpose, technical solution, and beneficial effects of the invention. It should be understood that the above description is merely a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A simulation method for short-range line-scan laser point cloud imaging under smoke interference environment, characterized in that: Includes the following steps, Step 1: Combining Mie scattering theory and lidar equations, establish a laser power model of target reflection under smoke interference environment and quantify the received power; Step 2: Combining the turbulence model and particle dynamics equations, establish a smoke diffusion model based on computational fluid dynamics to describe the diffusion characteristics of smoke under the influence of natural environmental wind fields; Step 3: Based on the principle of line scan imaging, establish a geometric measurement model and determine the imaging boundary through three-dimensional spatial mapping; couple the laser power model and the smoke diffusion model to establish a short-range line scan laser point cloud imaging model under smoke interference environment; Step 4: Perform multi-physics simulation based on the laser point cloud imaging model in a 3D virtual environment, that is, perform simultaneous simulation of smoke diffusion and line-scan laser imaging to obtain simulated laser point cloud data. Step 5: Establish evaluation indicators. Evaluate the accuracy of the simulated laser point cloud data in Step 4 using measured laser point cloud data to obtain an effective simulation model. Based on the effective simulation model, realize laser point cloud imaging simulation prediction.

2. The method as described in claim 1, characterized in that: It also includes step six, which uses the simulation model from step five to generate short-range line-scan laser point cloud imaging data under various smoke interference conditions by configuring different parameters, constructs a large-scale short-range line-scan laser point cloud imaging dataset under smoke interference conditions, and applies it to the training of target recognition algorithms under smoke interference conditions to improve the accuracy and efficiency of target recognition.

3. The method as described in claim 1, characterized in that: The laser power model described in step one is expressed as follows: In the formula, P t Let P(R) represent the transmit power, P(R) represent the receive power at a distance R, η represent the LiDAR system efficiency, O(R) represent the overlap rate between the transmit beam field of view and the receive field of view, c represent the speed of light, τ represent the pulse width, and A represent the area of ​​the receiver aperture; T(R), B(R), and H(R) represent the transmission term, backscattering term, and target reflection term, respectively.

4. The method as described in claim 2, characterized in that: The transmission term, T(R), describes the bidirectional attenuation along the light transmission path. According to Beer-Lambert's law, the attenuation of laser light passing through smoke aerosol is expressed as follows: In the formula, L R The extinction coefficient α represents the effective distance of the laser beam through the smoke at a distance R. In the formula, r a Q represents the average particle radius, N represents the particle number concentration, and Q represents the average particle radius. ext This represents the extinction efficiency factor.

5. The method as described in claim 2, characterized in that: The backscattering term is represented as follows: In the formula, the backscattering coefficient β is expressed as follows: In the formula, Q back This represents the backscattering efficiency factor.

6. The method as described in claim 2, characterized in that: The expression for the target reflection term is as follows: In the formula, θ and φ represent the zenith angle and azimuth angle, respectively; the subscripts i and r represent the incident light and reflected light, respectively; δ represents the Dirac function; R0 represents the distance to the target surface; and f(θ) i , θ r ; φ i ,φ r () represents the bidirectional reflection distribution function: In the formula, k b k r k d a, b, and d are parameters to be determined, γ represents the angle between the surface normal and the target surface normal, and ξ represents the angle between the incident light direction and the surface normal.

7. The method as described in claim 1, characterized in that: The implementation method for step two is as follows: The diffusion of smoke under the influence of wind is considered as a typical gas-solid two-phase flow. The spatial distribution of particle clusters affects the wind field changes, and the smoke particles are also affected by air forces to carry out diffusion motion. The gas phase model and particle phase model are established as follows: Gas-phase model; the standard k-ε two-equation Reynolds-averaged Navier-Stokes equations model RANS is adopted, and the flow field calculation is simplified by time averaging, utilizing an efficient turbulence model with closed equations; in the initial state, the smoke source is under constant wind and temperature fields, and the continuity equation and momentum equation of the RANS model are expressed as follows: , In the formula, the overline represents the ensemble average, u represents the velocity vector, x represents the tensor expression of the turbulence model, and i, j = 1, 2, or 3; ρ a ρ represents gas density, p represents gas pressure, and μ represents viscosity coefficient. The Reynolds stress term is expressed as follows: In the formula, μ t = ρ a C μ k 2 / ε, k, and ε represent the turbulent kinetic energy and turbulent energy dissipation rate, respectively. When i = j, δ ij = 1, otherwise δ ij = 0; The transport equations for k and ε are calculated as follows: ， In the formula, P k The term representing the turbulent kinetic energy production term satisfies P k =μ t S², where , The constant term is C μ =0.09, C ε1 =1.44, C ε2 =1.92, σ k =1 and σ ε =1.3; Particle phase model; particle motion is analyzed through Lagrange dynamics, and single-particle force balance equations are established to achieve particle trajectory tracking, as shown below: In the formula, u p U represents particle velocity. a F represents air velocity. D ·(u a -u p ) represents the drag force per unit mass of a particle, g·(ρ) p -ρ a ) / ρ p ρ represents the resultant force of gravity and buoyancy. p F represents the density of smoke particles. E F represents other external forces; in the drag force term, F D The calculation is as follows: In the formula, C D d represents the traction coefficient. p Re represents the diameter of the smoke particles, and Re represents the Reynolds number; Equations (8), (9), (11), (12) and (13) constitute the smoke diffusion model.

8. The method as described in claim 1, characterized in that: The implementation method for step three is as follows: The homogeneous form of the transformation matrix from the local coordinate system to the global coordinate system is as follows: In the formula, Rot and B represent rotation and translation transformations, respectively, determined by their relative spatial positions; in the global coordinate system, the three-dimensional coordinates of the nth frame linear point cloud are calculated as follows: In the formula, Let ϕ represent the distance vector, and let ϕ represent the field of view angle. m The angle between the m-th line beam and the central axis is represented, ranging from (-ϕ / 2, ϕ / 2). During relative motion, three-dimensional imaging is achieved by stitching together continuous linear point clouds in the x-direction, with a fixed total number of stitched frames n. s The current frame is selected as the upper boundary of the image, and the corrected coordinates in the x and z directions are as follows: In the formula, v m f represents the speed of the mobile platform. s ψ represents the sampling rate. a The forward tilt angle is the angle between the platform's direction of movement and the horizontal plane. Equations (16) and (17) constitute a geometric measurement model, which is coupled with the laser power model and the smoke diffusion model to establish a short-range linear scanning laser point cloud imaging model under smoke interference environment.

9. The method as described in claim 1, characterized in that: Step four is implemented by establishing a multiphysics simulation application in a 3D virtual environment, configuring scene elements and optical attribute data in the 3D virtual environment according to the actual application scenario, and configuring simulation parameters according to different smoke interference conditions and measurement conditions; using the parallel ray tracing method to solve the geometric measurement model obtained in step three to obtain spatial relative coordinate parameters, and simultaneously using the particle dynamics solution method to calculate the smoke diffusion model obtained in step two to obtain the wind field and particle state parameters; using the iterative radiative transfer calculation method to solve the laser power model obtained in step one to obtain simulated laser point cloud data; scene elements include terrain, vegetation, targets, and smoke interference sources.

10. The method as described in claim 1, characterized in that: In step five, the similarity between the simulated laser point cloud data and the measured laser point cloud data is comprehensively evaluated using the chamfer distance (CD), Hausdorff distance (HD), and bulldozing distance (EMD). When the CD, HD, and EMD values ​​all meet the preset validity requirements, an effective simulation model is obtained.

Images