Laser radar ranging process simulation method suitable for complex scene

Abstract

The invention discloses a laser radar ranging process simulation method suitable for a complex scene. The method belongs to the field of full-waveform laser radar ranging simulation. The method comprises the following steps: setting initial parameters, and calculating to obtain a backscatter cross section function of a complex scene; an ideal echo signal is obtained through convolution; generating random noise and superposing the random noise with the ideal echo signal; filtering the signal on which the noise is superposed; carrying out adaptive threshold detection on the filtered signal to judge whether an effective signal exists or not; determining an echo pulse arrival time by using a standard waveform fitting method, and calculating to obtain a distance measurement result; and carrying out multiple detection to obtain laser radar ranging statistical characteristics. According to the invention, the Monte Carlo ray tracing algorithm is introduced to realize the calculation of the backscattering cross section function of the complex scene, the transmission process of the light beam in the complex scene is truly restored, and the adaptive threshold detection method is adopted to effectively reduce the false alarm probability of the laser radar system and improve the detection probability.

Description

Technical Field

[0001] The present invention belongs to a full-waveform laser radar ranging simulation technology, in particular to a laser radar ranging process simulation method suitable for complex scenes. Background Art

[0002] LiDAR technology is an active remote sensing technology that measures the position and surface features of a target by emitting laser pulses and receiving echo signals reflected by the target. Due to its advantages such as short wavelength, high intensity and small divergence angle, it has higher spatial resolution and detection capabilities than traditional microwave and millimeter wave radars. It has been widely used in geographic surveying and mapping fields such as ground object elevation measurement, topography and geomorphology measurement, and forestry resource survey. Vegetation target detection has always been a difficult problem in this field due to the diversity of its appearance characteristics and the complexity of its spatial structure. How to describe the interaction between the light beam and the vegetation target and the transmission process of the light beam in complex scenes such as vegetation, and obtain the full waveform echo signal of the LiDAR has become the key to solving this problem. Through research, it was found that there are descriptions of the calculation methods of LiDAR echo signals in existing reports, among which the analytical method is widely used. However, for complex scenes such as vegetation targets, it is difficult to describe the specific form of the target surface. In addition, the interaction between the light beam and the target is very complex, including a series of scattering and absorption processes, and it is impossible to directly obtain the analytical solution of its echo signal. To this end, the present invention introduces a backward ray tracing algorithm, which effectively restores the transmission process of the light beam in the complex scene, and calculates the lidar echo signal in the complex scene. On this basis, it can analyze the statistical characteristics of lidar ranging in different complex scenes, which has important research significance for improving the lidar's ability to detect complex scenes. Summary of the invention

[0003] The present invention proposes a laser radar ranging process simulation method suitable for complex scenes, which can effectively obtain full-waveform laser radar echo signals in complex scenes, especially in different vegetation scenes, and perform ranging simulation to obtain the ranging statistical characteristics of the laser radar under different conditions.

[0004] The technical solution to achieve the purpose of the present invention is: a laser radar ranging process simulation method suitable for complex scenes, comprising the following steps:

[0005] Step 1: Set the initial parameters of LiDAR ranging;

[0006] Step 2: Calculate the backscattering cross-section function of the complex scene;

[0007] Step 3: Convolve the backscattering cross-section function with the laser pulse time distribution to obtain the ideal echo signal;

[0008] Step 4: Generate random noise and superimpose it with the ideal echo signal to obtain a noisy signal;

[0009] Step 5: Filter and reduce noise on the noisy signal to obtain a filtered and denoised echo signal;

[0010] Step 6: Perform adaptive threshold detection on the echo signal after filtering and noise reduction to obtain a valid signal;

[0011] Step 7: Determine the arrival time of the echo in the effective signal using the standard waveform fitting method;

[0012] Step 8: Calculate and obtain the single distance measurement result;

[0013] Step 9: Repeat steps 1 to 8 for a set number of times, and obtain the laser radar ranging statistical characteristic output based on multiple ranging results.

[0014] Preferably, the initial parameters of the laser radar ranging set in step 1 include: the number of detections N, the detection scene, and the laser radar system parameters. The laser radar system parameters include the laser pulse waveform, pulse width, power, and time distribution characteristics. At the same time, the coordinates of the scene objects and the laser radar are set.

[0015] Preferably, the specific method of calculating the backscattering cross-section function of the complex scene in step 2 is:

[0016] The light beam is emitted from the laser radar, enters the scene and hits r 0 ,r 1 ,r 2 Click finally to enter the LiDAR receiver;

[0017] The vertices of the path traced by the beam in the scene are r i ∈{r 0 ,r 1 ,r 2 ...,r n}, the path length of the laser light is expressed as Calculate the flight time t of the laser beam based on its path length p , specifically:

[0018]

[0019] Calculate the laser power received by the lidar system as:

[0020]

[0021] Among them, φ lidar is the total power received by the LiDAR system, D is the set of paths of all sampled rays, r is the complete path connecting the LiDAR system light source and receiver, Indicates the path The power contribution of is the product measure of the path areas, n is the number of intersections between the light and the scene. Substituting the flight time calculation expression into the laser power received by the lidar system, the backscattering cross-section function of the scene target is obtained:

[0022]

[0023] The backscattering cross-section function of the scene target is written in the form of Monte Carlo integration:

[0024]

[0025] where w k represents the weight of the kth optical path’s contribution to the laser power. When N rays are emitted, unbiased estimation of the laser radar return power is achieved when w = 1 / N, where The expression of is obtained from the light energy radiation transfer equation, and the final backscattering cross-section function of the scene target is:

[0026]

[0027] Preferably, the ideal echo signal l(t) in step 3 is specifically:

[0028]

[0029] Where S(t) is the laser pulse time distribution, Φ lidar is the backscattering cross section function.

[0030] Preferably, the random noise generated in step 4 is Gaussian noise simulating sunlight background noise and noise in the circuit.

[0031] Preferably, the filtering and noise reduction in step 5 adopts a matched filtering method.

[0032] Preferably, in step 6, the specific method of performing adaptive threshold detection on the filtered echo signal to obtain a valid signal is:

[0033] The noise level is estimated to obtain the noise root mean square and the mean, and the adaptive threshold is set according to the noise root mean square and the mean: th = 3m + σ, where th is the adaptive threshold, σ is the noise root mean square, and m is the noise mean.

[0034] Compare the peak value of the echo signal after filtering with ath. When the signal peak value is greater than ath, it is judged as a valid signal, and a is the judgment coefficient.

[0035] Preferably, the specific method of determining the arrival time of the echo in the effective signal by using the standard waveform fitting method in step 7 is:

[0036] The laser radar echo signal is reconstructed using the standard waveform fitting method, which is specifically expressed as:

[0037]

[0038] where f i (t) is the fitted standard waveform, n(t) represents Gaussian white noise, where M represents the number of reconstructed standard echo waveforms. When i=1, f 1 (t) is the first echo. When i=M, f N (t) is the last echo;

[0039] The time t corresponding to the first echo peak 2 as the echo arrival time.

[0040] Preferably, the specific method of calculating the single distance measurement result in step 8 is:

[0041] The laser radar pulse emission time t1 is taken as the echo start time, and the time t2 corresponding to the first echo peak position is taken as the echo arrival time, then the single ranging result d is specifically:

[0042]

[0043] Where c is the speed of light.

[0044] Preferably, the laser radar ranging statistical characteristics include the mean, variance and distribution of the ranging data

[0045] Compared with the prior art, the present invention has the following significant advantages: the present invention introduces a ray tracing algorithm, describes the interaction between the light beam and the scene target, deeply restores the transmission process of the light beam in a complex scene, and uses the radiation transmission principle to calculate the laser radar echo signal in a complex scene. The present invention overcomes the problem that the commonly used analytical method is difficult to accurately describe the interaction between the light beam and the target and is difficult to solve the analytical solution of the laser radar echo signal in a complex scene. On this basis, the adaptive threshold detection is used in combination with the standard waveform fitting method to obtain the statistical characteristics of the laser radar ranging in complex scenes, which is of great significance to improving the ranging performance of the laser radar when dealing with complex scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 The present invention is a flow chart of a laser radar ranging process simulation method suitable for complex scenes.

[0047] Figure 2 This is a histogram of the statistical characteristics of the lidar ranging obtained using the standard waveform fitting method. DETAILED DESCRIPTION

[0048] like Figure 1 As shown, a laser radar ranging process simulation method suitable for complex scenes mainly includes the following steps:

[0049] Step 1: Set the initial parameters of the laser radar ranging, including: detection times N, detection scene, laser radar system parameters, system parameters include laser pulse waveform, pulse width, power and time distribution characteristics, etc., and set the coordinates of the scene objects and the laser radar;

[0050] As a specific implementation method, the initial parameters in step 1 are set as follows: the number of ranging times N is set to 10000, the spatial distribution and temporal distribution characteristics of the laser pulse are both set to Gaussian functions, the wavelength is 905nm, the power is 70mv, the pulse width is 20ns, the scene is set to a single tree with different canopy shapes, leaf area indexes, and crown height ratios, and the laser radar detection height is 150m, located directly above the detection scene;

[0051] Step 2: Calculate the backscattering cross-section function of the complex scene;

[0052] As a specific implementation method, the backscattering cross-section function in step 2 is the impulse response of the target surface. The present invention uses the Monte Carlo ray tracing algorithm to describe the interaction between the light beam and the scene, and realizes the calculation of the pulse flight time and the return power, thereby obtaining the power distribution function over time, that is, the impulse response of the target surface. The transmission of light in the scene satisfies the light energy transmission equation, and the specific expression is as follows:

[0053] L(r,ω o )=L e (r,ω o )+∫f(r,ω i ,ω o )L i (r,-ω i )|cos(θ i )|dω i

[0054] This equation shows the radiance L(r,ω o ) is equal to the radiance L emitted by the surface element e (r,ω o ) plus the Yuan Dynasty ω o Directional diffuse radiation brightness L i (r,-ω i ). In the formula, θ i is the angle between the incident light and the normal vector of the surface element of the object, f(r,ω i ,ω o ) represents the bidirectional reflectance distribution function of the object surface.

[0055] Furthermore, the light beam is emitted from the laser radar, enters the scene and hits r 0,r 1 ,r 2 The point finally enters the lidar receiver. Without loss of generality, the vertices of the path traced by the beam in the scene are r i ∈{r 0 ,r 1 ,r 2 ...,r n}, then the path length of the laser light can be expressed as Then the flight time of the light is t p It can be written as follows:

[0056]

[0057] Furthermore, the laser power received by the lidar system can be characterized as the continuous integral of the power contribution of each part of the laser transmission path, as shown in the following formula:

[0058]

[0059] where φ lidar is the total power received by the LiDAR system. D is the set of paths of all sampled rays, r is the complete path connecting the LiDAR system light source and receiver, Indicates the path The power contribution of It can be understood as a product measure of the path area. Substituting the flight time calculation expression into it, we can get the backscattering cross-section function of the scene target:

[0060]

[0061] The above formula is difficult to solve directly, so it can be written as Monte Carlo integration:

[0062]

[0063] Among them, w k represents the weight of the kth optical path’s contribution to the laser power. When N rays are emitted, an unbiased estimate of the laser radar return power can be achieved when w = 1 / N. The expression of is obtained from the light energy radiation transfer equation:

[0064]

[0065] Step 3: Convolve the backscattering cross-section function with the laser pulse time distribution to obtain the ideal echo signal;

[0066] As a specific implementation method, the convolution in step 3 takes into account the time distribution characteristics of the laser pulse. The ideal echo signal l(t) is the convolution of the time distribution of the laser power and the time distribution of the emitted laser pulse, where the laser pulse time distribution S(t) is a Gaussian distribution. The ideal echo signal is specifically:

[0067]

[0068] Step 4: Generate random noise and superimpose it with the ideal echo signal to obtain a noisy signal;

[0069] As a specific implementation method, the noise generated in step 4 is Gaussian noise that can simulate sunlight background noise and noise in the circuit, and is superimposed with the ideal echo signal obtained in step 3.

[0070] Step 5: Filter and reduce noise on the noisy signal;

[0071] As a specific implementation method, the filtering and noise reduction in step 5 adopts a matched filtering method.

[0072] Step 6: Perform adaptive threshold detection on the filtered echo signal to obtain a valid signal;

[0073] As a specific implementation method, in step 6, adaptive threshold detection is performed on the filtered echo signal. First, the noise level is estimated to obtain the noise root mean square and standard deviation, and then the adaptive threshold is set according to the noise root mean square and mean. Finally, the signal peak value is compared with the value obtained by multiplying the threshold by the judgment coefficient to determine whether the signal is a valid signal. The specific process is as follows:

[0074] First, the noise level is estimated to obtain the noise root mean square σ and the mean m, and then the adaptive threshold th is set to 3 times the noise root mean square plus the mean, that is:

[0075] th=3m+σ

[0076] Finally, the judgment coefficient is taken as 3. When the signal peak value is greater than 3th, the signal is considered to be a valid signal.

[0077] Step 7: Determine the arrival time of the echo in the effective signal using the standard waveform fitting method;

[0078] As a specific implementation method, in step 7, a standard waveform fitting method is used to reconstruct the effective laser radar echo signal, and the echo arrival time is determined from the first echo. Furthermore, the standard waveform selects a Gaussian function waveform that is consistent with the spatial distribution characteristics of the laser radar, and its specific form is:

[0079]

[0080] Among them A i,μ i ,ω i They represent the amplitude, mean and standard deviation of the i-th Gaussian component respectively. Furthermore, the effective lidar echo signal P(t) reconstructed by the standard waveform fitting method can be expressed as follows:

[0081]

[0082] where f i (t) is the fitted standard waveform, n(t) represents Gaussian white noise, where M represents the number of reconstructed standard echo waveforms. When i=1, f 1 (t) is the first echo. When i=M, f N (t) is the last echo. By fitting this method, the lidar echo signal can be reconstructed into the superposition of the first echo, the middle echo and the last echo.

[0083] Step 8: Calculate the distance measurement result;

[0084] As a specific implementation method, in step 8, the laser radar pulse emission time t 1 As the echo start time, the time corresponding to the first echo peak is t 2 As the echo arrival time, the laser radar ranging formula is:

[0085]

[0086] In the formula, c is the speed of light, specifically 299792458 m / s.

[0087] Step 9: Perform multiple tests to obtain the statistical characteristics of the laser radar ranging output.

[0088] As a specific implementation method, step 9 repeats the measurement N times to obtain the statistical characteristics of the laser radar ranging. Furthermore, the description of the statistical characteristics includes the mean, variance and distribution of the ranging data, wherein the distribution of the ranging data often obeys a normal distribution.

[0089] like Figure 2 As shown in Figure 1, a schematic diagram of the interaction between the light beam and the scene is restored using the ray tracing method. The light is emitted from the laser radar light source, enters the scene and reaches r 0 ,r 1 ,r 2 , and finally enters the receiver, completing a complete light path sampling.

Claims

1. A laser radar ranging process simulation method suitable for complex scenes, characterized in that: The following steps are involved: Step 1: Set the initial parameters of LiDAR ranging; Step 2: Calculate the backscattering cross-section function of the complex scene; Step 3: Convolve the backscattering cross-section function with the laser pulse time distribution to obtain the ideal echo signal; Step 4: Generate random noise and superimpose it with the ideal echo signal to obtain a noisy signal; Step 5: Filter and reduce noise on the noisy signal to obtain a filtered and denoised echo signal; Step 6: Perform adaptive threshold detection on the echo signal after filtering and noise reduction to obtain a valid signal; Step 7: Determine the arrival time of the echo in the effective signal using the standard waveform fitting method; Step 8: Calculate and obtain the single distance measurement result; Step 9: Repeat steps 1 to 8 for a set number of times, and obtain the laser radar ranging statistical characteristic output based on multiple ranging results.

2. The laser radar ranging process simulation method applicable to complex scenes according to claim 1 is characterized in that: The initial parameters of the laser radar ranging set in step 1 include: number of detections N, detection scene, and laser radar system parameters. The laser radar system parameters include laser pulse waveform, pulse width, power, and time distribution characteristics. At the same time, the coordinates of the scene objects and the laser radar are set.

3. The laser radar ranging process simulation method applicable to complex scenes according to claim 1 is characterized in that: The specific method for calculating the backscattering cross-section function of the complex scene in step 2 is: The light beam is emitted from the LiDAR, enters the scene and hits points r0, r1, and r2 in sequence before entering the LiDAR receiver. The vertices of the path traced by the beam in the scene are r i ∈{r0,r1,r2...,r n }, the path length of the laser light is expressed as Calculate the flight time t of the laser beam based on its path length p , specifically: Calculate the laser power received by the lidar system as: Among them, φ lidar is the total power received by the LiDAR system, D is the set of paths of all sampled rays, r is the complete path connecting the LiDAR system light source and receiver, Indicates the path The power contribution of is the product measure of the path areas, n is the number of intersections between the light and the scene. Substituting the flight time calculation expression into the laser power received by the lidar system, the backscattering cross-section function of the scene target is obtained: The backscattering cross-section function of the scene target is written in the form of Monte Carlo integration: where w k represents the weight of the kth optical path’s contribution to the laser power. When N rays are emitted, unbiased estimation of the laser radar return power is achieved when w = 1 / N, where The expression of is obtained from the light energy radiation transfer equation, and the final backscattering cross-section function of the scene target is:

4. The laser radar ranging process simulation method applicable to complex scenes according to claim 1 is characterized in that: The ideal echo signal l(t) in step 3 is specifically: Where S(t) is the laser pulse time distribution, Φ lidar is the backscattering cross section function.

5. The laser radar ranging process simulation method applicable to complex scenes according to claim 1 is characterized in that: The random noise generated in step 4 is Gaussian noise that simulates the background noise of sunlight and the noise in the circuit.

6. The laser radar ranging process simulation method applicable to complex scenes according to claim 1 is characterized in that: In step 5, the filtering and noise reduction adopts the matched filtering method.

7. The laser radar ranging process simulation method applicable to complex scenes according to claim 1 is characterized in that: In step 6, the specific method of performing adaptive threshold detection on the filtered echo signal to obtain a valid signal is as follows: The noise level is estimated to obtain the noise root mean square and the mean, and the adaptive threshold is set according to the noise root mean square and the mean: th = 3m + σ, where th is the adaptive threshold, σ is the noise root mean square, and m is the noise mean. Compare the peak value of the echo signal after filtering with ath. When the signal peak value is greater than ath, it is judged as a valid signal, and a is the judgment coefficient.

8. The laser radar ranging process simulation method applicable to complex scenes according to claim 1 is characterized in that: The specific method of determining the echo arrival time in the effective signal using the standard waveform fitting method in step 7 is: The laser radar echo signal is reconstructed using the standard waveform fitting method, which is specifically expressed as: where f i (t) is the fitted standard waveform, n(t) is Gaussian white noise, M is the number of reconstructed standard echo waveforms, when i = 1, f1(t) is the first echo, when i = M, f N (t) is the last echo; The time t2 corresponding to the first echo peak is taken as the echo arrival time.

9. The laser radar ranging process simulation method applicable to complex scenes according to claim 1, characterized in that: The specific method of calculating the single distance measurement result in step 8 is: The laser radar pulse emission time t1 is taken as the echo start time, and the time t2 corresponding to the first echo peak position is taken as the echo arrival time, then the single ranging result d is specifically: Where c is the speed of light.

10. The laser radar ranging process simulation method applicable to complex scenes according to claim 1, characterized in that: The statistical characteristics of LiDAR ranging include the mean, variance and distribution of ranging data.

Images