Lidar aliasing waveform decomposition method based on waveform features

Abstract

The application provides a laser radar aliasing waveform decomposition method based on waveform characteristics, constructs a waveform mathematical model according to seed light characteristics of a laser radar, combines an LM curve fitting algorithm, decomposes aliasing echoes into multiple single echo data, and achieves the purpose of accurately extracting multi-target ranging information. The technical basis of the application is a laser radar full waveform detection technology, that is, a laser radar detection system can sample analog electric signals converted from optical signals into digital waveform signals, and the processing object of the application is a waveform sampling point sequence of seed light and return light detected by the laser radar.

Description

Technical Field

[0001] This invention relates to the field of visibility meters, and in particular to a method for decomposing aliased waveforms in lidar based on waveform features. Background Technology

[0002] LiDAR is an active detection technology widely used in remote sensing, surveying, and other fields. With the integration and miniaturization of laser sources and continuous advancements in detection technology, LiDAR technology has made significant progress in recent years. In fields such as driver assistance, robotics, security monitoring, and 3D measurement, LiDAR has become one of the common technical tools.

[0003] The performance requirements for lidar in the surveying and mapping industry differ from those in other technological fields, demanding higher standards for core technical indicators such as point frequency, ranging, accuracy, and penetration. When facing long-range, complex target scenarios, the echo signals from lidar are often complex. Taking pulsed lidar as an example, while the seed pulse signal is a relatively ideal pulse waveform, the echo signal is influenced by factors such as the target's spatial distribution characteristics, geometric characteristics, and reflection characteristics, resulting in complex and diverse shapes. This can easily lead to special echo signals such as multiple echoes, aliased echoes, and irregular echoes. These signals affect the lidar's ranging and intensity information calculations, causing noise in the lidar point cloud data and impacting measurement accuracy. Simultaneously, some useful signals are submerged by aliased and irregular echoes, resulting in incomplete target detection and affecting the lidar's penetration power. Summary of the Invention

[0004] The main objective of this invention is to provide a method for decomposing aliased waveforms in lidar based on waveform characteristics. This method addresses the issue that while the seed light pulse signal is a relatively ideal pulse waveform, the return light signal is often complex and diverse in shape due to factors such as the spatial distribution characteristics, geometric characteristics, and reflection characteristics of the target. This can easily lead to special echo signals such as multiple echoes, aliased echoes, and irregular echoes. These types of signals can affect lidar ranging and intensity information calculation, resulting in noise in the lidar point cloud data and affecting measurement accuracy.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by this invention is: a method for decomposing aliased waveforms in lidar based on waveform features. This method constructs a waveform mathematical model based on the seed light characteristics of the lidar, and combines it with an LM curve fitting algorithm to decompose the aliased echoes into multiple single-echo data, thereby achieving the goal of accurately extracting multi-target ranging information. The method includes:

[0006] S1. Obtain waveform sampling sequence: Obtain the waveform sampling sequence and sampling point time sequence of the lidar seed light / return light after AD sampling. The time interval between adjacent sampling points is a known fixed value.

[0007] S2. Calculate the noise floor value: Calculate the noise floor value based on the signal within a certain interval to the left and right of the waveform data, and take the average signal within the noise interval as the noise floor value, so that it changes dynamically with the waveform data.

[0008] S3. Waveform data smoothing: Algorithms such as mean filtering are used to smooth the original waveform data;

[0009] S4. Determine the effective waveform range: Take the smoothed waveform data. The starting point is the point where the value is greater than the noise floor value and the value increases for n consecutive times. The ending point is the point where the value decreases for n consecutive times. n is related to the characteristics of the lidar. The waveform sequence that participates in the subsequent calculation is the effective waveform range.

[0010] S5. Set the basic function model for waveform decomposition: The single pulse waveform function model is determined by the seed light emission characteristics and the response characteristics of the lidar photoelectric detection system. The practical model of this invention is a specific form. The parameters to be fitted characterize the pulse waveform intensity, peak position, and pulse width.

[0011] S6. Initial determination of the number of pulse waveforms: Determine the initial number of waveforms based on the pulse waveform sequence, and set the aliasing waveform function to be the superposition of the corresponding number of basic functions;

[0012] S7. Set the initial values ​​and search range of the fitting parameters: Determine the initial values ​​of the variables based on the waveform peak value, position, and seed light pulse width, and set the initial values ​​of the pulse width parameter and the search range of the fitting parameters.

[0013] S8. Using the LM algorithm for curve fitting: Use the LM algorithm to perform curve fitting and calculate the optimal parameter estimate;

[0014] S9. Verify the fitting accuracy: Substitute the fitting parameters into the model to calculate the fitted waveform sequence, and calculate the deviation by comparing it with the original waveform sequence. If the predetermined index is not met, return to step S6 to increase the number of pulse functions to be fitted. If the index is met, output the fitting result and determine the distance measurement and reflection intensity information of the reflecting target.

[0015] In the preferred scheme, steps S1-S2 are as follows:

[0016] S11. Let the data sequence output after sampling a pulse signal by an AD converter be:

[0017] ;

[0018] The corresponding sampling point time series is:

[0019] ;

[0020] The time interval between adjacent sampling points is the system AD sampling period. , is a known fixed value:

[0021] ;

[0022] S21. Calculate the noise floor value corresponding to the waveform data based on the signal values ​​within a certain range to the left and right of the waveform data. Although the noise floor of the echo signal of a fixed lidar detection system is generally at a fixed level, the noise floor of the waveform sequence is generally dynamically changing due to factors such as the influence of ambient light and the temperature characteristics of the hardware system itself. The noise floor calculation method described in this invention can estimate the noise floor of the current waveform data in real time, so that the noise floor value changes dynamically with the waveform data.

[0023] In the preferred scheme, steps S3-S4 are as follows:

[0024] S31. Waveform Data Smoothing: The original waveform data is smoothed. Smoothing methods can include filtering algorithms such as mean filtering. The smoothed waveform data sequence is as follows:

[0025] ;

[0026] S41. Determine the effective waveform range: The effective waveform is the data sequence that participates in the subsequent waveform decomposition calculation. The method for determining it is as follows: Take the smoothed waveform data, and the point where the value is greater than the noise floor value and the value increases for n consecutive times is the starting point of the effective waveform; the point where the value is greater than the noise floor value and the value decreases for n consecutive times is the ending point of the effective waveform; the value of n is related to the characteristics of the lidar. In the description of the claims, this invention uses n=4 as an example for explanation.

[0027] Starting point k Sure:

[0028] ;

[0029] End point j Sure:

[0030] ;

[0031] The waveform sequence used in subsequent calculations is as follows:

[0032] .

[0033] In the preferred embodiment, step S5 specifically involves the following steps:

[0034] S51. Set up the basic function model for waveform decomposition: According to the ranging principle of pulsed lidar, the seed light pulse waveform still retains the waveform characteristics of the seed light after being reflected by the target. However, due to the influence of reflection from adjacent targets, the backlight received by lidar is usually a superposition of multiple pulse signals. Waveform decomposition is to extract the multiple superimposed pulse waveforms and then obtain the ranging and intensity information of multiple adjacent targets.

[0035] The waveform function model of a single pulse is the basic function model of waveform decomposition, which can be determined by the emission characteristics of the seed light and the response characteristics of the lidar photoelectric detection system. The Gaussian function is an ideal lidar seed light pulse model, which is a symmetrical function model. The actual lidar seed light pulse waveform is affected by the characteristics of the laser and the photoelectric response characteristics of the lidar, and is deformed on the basis of the Gaussian model, with the rising and falling edges of the waveform being asymmetrical.

[0036] The basic model of the waveform decomposition function is:

[0037] .

[0038] In the preferred embodiment, step S6 specifically involves the following steps:

[0039] S61. Initial determination of pulse waveform quantity: Aliased waveforms are usually composed of multiple basic pulse waveforms superimposed, based on the pulse waveform sequence. Determining the initial number of waveforms facilitates faster convergence of waveform fitting calculations. The aliased waveform function to be fitted is then a superposition of N basic functions. Let:

[0040] ;

[0041] The initial form of the aliased waveform function is:

[0042] .

[0043] In the preferred embodiment, step S7 specifically involves the following steps:

[0044] S71. Set the initial values ​​and search range of waveform decomposition fitting parameters: As described in step S61, if the number of pulse waveforms initially determined is N, then it is necessary to determine the initial values ​​and search range of multiple variables, expressed by the formula:

[0045] ;

[0046] The initial values ​​of the variables are determined by the peak magnitude, peak position, and seed light pulse width of the waveform, with the pulse width parameter being... c The initial value can be set to the same pulse width parameter value as the seed light. :

[0047] ;

[0048] The search range for the fitting parameters is:

[0049] .

[0050] In the preferred embodiment, step S8 specifically involves the following steps:

[0051] S81. Using the LM algorithm to fit the aliased waveform curve: Based on the aforementioned function model of the aliased waveform... The initial values ​​of variables and the search range of variables are determined. The (LM) Levenberg-Marquardt algorithm is used for curve fitting to calculate the optimal parameter estimate under the current state.

[0052] In the preferred embodiment, step S9 specifically involves the following steps:

[0053] S91. Verify the fitting accuracy: Substitute the optimal parameter estimates obtained from the fitting algorithm described in step 81 into the aliasing waveform function model. Calculate the fitted waveform sequence:

[0054] ;

[0055] With the original aliased waveform sequence The difference is calculated to obtain the deviation between the fitted function and the original waveform sequence. :

[0056] ;

[0057] like If the predetermined target is not met, return to step 6 and increase the number of impulse functions to be fitted by one, i.e.:

[0058] ;

[0059] Repeat steps S6-S9;

[0060] like Once the predetermined target is reached, the fitting result is output, and the ranging and reflection intensity information of the reflecting target can be determined based on the multiple pulse waveforms obtained from the decomposition.

[0061] In the preferred scheme, step S8 using the LM algorithm specifically involves the following steps:

[0062] Based on the aliased waveform function model, initial variable values ​​(and variable search range) determined in step S71 above, set the initial parameter vector. The quantity is determined based on the number of pulse waveforms initially determined;

[0063] For a given waveform sampling point sequence and the corresponding sampling point time series Based on the current parameter vector Calculate model predictions ;

[0064] Calculate the residual vector ,in Indicates the sampling point number;

[0065] Calculate the Jacobian matrix Its elements For the model function with respect to the first The parameter in the first... Partial derivatives at each sampling point;

[0066] For the basic model of the waveform decomposition function in this invention, respectively calculate the values ​​related to... The partial derivatives;

[0067] Calculate the gradient vector ,in It is the transpose of the Jacobian matrix;

[0068] Calculate the approximate Hessian matrix ;

[0069] Calculate the parameter update amount based on the LM algorithm formula. :

[0070] First, calculate the damping factor. ;

[0071] calculate ,in It is the identity matrix;

[0072] Then calculate ;

[0073] Update parameter vector ;

[0074] Calculate the updated residual vector and the new objective function value;

[0075] Compare the new objective function value with the objective function value from the previous iteration. If the convergence condition is met, stop the iteration. The current parameter vector... This is the optimal parameter estimate;

[0076] Otherwise, return to the above steps and continue iterating. In subsequent iterations, the damping factor may need to be adjusted as needed. .

[0077] This invention provides a method for decomposing aliased waveforms in lidar based on waveform features. According to the seed light characteristics of lidar, a waveform mathematical model is constructed, and combined with the LM curve fitting algorithm, the aliased echo is decomposed into multiple single echo data, thereby achieving the purpose of accurately extracting multi-target ranging information.

[0078] The technical basis of this invention is lidar full waveform detection technology, that is, lidar detection system can sample analog electrical signals converted from optical signals into digital waveform signals. The processing object of this invention is the waveform sampling point sequence of seed light and return light detected by lidar.

[0079] By constructing a waveform mathematical model and combining it with the LM curve fitting algorithm, complex aliased echoes can be decomposed into multiple single echo data. This method can effectively extract multi-target ranging information when dealing with special echo signals such as multiple echoes and aliased echoes generated in long-range, complex target scenarios. For example, in actual mapping scenarios, echo signals reflected from multiple targets are often superimposed. This method can accurately separate the echoes of each target, thereby accurately obtaining the distance to each target and avoiding ranging errors caused by aliasing.

[0080] The complex morphology of lidar backlight signals, influenced by various factors, can introduce noise into point cloud data. This decomposition method effectively processes such signals, reducing the negative impact of noise on measurement accuracy. For example, in complex environments, noise signals generated by ambient light interference and changes in target reflectivity can be processed by this method to more accurately reflect the true target position, thus improving overall measurement accuracy.

[0081] When some useful signals are submerged by aliased or irregular echoes, the detection of the target becomes incomplete. This technology can decompose the aliased waveform, allowing the originally submerged useful signals to be recovered, thereby enabling complete target detection and improving the lidar's ability to detect targets in complex scenes. It's like obtaining target information more comprehensively and accurately when penetrating complex obstacles.

[0082] By calculating the noise floor value of the waveform sequence, the dynamic changes in noise floor caused by factors such as ambient light and the temperature characteristics of the hardware system itself are taken into account, allowing the noise floor value to be adjusted in real time according to the waveform data. This helps to more accurately identify effective signals under different environmental conditions, further improving the accuracy of waveform decomposition and ranging calculations. Attached Figure Description

[0083] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0084] Figure 1 This is a schematic diagram of the laser radar waveform sampling sequence of the present invention;

[0085] Figure 2 This is a schematic diagram of the noise floor of the lidar waveform of the present invention;

[0086] Figure 3 This is a schematic diagram of a typical seed light pulse waveform for lidar according to the present invention;

[0087] Figure 4 This is a schematic diagram illustrating the initial determination of the number of aliased waveforms in this invention;

[0088] Figure 5 This is a schematic diagram illustrating the selection of initial values ​​for the parameters to be fitted in the aliased waveform according to the present invention; Detailed Implementation

[0089] Example 1

[0090] like Figures 1-5 As shown, a method for decomposing aliased waveforms in lidar based on waveform features is proposed. Based on the seed light characteristics of the lidar, a mathematical model of the waveform is constructed. Combined with the LM curve fitting algorithm, the aliased echoes are decomposed into multiple single-echo data, achieving the goal of accurately extracting multi-target ranging information. This method includes:

[0091] S1. Obtain waveform sampling sequence: Obtain the waveform sampling sequence and sampling point time sequence of the lidar seed light / return light after AD sampling. The time interval between adjacent sampling points is a known fixed value.

[0092] S2. Calculate the noise floor value: Calculate the noise floor value based on the signal within a certain interval to the left and right of the waveform data, and take the average signal within the noise interval as the noise floor value, so that it changes dynamically with the waveform data.

[0093] S3. Waveform data smoothing: Algorithms such as mean filtering are used to smooth the original waveform data;

[0094] S4. Determine the effective waveform range: Take the smoothed waveform data. The starting point is the point where the value is greater than the noise floor value and the value increases for n consecutive times. The ending point is the point where the value decreases for n consecutive times. n is related to the characteristics of the lidar. The waveform sequence that participates in the subsequent calculation is the effective waveform range.

[0095] S5. Set the basic function model for waveform decomposition: The single pulse waveform function model is determined by the seed light emission characteristics and the response characteristics of the lidar photoelectric detection system. The practical model of this invention is a specific form. The parameters to be fitted characterize the pulse waveform intensity, peak position, and pulse width.

[0096] S6. Initial determination of the number of pulse waveforms: Determine the initial number of waveforms based on the pulse waveform sequence, and set the aliasing waveform function to be the superposition of the corresponding number of basic functions;

[0097] S7. Set the initial values ​​and search range of the fitting parameters: Determine the initial values ​​of the variables based on the waveform peak value, position, and seed light pulse width, and set the initial values ​​of the pulse width parameter and the search range of the fitting parameters.

[0098] S8. Using the LM algorithm for curve fitting: Use the LM algorithm to perform curve fitting and calculate the optimal parameter estimate;

[0099] S9. Verify the fitting accuracy: Substitute the fitting parameters into the model to calculate the fitted waveform sequence, and calculate the deviation by comparing it with the original waveform sequence. If the predetermined index is not met, return to step S6 to increase the number of pulse functions to be fitted. If the index is met, output the fitting result and determine the distance measurement and reflection intensity information of the reflecting target.

[0100] Example 2

[0101] Further explanation in conjunction with Example 1, such as Figure 1-5 The structure shown, steps S1-S2 are as follows:

[0102] S11. Let the data sequence output after sampling a pulse signal by an AD converter be:

[0103] ;

[0104] The corresponding sampling point time series is:

[0105] ;

[0106] The time interval between adjacent sampling points is the system AD sampling period. , is a known fixed value:

[0107]

[0108] as follows Figure 1 The image shows a schematic diagram of the waveform sequence obtained after a lidar samples a typical echo signal.

[0109] S21. Calculate the noise floor value corresponding to the waveform data based on the signal values ​​within a certain range to the left and right of the waveform data. Although the noise floor of the echo signal of a fixed lidar detection system is generally at a fixed level, the noise floor of the waveform sequence is generally dynamically changing due to factors such as the influence of ambient light and the temperature characteristics of the hardware system itself. The noise floor calculation method described in this invention can estimate the noise floor of the current waveform data in real time, so that the noise floor value changes dynamically with the waveform data.

[0110] as follows Figure 2 The image shows waveform data containing noise floor. The signal within a certain range before and after the waveform interval is defined as the noise interval. The average signal value within the noise interval is taken as the noise floor value, denoted as . .

[0111] Example 3

[0112] Further explanation in conjunction with Example 1, such as Figure 1-5 The structure shown, steps S3-S4 are as follows:

[0113] S31. Waveform Data Smoothing: The original waveform data is smoothed. Smoothing methods can include filtering algorithms such as mean filtering. The smoothed waveform data sequence is as follows:

[0114] ;

[0115] S41. Determine the effective waveform range: The effective waveform is the data sequence that participates in the subsequent waveform decomposition calculation. The method for determining it is as follows: Take the smoothed waveform data, and the point where the value is greater than the noise floor value and the value increases for n consecutive times is the starting point of the effective waveform; the point where the value is greater than the noise floor value and the value decreases for n consecutive times is the ending point of the effective waveform; the value of n is related to the characteristics of the lidar. In the description of the claims, this invention uses n=4 as an example for explanation.

[0116] Starting point k Sure:

[0117] ;

[0118] End point j Sure:

[0119] ;

[0120] The waveform sequence used in subsequent calculations is as follows:

[0121] .

[0122] In the preferred embodiment, step S5 specifically involves the following steps:

[0123] S51. Set up the basic function model for waveform decomposition: According to the ranging principle of pulsed lidar, the seed light pulse waveform still retains the waveform characteristics of the seed light after being reflected by the target. However, due to the influence of reflection from adjacent targets, the backlight received by lidar is usually a superposition of multiple pulse signals. Waveform decomposition is to extract the multiple superimposed pulse waveforms and then obtain the ranging and intensity information of multiple adjacent targets.

[0124] The waveform function model of a single pulse is the basic function model for waveform decomposition, which can be jointly determined by the emission characteristics of the seed light and the response characteristics of the lidar photoelectric detection system. The Gaussian function is an ideal lidar seed light pulse model, a symmetrical function model. However, the actual lidar seed light pulse waveform is affected by the laser characteristics and the lidar photoelectric response characteristics, resulting in deformation based on the Gaussian model. The rising and falling edges of the waveform are asymmetrical, as shown in the figure below, where the rising edge of the pulse waveform is steeper and the falling edge is gentler.

[0125] The basic model of the waveform decomposition function in this invention is as follows:

[0126] .

[0127] Example 3

[0128] Further explanation in conjunction with Example 1, such as Figure 1-5 The structure shown has the following specific steps: steps S3-S4 are steps 4, and step S6 is a specific step:

[0129] S61. Aliased waveforms are usually composed of multiple basic pulse waveforms superimposed, based on the pulse waveform sequence. Determining the initial number of waveforms facilitates faster convergence of waveform fitting calculations, as follows: Figure 4 The waveform sequence shown can quickly detect three local peaks. Therefore, the initial number of waveforms is set to , and the aliased waveform function to be fitted is the superposition of three basic functions. Let:

[0130] ;

[0131] The initial form of the aliased waveform function is:

[0132] .

[0133] In the preferred embodiment, step S7 specifically involves the following steps:

[0134] S71. Set the initial values ​​and search range of waveform decomposition fitting parameters: As described in step S61, if the number of pulse waveforms initially determined is N, then the initial values ​​and search ranges of multiple variables need to be determined; if the number of pulse waveforms initially determined is 3, then the initial values ​​and search ranges of 9 variables need to be determined. The formula is expressed as:

[0135] ;

[0136] The initial values ​​of the variables are determined by the peak magnitude, peak position, and seed pulse width of the waveform, as described in step S6. Figure 5 As shown:

[0137] The initial values ​​of the variables are determined by the peak magnitude, peak position, and seed light pulse width of the waveform, with the pulse width parameter being... c The initial value can be set to the same pulse width parameter value as the seed light. :

[0138] ;

[0139] The search range for the fitting parameters is:

[0140] .

[0141] In the preferred embodiment, step S8 specifically involves the following steps:

[0142] S81. Using the LM algorithm to fit the aliased waveform curve: Based on the aforementioned function model of the aliased waveform... The initial values ​​of variables and the search range of variables are determined. The (LM) Levenberg-Marquardt algorithm is used for curve fitting to calculate the optimal parameter estimate under the current state.

[0143] In the preferred embodiment, step S9 specifically involves the following steps:

[0144] S91. Verify the fitting accuracy: Substitute the optimal parameter estimates obtained from the fitting algorithm described in step 81 into the aliasing waveform function model. Calculate the fitted waveform sequence:

[0145] ;

[0146] With the original aliased waveform sequence The difference is calculated to obtain the deviation between the fitted function and the original waveform sequence. :

[0147] ;

[0148] like If the predetermined target is not met, return to step 6 and increase the number of impulse functions to be fitted by one, i.e.:

[0149] ;

[0150] Repeat steps S6-S9;

[0151] like Once the predetermined target is reached, the fitting result is output, and the ranging and reflection intensity information of the reflecting target can be determined based on the multiple pulse waveforms obtained from the decomposition.

[0152] In the preferred scheme, step S8 using the LM algorithm specifically involves the following steps:

[0153] Based on the aliased waveform function model, initial variable values ​​(and variable search range) determined in step S71 above, set the initial parameter vector. The quantity is determined based on the number of pulse waveforms initially determined;

[0154] For a given waveform sampling point sequence and the corresponding sampling point time series Based on the current parameter vector Calculate model predictions ;

[0155] Calculate the residual vector ,in Indicates the sampling point number;

[0156] Calculate the Jacobian matrix Its elements For the model function with respect to the first The parameter in the first... Partial derivatives at each sampling point;

[0157] For the basic model of the waveform decomposition function in this invention, respectively calculate the values ​​related to... The partial derivatives;

[0158] Calculate the gradient vector ,in It is the transpose of the Jacobian matrix;

[0159] Calculate the approximate Hessian matrix ;

[0160] Calculate the parameter update amount based on the LM algorithm formula. :

[0161] First, calculate the damping factor. ;

[0162] calculate ,in It is the identity matrix;

[0163] Then calculate ;

[0164] Update parameter vector ;

[0165] Calculate the updated residual vector and the new objective function value;

[0166] Compare the new objective function value with the objective function value from the previous iteration. If the convergence condition is met, stop the iteration. The current parameter vector... This is the optimal parameter estimate;

[0167] Otherwise, return to the above steps and continue iterating. In subsequent iterations, the damping factor may need to be adjusted as needed. Through continuous iterative optimization using the above steps, the optimal estimated values ​​of the parameters that meet the conditions are finally calculated, thus achieving curve fitting of the aliased waveform.

[0168] The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for decomposing aliased waveforms in lidar based on waveform features, characterized by: Based on the seed light characteristics of lidar, a waveform mathematical model is constructed. Combined with the LM curve fitting algorithm, the aliased echo is decomposed into multiple single-echo data, achieving the goal of accurately extracting multi-target ranging information. This method includes: S1. Obtain the waveform sampling sequence: Obtain the waveform sampling sequence and sampling point time sequence of the lidar seed light / return light after AD sampling. The time interval between adjacent sampling points is a known fixed value. Given a pulse signal, obtain the data sequence output after AD sampling and the corresponding sampling point time sequence. The time interval between adjacent sampling points is the system AD sampling period. Given a known fixed value, the noise floor value corresponding to the waveform data is calculated based on the signal values ​​within a certain range to the left and right of the waveform data. Although the noise floor of the echo signal of a fixed lidar detection system is generally at a fixed level, the noise floor of the waveform sequence is generally dynamically changing due to the influence of ambient light and the temperature characteristics of the hardware system itself. The noise floor calculation method is used to estimate the noise floor of the current waveform data in real time, so that the noise floor value changes dynamically with the waveform data. S2. Calculate the noise floor value: Calculate the noise floor value based on the signal within a certain interval to the left and right of the waveform data, and take the average signal within the noise interval as the noise floor value, so that it changes dynamically with the waveform data. S3. Waveform data smoothing: The original waveform data is smoothed using a mean filtering algorithm. The smoothing method can be a mean filtering algorithm. The smoothed waveform data sequence is as follows: S4. Determine the effective waveform range: Take the smoothed waveform data. The starting point is the point where the value is greater than the noise floor value and the value increases for n consecutive times. The ending point is the point where the value decreases for n consecutive times. n is related to the characteristics of the lidar. The waveform sequence that participates in the subsequent calculation is the effective waveform range. S5. Set the basic function model for waveform decomposition: The single pulse waveform function model is determined by the seed light emission characteristics and the response characteristics of the lidar photoelectric detection system. The practical model is a specific form, and the parameters to be fitted characterize the pulse waveform intensity, peak position, and pulse width. S6. Initial determination of pulse waveform quantity: Determine the initial number of waveforms based on the pulse waveform sequence, and set the aliasing waveform function to be the superposition of the corresponding number of basic functions; S7. Set the initial values ​​and search range of the fitting parameters: Determine the initial values ​​of the variables based on the waveform peak value, position, and seed light pulse width, and set the initial values ​​of the pulse width parameter and the search range of the fitting parameters. S8. Using the LM algorithm for curve fitting: Use the LM algorithm to perform curve fitting and calculate the optimal parameter estimate; Using the LM algorithm to fit the aliased waveform curve: based on the aforementioned function model of the aliased waveform. Initial values ​​of variables and search range of variables are determined, and curve fitting is performed using the (LM) Levenberg-Marquardt algorithm to calculate the optimal parameter estimate under the current state. S9. Verify the fitting accuracy: Substitute the fitting parameters into the model to calculate the fitted waveform sequence, and calculate the deviation by comparing it with the original waveform sequence. If the predetermined index is not met, return to step S6 to increase the number of pulse functions to be fitted. If the index is met, output the fitting result and determine the distance measurement and reflection intensity information of the reflecting target. Step S5 is as follows: S51. Set up the basic function model for waveform decomposition: According to the ranging principle of pulsed lidar, the seed light pulse waveform still retains the waveform characteristics of the seed light after being reflected by the target. However, due to the influence of reflection from adjacent targets, the backlight received by lidar is usually a superposition of multiple pulse signals. Waveform decomposition is to extract the multiple superimposed pulse waveforms and then obtain the ranging and intensity information of multiple adjacent targets. The waveform function model of a single pulse is the basic function model of waveform decomposition, which can be determined by the emission characteristics of the seed light and the response characteristics of the lidar photoelectric detection system. The Gaussian function is an ideal lidar seed light pulse model, which is a symmetrical function model. The actual lidar seed light pulse waveform is affected by the characteristics of the laser and the photoelectric response characteristics of the lidar, and is deformed on the basis of the Gaussian model, with the rising and falling edges of the waveform being asymmetrical. The basic model of the waveform decomposition function is: 。 2. The method for decomposing aliased waveforms in lidar based on waveform features according to claim 1, characterized in that: step The specific steps for S1-S2 are as follows: S11. Let the data sequence output after sampling a pulse signal by an AD converter be: ； The corresponding sampling point time series is: ； The time interval between adjacent sampling points is the system AD sampling period. , is a known fixed value: 。 3. The lidar aliasing waveform decomposition method based on waveform features according to claim 1, characterized in that: The specific steps in steps S3-S4 are as follows: S31. Waveform data smoothing: The smoothed waveform data sequence is as follows: 。 4. The lidar aliasing waveform decomposition method based on waveform features according to claim 1, characterized in that: Step S6 is as follows: S61. Initial determination of pulse waveform quantity: Aliased waveforms are usually composed of multiple basic pulse waveforms superimposed, based on the pulse waveform sequence. Determining the initial number of waveforms facilitates faster convergence of waveform fitting calculations. The aliased waveform function to be fitted is then a superposition of N basic functions. Let: ； The initial form of the aliased waveform function is: 。 5. The lidar aliasing waveform decomposition method based on waveform features according to claim 4, characterized in that: Step S7 is as follows: S71. Set the initial values ​​and search range of waveform decomposition fitting parameters: As described in step S61, if the number of pulse waveforms initially determined is N, then it is necessary to determine the initial values ​​and search range of multiple variables, expressed by the formula: ； The initial values ​​of the variables are determined by the peak magnitude, peak position, and seed light pulse width of the waveform, with the pulse width parameter being... c The initial value can be set to the same pulse width parameter value as the seed light. : ； The search range for the fitting parameters is: 。 6. The lidar aliasing waveform decomposition method based on waveform features according to claim 1, characterized in that: Step S9 is as follows: S91. Verify the fitting accuracy: Substitute the optimal parameter estimates obtained from the fitting algorithm described in step 81 into the aliasing waveform function model. Calculate the fitted waveform sequence: ； With the original aliased waveform sequence The difference is calculated to obtain the deviation between the fitted function and the original waveform sequence. : ； like If the predetermined target is not met, return to step 6 and increase the number of impulse functions to be fitted by one, i.e.: ； Repeat steps S6-S9; like Once the predetermined target is reached, the fitting result is output, and the ranging and reflection intensity information of the reflecting target can be determined based on the multiple pulse waveforms obtained from the decomposition.

7. The lidar aliasing waveform decomposition method based on waveform features according to claim 1, characterized in that: Step S8, using the LM algorithm, involves the following steps: Based on the aliased waveform function model, initial variable values, and variable search range determined in step S71, set the initial parameter vector. The quantity is determined based on the number of pulse waveforms initially determined; For a given waveform sampling point sequence and the corresponding sampling point time series Based on the current parameter vector Calculate model predictions ; Calculate the residual vector ,in Indicates the sampling point number; Calculate the Jacobian matrix Its elements For the model function with respect to the first The parameter in the first... Partial derivatives at each sampling point; For the basic model of waveform decomposition function, calculate the following respectively: The partial derivatives; Calculate the gradient vector ,in It is the transpose of the Jacobian matrix; Calculate the approximate Hessian matrix ; Calculate the parameter update amount based on the LM algorithm formula. : First, calculate the damping factor. ; calculate ,in It is the identity matrix; Then calculate ; Update parameter vector ; Calculate the updated residual vector and the new objective function value; Compare the new objective function value with the objective function value from the previous iteration. If the convergence condition is met, stop the iteration. The current parameter vector... This is the optimal parameter estimate; Otherwise, return to the above steps and continue iterating. In subsequent iterations, the damping factor may need to be adjusted as needed. .

Images