Underwater multispectral full-waveform laser radar robust ranging method and device and medium

Abstract

The invention discloses an underwater multispectral full-waveform laser radar robust ranging method and device and a medium. The method comprises the steps of S1, data acquisition; s2, waveform preprocessing; s3, constructing a target echo model; s4, target echo detection; s5, target echo feature extraction; and S6, peak value estimation and distance calculation. According to the method, robust time positioning and distance estimation under the conditions of interface echo peak top saturation clipping and peak back trailing right deviation are realized, systematic deviation introduced by peak point distortion and trailing superposition is inhibited, and the consistency and comparability of multi-wavelength distance measurement results are improved; and a new technical path and an application prospect are provided for accurate distance measurement of the underwater multispectral laser radar.

Description

Robust ranging methods, devices and media for underwater multispectral full-waveform lidar Technical Field

[0001] This invention relates to the field of multispectral lidar signal processing and ranging calibration technology, specifically to a robust ranging method, device, and medium for underwater multispectral full-waveform lidar. Background Technology

[0002] Multispectral lidar acquires complete waveforms of echoes over time at multiple wavelengths, simultaneously obtaining target distance information and wavelength-dependent reflection / scattering differences. It is widely used in underwater target detection, shallow-sea depth sounding, environmental monitoring, and multi-parameter inversion. Compared to pulse ranging, which only outputs single-point distance or a few features, full waveform data can characterize the morphology and structure of the echoes, improving ranging accuracy and robustness in complex media.

[0003] However, in underwater multispectral lidar scenarios, water volume scattering, backscattering of suspended particles, and interface-related multipath propagation introduce multiple echo components with different propagation paths. These late-arriving components concentrate towards the side with the larger time delay in the time domain and superimpose with the attenuation segment of the main echo, causing the waveform's falling edge to decay more slowly, deflect to the right, and develop a long tail. This, in turn, leads to a systematic shift in time positioning based on the peak point. Simultaneously, interface echoes in practical systems often exhibit a clipped saturation plateau due to receiver link saturation or analog-to-digital converter range limitations, distorting the peak shape and causing unreliable detection of the peak moment. If the peak point is still used as the ranging reference, the ranging stability and consistency will be significantly reduced.

[0004] Furthermore, multispectral systems exhibit differences in transmit energy, receive response, gain, and system delay across different wavelength channels, potentially leading to inconsistent distance measurements at the same actual distance. Additionally, multiple acquisitions at the same wavelength can introduce random fluctuations and system bias. Common existing techniques include Gaussian or multi-Gaussian fitting of single waveforms, compensation for saturated waveforms, or threshold intersection location. However, under conditions of significant tailing and interface echo saturation, peaks become unusable, falling edge distortion is severe, and fitting stability and cross-wavelength consistency remain difficult to guarantee.

[0005] Therefore, a robust ranging and calibration method is needed that can stably extract time features under conditions of peak saturation clipping and post-peak tail distortion of the interface echo, and achieve unified correction for both same-wavelength and cross-wavelength measurements by combining known absolute distances. Summary of the Invention

[0006] The purpose of this invention is to provide a robust ranging method, device, and medium for underwater multispectral full-waveform lidar, in order to solve problems in the prior art such as unusable peak points due to interface echo saturation clipping, peak point time shift due to rightward tailing, and inconsistent ranging results for different wavelength channels that are difficult to calibrate uniformly.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a robust ranging method for underwater multispectral full-waveform lidar, comprising the following steps: S1, data acquisition: repeatedly acquiring multiple sets of multi-wavelength echo data within a preset wavelength range and storing them in a storage medium for subsequent reading and processing, and acquiring calibration waveforms at at least one known distance; S2, waveform preprocessing: performing Gaussian smoothing filtering on the original echoes to suppress noise and obtain smoothed echoes; S3, target echo model construction: extracting target echo segments within a preset time window from the calibration waveforms, performing peak alignment and amplitude normalization on each echo segment, and averaging to generate a shared target echo reference model; S4, target echo detection: using sliding window normalized cross-correlation for template matching to obtain phase... Similarity sequence: Select the echo segments corresponding to the highest preset number of local peaks as candidates, and after removing noise by amplitude significance judgment, select the latest arriving peak as the target echo arrival point according to time sequence; S5, Time feature extraction: Locate the peak point in the target echo interval and calculate the slope on the rising edge to extract the maximum positive slope point. Perform saturation plateau detection on the interface echo and extract its maximum positive slope point on the rising edge; S6, Peak estimation and distance calculation: Use the sampling offset between the target echo peak point and the maximum slope point as the system offset, back-calculate the equivalent peak point of the interface echo, and combine the sampling interval to convert the sampling point difference between the equivalent peak point of the interface and the target peak point into the time difference, and calculate the target distance according to the time-of-flight ranging relationship.

[0008] Further, S1 includes the following step: setting a wavelength set within a preset wavelength range at preset wavelength intervals: Where Λ is the set of wavelengths, and λ i Let λ be the i-th wavelength, and N be the number of wavelengths; for each wavelength λ i Echo sequences of a preset number of groups are acquired at preset sampling intervals and stored in a three-dimensional structure of wavelength index, group number index, and time sampling point index, and are kept at at least one known calibration distance d. cal Under certain conditions, calibration waveforms are collected for template construction.

[0009] Further, S2 includes the following steps: S2.1, processing the original echo x i,m [k] Perform Gaussian smoothing filtering to obtain the smoothed echo y i,m [k], where x i,m [k] represents the original echo amplitude of the k-th sampling point under the condition that the wavelength index is i and the group number index is m; yi,m [k] represents the smoothed echo amplitude at the k-th sampling point under the condition that the wavelength index is i and the group number index is m; S2.2, the smoothed echo y i,m [k] serves as the input for subsequent sliding window cross-correlation matching, peak localization, and slope calculation.

[0010] Further, S3 includes the following steps: S3.1, with the known calibration distance d cal Under these conditions, for each wavelength λ i Extract a target echo band with a fixed time window of length W; S3.2, denote the peak index of the i-th target echo band as k. p,i And shift each segment so that its peak index is unified with the reference peak index k. p S3.3, average the normalized echo bands of the entire wavelength to generate a shared target echo reference template r[g], where g is the index of the sampling point within the target echo reference template.

[0011] Further, S4 includes the following steps: S4.1, smoothing the echo y i,m [k] is subjected to sliding window normalized cross-correlation calculation with the shared target echo reference template r[g] to obtain the similarity sequence c[τ], where τ is the index of the sliding window start position; the echo segments corresponding to the P local peaks with the highest scores are selected from the similarity sequence as the candidate set; S4.2, the mean noise baseline μ is calculated within the preset noise baseline interval. n Standard deviation of noise baseline σ n And use the amplitude determination threshold T=μ n +ασ n Amplitude significance is determined, where α is a preset coefficient; S4.3, the candidate echo set that passes the determination is sorted in chronological order, and the peak with the latest arrival time is selected as the final target echo arrival point.

[0012] Further, S5 includes the following steps: S5.1, calculating a discrete slope sequence within the rising edge interval of the target echo according to the sampling interval Δt: , where s i,m [k] represents the discrete slope of the k-th sampling point given wavelength index i and group number index m. i,m [k+1], y i,m [k-1] represents the smoothed echo, k+1 represents the index of the next sampling point after the k-th sampling point, and k-1 represents the index of the previous sampling point after the k-th sampling point. a ,k b [ ] represents the index interval of the rising edge of the target echo; the point with the maximum positive slope is taken as the robust time delay feature point of the target echo; S5.2, saturation plateau detection is performed on the interface echo, when L consecutive sampling points satisfy y[k]≥V satWhen a point is identified as a saturation plateau, L is a preset threshold for the number of consecutive points, and V... sat The system saturation amplitude threshold is used; the discrete slope s is calculated within the rising edge interval before the saturation plateau. i,m [k], extracts the point with the maximum positive slope on the interface.

[0013] Further, S6 includes the following steps: S6.1, calculating the sampling offset Δk between the target echo peak point and the target echo maximum positive slope point, and using the sampling offset Δk to back-calculate the interface echo equivalent peak point; S6.2, based on the sampling point difference Δk between the target echo peak point and the interface echo equivalent peak point... T Converting time difference The distance is calculated based on the time-of-flight ranging relationship, where Δt is the sampling interval.

[0014] The present invention also provides a robust ranging device for underwater multispectral full-waveform lidar, comprising one or more processors and a memory, wherein the memory stores instructions, and when the instructions are executed by the processor, the robust ranging method for underwater multispectral full-waveform lidar as described above is implemented.

[0015] The present invention also provides a readable storage medium having a program stored thereon, which, when executed by a processor, implements the robust ranging method for underwater multispectral full-waveform lidar as described above.

[0016] Compared with the prior art, the present invention has the following beneficial effects: (1) Under the condition of saturated clipping at the peak of the interface echo, the present invention does not rely on the peak point of the interface echo, but realizes the peak time back-pushing based on the stable characteristics of the rising edge of the interface echo, thereby improving the reliability of the interface echo time positioning; (2) In view of the right-biased distortion of the trailing edge, the present invention suppresses the false detection caused by late scattering and long tail superposition through target echo reference construction and dual-condition detection, thereby improving the stability of target echo positioning; (3) The method of the present invention can be implemented in software in the multispectral lidar ranging and calibration process, which is convenient for engineering deployment and provides a new technical path and application prospect for the accurate ranging and calibration of underwater multispectral lidar. Attached Figure Description

[0017] Figure 1 is a flowchart of a robust ranging method for underwater multispectral full-waveform lidar according to the present invention.

[0018] Figure 2 is a schematic diagram of the intensity-time real waveform of multiple wavelengths, showing 14 wavelengths.

[0019] Figure 3 shows a comparison of distance estimation using different ranging methods on 200 sets of repeated measurement data when the actual distance is 20 m, where: (a) Gaussian fitting method; (b) exponential Gaussian fitting method; (c) linear fitting method; (d) the robust ranging method of underwater multispectral lidar of the present invention.

[0020] Figure 4 is a schematic diagram comparing the average absolute distance deviation of the robust ranging method of underwater multispectral full-waveform lidar of the present invention with that of Gaussian fitting, exponential Gaussian fitting and linear fitting methods at real distances of 20m, 22m, 24m, 26m, 28m and 30m after repeated measurements of 200 sets of data.

[0021] Figure 5 is a schematic diagram of the robust ranging device of underwater multispectral full-waveform lidar according to the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, other embodiments obtained by those of ordinary skill in the art without creative effort are all within the protection scope of the present invention.

[0023] As shown in Figure 1, a robust ranging method for underwater multispectral full-waveform lidar is achieved through the following steps: S1, Data acquisition: In the underwater multispectral lidar system, the transmitter operates within a preset wavelength range (450nm-580nm) at preset wavelength intervals (10nm), the receiver repeatedly samples and stores the echo, and simultaneously acquires calibration waveforms under at least one known distance condition (20m), as shown in Figure 2.

[0024] The wavelength set is represented as: Where Λ represents the set of wavelengths involved in the measurement; λ i This represents i working wavelengths; N represents the number of wavelengths.

[0025] For each wavelength λ i M groups (200 groups) of echo sequences were acquired and stored in a three-dimensional structure based on wavelength index, group number index, and time sampling point index. The original echo can be denoted as x. i,m [k], where x i,m [k] represents the original echo amplitude of the kth sampling point under the condition that the wavelength index is i and the group number index is m, with a sampling interval of Δt (0.1ns).

[0026] The known distance calibration waveform can be obtained by placing a planar calibration target within the system's field of view.

[0027] S2, Waveform preprocessing aims to suppress noise and improve the stability of template matching and time feature extraction. In this embodiment, the original echo is subjected to Gaussian smoothing filtering to obtain a smoothed echo. Smoothed echo y i,m [k] serves as the input for subsequent sliding window normalized cross-correlation matching, peak localization, and slope calculation to reduce the impact of noise on temporal feature extraction, where y i,m [k] represents the smoothed echo amplitude of the kth sampling point under the condition that the wavelength index is i and the group number index is m.

[0028] S3, Target echo model construction: In the calibration waveform, the target echo segment within the preset time window is extracted, and the peak value of each echo segment is aligned and the amplitude is normalized and averaged to generate a shared target echo reference model.

[0029] S3.1, Target echo band interception: Under at least one known distance condition (20m), for each wavelength λ i The calibration waveform is used to extract a target echo segment of length W from a preset time window, denoted as: , where u i [k] represents the wavelength λ. i The corresponding echo band to be captured; W is the capture window length.

[0030] S3.2, Peak Alignment: First, find the peak index within each echo band: , where k p,i wavelength λ i The echo band peak position index, k represents the kth sampling point of the target echo band, and argmax represents the independent variable that makes the objective function reach its maximum value.

[0031] Align each echo band to a unified peak index k based on its peak value. p The aligned echo band can be denoted as .

[0032] S3.3, Amplitude Normalization and Averaging: To eliminate energy differences between different wavelengths or different sampling times, amplitude normalization is performed after alignment. ,in, This represents the normalized echo band; max indicates taking the maximum value of the objective function, i.e., the peak amplitude of this echo band.

[0033] Averaging across the normalized echo bands across the entire wavelength range yields a shared target echo reference template: , where r[g] is the shared target echo reference model, and N represents the number of wavelengths.

[0034] S4, Target echo detection: In this embodiment, sliding window normalized cross-correlation is used to perform template matching to obtain a similarity sequence. The echo segments corresponding to the highest preset number of local peaks are selected as candidates. After removing noise by amplitude significance determination, the latest peak is selected as the target echo arrival point in chronological order.

[0035] S4.1 Sliding window normalized cross-correlation and candidate peak selection: for the smoothed echo y i,m Calculate the similarity sequence between [k] and the shared target echo reference template r[g]: , where c i,m [τ] represents the similarity sequence at wavelength i and group number m; τ is the index of the sliding window starting position. y is the mean echo within the sliding window. i,m [k] represents the smoothed echo amplitude at the k-th sampling point, given wavelength index i and group number index m, where W is the truncation window length. The mean of the target echo reference model.

[0036] In c i,m Local maxima are detected in [τ], and the P local peaks with the highest similarity scores are selected as the candidate set, where P is a preset number.

[0037] S4.2, Amplitude Significance Determination: Within the preset noise baseline range Ω n Within, calculate the noise baseline mean μ n Standard deviation of noise baseline σ n : , where Ω n For the set of noise baseline sampling points, It is the number of its elements.

[0038] Amplitude determination threshold T is used:

[0039] Where α is a preset coefficient.

[0040] For each candidate echo band, calculate its peak amplitude A. j , when A j If the value is greater than or equal to T, the candidate is determined to be a valid target echo candidate; otherwise, it is discarded as a noise candidate.

[0041] S4.3, Arrival Point Selection: Sort the candidates that have passed the judgment in chronological order, and select the candidate with the latest arrival time as the final target echo arrival point. The latest arrival time can be calculated using the peak index within the window.

[0042] S5, time feature extraction within the target echo interval. In this embodiment, the target echo peak point and the point with the maximum positive slope on the rising edge are extracted simultaneously. In case of possible saturation of the interface echo, saturation plateau detection is performed first, and then the point with the maximum positive slope is extracted on the rising edge before the saturation plateau.

[0043] S5.1, Target echo peak point location: Within the final target echo interval, locate the peak index: , where Ω t k represents the set of sampling points for the target echo search interval; t,p is the index of the target echo peak point, W is the length of the truncation window, and argmax represents the independent variable that makes the objective function reach its maximum value.

[0044] S5.2, Maximum positive slope point of the target echo rising edge: Calculate the discrete slope sequence within the target echo rising edge interval according to the sampling interval Δt: , where s i,m [k] represents the discrete slope of the k-th sampling point given wavelength index i and group number index m; [k] a ,k b [k] represents the index interval of the rising edge of the target echo. a k b These are the start and end indices of the rising edge interval, respectively.

[0045] Take the point with the maximum positive slope as the target echo time feature point: , where k t,s Index the point with the maximum positive slope on the rising edge of the target echo.

[0046] S5.3, Interface echo saturation plateau detection and maximum slope point: For the interface echo interval Ω I Saturation plateau detection: When there exists an index k0 such that L consecutive sampling points satisfy: Then determine [k0,k 0+L-1 [This is a saturated platform.]

[0047] Among them, V sat L is the system saturation amplitude threshold; L is the preset continuous point threshold.

[0048] The rising edge interval [k] before the saturation plateau I,a ,k I,b The slope sequence is also calculated within [the same area]. And extract the point with the maximum positive slope on the interface: , where k I,s This is the index of the point with the maximum positive slope on the rising edge of the interface echo.

[0049] S6. Peak estimation and distance calculation are to maintain a consistent ranging benchmark across wavelengths even when the peak cannot be directly and reliably located due to interface echo saturation. In this embodiment, the sampling offset between the target echo peak point and the maximum slope point is used as the system offset to back-calculate the equivalent peak point of the interface echo. The sampling interval is combined with the sampling point difference between the equivalent peak point of the interface and the target peak point to convert the time difference. The target distance is calculated according to the time-of-flight ranging relationship.

[0050] S6.1, System Offset Estimation and Interface Equivalent Peak Backtracking: Calculate the sampling offset between the target echo peak point and the target echo maximum positive slope point: , where △k is the sampling offset (in units of sampling points), which reflects the relative positional relationship between the slope feature point and the peak point under the same system conditions.

[0051] Using this offset, the point of maximum slope of the interface echo is used to deduce the equivalent peak point of the interface echo: ,in, Index of the equivalent peak points of the interface echo.

[0052] S6.2, Time Difference Conversion: Calculate the sampling point difference between the target echo peak point and the equivalent peak point of the interface echo: , where △k T This represents the difference in sampling points between two feature points.

[0053] Converted to time difference: , where △T is the time difference between the target echo and the interface echo; △t is the sampling interval.

[0054] S6.3, Time-of-Flight Ranging: Target distance can be calculated using the time-of-flight ranging relationship: Where D is the one-way distance from the interface to the target, and v is the equivalent propagation speed of light in the medium.

[0055] In underwater scenarios, if propagation primarily occurs within the water medium, then the following can be considered: Where c0 is the speed of light in a vacuum, n w Let be the refractive index of water. In cross-wavelength distance measurement, the refractive index can be further written as n. w (λ i This allows us to obtain the distances at various wavelengths. , where D i Indicates wavelength λ i The distance calculated below; △T i Indicates wavelength λ i The time difference obtained.

[0056] Using the above method, consistency between target echo detection and time reference extraction can be maintained under different wavelengths, noise levels, and interface saturation conditions, thereby achieving robust ranging. Figure 3 shows the ranging results of Gaussian fitting, exponential Gaussian fitting, linear fitting methods, and the method of this invention at a real distance of 20m, with 200 sets of repeated measurements at wavelengths of 450-580nm (10nm intervals). Figure 4 shows the average absolute distance deviation of Gaussian fitting, exponential Gaussian fitting, linear fitting methods, and the method of this invention at different real distances of 20m, 22m, 24m, 26m, 28m, and 30m, with 200 sets of repeated measurements at wavelengths of 450-580nm (10nm intervals).

[0057] The specific quantitative evaluation results are shown in Table 1, which compares the performance of Gaussian fitting, exponential Gaussian fitting, linear fitting methods, and the method of this invention. The average absolute distance deviation of the method of this invention is 0.1792 m, which is a reduction of 0.3039 m, 0.2572 m, and 0.3184 m compared to Gaussian fitting, exponential Gaussian fitting, and linear fitting methods, respectively. The average relative distance deviation of the method of this invention is 0.0090, which is a reduction of 0.0152 m, 0.0128 m, and 0.0159 m compared to Gaussian fitting, exponential Gaussian fitting, and linear fitting methods, respectively. The root mean square error of the method of this invention is 0.1806 m, which is a reduction of 0.3035 m, 0.2574 m, and 0.3182 m compared to Gaussian fitting, exponential Gaussian fitting, and linear fitting methods, respectively.

[0058] Table 1. Comparison of ranging errors of Gaussian fitting, exponential Gaussian fitting, linear fitting, and the method of this invention.

[0059] Referring to Figure 5, an embodiment of the present invention provides a robust ranging device for underwater multispectral full-waveform lidar, including one or more processors, for implementing a robust ranging method for underwater multispectral full-waveform lidar in the above embodiment.

[0060] The robust ranging device for underwater multispectral full-waveform lidar of this invention can be applied to any device with data processing capabilities, such as a computer. The device can be implemented in software, hardware, or a combination of both. Taking software implementation as an example, as a logical device, it is formed by the processor of the device with data processing capabilities reading the corresponding computer program instructions from non-volatile memory into memory and running them. From a hardware perspective, as shown in Figure 5, which illustrates a hardware structure of any device with data processing capabilities, in addition to the processor, memory, network interface, and non-volatile memory shown in Figure 5, the device in the embodiment may also include other hardware depending on its actual function; these will not be elaborated further.

[0061] The specific implementation process of the functions and roles of each unit in the above device can be found in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0062] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0063] This invention also provides a readable storage medium storing a program that, when executed by a processor, implements a robust underwater multispectral full-waveform lidar ranging method described in the above embodiments.

[0064] The readable storage medium can be an internal storage unit of any data processing device described in any of the foregoing embodiments, such as a hard disk or memory. The readable storage medium can also be an external storage device, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc., equipped on the device. Furthermore, the readable storage medium can include both internal storage units of any data processing device and external storage devices. The readable storage medium is used to store the computer program and other programs and data required by the data processing device, and can also be used to temporarily store data that has been output or will be output.

[0065] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A robust ranging method for underwater multispectral full-waveform lidar, characterized in that, Includes the following steps: S1, Data Acquisition: Repeatedly acquire multiple sets of multi-wavelength echo data within a preset wavelength range and store them in a storage medium for subsequent reading and processing; acquire calibration waveforms at at least one known distance. S2, Waveform Preprocessing: Perform Gaussian smoothing filtering on the original echoes to suppress noise and obtain smoothed echoes. S3, Target Echo Model Construction: Extract target echo segments within a preset time window from the calibration waveforms; perform peak alignment and amplitude normalization on each echo segment and calculate the average to generate a shared target echo reference model. S4, Target Echo Detection: Use sliding window normalized cross-correlation for template matching to obtain a similarity sequence; select the preset number of local peaks with the highest scores corresponding to the echoes. S5, Time Feature Extraction: Locate the peak point within the target echo interval and extract the maximum positive slope point by calculating the slope on the rising edge. Perform saturation plateau detection on the interface echo and extract its maximum positive slope point on the rising edge. S6, Peak Estimation and Distance Calculation: Use the sampling offset between the target echo peak point and the maximum slope point as the system offset to back-calculate the equivalent peak point of the interface echo. Combine the sampling interval with the sampling point difference between the equivalent peak point of the interface and the target peak point to calculate the time difference. Calculate the target distance according to the time-of-flight ranging relationship.

2. The robust ranging method for underwater multispectral full-waveform lidar as described in claim 1, characterized in that, S1 includes the following steps: setting a wavelength set within a preset wavelength range at preset wavelength intervals: Where Λ is the set of wavelengths, and λ i Let λ be the i-th wavelength, and N be the number of wavelengths; for each wavelength λ i Echo sequences of a preset number of groups are acquired at preset sampling intervals and stored in a three-dimensional structure of wavelength index, group number index, and time sampling point index, and are kept at at least one known calibration distance d. cal Under certain conditions, calibration waveforms are collected for template construction.

3. The robust ranging method for underwater multispectral full-waveform lidar as described in claim 1, characterized in that, S2 includes the following steps: S2.1, processing the original echo x i,m [k] Perform Gaussian smoothing filtering to obtain the smoothed echo y i,m [k], where x i,m [k] represents the original echo amplitude of the k-th sampling point under the condition that the wavelength index is i and the group number index is m; y i,m [k] represents the smoothed echo amplitude at the k-th sampling point under the condition that the wavelength index is i and the group number index is m; S2.2, the smoothed echo y i,m [k] serves as the input for subsequent sliding window cross-correlation matching, peak localization, and slope calculation.

4. The robust ranging method for underwater multispectral full-waveform lidar as described in claim 1, characterized in that, S3 includes the following steps: S3.1, with a known calibration distance d cal Under these conditions, for each wavelength λ i Extract a target echo band with a fixed time window of length W; S3.2, denote the peak index of the i-th target echo band as k. p,i And shift each segment so that its peak index is unified with the reference peak index k. p S3.3, average the normalized echo bands of the entire wavelength to generate a shared target echo reference template r[g], where g is the index of the sampling point within the target echo reference template.

5. The robust ranging method for underwater multispectral full-waveform lidar as described in claim 1, characterized in that, S4 includes the following steps: S4.1, smoothing the echo y i,m [k] is subjected to sliding window normalized cross-correlation calculation with the shared target echo reference template r[g] to obtain the similarity sequence c[τ], where τ is the index of the sliding window start position; the echo segments corresponding to the P local peaks with the highest scores are selected from the similarity sequence as the candidate set; S4.2, the mean noise baseline μ is calculated within the preset noise baseline interval. n Standard deviation of noise baseline σ n And use the amplitude determination threshold T=μ n +ασ n Amplitude significance is determined, where α is a preset coefficient; S4.3, the candidate echo set that passes the determination is sorted in chronological order, and the peak with the latest arrival time is selected as the final target echo arrival point.

6. The robust ranging method for underwater multispectral full-waveform lidar as described in claim 1, characterized in that, S5 includes the following steps: S5.1, calculating the discrete slope sequence within the rising edge interval of the target echo according to the sampling interval Δt: , where s i,m [k] represents the discrete slope of the k-th sampling point given wavelength index i and group number index m. i,m [k+1], y i,m [k-1] represents the smoothed echo, k+1 represents the index of the next sampling point after the k-th sampling point, and k-1 represents the index of the previous sampling point after the k-th sampling point. a ,k b [ ] represents the index interval of the rising edge of the target echo; the point with the maximum positive slope is taken as the robust time delay feature point of the target echo; S5.2, saturation plateau detection is performed on the interface echo, when L consecutive sampling points satisfy y[k]≥V sat When a point is identified as a saturation plateau, L is a preset threshold for the number of consecutive points, and V... sat The system saturation amplitude threshold is used; the discrete slope s is calculated within the rising edge interval before the saturation plateau. i,m [k], extracts the point with the maximum positive slope on the interface.

7. The robust ranging method for underwater multispectral full-waveform lidar as described in claim 1, characterized in that, S6 includes the following steps: S6.1, calculating the sampling offset Δk between the target echo peak point and the target echo maximum positive slope point, and using the sampling offset Δk to back-calculate the interface echo equivalent peak point; S6.2, based on the sampling point difference Δk between the target echo peak point and the interface echo equivalent peak point... T Converting time difference The distance is calculated based on the time-of-flight ranging relationship, where Δt is the sampling interval.

8. A robust ranging device for underwater multispectral full-waveform lidar, characterized in that, It includes one or more processors and a memory, wherein the memory stores instructions that, when executed by the processor, implement the robust ranging method of underwater multispectral full-waveform lidar as described in any one of claims 1-7.

9. A readable storage medium, characterized in that, It stores a program that, when executed by a processor, implements the robust ranging method for underwater multispectral full-waveform lidar as described in any one of claims 1-7.

Images