A vegetation height extraction method, system and device of a TECIS full waveform laser radar

Abstract

The application discloses a TECIS full waveform laser radar vegetation height extraction method, system and equipment, relates to the field of vegetation height calculation, and comprises the following steps: performing primary signal interception on the TECIS full waveform laser radar echo waveform of a target vegetation area to reduce data redundancy, and then performing secondary signal interception to preliminarily determine a signal search range; in combination with a preset multi-scale noise signal discrimination condition set and an adaptive noise coefficient selection method, an optimal signal threshold is adaptively determined, and then a final waveform effective signal range is determined, finally, a ground position and a crown position are determined, and after height difference calculation, the vegetation height is obtained. The adaptive threshold determination and waveform effective signal range extraction are realized based on the noise coefficient adaptive selection method, the limitation that an experienced noise coefficient selection depends on a large range of synchronous reference data can be eliminated, the applicability is improved, the vegetation height estimation precision is improved, and the production of a TECIS data large-range vegetation height product can be realized.

Description

Technical Field

[0001] This application relates to the field of vegetation height calculation, and in particular to a method, system and device for extracting vegetation height using a TECIS full-waveform lidar. Background Technology

[0002] Spaceborne large-spot full-waveform lidar (hereinafter referred to as spaceborne lidar) can be used for large-scale forest height estimation. By recording the time interval between transmitted and received pulses, it can accurately infer the ground position and canopy top from the echo signal, thus providing a direct estimate of global or near-global vegetation height. Forest height estimation based on spaceborne lidar echo signals depends on the identification of effective echo signals.

[0003] In existing forest height estimation using sensors, effective signal identification often employs empirical thresholding methods. However, the application of such methods is limited by the lack of large-scale synchronous reference data. Furthermore, noise figures vary significantly among different sensors. Simultaneously, even with the same sensor, different laser emitters and observation conditions also influence the selection of the noise figure, and current forest height estimation methods cannot address this applicability issue. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, and device for extracting vegetation height from a TECIS full-waveform lidar, which can eliminate the limitation of relying on large-scale synchronous reference data for the selection of empirical noise figure, improve the applicability of the method, enhance the accuracy of vegetation height estimation, and help realize the production of large-scale vegetation height products from TECIS data.

[0005] To achieve the above objectives, this application provides the following solution:

[0006] Firstly, this application provides a method for extracting vegetation height using a TECIS full-waveform lidar, comprising:

[0007] A signal truncation is performed on the TECIS full-waveform lidar echo waveform of the target vegetation area to determine the range of the waveform signal to be processed, the mean and standard deviation of the noise;

[0008] A threshold is determined based on a preset noise figure, the mean and standard deviation of the noise, and a secondary signal truncation is performed on the range of the waveform signal to be processed based on the threshold to determine the signal search range;

[0009] For waveform signals within the signal search range, a preset multi-scale noise signal discrimination condition set and an adaptive noise figure selection method are used to determine the optimal signal threshold based on the mean and standard deviation of the noise, thereby determining the effective signal range of the final waveform.

[0010] Based on the waveform signal within the effective signal range of the final waveform, the ground position and crown position are determined, and then the elevation difference is calculated to obtain the vegetation height.

[0011] Secondly, this application provides a vegetation height extraction system for a TECIS full-waveform lidar, comprising:

[0012] The primary signal interception module is used to perform a primary signal interception on the TECIS full-waveform lidar echo waveform of the target vegetation area to determine the range of the waveform signal to be processed, the mean and standard deviation of the noise;

[0013] The secondary signal interception module is used to determine a threshold based on a preset noise coefficient, the mean and standard deviation of the noise, and to perform secondary signal interception on the range of the waveform signal to be processed based on the threshold to determine the signal search range;

[0014] The adaptive threshold-based range determination module is used to determine the optimal signal threshold for waveform signals within the signal search range by using a preset multi-scale noise signal discrimination condition set and an adaptive noise coefficient selection method, combined with the mean and standard deviation of the noise, and then determining the final effective signal range of the waveform.

[0015] The vegetation height calculation module is used to determine the ground position and crown position based on the waveform signal within the effective signal range of the final waveform, and then perform elevation difference calculation to obtain the vegetation height.

[0016] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a vegetation height extraction method for TECIS full-waveform lidar.

[0017] According to the specific embodiments provided in this application, the following technical effects are disclosed: First, this application obtains the signal search range through two signal interceptions, providing an accurate data foundation for subsequent processing and improving computational efficiency. Second, by combining a preset multi-scale noise signal discrimination condition set and an adaptive noise figure selection method, the optimal signal threshold is determined based on the mean and standard deviation of the noise. Thus, this application achieves automated processing of effective signal identification through adaptive selection. Compared with existing empirical threshold methods that rely on ground observation or airborne lidar data, this application does not require reference data, improving the method's wide applicability and reducing its dependence on empirical parameters. Then, the ground position and crown position are determined based on the waveform signal within the effective signal range of the final waveform, and the vegetation height is obtained by calculating the elevation difference.

[0018] In summary, this application adaptively selects the optimal noise figure, fully considering the characteristics of the sensor (different transmitted pulse energies and energy attenuation levels) and the observation environment (different observation times, laser pulse transmission environment, etc.) during data acquisition. This effectively eliminates the limitation of relying on large-scale synchronous reference data for the selection of empirical noise figures, and also avoids the estimation deviation of the effective signal range or the top of the canopy caused by using the same noise figure throughout the entire area in traditional methods, thereby reducing the estimation error of the top of the forest canopy. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the vegetation height extraction method of TECIS full-waveform lidar in one embodiment of this application.

[0021] Figure 2 This is a schematic diagram of the loss function.

[0022] Figure 3 This is a flowchart illustrating the vegetation height extraction method of TECIS full-waveform lidar in another embodiment of this application.

[0023] Figure 4 This is a comparative schematic diagram of the vegetation height extraction method of TECIS full-waveform lidar in the embodiments of this application compared with traditional methods.

[0024] Figure 5 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] To break free from the dependence of traditional methods on empirical parameters and improve the applicability of the method under different sensors and observation conditions, this application proposes a vegetation height extraction method based on adaptive threshold selection for TECIS full-waveform lidar.

[0027] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] In one exemplary embodiment, such as Figure 1 As shown, a method for extracting vegetation height using TECIS full-waveform lidar is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method includes steps 101 to 104.

[0029] The majority of the echo signals from the China Terrestrial Ecosystem Carbon Monitoring Satellite (TECIS) are Gaussian white noise, with only a small portion being valid signals. For example, a 100-meter-long tree corresponds to 800 valid frames of echo signal with a vertical resolution of 0.125 meters, but the actual echo record is 8000. To improve computational efficiency, the original waveform signal must be truncated.

[0030] Step 101: Perform a signal truncation on the TECIS full-waveform lidar echo waveform of the target vegetation area to determine the range of the waveform signal to be processed, and the mean and standard deviation of the noise. Considering that the noise signal intensity is much lower than the effective signal, the signal location can be initially determined by searching for the maximum signal value. This specifically includes the following steps:

[0031] (11) For the TECIS full waveform lidar echo waveform of the target vegetation area, perform a maximum signal search to determine the location of the maximum signal.

[0032] (12) Extend the waveform signal forward and backward by a first preset number of frames, centered on the maximum signal position, to determine the first waveform signal search range. In one application, to ensure that the captured waveform data completely covers the signal and noise, capture 1500 frames before and after the maximum signal position (i.e., the first preset number of frames) as the initial range for waveform signal search (i.e., the first waveform signal search range).

[0033] (13) Taking the start point and end point of the first waveform signal search range as the initial point, extend the waveform signal forward and backward by a second preset number of frames respectively to determine the second waveform signal search range. In one application, the second preset number of frames can be 100.

[0034] (14) Based on the waveform signal within the search range of the second waveform signal, calculate the mean and standard deviation of the noise; corresponding to the previous step, use 100 frames before and after the extracted waveform signal to calculate the mean of the noise. noise and standard deviation σ noise , where forward and backward represent the direction of the crown and the direction of the ground, respectively.

[0035] (15) Based on the mean and standard deviation of the noise, Gaussian filtering is used to smooth the waveform signal within the first waveform signal search range, and then the first waveform signal search range is marked as the waveform signal range to be processed.

[0036] To eliminate waveform noise to the greatest extent possible while preserving the waveform signal, a standard Gaussian function is used as a filter to smooth the waveform signal within the first waveform signal search range. The Gaussian function used is:

[0037] .

[0038] An excessively large smoothing filter width will result in over-smoothing, while a narrow width may fail to effectively suppress noise. Therefore, this application sets the smoothing window width to 6.5σ. t , σ t The calculation formula is:

[0039] .

[0040] Where μ is the center position of the smoothing window, σ is 1, and FWHM is the half-width at half maximum (FWHM) of the transmitted pulse.

[0041] Step 102: Determine a threshold based on the preset noise coefficient, the mean and standard deviation of the noise, and perform secondary signal truncation on the range of the waveform signal to be processed based on the threshold to determine the signal search range.

[0042] Narrowing the signal search range helps reduce the error rate of noise being misdiagnosed as vegetation or ground signals, also known as false alarms. When false alarms occur, the top of the vegetation canopy is overestimated while the ground location is underestimated. To reduce the interference of false alarms on the identification of valid signals, a threshold can be set to identify the signal search range. The formula for determining the threshold is:

[0043] .

[0044] Among them, T k is the threshold, specifically the noise threshold; k is the noise coefficient.

[0045] When a signal in the waveform exceeds a threshold, it is identified as a signal endpoint. To determine the signal search range, k is set to 20, and the initial start and end points of the valid signal are calculated. Correspondingly, step 102 includes the following steps:

[0046] (21) Based on the relationship between the waveform signal within the range to be processed and the threshold, determine the cutoff position of the signal search range; the formula for determining the cutoff position bin is:

[0047] .

[0048] Where, where() is the position determination function, d represents the direction, and I is the intensity of the waveform signal.

[0049] (22) Taking the start and end points of the truncation position of the signal search range as initial points, respectively, extend the waveform signal forward and backward by a third preset number of frames to determine the final signal search range. In one application, the third preset number of frames can be 400, that is, extend outward by 400 frames from the start and end points of the truncation position as the signal search range to ensure that the echo signal contains the complete forest echo.

[0050] Identifying the effective signal range of the TECIS full-waveform lidar is one of the most crucial steps in forest vegetation height estimation, determined by a threshold. Setting the threshold too high or too low directly impacts the accuracy of forest parameter estimation. In traditional threshold methods, the noise figure k relies on manual setting, reducing the method's wide applicability. How to adaptively select the optimal noise figure for signal threshold calculation has long been neglected. To address these issues, this application employs an adaptive noise figure selection method in step 103 to improve the method's automation, reduce reliance on empirical data, and enhance the accuracy of large-scale forest vegetation height estimation.

[0051] Step 103: For the waveform signal within the signal search range, a preset multi-scale noise signal discrimination condition set and an adaptive noise figure selection method are used, combined with the mean and standard deviation of the noise, to determine the optimal signal threshold, and then determine the effective signal range of the final waveform.

[0052] In practical applications, this application first sets a loose noise figure range, with values ​​ranging from 2 to 20, in intervals of 2. A threshold can be calculated for each noise figure to determine the start and end positions of the echo signal. Ideally, the cutoff position corresponding to the noise figure corresponds to noise and signal on either side, with a significant difference in the mean echo intensity between the two. To avoid the influence of local anomalous noise, three scale windows—short, medium, and long—are set for significance testing, with lengths of FWHM / 2, FWHM, and 2xFWHM, respectively, where FWHM is the full width at half maximum (FWHM) of the transmitted pulse. Due to the significant variability in the vertical distribution of surface echo energy, the mean signal intensity within different scale signal windows also exhibits significant differences. These signal intensity differences can be judged using a t-test, with a value of 1 indicating significant echo intensity and 0 otherwise.

[0053] Theoretically, when the noise figure is very small, the threshold is at the noise level, and the means of the noise end and the signal end do not have significant differences at different scales. At this time, the cumulative value of the t-test significance indicator is the smallest. As the noise figure increases to the noise figure corresponding to the optimal threshold, the means at different scales begin to show significant differences, and the cumulative value of the t-test significance indicator gradually increases to the maximum value. After reaching the optimal value, if the noise figure continues to increase, there may be no significant difference between the echo energy of the noise and signal ends, and the cumulative value of the t-test significance indicator begins to decrease. Based on this, the preset multi-scale noise signal discrimination condition set in this application includes six discrimination conditions for the significance test between noise and signal and the significance test between signals at three preset scales: short, medium, and long. Based on this, step 103 includes the following steps:

[0054] (31) Based on the preset multi-scale noise signal discrimination condition set, a loss function is constructed according to the significance test at different scales; to reduce the random error in calculating a single echo, the loss function calculation unit is set as a set of echo signals from the entire track or a certain spatial range. Figure 2 The diagram shown illustrates the loss function for the TECIS full-waveform lidar track alignment calculation. The formula for the loss function is:

[0055] .

[0056] Where n is the number of whole-track echoes, and J is the number of preset multi-scale noise signal discrimination conditions; Con i,j For the first i The significance discriminant function for the j-th discrimination condition of each echo.

[0057] (32) For the waveform signal within the signal search range, calculate the minimum value of the loss function corresponding to the preset noise figure range, and determine the optimal noise figure. Specifically, when calculating the minimum value of the loss function corresponding to the preset noise figure range, to avoid the problem of the noise figure continuously increasing while a significant difference exists between the noise and the signal, resulting in the MSD change approaching 0 and the noise figure being overestimated, the MSD change percentage is increased to determine the convergence of the loss function when constructing the objective function. The objective function is as follows:

[0058] .

[0059] Where, k selected The noise coefficient is selected when the objective function is optimal, representing the optimal noise coefficient; min() and max() are the minimum and maximum value functions, respectively; F is the rate of change threshold, which has been experimentally verified to be 5%. Furthermore, during the solution of the objective function, the value of k increases from small to large.

[0060] (33) Based on the optimal noise figure, and combined with the mean and standard deviation of the noise, calculate the optimal signal threshold.

[0061] (34) Based on the optimal signal threshold, the signal search range is truncated to determine the effective signal range of the final waveform. Specifically, the forward and backward thresholds are calculated using the optimal noise figure, and the start and end positions of the signal are determined by using five consecutive frames of waveform intensity greater than the threshold, so as to obtain the effective signal range.

[0062] Step 104: Determine the ground position and crown position based on the waveform signal within the effective signal range of the final waveform, and then calculate the elevation difference to obtain the vegetation height.

[0063] The derivative method is a commonly used method for waveform ground location detection. This method determines the peak position by the first derivative of the waveform and uses the last peak as the ground position. However, when vegetation components overlap with ground echoes to produce aliased waveforms, the ground peaks disappear, leading to errors in ground location detection. Waveform decomposition methods are relatively mature in hidden peak detection applications, but they generally suffer from problems such as complex calculation processes, high computational resource consumption, and results being affected by algorithm parameters. To solve this problem, step 104 of this application includes the following specific steps:

[0064] (41) For the waveform signal within the effective signal range of the final waveform, Gaussian sharpening is applied for waveform sharpening; wherein, waveform sharpening is a fast and low-power hidden peak detection method, which is essentially a differential processing of the signal. Compared with the traditional differential sharpening method, Gaussian sharpening can not only sharpen the signal, but also suppress noise through convolution operation, and its expression is as follows: .in, This is the sharpened waveform signal. The waveform signal is within the effective signal range of the final waveform. This is the Gaussian sharpening operator. When n is 2, the Gaussian function formula represents the second derivative, expressed as: r is the Gaussian sharpening multiplier (usually 40). Refers to the Gaussian function formula; to achieve waveform sharpening, The half-width at half maximum (FWHM) is set to half the width of the emitted pulse to highlight the overlapping peaks.

[0065] To ensure that the peak area remains unchanged before and after signal sharpening, the Gaussian function integral of the Gaussian sharpening operator satisfies the following condition: .

[0066] (42) For the waveform signal after waveform sharpening, the last peak position extracted by the first derivative will be used as the ground position; the elevation corresponding to the ground position is the ground elevation.

[0067] In this application, to prevent noise from interfering with the ground location, this application further limits the waveform intensity of the ground wave peak to be greater than 10% of the maximum signal value. If this condition is not met, it is considered noise, and the ground wave peak is iteratively searched. Specifically, for the waveform signal within the range of the final waveform signal after sharpening, from back to front, ground wave peak identification is performed based on the judgment condition that "the waveform intensity of the wave peak is greater than 10% of the maximum signal value". When a wave peak meets the condition, the iteration stops, and the wave peak is regarded as the ground wave peak, and its location is the ground location.

[0068] (43) Determine the crown position based on the cumulative energy of the waveform signal within the effective signal range of the final waveform; the elevation corresponding to the crown position is the crown elevation; specifically, the position corresponding to 98% of the cumulative energy of the waveform in the effective signal is taken as the crown position.

[0069] (44) Calculate the difference between the crown elevation and the ground elevation to obtain the vegetation height.

[0070] like Figure 3 As shown, this application also provides an application example, including the following steps: Acquiring the full-waveform lidar echo waveform. Calculating the mean and standard deviation of noise for 100 frames at each end of the echo signal. Constructing a Gaussian smoothing operator with a window width of 6.5 times the standard deviation of the transmitted pulse to smooth the echo signal. To reduce the error of noise being diagnosed as vegetation or ground signals, using the noise mean plus 10 times the noise standard deviation as a threshold, searching for the initial start and end points of the effective signal truncation position; extending outwards from the start and end points by 400 frames (corresponding to 60 m at a 0.15 m vertical resolution) as the effective signal search range. By setting six discrimination conditions, including a significance test of the noise and signal intensity mean values ​​at short, medium, and long scales with the effective signal start and end positions as boundaries under specific noise coefficient values, a loss function based on the significant difference between signal and noise intensity at different scales is constructed; based on this, by calculating the minimum value of the loss function corresponding to the noise coefficient range of 2~20 with a 2-fold interval, adaptive selection of noise coefficients and identification of effective signals can be achieved. Using the effective signal as input, Gaussian sharpening is employed to enhance the echo signal. Then, using the enhanced waveform as input, the derivative method is used to identify the last peak within the effective signal range to pinpoint the ground location. The location where 98% of the cumulative energy of the effective signal is taken as the canopy top, and the ground peak as the ground location, to calculate the forest vegetation height.

[0071] like Figure 4 The diagram shows the accuracy achieved by using vegetation height extracted from airborne lidar data in five selected experimental areas across China as reference data, compared to the typical empirical threshold method which requires high-precision ground data or airborne lidar data. Figure 4 (a) and (d) in the figure are schematic diagrams showing the accuracy of extracting the top elevation (i.e., crown elevation) of the airborne LiDAR canopy using the method of this application and the comparative method, respectively. Figure 4 In the diagrams, (b) and (e) are schematic diagrams illustrating the accuracy of extracting airborne LiDAR ground elevation using the method of this application and the comparative method, respectively. Figure 4 Figures (c) and (f) show the accuracy of extracting vegetation height using the method of this application and the comparative method, respectively. Through comparison of the above figures, it can be seen that this application effectively improves the wide applicability of waveform effective signal identification and enhances the accuracy of vegetation height estimation.

[0072] Furthermore, in practical applications, the method described in this application can also be used to extract vegetation height for the Ice Cloud and land Elevation Satellite (ICESat GLAS), launched by NASA for monitoring glacier melting, and the Global Ecosystem Dynamics Investigation (GEDI) system. However, in practical applications, step 101 is not required, i.e., a first signal interception is not necessary; data processing can begin from the second signal interception.

[0073] In summary, this application has the following advantages:

[0074] (1) This application constructs a loss function with adaptive noise figure selection, which can adaptively extract the effective signal range of the waveform, thereby reducing the dependence on ground observation or airborne lidar data in the identification of effective signals and improving the applicability of the method in large-scale lidar echo signal processing.

[0075] This application constructs a loss function for adaptive noise figure selection. Using single-track laser data as the statistical unit, the noise figure corresponding to the waveform signal and the noise truncation threshold is determined in the form of an objective function, thus realizing automated processing for effective signal identification. Compared with existing empirical threshold methods that rely on ground observation or airborne lidar data, the proposed method does not require reference data, improving its wide applicability and reducing its dependence on empirical parameters.

[0076] (2) This application improves the accuracy of ground position detection by introducing echo signal enhancement. Compared with the existing derivative method, the ground elevation estimation accuracy is higher in the echo waveform of ground and vegetation overlap.

[0077] The echo signal enhancement in this application primarily targets waveforms where ground echo signals are weak or where vegetation echoes overlap with ground echoes, including surfaces with low vegetation or dense canopies. Compared to existing studies that initialize waveform decomposition algorithm parameters through waveform sharpening, this application directly leverages the enhancement effect of waveform sharpening on weak signals to extract ground elevation. Its advantage lies in improving waveform detection capabilities while reducing the computational complexity of waveform decomposition, thus increasing data processing efficiency. Compared to existing derivative methods and Gaussian decomposition methods, this application can effectively improve the accuracy of ground elevation estimation.

[0078] (3) This application constructs a loss function to adaptively identify the effective signal range for estimating the top of the forest canopy, and at the same time improves the ground position detection capability based on echo signal enhancement, which can effectively improve the estimation accuracy of forest height.

[0079] This application adaptively selects the noise figure based on the characteristics of monorail lidar echo signals. This fully considers the characteristics of the sensor (different transmitted pulse energies and energy attenuation levels) and the observation environment (different observation times, laser pulse transmission environment, etc.) during data acquisition. It effectively avoids the estimation deviation of the effective signal range or canopy top caused by using the same noise figure across the entire area in traditional methods, thus reducing the estimation error of the forest canopy top. Based on this, forest height estimation is performed using the ground position extracted after signal enhancement. Compared with the empirical threshold method and Gaussian decomposition method, it has higher accuracy, and the root mean square error can be reduced by 0.06~0.27 meters.

[0080] Based on the same inventive concept, this application also provides a system for implementing the methods described above. The solution provided by this system is similar to the solution described in the above methods; therefore, the specific limitations in one or more system embodiments provided below can be found in the limitations of the methods described above, and will not be repeated here. In an exemplary embodiment, a TECIS full-waveform lidar vegetation height extraction system is provided, comprising:

[0081] The primary signal interception module is used to perform a primary signal interception on the TECIS full-waveform lidar echo waveform of the target vegetation area to determine the range of the waveform signal to be processed, the mean and standard deviation of the noise.

[0082] The secondary signal interception module is used to determine a threshold based on a preset noise coefficient, the mean and standard deviation of the noise, and to perform secondary signal interception on the range of the waveform signal to be processed based on the threshold, so as to determine the signal search range.

[0083] The adaptive threshold-based range determination module is used to determine the optimal signal threshold for waveform signals within the signal search range by using a preset multi-scale noise signal discrimination condition set and an adaptive noise coefficient selection method, based on the mean and standard deviation of the noise, and then determine the final effective signal range of the waveform.

[0084] The vegetation height calculation module is used to determine the ground position and crown position based on the waveform signal within the effective signal range of the final waveform, and then perform elevation difference calculation to obtain the vegetation height.

[0085] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 5 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the vegetation height extraction method of the TECIS full-waveform lidar.

[0086] Those skilled in the art will understand that Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0087] In one exemplary embodiment, a computer device is also provided, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in the above-described method embodiments.

[0088] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0089] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0090] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0091] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0092] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0093] 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.

[0094] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for extracting vegetation height using a TECIS full-waveform lidar, characterized in that, The method includes: A signal truncation is performed on the TECIS full-waveform lidar echo waveform of the target vegetation area to determine the range of the waveform signal to be processed, the mean and standard deviation of the noise; A threshold is determined based on a preset noise figure, the mean and standard deviation of the noise, and a secondary signal truncation is performed on the range of the waveform signal to be processed based on the threshold to determine the signal search range; For waveform signals within the signal search range, a preset multi-scale noise signal discrimination condition set and an adaptive noise figure selection method are used, combined with the mean and standard deviation of the noise, to determine the optimal signal threshold, thereby determining the final effective signal range of the waveform; including: Based on the preset multi-scale noise signal discrimination condition set, a loss function is constructed according to the significance test at different scales; for the waveform signal within the signal search range, the minimum value of the loss function corresponding to the preset noise coefficient range is calculated, and the optimal noise coefficient is determined; based on the optimal noise coefficient, combined with the mean and standard deviation of the noise, the optimal signal threshold is calculated; based on the optimal signal threshold, the signal search range is truncated to determine the final effective signal range of the waveform. The preset multi-scale noise signal discrimination condition set includes noise saliency tests and signal saliency tests at three preset scales: short, medium, and long; the loss function is: ； Where n is the number of whole-track echoes, J is the number of preset multi-scale noise signal discrimination conditions; k is the noise figure; Con i,j For the first i The significance discriminant function for the j-th discrimination condition of the echo; When calculating the minimum value of the loss function within a preset noise figure range, the convergence of the loss function is judged based on the rate of change of the loss function value, and the following objective function is constructed: ； Where, k selected The noise coefficient selected when the objective function is optimal represents the optimal noise coefficient; where() is the location discrimination function, min() and max() are the minimum and maximum value functions respectively, and F is the rate of change discrimination threshold; Based on the waveform signal within the effective signal range of the final waveform, the ground position and crown position are determined, and then the elevation difference is calculated to obtain the vegetation height.

2. The vegetation height extraction method of TECIS full-waveform lidar according to claim 1, characterized in that, A signal truncation is performed on the TECIS full-waveform lidar echo waveform of the target vegetation area to determine the range of the waveform signal to be processed, the mean and standard deviation of the noise, including: For the TECIS full-waveform lidar echo waveform of the target vegetation area, a maximum signal value search is performed to determine the location of the maximum signal; With the maximum signal position as the center, the waveform signal is extended forward and backward by a first preset number of frames to determine the first waveform signal search range; The second waveform signal search range is determined by extending the waveform signal forward and backward by a second preset number of frames, respectively, with the starting point and ending point of the first waveform signal search range as the initial points. Based on the waveform signals within the search range of the second waveform signal, calculate the mean and standard deviation of the noise; Based on the mean and standard deviation of the noise, Gaussian filtering is used to smooth the waveform signal within the first waveform signal search range, and then the first waveform signal search range is marked as the waveform signal range to be processed.

3. The vegetation height extraction method of TECIS full-waveform lidar according to claim 1, characterized in that, Based on the threshold, a secondary signal truncation is performed on the range of the waveform signal to be processed to determine the signal search range, including: Based on the relationship between the waveform signal within the range of the waveform signal to be processed and the threshold, the cutoff position of the signal search range is determined; Using the start and end points of the cutoff position of the signal search range as initial points, the waveform signal is extended forward and backward by a third preset number of frames respectively to determine the final signal search range.

4. The vegetation height extraction method of TECIS full-waveform lidar according to claim 1, characterized in that, Based on the waveform signal within the effective signal range of the final waveform, the ground position and crown position are determined, and then the elevation difference is calculated to obtain the vegetation height, including: For the waveform signals within the effective signal range of the final waveform, Gaussian sharpening is applied for waveform sharpening. For the waveform signal after waveform sharpening, the position of the last peak extracted by the first derivative is used as the ground position; the elevation corresponding to the ground position is the ground elevation. The crown position is determined based on the cumulative energy of the waveform signal within the effective signal range of the final waveform; the elevation corresponding to the crown position is the crown elevation. The difference between the crown elevation and the ground elevation is calculated to obtain the vegetation height.

5. The vegetation height extraction method of TECIS full-waveform lidar according to claim 3, characterized in that, The formula for determining the threshold is: ； Among them, T k For the threshold, mean noise Let σ be the mean of the noise. noise Let be the standard deviation of the noise, and k be the noise figure. The formula for determining the truncation position bin is: ； Where, where() is the position determination function, d represents the direction, and I is the intensity of the waveform signal.

6. The vegetation height extraction method of TECIS full-waveform lidar according to claim 4, characterized in that, When performing Gaussian sharpening on the waveform signal within the effective signal range of the final waveform, the following Gaussian function formula is used: ;in, This is the sharpened waveform signal. The waveform signal is within the range of the final waveform signal. For Gaussian sharpening operators; When n is 2, the Gaussian function formula represents the second derivative, expressed as: Where r is the Gaussian sharpening multiplier, Refers to the Gaussian function formula; the Gaussian function integral of the Gaussian sharpening operator satisfies the following conditions: .

7. A vegetation height extraction system for a TECIS full-waveform lidar, employing the vegetation height extraction method for the TECIS full-waveform lidar as described in any one of claims 1-6, characterized in that, The system includes: The primary signal interception module is used to perform a primary signal interception on the TECIS full-waveform lidar echo waveform of the target vegetation area to determine the range of the waveform signal to be processed, the mean and standard deviation of the noise; The secondary signal interception module is used to determine a threshold based on a preset noise coefficient, the mean and standard deviation of the noise, and to perform secondary signal interception on the range of the waveform signal to be processed based on the threshold to determine the signal search range; The adaptive threshold-based range determination module is used to determine the optimal signal threshold for waveform signals within the signal search range by using a preset multi-scale noise signal discrimination condition set and an adaptive noise coefficient selection method, combined with the mean and standard deviation of the noise, and then determining the final effective signal range of the waveform. The vegetation height calculation module is used to determine the ground position and crown position based on the waveform signal within the effective signal range of the final waveform, and then perform elevation difference calculation to obtain the vegetation height.

8. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the vegetation height extraction method of the TECIS full-waveform lidar according to any one of claims 1-6.

Images