Laser area array radar high-frequency tree height measuring system

Abstract

The invention belongs to the technical field of tree height measurement, and particularly relates to a laser area array radar high-frequency tree height measurement system which comprises a monitoring device, the monitoring device comprises a movable frame body, a computer is fixedly arranged on a bottom plate of the movable frame body, the movable frame body is movably provided with a mounting plate, and a sensor assembly is mounted on the mounting plate; the high-frequency tree height measurement system comprises an acquisition module, a processing module and an output module. The acquisition module emits diffused wide-angle laser pulses through a sensor assembly, then uses an SPAD array as a detector to receive pixel points, and finally obtains complete point cloud data of a sapling; and finally, calculating and outputting the sapling height according to the point cloud data. The sapling height can be quickly output based on the three-dimensional point cloud.

Description

Technical Field

[0001] This invention belongs to the field of tree height measurement technology, specifically relating to a high-frequency tree height measurement system using laser array radar. Background Technology

[0002] After transplanting, the height of the saplings needs to be measured to observe their subsequent growth. Traditional methods for measuring sapling height use a flexible measuring tape or tape measure, which is fast but lacks accuracy. When measuring the height of saplings transplanted with leaves, water shortages may occur during transport, causing the leaves and tips to soften and potentially bend to one side. Measuring bent saplings with a tape measure or flexible measuring tape is inconvenient and affects the accuracy of the results.

[0003] In the prior art, for example, Chinese Patent Publication No. CN117268333B discloses a seedling height measuring device, including: a base and a seedling body, and further including: multiple cylinders fixed to the bottom of the base; an electronic level fixed to the upper side of the base; a first support frame sliding on the base; a controller fixed to the first support frame; and a straightening assembly disposed on the first support frame. The straightening assembly includes a lifting telescopic rod fixed to the first support frame. This application uses a clamp to fit over the trunk of the seedling, and then, in conjunction with the straightening assembly and multiple cylinders, straightens the bent seedling. The seedling height is then calculated using an angle measuring device and a length measuring device. In actual use, the device needs to be moved to the seedling, its position adjusted, and the top and bottom ends of the seedling clamped before straightening it and finally measuring. Firstly, this measurement method is very cumbersome, and secondly, straightening the seedling before measuring will inevitably damage it. Therefore, a laser array radar high-frequency tree height measurement system is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide a high-frequency tree height measurement system using laser array radar, which can quickly output the height of seedlings based on three-dimensional point clouds.

[0005] The specific technical solution adopted by this invention is as follows: A high-frequency tree height measurement system using a laser array radar includes a monitoring device. The monitoring device includes a mobile frame, on which a computer is fixedly mounted. A mounting plate is movably mounted on the mobile frame, and sensor components are mounted on the mounting plate. The high-frequency tree height measurement system includes a data acquisition module, a processing module, and an output module. The acquisition module emits and diffuses wide-angle laser pulses through the sensor components, and then uses a SPAD array as a detector to receive pixel points, ultimately acquiring complete point cloud data of the seedling; The processing module reads the complete point cloud data of the seedlings acquired by the acquisition module through the computer. The processing module's processing procedure includes the following steps: Step a; Preprocess the original point cloud by denoising filtering, downsampling, and gravity orientation calibration; Step b: Fit the ground plane using the RANSAC algorithm and determine the reference height, then remove ground points through height filtering; Step c; Use DBSCAN clustering to separate the main body and remove weed interference; Step d: Detect the crown vertices in the point cloud after segmenting the seedling point cloud; then calculate the base height by subtracting the reference plane height from the Z coordinate of the crown vertex. Step e; Finally, apply penetration compensation based on canopy density and tilt compensation based on trunk inclination angle to obtain the final height value; The output module outputs the final height.

[0006] Preferably, the sensor assembly uses a flash lidar, employing a VCSEL laser array to emit diffused wide-angle laser pulses; wavelength: 905nm; a single pulse covers the entire field of view, with an angle of 60°×45°; single-photon avalanche diodes are used as SPAD arrays to complete the detector receiving pixels, with a resolution of 320×240.

[0007] Preferably, in step a, points outside the mean of 1.5σ are removed by statistical filtering to complete filtering and noise reduction, and voxel lattice filtering is used to reduce the number of points while maintaining features to complete downsampling. Finally, the vertical direction is determined by PCA principal component analysis to complete gravity alignment.

[0008] Preferably, in step b, the RANSAC random sampling consensus algorithm is first used to robustly fit a ground plane model from the original point cloud; the RANSAC random sampling consensus algorithm iteratively calculates the plane equation ax+by+cz+d=0 through a random sampling point set, filters the interior point set that meets the threshold, and finally outputs the optimal plane parameters; the ground reference height is extracted based on the fitted plane; then height filtering is performed; with the reference height as a reference, all point clouds with Z coordinates lower than the reference height plus a safety threshold are removed, thereby separating the ground surface from the target seedling point cloud.

[0009] Preferably, in step c, specifically: the DBSCAN clustering algorithm is used to divide the point cloud above the ground into several clusters by setting a neighborhood radius and a minimum point count threshold. Density reachability is used to separate the main seedling clusters from weed and noise discrete point clusters. Then, dual filtering is performed: first, height filtering; clusters below the minimum effective height are removed, and short weeds are excluded; then, density verification; sparse clusters with insufficient points per unit volume are removed, eliminating interference from high-level weeds, and finally, a dense and complete main seedling point cloud is output. This method is highly robust to irregular tree shapes and complex backgrounds and can adapt to the natural undulations of the nursery terrain.

[0010] Preferably, in step d, the area near the highest point in space is first extracted, and density weighting is performed by calculating the local point cloud density. The highest point in the high-density area is selected as the effective crown vertex. Then, the base height is calculated: the Z coordinate value of the crown vertex is subtracted from the ground reference height fitted in step b determined by RANSAC, and the vertical spatial distance from the ground surface to the crown top of the sapling is directly obtained. This base height has eliminated the influence of terrain undulations and provides core data for subsequent compensation and correction. The typical accuracy can reach ±1.5cm.

[0011] Preferably, in step e, firstly, penetration compensation is applied based on the density of the canopy point cloud; the lower the density, the greater the compensation, used to offset the underestimation of the laser caused by scattering by the leaves; secondly, tilt compensation is calculated based on the tilt angle of the central axis fitted by the trunk point cloud, using a geometric correction formula. △H = h(1 -cosθ) Wherein, △H is the tilt compensation amount; h is the uncorrected base height, i.e., the straight-line distance between the crown apex and the reference plane; θ is the angle between the trunk and the vertical direction, eliminating the vertical projection error caused by the sapling's tilt; the two compensations are superimposed on the base height, and finally output the accurate true vertical height of the sapling, reducing the overall error from ±3cm to ±0.8cm.

[0012] Preferably, the processing module outputs results after verifying point cloud integrity, height rationality, and structural continuity. Specifically, it first verifies point cloud integrity, requiring at least 1000 valid points, while also detecting projection coverage density and whether there are voids or defects in key areas. Then, it verifies height rationality, triggering an alert if the output deviates from historical data by more than 15%. Finally, it verifies structural continuity by detecting faults exceeding 20cm through vertical layered slicing analysis. Once all verifications pass, the system outputs a precise height value with a confidence rating. If verification fails, an early warning mechanism is activated based on the problem level, and the system automatically retests. This process reduces the error output rate to below 0.5%, processes a single plant in just 0.3 seconds, and can effectively handle weed coverage rates as high as 40% and strong winds of up to 8 m / s.

[0013] The technical effects achieved by this invention are as follows: In this invention, the acquisition module emits and diffuses wide-angle laser pulses through the sensor component 4, and then uses a SPAD array as a detector to receive pixel points, finally acquiring complete point cloud data of the seedling; the processing module reads the complete point cloud data of the seedling acquired by the acquisition module through the computer 2, and finally outputs the seedling height based on the point cloud data. For seedlings that are tilted, the accurate seedling height can be output.

[0014] This invention utilizes a high-frequency laser array radar to measure seedling height. The key lies in high-resolution point cloud acquisition, accurate ground segmentation and seedling point cloud segmentation, and reasonable height calculation to avoid mistaking the lowest point of branches and leaves for a ground point. Through point cloud processing technology, automated seedling height measurement can be achieved. Attached Figure Description

[0015] Figure 1 This is a system block diagram of a high-frequency tree height measurement system for laser array radar according to the present invention; Figure 2 This is a flowchart of the operation of a high-frequency tree height measurement system for a laser array radar according to the present invention; Figure 3 This is an overall structural diagram of the monitoring device in a high-frequency tree height measurement system for laser array radar according to the present invention.

[0016] In this invention, 1. a mobile frame; 2. a computer; 3. a mounting plate; and 4. a sensor assembly. Detailed Implementation

[0017] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0018] like Figures 1-3 As shown, a high-frequency tree height measurement system for laser array radar includes a monitoring device, which includes a mobile frame 1. A computer 2 is fixedly mounted on the base plate of the mobile frame 1, and a mounting plate 3 is movably mounted on the mobile frame 1. A sensor assembly 4 is mounted on the mounting plate 3. The high-frequency tree height measurement system includes an acquisition module, a processing module, and an output module. The acquisition module emits and diffuses wide-angle laser pulses through sensor component 4, and then uses SPAD array as a detector to receive pixel points, finally acquiring complete point cloud data of the seedling; The processing module reads the complete point cloud data of the seedlings acquired by the acquisition module through computer 2. The processing module's process includes the following steps: Step a; Preprocess the original point cloud by denoising filtering, downsampling, and gravity orientation calibration; Step b: Fit the ground plane using the RANSAC algorithm and determine the reference height, then remove ground points through height filtering; Step c; Use DBSCAN clustering to separate the main body and remove weed interference; Step d: Detect the crown vertices in the point cloud after segmenting the seedling point cloud; then calculate the base height by subtracting the reference plane height from the Z coordinate of the crown vertex. Step e; Finally, apply penetration compensation based on canopy density and tilt compensation based on trunk inclination angle to obtain the final height value; The output module outputs the final height.

[0019] Preferably, the sensor assembly 4 adopts a flash lidar, which uses a VCSEL laser array to emit diffused wide-angle laser pulses; wavelength: 905nm; a single pulse covers the entire field of view, with an angle of 60°×45°; single-photon avalanche diodes are used as SPAD arrays to complete the detector receiving pixels, with a resolution of 320×240.

[0020] This invention utilizes a high-frequency laser array radar to measure seedling height. The key lies in high-resolution point cloud acquisition, accurate ground segmentation and seedling point cloud segmentation, and reasonable height calculation to avoid mistaking the lowest point of branches and leaves for a ground point. Through point cloud processing technology, automated seedling height measurement can be achieved.

[0021] Preferably, in step a, points outside the mean of 1.5σ are removed by statistical filtering to complete filtering and noise reduction, and voxel lattice filtering is used to reduce the number of points while maintaining features to complete downsampling. Finally, the vertical direction is determined by PCA principal component analysis to complete gravity alignment.

[0022] Preferably, in step b, the RANSAC random sampling consensus algorithm is first used to robustly fit a ground plane model from the original point cloud; the RANSAC random sampling consensus algorithm iteratively calculates the plane equation ax+by+cz+d=0 through a random sampling point set, filters the set of interior points that meet the 2cm threshold, and finally outputs the optimal plane parameters; the ground reference height is extracted based on the fitted plane; then height filtering is performed; with the reference height as a reference, all point clouds with Z coordinates lower than the reference height, such as a 10cm safety threshold, are removed, thereby separating the ground surface from the target seedling point cloud.

[0023] Compared to traditional altimeters, this invention uses an area array radar to simultaneously acquire three-dimensional morphological parameters, including crown width and branch angles, without requiring contact with the seedlings to avoid damage.

[0024] Preferably, in step c, specifically: the DBSCAN clustering algorithm is used, and by setting a neighborhood radius and a minimum point count threshold; a typical value for the neighborhood radius is 10cm; the minimum point count threshold is 20 points, to divide the point cloud above the ground into several clusters. Density reachability is used to separate the main seedling clusters from weed and noise discrete point clusters. Then, dual filtering is performed: first, height filtering; clusters below a minimum effective height of 30cm are removed, eliminating short weeds; then density verification; sparse clusters with less than 50 points / m³ per unit volume are removed, eliminating interference from high-level weeds, and finally outputting a dense and complete main seedling point cloud. This method is highly robust to irregular tree shapes and complex backgrounds and can adapt to the natural undulations of the nursery terrain.

[0025] Preferably, in step d, the area near the highest point in space is first extracted; for example, the top 10cm height zone. Density weighting is performed by calculating the local point cloud density of neighboring points within a radius of 5cm, and the highest point in the high-density area is selected as the effective crown vertex to avoid single-point noise interference. Then, the base height is calculated: the Z coordinate value of the crown vertex is subtracted from the ground reference height fitted in step b determined by RANSAC; that is, the average height of the lowest 5% of points of the fitted plane in step b. The vertical spatial distance from the ground surface to the crown top of the sapling is directly obtained. This base height has eliminated the influence of terrain undulations and provides core data for subsequent compensation and correction. The typical accuracy can reach ±1.5cm.

[0026] Preferably, in step e, penetration compensation is first applied based on the density of the canopy point cloud; typically, the top 30cm region is calculated; the lower the density, the greater the compensation; for example, an increase of 0.5cm for every 10% decrease in density, with a maximum compensation of 5cm, to compensate for the underestimation caused by laser scattering by leaves; secondly, tilt compensation is calculated based on the tilt angle of the central axis fitted to the trunk point cloud, using a geometric correction formula: △H = h(1 -cosθ) Wherein, △H is the tilt compensation amount; h is the uncorrected base height, i.e., the straight-line distance between the crown apex and the reference plane; θ is the angle between the trunk and the vertical direction, eliminating the vertical projection error caused by the sapling's tilt; the two compensations are superimposed on the base height, and finally output the accurate true vertical height of the sapling, reducing the overall error from ±3cm to ±0.8cm.

[0027] Preferably, the processing module outputs results after verifying point cloud integrity, height rationality, and structural continuity. Specifically, it first verifies point cloud integrity, requiring at least 1000 valid points, while also detecting projection coverage density and whether there are voids or defects in key areas. Then, it verifies height rationality, triggering an alert if the output deviates from historical data by more than 15%. Finally, it verifies structural continuity by detecting faults exceeding 20cm through vertical layered slicing analysis. Once all verifications pass, the system outputs a precise height value with a confidence rating. If verification fails, an early warning mechanism is activated based on the problem level, and the system automatically retests. This process reduces the error output rate to below 0.5%, processes a single plant in just 0.3 seconds, and can effectively handle weed coverage rates as high as 40% and strong winds of up to 8 m / s.

[0028] In this invention, the acquisition module emits and diffuses wide-angle laser pulses through sensor component 4, and then uses a SPAD array as a detector to receive pixel points, finally acquiring complete point cloud data of the seedling; the processing module reads the complete point cloud data of the seedling acquired by the acquisition module through computer 2, and finally outputs the seedling height based on the point cloud data. For seedlings that are tilted, it can output accurate seedling height.

[0029] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A laser array radar high-frequency tree height measurement system, comprising a monitoring device, the monitoring device comprising a mobile frame (1), a computer (2) fixedly mounted on the base plate of the mobile frame (1), and a mounting plate (3) movably mounted on the mobile frame (1), characterized in that: The mounting plate (3) is equipped with a sensor assembly (4); the high-frequency tree height measurement system includes an acquisition module, a processing module and an output module; The acquisition module emits a diffused wide-angle laser pulse through the sensor component (4), and then uses a SPAD array as a detector to receive pixel points, and finally obtains complete point cloud data of the seedling; The processing module reads the complete point cloud data of the seedlings acquired by the acquisition module through the computer (2), and the processing module's processing procedure includes the following steps: Step a; Preprocess the original point cloud by denoising filtering, downsampling, and gravity orientation calibration; Step b: Fit the ground plane using the RANSAC algorithm and determine the reference height, then remove ground points through height filtering; Step c; Use DBSCAN clustering to separate the main body and remove weed interference; Step d: Detect the crown vertices in the point cloud after segmenting the seedling point cloud; then calculate the base height by subtracting the reference plane height from the Z coordinate of the crown vertex. Step e; Finally, apply penetration compensation based on canopy density and tilt compensation based on trunk inclination angle to obtain the final height value; The output module outputs the final height.

2. The high-frequency tree height measurement system for laser array radar according to claim 1, characterized in that: The sensor assembly (4) adopts a flash lidar, which uses a VCSEL laser array to emit diffused wide-angle laser pulses; Wavelength: 905nm; a single pulse covers the entire field of view at an angle of 60°×45°; single-photon avalanche diodes are used as SPAD arrays to complete the detector's pixel reception, with a resolution of 320×240.

3. The laser array radar high-frequency tree height measurement system according to claim 2, characterized in that: In step a, points outside the mean of 1.5σ are removed by statistical filtering to remove noise, and voxel lattice filtering is used to reduce the number of points while preserving features to achieve downsampling. Finally, PCA principal component analysis is used to determine the vertical direction to achieve gravity alignment.

4. The high-frequency tree height measurement system for laser array radar according to claim 3, characterized in that: In step b, the RANSAC random sampling consensus algorithm is first used to fit a ground plane model from the original point cloud; the RANSAC random sampling consensus algorithm iteratively calculates the plane equation ax+by+cz+d=0 through a random sampling point set, filters the set of interior points that meet the threshold, and finally outputs the optimal plane parameters; the ground reference height is extracted based on the fitted plane. Then, height filtering is performed; using the reference height as a reference, all point clouds with Z coordinates lower than the reference height plus a safety threshold are removed, thereby separating the ground surface from the target sapling point cloud.

5. The laser array radar high-frequency tree height measurement system according to claim 4, characterized in that: In step c, specifically: the DBSCAN clustering algorithm is used to divide the point cloud above the ground into several clusters by setting the neighborhood radius and the minimum number of points threshold. The density reachability is used to separate the main seedling cluster from the discrete point clusters of weeds and noise. Then, double filtering is performed: first, height filtering; clusters below the minimum effective height are removed, and short weeds are excluded; then density verification; sparse clusters with insufficient points per unit volume are removed, and interference from high-level weeds is eliminated. Finally, a dense and complete main seedling point cloud is output.

6. The laser array radar high-frequency tree height measurement system according to claim 5, characterized in that: In step d, the area near the highest point in space is first extracted, and density weighting is performed by calculating the local point cloud density. The highest point in the high-density area is selected as the effective crown vertex. Then, the base height is calculated: the Z coordinate value of the crown vertex is subtracted from the ground reference height fitted in step b determined by RANSAC, and the vertical spatial distance from the ground surface to the crown top of the sapling is directly obtained.

7. The high-frequency tree height measurement system for laser array radar according to claim 6, characterized in that: In step e, firstly, penetration compensation is applied based on the canopy point cloud density to offset the underestimation of the laser caused by scattering from the leaves; secondly, tilt compensation is calculated based on the tilt angle of the central axis fitted by the trunk point cloud, using a geometric correction formula. △H = h(1 -cosθ) Where △H is the tilt compensation amount; h is the uncorrected base height, i.e., the straight-line distance between the crown apex and the reference plane; θ is the angle between the trunk and the vertical direction, eliminating the vertical projection error caused by the sapling's tilt; the two compensations are superimposed on the base height to finally output the accurate true vertical height of the sapling.

8. The high-frequency tree height measurement system for laser array radar according to claim 7, characterized in that: The processing module outputs results after verifying point cloud integrity, height rationality, and structural continuity. Specifically, it first verifies point cloud integrity, requiring at least 1000 valid points, while also detecting projection coverage density and whether there are voids or defects in key areas. Then, it verifies height rationality, triggering an alert if the output deviates from historical data by more than 15%. Finally, it verifies structural continuity by detecting faults exceeding 20cm through vertical layered slicing analysis. Once all verifications are successful, the system outputs a precise height value with a confidence rating. If verification fails, an early warning mechanism is activated based on the problem level, and the system automatically retests. This process reduces the error output rate to below 0.5%, and the processing time for a single plant is only 0.3 seconds. It can effectively cope with weed coverage rates as high as 40% and strong wind interference environments of 8 m / s.

Images