Effective leaf area index extraction method and system based on point cloud data

Abstract

The present invention discloses a method and system for extracting the effective leaf area index (ELAI) based on point cloud data. First, three-dimensional point cloud data containing planted areas is acquired. Then, using the color and geometric information of the point cloud data, the color index (EXG) of the point cloud data is calculated, along with a set of height and slope thresholds. The Otsu filter method, combined with an improved slope-based filtering method, is used to filter out exposed ground points, thereby obtaining a crop point cloud. The crop point cloud is then projected onto an upper hemisphere with a radius of 1, then onto a plane and binarized to obtain a hemispherical image. Finally, the hemispherical image is divided into 18 concentric rings at a zenithal perspective, and the porosity at specific perspectives is calculated to estimate the ELAI. This method filters out remaining exposed ground points, avoiding the problem of using only the Otsu filter method to automatically determine the color index (EXG) threshold to filter out ground points, which can result in a small number of ground points remaining and thus lead to an overestimation of the ELAI.

Description

Technical Field

[0001] The present invention relates to the technical field of point cloud data processing, and in particular to a method and system for extracting effective leaf area index based on point cloud data. Background Art

[0002] The leaf area index (LAI) is a key parameter in vegetation photosynthesis and evapotranspiration models, playing a crucial role in the exchange of matter and energy between the soil, vegetation, and atmosphere. Therefore, LAI is of great significance for crop growth monitoring, crop yield prediction, and plant growth model simulation. The leaf area index, obtained under the assumption that leaf elements are spatially randomly distributed, is called the effective leaf area index (LAIe). Traditional ground-based methods for measuring LAI primarily rely on direct and indirect measurements, but these methods are time-consuming, tedious, destructive, and unsuitable for large areas. Remote sensing, with its non-contact nature and wide coverage, is the primary means of producing regional and global LAI products and has garnered widespread attention in agricultural applications such as crop management and yield estimation. Remote sensing methods for obtaining LAI primarily include those based on statistical models and those based on physical models. Statistical methods derive vegetation indices by measuring parameters closely related to the LAI. Single-factor or multi-factor models combining the vegetation index and LAI are then used to predict LAI. However, these methods suffer from unclear mechanisms and regional variations in the obtained models. The physical model-based method mainly considers the non-Lambertian and bidirectional reflectance characteristics of vegetation canopy reflectance, and inverts LAI by constructing models of biophysical parameters such as surface reflectance and leaf area index. However, there are problems such as the inversion process is complex and requires a large number of input parameters.

[0003] In recent years, with the continuous improvement in the stability, safety, and controllability of unmanned aerial vehicle (UAV) flight platforms, their applications for crop parameter monitoring have rapidly grown. UAV platforms provide a variety of cost-effective remote sensing data types with very high spatial resolution and flexible acquisition cycles, including RGB, multispectral, hyperspectral, LiDAR, microwave, and thermal infrared data. LiDAR is an active remote sensing technology that has rapidly developed internationally in recent years. Many studies have begun to use LiDAR data to infer forest porosity and LAI. The point cloud data generated by LiDAR scans includes information such as 3D coordinates, color, reflection intensity, and echo count. Furthermore, point cloud data can be generated from RGB images using SfM technology. Due to the high cost of acquiring point cloud data from LiDAR and the complexity of data processing, low-cost consumer UAV systems consisting of RGB digital cameras and lightweight drones have attracted widespread attention. Therefore, generating point cloud data from RGB images has become an effective and low-cost alternative.

[0004] LAI is primarily estimated from point cloud data through correlation with porosity. For example, LAI is estimated by calculating porosity based on information such as reflection intensity and echo count from point clouds acquired by LiDAR. While this method can achieve good results, LiDAR data acquisition is expensive. Alternatively, LAI can be estimated using statistical models that correlate plant geometry and structural parameters derived from point cloud data with field-measured LAI. However, these methods require additional field-measured LAI values ​​to establish an empirical relationship, and the parameterization for specific objects and locations may not be applicable to other scenarios.

[0005] RGB images acquired by drones equipped with digital cameras can be used to generate cost-effective 3D point cloud data using the SfM technique, a low-cost remote sensing alternative to airborne and ground-based LiDAR. Furthermore, this method offers an unparalleled combination of temporal and spatial resolution, effectively assessing the spatiotemporal variability of leaf area index (LAI) across large agricultural fields. To our knowledge, few studies have used this approach to derive the effective leaf area index (LAIe) of winter wheat from information derived from SfM point clouds. Therefore, this study used SfM 3D point clouds generated from drone RGB imagery to predict LAIe by calculating porosity based on the structural and color information of the crop point cloud data. Summary of the Invention

[0006] In view of this, the purpose of the present invention is to provide a method and system for extracting the effective leaf area index based on point cloud data. The method uses point cloud data collected by drones to extract the effective leaf area index of crops.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] The effective leaf area index extraction method based on point cloud data provided by the present invention comprises the following steps:

[0009] Obtain three-dimensional point cloud data of the planted area;

[0010] Use the color information and geometric information of the point cloud data to calculate and obtain crop point cloud data;

[0011] Project the crop point cloud data onto the upper hemisphere, then project it onto the plane and binarize it to obtain a hemispherical image;

[0012] The hemispherical image is divided into several concentric rings at the zenithal angle of view, and the porosity at each zenithal angle of view is calculated;

[0013] The effective leaf area index was estimated based on the porosity.

[0014] Furthermore, the effective leaf area index is estimated using a single-angle inversion method or a multi-angle inversion method.

[0015] Furthermore, the method of using the color information and geometric information of the point cloud data to calculate and obtain the crop point cloud data is performed according to the following steps:

[0016] Calculate the EXG index based on the color information of the point cloud data;

[0017] The ground points are filtered out using the Otsu method based on the EXG index of the point cloud data;

[0018] Determine the height threshold set and slope threshold set of the plant planting area according to the geometric information of the point cloud data;

[0019] According to the height threshold set and slope threshold set of the plant area, an improved algorithm based on slope change is used to filter out the residual ground points and obtain the crop point cloud data.

[0020] Furthermore, the hemispherical image is specifically processed according to the following steps:

[0021] Use coordinate system transformation to transform the crop point cloud data in the 3D rectangular coordinate system into the spherical coordinate system and project it onto a hemisphere with a radius of 1;

[0022] Project the spherical point cloud onto a plane using equal-area azimuthal or stereographic projection;

[0023] The hemispherical image is binarized to obtain a binary hemispherical image.

[0024] Furthermore, the porosity is specifically determined by the following steps:

[0025] The binary hemispherical image is divided into several concentric rings. Each concentric ring corresponds to a zenithal view. The median value of each ring's view represents the zenithal view value of the ring. The porosity of each zenithal view is calculated using the following formula:

[0026]

[0027] Among them, N l (θ) is the leaf pixel content under viewing angle θ, N s (θ) is the background pixel content at viewing angle θ.

[0028] Furthermore, the single-view inversion method estimation of the effective leaf area index is calculated according to the following formula:

[0029]

[0030] Where θ is the zenith angle of the incident light, G(θ) represents the projected area of ​​the unit blades randomly distributed in space under the viewing angle θ, and P(θ) is the porosity under the zenith angle of the incident light.

[0031] Furthermore, the multi-view inversion method for estimating the effective leaf area index is calculated according to the following formula:

[0032]

[0033] Among them, n is the number of concentric rings divided, P(θ i ) is the porosity of the i-th concentric ring, and Δθ is the zenith angle difference between adjacent rings.

[0034] The effective leaf area index extraction system based on point cloud data provided by the present invention includes a memory, a processor and a computer program stored in the memory and executable on the processor, and the processor executes the above method when executing the program.

[0035] The effective leaf area index extraction system based on point cloud data provided by the present invention includes a data acquisition unit, an Otsu combined with an improved filter based on slope change, a point cloud spherical projection unit, a point cloud plane projection unit, a porosity calculation unit, and a crop effective leaf area index estimation unit;

[0036] The data acquisition unit is used to acquire drone images of the crop planting area and generate a three-dimensional point cloud using the SfM algorithm;

[0037] The Otsu method, combined with an improved slope-change-based filter, is used to preprocess the original point cloud. The point cloud is sampled to obtain a square sampling area based on the actual crop planting spacing. The EXG index is calculated based on the color information of the point cloud data, and the Otsu filtering method is then used to preliminarily filter out most ground points. Based on the geometric information of the point cloud data, the height threshold set and the slope threshold set of the plant area are calculated. The improved slope-change-based filtering method is used to filter out the remaining ground points to obtain the crop point cloud.

[0038] The point cloud spherical projection unit is used to project the point cloud data in the three-dimensional Cartesian coordinate system onto a sphere with a radius of 1 to obtain a spherical point cloud;

[0039] The point cloud plane projection unit is used to project the spherical point cloud onto a plane using equal-area azimuthal projection and stereographic plane projection and perform binarization to obtain a binary hemispherical image;

[0040] The porosity calculation unit is used to calculate the porosity at 18 zenith viewing angles according to the porosity calculation formula.

[0041] The crop effective leaf area index estimation unit is used to estimate the effective leaf area index using a single-angle inversion method and a multi-angle inversion method according to the calculated porosity.

[0042] Furthermore, the porosity is specifically determined by the following steps:

[0043] The binary hemispherical image is divided into several concentric rings. Each concentric ring corresponds to a zenithal view. The median value of each ring's view represents the zenithal view value of the ring. The porosity of each zenithal view is calculated using the following formula:

[0044]

[0045] Among them, N l (θ) is the leaf pixel content under viewing angle θ, N s (θ) is the background pixel content at viewing angle θ;

[0046] The effective leaf area index is calculated according to the following formula:

[0047]

[0048] Where θ is the zenith angle of the incident light, G(θ) represents the projected area of ​​the unit blades randomly distributed in space under the viewing angle θ, and P(θ) is the porosity under the zenith angle of the incident light.

[0049] The beneficial effects of the present invention are:

[0050] The present invention provides a method and system for extracting the effective leaf area index of plant species based on point cloud data. Compared to crop point clouds generated by laser radar scanning, imagery captured by drone-mounted digital cameras can also generate similar point clouds at a significantly lower cost. This method uses Otsu combined with an improved slope-based filtering method to filter ground points from the point cloud. By leveraging the color and geometric information of the point cloud data, bare ground points are filtered out to the greatest extent possible, overcoming the error caused by these ground points being misinterpreted as crop points in extracting the effective leaf area index. After obtaining the crop point cloud without the bare ground points, the method, based on the principles of digital hemispherical photography, converts the crop point cloud from a three-dimensional rectangular coordinate system to a spherical coordinate system. Specifically, the point cloud is projected onto a hemisphere with a radius of 1. The point cloud on the hemisphere is then projected onto a plane, and finally binarized to obtain a hemispherical image. After obtaining the hemispherical image, the porosity corresponding to the viewing angle is calculated, and the effective leaf area index is then calculated using single-angle and multi-angle inversion methods.

[0051] The improved filtering method based on slope change proposed in the present invention can utilize the geometric information of point cloud data to further filter out ground points by obtaining a height difference threshold set and a slope threshold set based on the point cloud data, thereby reducing the problem of some ground points not being filtered out due to the Otsu filtering method automatically determining the threshold using color information, that is, reducing the problem of ground points being treated as crop points, resulting in an overestimation of the effective leaf area index.

[0052] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to make the purpose, technical solutions and beneficial effects of the present invention more clear, the present invention provides the following drawings for illustration:

[0054] Figure 1 Schematic diagram of the overall method of the present invention.

[0055] Figure 2 Schematic diagram of a winter wheat farmland research area in a specific embodiment of the present invention. The selected research area is a winter wheat planting area near Melbourne in southeastern Ontario, Canada.

[0056] Figure 3 Schematic diagram of the Otsu method combined with an improved filter based on slope change used in a specific embodiment of the present invention.

[0057] Figure 4 Schematic diagram of the process of extracting the spherical projection of the point cloud and the planar projection of the spherical point cloud in a specific embodiment of the present invention.

[0058] Figure 5 Schematic diagram of dividing a hemispherical image into 18 concentric rings in a specific embodiment of the present invention.

[0059] Figure 6 The figure is a scatter plot of the correlation between the estimated effective leaf area index of winter wheat and the measured leaf area index obtained in a specific embodiment of the present invention.

[0060] Figure 7 A graph is prepared for estimating the effective leaf area index of winter wheat in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0061] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0062] Example 1

[0063] like Figure 1 As shown, the effective leaf area index extraction method based on point cloud data provided in this embodiment includes the following steps:

[0064] Obtain three-dimensional point cloud data of the planted area;

[0065] Calculate the EXG index based on the color information of the point cloud data;

[0066] The Otsu method is used to filter out most of the ground points based on the EXG index of the point cloud data;

[0067] Determine the height threshold set and slope threshold set of the plant planting area according to the geometric information of the point cloud data;

[0068] According to the height threshold set and slope threshold set of the plant area, an improved algorithm based on slope change is used to filter out the remaining ground points to obtain the plant canopy data information;

[0069] The plant canopy data information in this embodiment is determined according to the following steps:

[0070] According to the color information of the plant point cloud data, the EXG index is calculated by the following formula:

[0071] EXG=2G-BR

[0072] In this embodiment, the residual ground point data is filtered out according to the following steps:

[0073] The point cloud data is divided into 100×100 grids and numbered. The lowest point in each grid is used as the reference point of the grid. The height difference Δh and the slope value slope of other points in the grid are calculated.

[0074] In this embodiment, the height difference Δh and the slope value slope are calculated according to the following formula:

[0075] Δh i =z i -z0

[0076] Where Δh i For all points p in the grid except the lowest point i The height difference from the lowest point p0, z i For p i The z coordinate value of p0, z0 is the z coordinate value of p0.

[0077]

[0078] Among them, slope i For all points p in the grid except the lowest point i The slope value with respect to the lowest point p0, x i 、y i For p i x0 and y0 are the x and y coordinate values ​​of p0, respectively.

[0079] In this embodiment, the average value of the height difference and slope value of all points in each grid of the point cloud data is used as the height difference threshold and slope threshold for separating the crop point cloud and the ground point cloud in each grid. Points with a height difference lower than the height difference threshold and a slope lower than the slope threshold will be filtered out as ground points.

[0080] The hemispherical image in this embodiment is obtained according to the following steps: based on the obtained crop point cloud data, the point cloud data is projected onto an upper hemisphere with a radius of 1; based on the projection, a spherical point cloud is obtained, and it is projected onto a plane using equal-area azimuthal projection and stereographic projection, and then binarized to obtain a hemispherical image.

[0081] In this embodiment, the effective leaf area index is estimated by calculating the porosity based on the obtained binary hemispherical image.

[0082] In this embodiment, the porosity is determined according to the following steps:

[0083] The binary hemispherical image is divided into 18 concentric rings, each of which corresponds to a 5° zenith angle. The median value of each ring angle represents the zenith angle value of that ring. The porosity of each zenith angle is calculated using the following formula:

[0084]

[0085] Among them, N l (θ) is the leaf pixel content under viewing angle θ, N s (θ) is the background (sky or soil) pixel content at viewing angle θ.

[0086] The porosity corresponding to the zenithal viewing angle is calculated based on the binary hemispherical image, and the effective leaf area index is calculated using the single-angle inversion method or the multi-angle inversion method.

[0087] According to the calculated porosity, the effective leaf area index is calculated by the following two formulas:

[0088]

[0089] This formula is called the single-angle inversion method, where P(58°) is the porosity at a zenith angle of 58°, which is approximately the porosity for the viewing angle interval of the 12th concentric ring, corresponding to the viewing angle interval of 55° to 60°. This formula is derived from the Beer-Lambert law. When θ is 58°, the projection function G(θ) is independent of the blade inclination angle and is always equal to 0.5.

[0090] The Beer-Lambert law for calculating the effective leaf area index is as follows:

[0091]

[0092] Where θ is the zenith angle of the incident light, which refers to the angle between the incident direction of the light and the zenith direction. From the perspective of hemispherical photography, this is the viewing angle of the shot. G(θ) is the projection function, which is related to the zenith angle and the leaf inclination angle. It represents the projected area of ​​unit leaves randomly distributed in space under the viewing angle of θ. P(θ) is the porosity under the zenith angle of the incident light, which is the probability that light will pass through the canopy and reach the surface without being intercepted.

[0093]

[0094] This formula is called the multi-angle inversion method, where n is the number of concentric rings, P(θ i ) is the porosity of the i-th concentric ring, Δθ is the zenith angle difference between adjacent rings, and when the binary hemispherical image is divided into 18 concentric rings, n = 18 and Δθ = 5°.

[0095] According to the above operation, all sampling areas are traversed to obtain the effective leaf area index of all sampling points in the plant area.

[0096] Example 2

[0097] The effective leaf area index extraction method based on point cloud data provided in this embodiment includes the following steps:

[0098] First, obtain the three-dimensional point cloud data of the plant area;

[0099] Then, the bare ground point cloud is filtered out based on the point cloud data to obtain the crop point cloud; the crop point cloud data is determined according to the following steps:

[0100] According to the color information of the plant point cloud data, the color index EXG is calculated, and the Otsu method is used to automatically determine the threshold between the ground point cloud and the crop point cloud to filter out most of the ground points.

[0101] Based on the point cloud data of the early stage of crop growth, the point cloud data was divided into 100×100 grids. The lowest point in each grid was used as the reference point of the grid, and the height difference and slope values ​​between the other points of the grid and the lowest point were calculated. The average of the height difference and slope values ​​of all points in each grid of the point cloud data was used as the height difference threshold and slope threshold for separating the crop point cloud from the ground point cloud in each grid.

[0102] Based on the obtained height difference threshold set and slope threshold set containing crop planting areas, they are applied to the point cloud data of other dates. Points with height difference lower than the height difference threshold and slope lower than the slope threshold will be filtered out as ground points.

[0103] Next, a hemispherical image of the crop is obtained based on the crop point cloud after filtering out the ground points. The hemispherical image is determined according to the following steps:

[0104] Use coordinate system transformation to transform the crop point cloud data in the 3D rectangular coordinate system into the spherical coordinate system and project it onto a hemisphere with a radius of 1.

[0105] Project a spherical point cloud onto a plane using either equal-area azimuthal or stereographic projection.

[0106] The hemispherical image is binarized to obtain a binary hemispherical image.

[0107] Finally, the effective leaf area index is estimated by calculating the porosity based on the acquired hemispherical image. The estimation of the effective leaf area index is determined according to the following steps:

[0108] The acquired hemispherical image is divided into 18 concentric rings of zenith angle of view.

[0109] The porosity at the zenith angle corresponding to the 12th ring was calculated, and the effective leaf area index was estimated using the single-angle inversion method.

[0110] The porosity of all rings corresponding to the zenithal perspective was calculated, and the effective leaf area index was estimated using the multi-angle inversion method.

[0111] Example 3

[0112] like Figure 1 and Figure 2 As shown, Figure 1 This is a schematic flow chart of the overall method of the present invention, Figure 2 The schematic diagram of the winter wheat farmland research area in the specific embodiment of the present invention is selected as the winter wheat planting area near London in southeastern Ontario, Canada. Figure 2 The leaf area index of winter wheat was actually measured at different sampling points on different observation dates.

[0113] The method for extracting effective leaf area index of crops based on drone point cloud provided in this embodiment includes the following steps:

[0114] Step 1: 32 sampling points were evenly spaced every 30 meters across the study area to measure the leaf area index of winter wheat throughout its growth cycle. During this process, drone digital images of the experimental area were captured simultaneously, and the SfM algorithm was used to generate point cloud data containing 3D coordinates from the drone images.

[0115] Step 2: Preprocess the original UAV point cloud, crop the 90m×210m crop area point cloud, and extract the 3D coordinates.

[0116] Step 3: Calculate the color index EXG. Based on the color information of the point cloud data, use the following formula to calculate EXG:

[0117] EXG=2G-BR

[0118] Step 4: Use the Otsu method to automatically determine the threshold of EXG between the bare ground point cloud and the winter wheat point cloud, and filter out most of the bare ground point cloud. Figure 3 As shown, Figure 3 Schematic diagram of the Otsu method combined with an improved slope change-based filter used in a specific embodiment of the present invention. (a), (d), (g), (j), (m), and (p) show the original point clouds on May 11, May 16, May 21, May 26, June 3, and June 11. (b), (e), (h), (k), (n), and (q) show the point clouds on May 11, May 16, May 21, May 26, June 3, and June 11 after filtering using the Otsu method.

[0119] Step 5: Obtain the height difference threshold set and the slope threshold set. Based on the point cloud data of the early growth period of winter wheat, first divide the point cloud data into 100×100 grids and number them. Use the lowest point cloud in each grid as the reference point of the grid, and calculate the height difference Δh and slope value between other points in the grid and the lowest point. The calculation formula is as follows:

[0120] Δh i =z i -z0 (2)

[0121] Where Δh i For all points p in the grid except the lowest point i The height difference from the lowest point p0, z i For p i The z coordinate value of p0, z0 is the z coordinate value of p0.

[0122]

[0123] Among them, slope i For all points p in the grid except the lowest point i The slope value with respect to the lowest point p0, x i 、y i For p i x0 and y0 are the x and y coordinate values ​​of p0, respectively.

[0124] The average value of the height difference and slope value of all points in each grid of the point cloud data is used as the height difference threshold and slope threshold for separating the winter wheat point cloud and the ground point cloud in each grid. This is applied to the point cloud data of other dates. Points with a height difference lower than the height difference threshold and a slope lower than the slope threshold will be filtered out as ground points. Finally, the point cloud after filtering out the ground points is obtained. Figure 3 As shown, Figure 3This is a schematic diagram of the Otsu combined with the improved slope change-based filter applied in the specific implementation plan of the present invention. (c), (f), (i), (l), (o), and (r) show the point clouds on May 11, May 16, May 21, May 26, June 03, and June 11 after using the Otsu filtering method combined with the improved slope change-based filtering method.

[0125] Step 6: Convert the point cloud data in the 3D rectangular coordinate system to the spherical coordinate system, that is, project it onto a hemisphere with a radius of 1. First, convert the 3D rectangular coordinate system to the spherical coordinate system:

[0126]

[0127] Then let r = 1 and transform to a three-dimensional rectangular coordinate system:

[0128]

[0129] The point cloud can be projected onto the upper hemisphere with a radius of 1.

[0130] like Figure 4 As shown; Figure 4 Schematic diagram of the process of extracting spherical and planar projections of a point cloud in a specific embodiment of the present invention. (a) and (b) are point clouds in a 3D rectangular coordinate system. (c) and (d) are point clouds projected onto the upper hemisphere with a radius of 1.

[0131] Step 7: Project the spherical point cloud onto a plane. There are two common methods for projecting a spherical point cloud onto a plane: equal-area azimuthal projection and stereographic projection. The relationship between the coordinates before and after the equal-area azimuthal projection is:

[0132]

[0133] The relationship between the coordinates before and after stereographic projection is:

[0134]

[0135] like Figure 5 As shown; Figure 5 Schematic diagram of the process of extracting spherical projections of a point cloud and planar projections of a spherical point cloud in a specific embodiment of the present invention. (e) shows the binary hemispherical image obtained using the equal-area azimuthal projection method. (f) shows the binary hemispherical image obtained using the stereographic plane projection method.

[0136] Step 8: Divide the hemispherical image into 18 concentric rings at different zenithal angles. The hemispherical image contains porosity information from 0° to 90°, so each concentric ring corresponds to a 5° viewing angle interval. The median value of each concentric ring's viewing angle interval represents the zenithal angle of that concentric ring.

[0137] like Figure 6 As shown; Figure 6 The concentric rings of the hemispherical image are divided under 18 zenith angles.

[0138] Step 9: Calculate the porosity at 18 zenith angles. The porosity calculation formula is as follows:

[0139]

[0140] Step 10: Estimate the effective leaf area index using the single-view inversion method and the multi-view inversion method. The single-view inversion method estimation formula is:

[0141]

[0142] Among them, P(58°) can be approximated as the porosity of the viewing angle interval of the 12th concentric ring, and the corresponding viewing angle interval is 55° to 60°.

[0143] The estimation formula of the multi-view inversion method is:

[0144]

[0145] Among them, n is the number of concentric rings divided, P(θ i ) is the porosity of the i-th concentric ring, and Δθ is the zenith angle difference between adjacent rings.

[0146] like Figure 6 As shown; Figure 6 A map of the effective leaf area index of winter wheat obtained in a specific embodiment of the present invention is shown. (a), (b), (c), (d), (e), and (f) are maps of the effective leaf area index for six dates, using stereographic projection and multi-angle inversion methods based on a three-dimensional point cloud dataset from an unmanned aerial vehicle. (a) is May 11, (b) is May 16, (c) is May 21, (d) is May 27, (e) is June 3, and (f) is June 11.

[0147] Step 11: Compare the effects of separating the winter wheat point cloud from the bare ground point cloud and different filtering methods on the LAIe estimation accuracy. This study uses the LAIe values ​​measured on site to evaluate the LAIe estimation accuracy of the UAV-based point cloud data under three conditions: no filtering, only using the Otsu filtering method, and using the Otsu filtering method combined with an improved slope-based filtering method. The R-squared value is used to describe the linear fit under the same estimation process over these six days.

[0148] like Figure 7 As shown; Figure 7is a scatter plot of the correlation between the estimated effective leaf area index and the measured leaf area index of winter wheat obtained in this embodiment under three filtering conditions; Figure 6 In, R 2 Indicates the degree of fit of the regression line to the estimated effective leaf area index of winter wheat. (a) is the case without filtering, (b) is the case using only the Otsu filtering method, and (c) is the case using the Otsu filtering method combined with the improved slope-based filtering method.

[0149] Example 4

[0150] The effective leaf area index extraction system based on point cloud data provided in this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the above method is implemented.

[0151] The effective leaf area index extraction system based on point cloud data provided in this embodiment includes a data acquisition unit, an Otsu-based filter based on improved slope changes, a point cloud spherical projection unit, a point cloud plane projection unit, a porosity calculation unit, and a crop effective leaf area index estimation unit;

[0152] The data acquisition unit is used to acquire drone images of the crop planting area and generate a three-dimensional point cloud using the SfM algorithm;

[0153] The Otsu method, combined with an improved slope-change-based filter, is used to preprocess the original point cloud. The point cloud is sampled to obtain a square sampling area based on the actual crop planting spacing. The EXG index is calculated based on the color information of the point cloud data, and the Otsu filtering method is then used to preliminarily filter out most ground points. Based on the geometric information of the point cloud data, the height threshold set and slope threshold set of the point cloud data are calculated. The improved slope-change-based filtering method is used to filter out the remaining ground points to obtain the crop point cloud.

[0154] The point cloud spherical projection unit is used to project the point cloud data in the three-dimensional Cartesian coordinate system onto an upper hemisphere with a radius of 1 to obtain a spherical point cloud;

[0155] The point cloud plane projection unit is used to project the spherical point cloud onto a plane using equal-area azimuthal projection and stereographic plane projection and perform binarization to obtain a binary hemispherical image;

[0156] The porosity calculation unit is used to calculate the porosity at 18 zenith viewing angles according to the porosity calculation formula.

[0157] The crop effective leaf area index estimation unit is used to estimate the effective leaf area index using a single-angle inversion method and a multi-angle inversion method according to the calculated porosity.

[0158] The above embodiments are merely preferred embodiments for the purpose of fully illustrating the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are within the scope of protection of the present invention. The scope of protection of the present invention shall be subject to the claims.

Claims

1. The effective leaf area index extraction method based on point cloud data is characterized by: The following steps are involved: Obtain three-dimensional point cloud data of the planted area; Use the color information and geometric information of the point cloud data to calculate and obtain crop point cloud data; Project the crop point cloud data onto the upper hemisphere, then project it onto the plane and binarize it to obtain a hemispherical image; The hemispherical image is divided into several concentric rings at the zenithal angle of view, and the porosity at each zenithal angle of view is calculated; Estimation of effective leaf area index based on porosity; The method of using the color information and geometric information of the point cloud data to calculate and obtain the crop point cloud data is carried out according to the following steps: Calculate the EXG index based on the color information of the point cloud data; The ground points are filtered out using the Otsu method based on the EXG index of the point cloud data; Determine the height threshold set and slope threshold set of the plant planting area according to the geometric information of the point cloud data; According to the height threshold set and slope threshold set of the plant area, an improved slope change-based algorithm is used to filter out the remaining ground points to obtain crop point cloud data; The hemispherical image is specifically obtained by the following steps: Use coordinate system transformation to transform the crop point cloud data in the 3D rectangular coordinate system into the spherical coordinate system and project it onto a hemisphere with a radius of 1; Project the spherical point cloud onto a plane using equal-area azimuthal or stereographic projection; Binarizing the hemispherical image to obtain a binary hemispherical image; The porosity is specifically determined by the following steps: The binary hemispherical image is divided into several concentric rings. Each concentric ring corresponds to a zenithal view. The median value of each ring's view represents the zenithal view value of the ring. The porosity of each zenithal view is calculated using the following formula: in, Perspective Lower leaf pixel content, Perspective Lower background pixel content; The single-view inversion method for estimating the effective leaf area index is calculated according to the following formula: in, is the zenith angle of the incident light, It means the viewing angle is The projected area of ​​the unit leaves under the condition of random distribution in space, The zenith angle of the incident light is The porosity below.

2. The effective leaf area index extraction method based on point cloud data according to claim 1, characterized in that: The effective leaf area index is estimated using a single-angle inversion method or a multi-angle inversion method.

3. The effective leaf area index extraction method based on point cloud data according to claim 1, characterized in that: The multi-view inversion method for estimating the effective leaf area index is calculated according to the following formula: Where n is the number of concentric rings divided, is the porosity of the i-th concentric ring, is the zenith angle difference between adjacent rings.

4. A system for extracting effective leaf area index based on point cloud data, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 3 is implemented.

5. Effective leaf area index extraction system based on point cloud data, characterized by: It includes a data acquisition unit, an Otsu-combined improved filter based on slope change, a point cloud spherical projection unit, a point cloud plane projection unit, a porosity calculation unit, and a crop effective leaf area index estimation unit; The data acquisition unit is used to acquire drone images of the crop planting area and generate a three-dimensional point cloud using the SfM algorithm; The Otsu method, combined with an improved slope-based filter, is used to preprocess the original point cloud. The point cloud is sampled to obtain a square sampling area based on the actual crop planting spacing. The EXG index is calculated based on the color information of the point cloud data, and the Otsu filtering method is then used to initially filter out most ground points. Based on the geometric information of the point cloud data, the improved slope-based filtering method is used to filter out the remaining ground points to obtain a crop point cloud. The point cloud spherical projection unit is used to project the point cloud data in the three-dimensional Cartesian coordinate system onto a sphere with a radius of 1 to obtain a spherical point cloud; The point cloud plane projection unit is used to project the spherical point cloud onto a plane using equal-area azimuthal projection and stereographic plane projection and perform binarization to obtain a binary hemispherical image; The porosity calculation unit is used to calculate the porosity at 18 zenith angles according to the porosity calculation formula; The crop effective leaf area index estimation unit is used to estimate the effective leaf area index using a single-angle inversion method and a multi-angle inversion method according to the calculated porosity; The porosity is specifically determined by the following steps: The binary hemispherical image is divided into several concentric rings. Each concentric ring corresponds to a zenithal view. The median value of each ring's view represents the zenithal view value of the ring. The porosity of each zenithal view is calculated using the following formula: in, Perspective Lower leaf pixel content, Perspective Lower background pixel content; The effective leaf area index is calculated according to the following formula: in, is the zenith angle of the incident light, It means the viewing angle is The projected area of ​​the unit leaves under the condition of random distribution in space, The zenith angle of the incident light is The porosity below.

Images