Vegetation leaf area index inversion method based on unmanned aerial vehicle laser radar data

Abstract

The embodiment of the invention provides a vegetation leaf area index inversion method based on unmanned aerial vehicle laser radar data, and is applied to the field of agricultural remote sensing monitoring. The method comprises the steps of preprocessing point cloud data of a target collection area, inputting the preprocessed point cloud data into a smooth constraint Markov random field model, obtaining ground point cloud data and point cloud data of each vegetation point to construct a canopy height model, and calculating the canopy height of the target collection area according to the preprocessed point cloud data and the canopy height model. The method comprises the following steps: constructing four-dimensional point cloud data of each vegetation point, dividing a vegetation region formed by each vegetation point into a plurality of subspaces according to a K-D tree algorithm and the dimension of the four-dimensional point cloud data of each vegetation point, performing nearest neighbor search on each subspace, estimating the density and volume of the vegetation region, and constructing a leaf area index inversion model by using a leaf area index formula. And obtaining a vegetation leaf area index. In this way, the vegetation leaf area index of the point cloud data coverage area can be rapidly and accurately obtained, and the inversion precision is improved.

Description

Technical Field

[0001] The present disclosure relates to the field of agricultural remote sensing monitoring, and particularly to a method for retrieving vegetation leaf area index based on unmanned aerial vehicle lidar data. Background Art

[0002] The leaf area index is an important indicator for evaluating the growth of farmland crops. It is defined as the ratio of the total single-sided area of crop leaves to the land area per unit land area, and is closely related to the physiological processes of crop growth. It is of great significance for research such as crop growth monitoring, yield estimation, and growth assessment. Traditional single-point leaf area index measurement methods include direct measurement methods and indirect optical instrument measurement methods. These methods have limitations such as a small measurement range, low efficiency, and high labor intensity, and are difficult to meet the needs of large-area, rapid, and accurate measurement. Unmanned aerial vehicle lidar technology, as an active remote sensing monitoring technology, has advantages such as high resolution, high precision, and the ability to penetrate the crop canopy to obtain three-dimensional information, providing a new technical means for the retrieval of the leaf area index.

[0003] Currently, the methods for estimating the leaf area index (LAI) using remote sensing technology are mainly divided into two categories: optical remote sensing-based and lidar remote sensing-based. Optical remote sensing shows good estimation ability at high LAI levels through the correlation between vegetation indices (such as the normalized difference vegetation index NDVI) and LAI; while lidar remote sensing can provide detailed vertical structure information of the crop canopy, providing higher precision support for the estimation of canopy LAI.

[0004] With the continuous improvement of the payload capacity of unmanned aerial vehicles (UAVs), the technology of UAV-mounted lidar (LiDAR) has gradually become a research hotspot. The point cloud density in LiDAR technology is a key parameter, which reflects the quality of lidar hardware equipment and data acquisition efficiency, and at the same time affects the selection and processing efficiency of subsequent data processing methods. The point cloud density is usually expressed as the number of laser foot points per unit area. By statistically analyzing the point cloud density value of the horizontal plane projection, the global density characteristics of the point cloud can be revealed. Therefore, how to use the point cloud data obtained by LiDAR technology to determine the correlation between the geometric characteristics of the point cloud data and the vegetation biological characteristics, and to achieve accurate estimation of the vegetation leaf area index has very important research significance. Summary of the Invention

[0005] The present disclosure provides a method, device, equipment, and storage medium for retrieving vegetation leaf area index based on unmanned aerial vehicle lidar data.

[0006] According to the first aspect of the present disclosure, a method for retrieving vegetation leaf area index based on unmanned aerial vehicle lidar data is provided. The method includes:

[0007] Preprocess the point cloud data of the target acquisition area, and input the preprocessed point cloud data into the smooth constraint Markov random field model, including the ground point cloud data and the point cloud data of each vegetation point;

[0008] Based on the ground point cloud data and the point cloud data of each vegetation point, construct a canopy height model, and according to the preprocessed point cloud data and the canopy height model, construct the four-dimensional point cloud data of each vegetation point;

[0009] According to the K-D tree algorithm, divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point, perform nearest neighbor search on each subspace, and estimate the density and volume of the vegetation area;

[0010] According to the density and volume of the vegetation area, use the leaf area index formula to construct a leaf area index inversion model and obtain the vegetation leaf area index.

[0011] In some realizable ways of the first aspect, the preprocessing of the point cloud data of the target acquisition area includes:

[0012] Perform adaptive filtering downsampling on the point cloud data of the target acquisition area to obtain sampled point cloud data, and perform singular value rejection and spatial median filtering operations on the sampled point cloud data.

[0013] In some realizable ways of the first aspect, the inputting the preprocessed point cloud data into the smooth constraint Markov random field model to obtain the ground point cloud data and the point cloud data of each vegetation point includes:

[0014] Minimize the energy function through the smooth constraint Markov random field model to label each input point, and obtain the point cloud data labeled as the ground and the point cloud data labeled as vegetation; where

[0015] The energy function includes a data term and a penalty term; where

[0016] The data term represents the fitting degree between the input point and the vegetation labeled point;

[0017] The penalty term represents the penalty for adjacent input points with different labels.

[0018] In some realizable ways of the first aspect, the constructing a canopy height model based on the obtained point cloud data labeled as the ground and the point cloud data labeled as vegetation includes:

[0019] Construct a digital elevation model according to the ground point cloud data;

[0020] Construct a digital surface model according to the point cloud data of each vegetation point;

[0021] Construct a canopy height model based on the digital surface model and the digital elevation model.

[0022] In some realizable ways of the first aspect, constructing the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model includes:

[0023] Extract the coordinates of each vegetation point on the x-axis, y-axis, and z-axis from the canopy height model;

[0024] Extract the echo intensity of each vegetation point from the preprocessed point cloud data;

[0025] Construct the four-dimensional point cloud data of each vegetation point according to the coordinates of each vegetation point on the x, y, and z axes and the echo intensity.

[0026] In some realizable ways of the first aspect, dividing the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point includes:

[0027] Select one dimension from the dimensions of the four-dimensional point cloud data of each vegetation point, use the median of the four-dimensional point cloud data on the selected dimension as the splitting point, and divide the vegetation area composed of each vegetation point into two subspaces according to the splitting point;

[0028] For each subspace, reselect one dimension from the remaining dimensions of the four-dimensional point cloud data, use the median of the four-dimensional point cloud data of each subspace on the reselected dimension as the new splitting point of the corresponding subspace, and divide the corresponding subspace into two new subspaces according to the new splitting point;

[0029] Repeat the subspace division operation until all dimensions of the four-dimensional point cloud data are used for the division operation or the number of points in the divided subspaces is less than the preset threshold.

[0030] In some realizable ways of the first aspect, performing a nearest neighbor search on each subspace to estimate the vegetation area density and volume includes:

[0031] Perform a nearest neighbor search on each subspace according to the K-D tree algorithm to obtain the number of points in each searched subspace;

[0032] Calculate the vegetation area density and volume according to the number of points in each searched subspace and the volume of each subspace obtained according to the canopy height model.

[0033] In some realizable ways of the first aspect, the leaf area index formula includes:

[0034]

[0035] Among them, LAI represents the vegetation leaf area index, V, ρ, and A respectively represent the volume, density, and ground area of the vegetation area, and k1 represents the regression coefficient.

[0036] In some realizable ways of the first aspect, the method further includes:

[0037] Perform a correlation analysis on the obtained vegetation leaf area index and the actually measured vegetation leaf area index of the same target collection area in the same period, and calculate the inversion accuracy of the leaf area index inversion model.

[0038] According to the second aspect of the present disclosure, there is provided a device for inverting the vegetation leaf area index based on unmanned aerial vehicle lidar data. The device includes:

[0039] A vegetation recognition module, configured to preprocess the point cloud data of the target collection area, input the preprocessed point cloud data into a smooth constraint Markov random field model, and obtain the point cloud data of the ground point cloud and each vegetation point;

[0040] A four-dimensional point cloud data construction module, configured to construct a canopy height model based on the point cloud data of the ground point cloud and each vegetation point, and construct the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model;

[0041] A vegetation area density and volume estimation module, configured to divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point by using the K-D tree algorithm, perform a nearest neighbor search on each subspace, and estimate the vegetation area density and volume;

[0042] A leaf area index inversion module, configured to construct a leaf area index inversion model according to the vegetation area density and volume by using the leaf area index formula, and obtain the vegetation leaf area index.

[0043] According to the third aspect of the present disclosure, there is provided an electronic device. The electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method as described above.

[0044] According to the fourth aspect of the present disclosure, there is provided a non-transitory computer-readable storage medium storing computer instructions, and the computer instructions are used to cause a computer to execute the method as described above.

[0045] In the present disclosure, a method for inverting the leaf area index of vegetation based on unmanned aerial vehicle lidar data is proposed. The method is as follows: preprocess the point cloud data of the target acquisition area, input the preprocessed point cloud data into a smooth constraint Markov random field model to obtain the ground point cloud data and the point cloud data of each vegetation point; based on the ground point cloud data and the point cloud data of each vegetation point, construct a canopy height model, and construct four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model; according to the K-D tree algorithm, divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point, perform nearest neighbor search on each subspace, and estimate the density and volume of the vegetation area; according to the density and volume of the vegetation area, use the leaf area index formula to construct a leaf area index inversion model to obtain the leaf area index of the vegetation. In this way, the leaf area index of the vegetation in the area covered by the point cloud data can be obtained quickly and accurately, improving the inversion accuracy, and the method has stronger anti-noise ability and higher calculation efficiency, significantly enhancing the practicability and reliability of the leaf area index inversion.

[0046] It should be understood that the content described in the summary of the invention section is not intended to limit the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. Brief Description of the Drawings

[0047] Combined with the drawings and referring to the following detailed description, the above and other features, advantages and aspects of the embodiments of the present disclosure will become more obvious. The drawings are used to better understand the solution and do not constitute a limitation to the present disclosure. In the drawings, the same or similar reference numerals represent the same or similar elements, where:

[0048] Figure 1 Shows a flowchart of a method for inverting the leaf area index of vegetation based on unmanned aerial vehicle lidar data provided by an embodiment of the present disclosure;

[0049] Figure 2 Shows a comparison chart between the leaf area index of vegetation calculated by the leaf area index inversion model provided by an embodiment of the present disclosure and the leaf area index obtained by actual measurement;

[0050] Figure 3 Shows a structural diagram of a device for inverting the leaf area index of vegetation based on unmanned aerial vehicle lidar data provided by an embodiment of the present disclosure;

[0051] Figure 4 Shows a structural diagram of an exemplary electronic device capable of implementing the embodiments of the present disclosure. Detailed Description of the Embodiments

[0052] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present disclosure without creative efforts shall fall within the scope of protection of the present disclosure.

[0053] In addition, the term "and / or" in this document is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after.

[0054] In response to the problems in the background art, the embodiments of the present disclosure provide a method for inverting the leaf area index of vegetation based on unmanned aerial vehicle lidar data. The method is as follows: preprocess the point cloud data of the target acquisition area, input the preprocessed point cloud data into a smooth constraint Markov random field model to obtain the ground point cloud data and the point cloud data of each vegetation point; based on the ground point cloud data and the point cloud data of each vegetation point, construct a canopy height model, and construct the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model; according to the K-D tree algorithm, divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point, perform nearest neighbor search on each subspace, and estimate the density and volume of the vegetation area; according to the density and volume of the vegetation area, use the leaf area index formula to construct a leaf area index inversion model to obtain the leaf area index of the vegetation. In this way, the leaf area index of the vegetation in the area covered by the point cloud data can be quickly and accurately obtained, the inversion accuracy can be improved, and the method has stronger anti-noise ability and higher calculation efficiency, significantly improving the practicability and reliability of the leaf area index inversion.

[0055] The following will, with reference to the accompanying drawings, describe in detail the method and device for inverting the leaf area index of vegetation based on unmanned aerial vehicle lidar data provided by the embodiments of the present disclosure through specific embodiments.

[0056] Figure 1 The flowchart of a method for inverting the leaf area index of vegetation based on unmanned aerial vehicle lidar data provided by the embodiments of the present disclosure is shown. Method 100 includes the following steps:

[0057] S110, preprocess the point cloud data of the target acquisition area, input the preprocessed point cloud data into a smooth constraint Markov random field model, and obtain the ground point cloud data and the point cloud data of each vegetation point.

[0058] In some embodiments, a lidar sensor carried by a multi-rotor unmanned aerial vehicle is used to collect point cloud data for a target acquisition area. Compared with the conventional method of using spectral imaging technology to collect point cloud data, the lidar on the unmanned aerial vehicle can quickly and accurately obtain high-precision point cloud data, provide richer and more accurate vegetation structure information, and overcome the limitations of traditional two-dimensional observations.

[0059] In some embodiments, the preprocessing of the point cloud data for the target acquisition area includes:

[0060] Performing adaptive filtering downsampling on the point cloud data of the target acquisition area to obtain sampled point cloud data, and performing singular value rejection and spatial median filtering operations on the sampled point cloud data to eliminate noise and correct errors;

[0061] Furthermore, according to factors such as the point density, geometric structure complexity, and information entropy of different regions in the point cloud data, adaptive filtering downsampling is performed on the point cloud data of the target acquisition area to adaptively adjust the sampling interval and obtain sampled point cloud data. This can selectively retain representative points according to the spatial distribution, reduce the data volume without losing key spatial information and retain the maximum amount of information, which helps to maintain the integrity and accuracy of the data in subsequent analysis.

[0062] In some embodiments, inputting the preprocessed point cloud data into a smooth constrained Markov random field model, the ground point cloud data and the point cloud data of each vegetation point include:

[0063] By minimizing the energy function through the smooth constrained Markov random field model, each input point is marked to obtain the point cloud data marked as the ground and the point cloud data marked as vegetation; where

[0064] The energy function includes a data term and a penalty term; where

[0065] The data term represents the fitting degree between the input point and the vegetation marked point;

[0066] The penalty term represents the penalty for adjacent input points with different marks.

[0067] In some embodiments, the expression of the energy function E(X) is as follows:

[0068]

[0069] Where D i (x i ) represents the data term corresponding to the i-th input point x i , V i,j (x i , x j ) represents the adjacent input points xi , x j (the penalty term corresponding to the j-th input point), where λ represents the weight of the penalty term.

[0070] In some embodiments, a clustering algorithm is used to obtain the point cloud data labeled as vegetation. The Smooth Constrained Markov Random Field Model (SCMRF) can enhance the ability to capture the edges and detailed information of vegetation by adjusting the weight of the penalty term, enabling the SCMRF to better obtain the actual shape and structure of vegetation and optimize the performance of the SCMRF model in complex scenarios. In this embodiment, the SCMRF model can flexibly adjust the weight of the penalty term according to the actual situation.

[0071] In some embodiments, the efficiency of point cloud data processing and the accuracy of vegetation recognition are improved through adaptive filtering downsampling and the Smooth Constrained Markov Random Field Model.

[0072] S120. Based on the ground point cloud data and the point cloud data of each vegetation point, construct a canopy height model. According to the preprocessed point cloud data and the canopy height model, construct the four-dimensional point cloud data of each vegetation point.

[0073] In some embodiments, the constructing of the canopy height model based on the ground point cloud data and the point cloud data of each vegetation point includes:

[0074] Construct a Digital Terrain Model (DTM) based on the ground point cloud data;

[0075] Construct a Digital Surface Model (DSM) based on the point cloud data of each vegetation point;

[0076] Construct a Canopy Height Model (CHM) based on the Digital Surface Model and the Digital Terrain Model;

[0077] Furthermore, CHM = DSM - DTM.

[0078] In some embodiments, the constructing of the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model includes:

[0079] Extract the coordinates of each vegetation point on the x-axis, y-axis, and z-axis from the canopy height model;

[0080] Extract the echo intensity of each vegetation point from the preprocessed point cloud data; specifically, since the echo intensities of vegetation points have similar characteristics, a clustering algorithm is used to extract the echo intensities of each vegetation point;

[0081] Construct the four-dimensional point cloud data of each vegetation point according to the coordinates of each vegetation point on the x, y, and z axes and the echo intensity, that is, the four-dimensional point cloud data includes the data of the coordinates of the vegetation point on the x, y, and z axes and the echo intensity in these four dimensions.

[0082] In some embodiments, the echo intensity of the vegetation points endows the vegetation points with more characteristic information, enabling more accurate differentiation of different types of vegetation points and more accurate estimation of the leaf area index of the vegetation area when constructing a K-D tree and calculating the leaf area index of the vegetation.

[0083] S130. According to the K-D tree algorithm, divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point, perform nearest neighbor search on each subspace, and estimate the density and volume of the vegetation area.

[0084] In some embodiments, the step of dividing the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point includes:

[0085] Select one dimension from the dimensions of the four-dimensional point cloud data of each vegetation point, use the median of the four-dimensional point cloud data on the selected dimension as the splitting point, and divide the vegetation area composed of each vegetation point into two subspaces according to the splitting point;

[0086] For each subspace, reselect one dimension from the remaining dimensions of the four-dimensional point cloud data, use the median of the four-dimensional point cloud data of each subspace on the reselected dimension as the new splitting point of the corresponding subspace, and divide the corresponding subspace into two new subspaces according to the new splitting point;

[0087] Repeat the subspace splitting operation until all dimensions of the four-dimensional point cloud data are used for the splitting operation or the number of points in the split subspaces is less than a preset threshold;

[0088] For example, if there are four dimensions in the four-dimensional point cloud data of each vegetation point, namely the x-axis, y-axis, z-axis, and echo intensity, the first selected dimension is the x-axis. Use the median of the four-dimensional point cloud data of each vegetation point on the x-axis as the splitting point, and divide the four-dimensional point cloud data of the vegetation points into two subsets according to this splitting point. Each subset of points is located in a subspace, that is, one subset corresponds to one subspace. Therefore, the vegetation area is divided into two subspaces. Suppose these two subspaces are subspace A and subspace B respectively. For each subspace, taking subspace A as an example, reselect one dimension from the remaining dimensions (y-axis, z-axis, echo intensity). Suppose the selected dimension is the z-axis. Use the median of the subset corresponding to subspace A on the z-axis as the splitting point of subspace A, and divide subspace A into two subspaces again according to this splitting point. According to the above method, repeat the subspace splitting operation until all dimensions of the four-dimensional point cloud data are used for the splitting operation or the number of points in the split subspaces is less than a preset threshold.

[0089] In some embodiments, the step of performing nearest neighbor search on each subspace and estimating the density and volume of the vegetation area includes:

[0090] Perform nearest neighbor search on each subspace according to the K-D tree algorithm to obtain the number of points in each searched subspace.

[0091] Calculate the vegetation area density and volume based on the number of points in each searched subspace and the volume of each subspace obtained according to the canopy height model.

[0092] In some embodiments, performing nearest neighbor search on each subspace according to the K-D tree algorithm to obtain the number of points in each searched subspace includes:

[0093] Determine the query point of each subspace according to a preset search rule (for example, specifying the center point of the subspace as the query point of the subspace). For the query point, perform nearest neighbor search on the query point according to the K-D tree algorithm, and determine whether to search the left subtree or the right subtree according to the value of the query point in the splitting dimension. Count the number of points located in the subspace where the query point is located in the search, and then obtain the number of points in each searched subspace.

[0094] In some embodiments, calculating the vegetation area density and volume based on the number of points in each searched subspace and the volume of each subspace obtained according to the canopy height model includes:

[0095] Obtain the volume elements of each point in each subspace from the canopy height model. For each subspace, accumulate the volume elements of each point in the subspace to obtain the volume of the subspace.

[0096] Divide the number of points located in the subspace searched to the volume of the subspace to obtain the density of the subspace.

[0097] Perform arithmetic mean on the densities of each subspace to obtain the vegetation area density.

[0098] Accumulate the volumes of each subspace to obtain the vegetation area volume.

[0099] In some embodiments, performing nearest neighbor search on each subspace according to the K-D tree algorithm reduces unnecessary traversals. It is not necessary to compare and calculate each point one by one. Only a small number of subspaces related to the query point need to be traversed to find the nearest neighbor point. Compared with traditional traversal calculations, the calculation efficiency is greatly improved and the calculation time is reduced. When dealing with large-scale point cloud data, the density and volume of the vegetation area can be quickly and accurately estimated.

[0100] In some embodiments, using the K-D tree algorithm for vegetation area segmentation and nearest neighbor search realizes efficient calculation of the vegetation area density and volume.

[0101] S140. According to the vegetation area density and volume, use the leaf area index formula to construct a leaf area index inversion model and obtain the vegetation leaf area index.

[0102] In some embodiments, the leaf area index formula includes:

[0103]

[0104] where LAI represents the vegetation leaf area index, V, ρ, and A represent the volume, density, and ground area of the vegetation area respectively, and k1 represents the regression coefficient, which is usually obtained by fitting empirical values or experimental data.

[0105] In some embodiments, the volume of the vegetation area is estimated by a canopy height model, which reflects the occupation degree of the vegetation in space and has an inherent relationship with LAI. Generally speaking, the larger the volume of the vegetation area, the larger the vegetation leaf area index; the K-D tree algorithm is used to estimate the density of the vegetation area, considering the density of the vegetation distribution in the vegetation area. Factors such as the volume of the vegetation area and the density of the vegetation area restrict and correlate with each other, jointly establishing a systematic and comprehensive leaf area index inversion model. Compared with the model that simply relies on empirical formulas or a small number of influencing factors, the leaf area index inversion model can more accurately describe the actual situation of the vegetation and improve the inversion accuracy.

[0106] In some embodiments, by constructing a leaf area index inversion model, the practicability and popularity of the leaf area index inversion are significantly improved.

[0107] In some embodiments, method 100 further includes:

[0108] Performing a correlation analysis on the obtained vegetation leaf area index and the vegetation leaf area index actually measured in the same target acquisition area at the same time, and calculating the inversion accuracy of the leaf area index inversion model;

[0109] Furthermore, the inversion accuracy can be represented by the coefficient of determination R 2 , root mean square error, mean error, etc.;

[0110] For example, Figure 2 shows a comparison chart between the vegetation leaf area index calculated by the leaf area index inversion model and the vegetation leaf area index actually measured. The circled part in the figure represents the fitting situation between the vegetation leaf area index output during the training of the leaf area index inversion model and the vegetation leaf area index actually measured in the same target acquisition area at the same time. At this time, the calculated coefficient of determination R 2 is 0.588, and the black straight line represents the fitting situation between the vegetation leaf area index output during the verification stage of the leaf area index inversion model and the vegetation leaf area index actually measured in the same target acquisition area at the same time. At this time, the calculated coefficient of determination R 2 is 0.895. Compared with the existing vegetation leaf area index inversion methods, method 100 improves the inversion accuracy.

[0111] In some embodiments, a portable leaf area index meter LAI-2200C is used to actually measure the vegetation leaf area index of the same target acquisition area in the same period.

[0112] In some embodiments, Method 100 is generally applicable to the inversion methods of vegetation leaf area index such as wheat, forest, summer maize, etc.

[0113] According to an embodiment of the present disclosure, a method for inverting the vegetation leaf area index based on unmanned aerial vehicle lidar data is proposed. The method is as follows: preprocess the point cloud data of the target acquisition area, input the preprocessed point cloud data into a smooth constraint Markov random field model to obtain the ground point cloud data and the point cloud data of each vegetation point; based on the ground point cloud data and the point cloud data of each vegetation point, construct a canopy height model, and according to the preprocessed point cloud data and the canopy height model, construct the four-dimensional point cloud data of each vegetation point; according to the K-D tree algorithm, divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point, perform nearest neighbor search on each subspace, and estimate the density and volume of the vegetation area; according to the density and volume of the vegetation area, use the leaf area index formula to construct a leaf area index inversion model to obtain the vegetation leaf area index. In this way, the vegetation leaf area index of the area covered by the point cloud data can be obtained quickly and accurately, the inversion accuracy can be improved, and the method has stronger anti-noise ability and higher calculation efficiency, significantly improving the practicability and reliability of the leaf area index inversion.

[0114] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that the present disclosure is not limited by the described action sequence, because according to the present disclosure, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to the present disclosure.

[0115] The above is the introduction of the method embodiments. The following further illustrates the solution of the present disclosure through device embodiments.

[0116] Figure 3 The structural diagram of a device for inverting the vegetation leaf area index based on unmanned aerial vehicle lidar data provided by an embodiment of the present disclosure is shown. Device 300 includes:

[0117] A vegetation recognition module 310, configured to preprocess the point cloud data of the target acquisition area, input the preprocessed point cloud data into a smooth constraint Markov random field model, and obtain the ground point cloud data and the point cloud data of each vegetation point.

[0118] In some embodiments, the vegetation recognition module 310 is specifically configured to:

[0119] Preprocess the point cloud data of the target acquisition area, including:

[0120] Perform adaptive filtering downsampling on the point cloud data of the target acquisition area to obtain sampled point cloud data, and perform singular value rejection and spatial median filtering operations on the sampled point cloud data.

[0121] In some embodiments, the vegetation recognition module 310 is further specifically configured to:

[0122] Input the preprocessed point cloud data into a smooth constrained Markov random field model to obtain ground point cloud data and point cloud data of each vegetation point, including:

[0123] Minimize the energy function through the smooth constrained Markov random field model, mark each input point, and obtain the point cloud data marked as the ground and the point cloud data marked as vegetation; wherein,

[0124] The energy function includes a data term and a penalty term; wherein,

[0125] The data term represents the fitting degree between the input point and the vegetation marked point;

[0126] The penalty term represents the penalty for adjacent input points with different marks.

[0127] The four-dimensional point cloud data construction module 320 is configured to construct a canopy height model based on the ground point cloud data and the point cloud data of each vegetation point, and construct the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model.

[0128] In some embodiments, the four-dimensional point cloud data construction module 320 is specifically configured to:

[0129] Construct a canopy height model from the ground point cloud data and the point cloud data of each vegetation point, including:

[0130] Construct a digital elevation model based on the ground point cloud data;

[0131] Construct a digital surface model based on the point cloud data of each vegetation point;

[0132] Construct a canopy height model according to the digital surface model and the digital elevation model.

[0133] In some embodiments, the four-dimensional point cloud data construction module 320 is further specifically configured to:

[0134] Construct the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model, including:

[0135] Extract the coordinates of each vegetation point on the x-axis, y-axis, and z-axis from the canopy height model;

[0136] Extract the echo intensity of each vegetation point from the preprocessed point cloud data;

[0137] Construct the four-dimensional point cloud data of each vegetation point according to the coordinates of each vegetation point on the x, y, and z axes and the echo intensity.

[0138] The vegetation area density and volume estimation module 330 is used to divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point by the K-D tree algorithm, perform nearest neighbor search on each subspace, and estimate the vegetation area density and volume.

[0139] In some embodiments, the vegetation area density and volume estimation module 330 is specifically used for:

[0140] The dividing the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point includes:

[0141] Select a dimension from the dimensions of the four-dimensional point cloud data of each vegetation point, use the median of the four-dimensional point cloud data on the selected dimension as the segmentation point, and divide the vegetation area composed of each vegetation point into two subspaces according to the segmentation point;

[0142] For each subspace, reselect a dimension from the remaining dimensions of the four-dimensional point cloud data, use the median of the four-dimensional point cloud data of each subspace on the reselected dimension as the new segmentation point of the corresponding subspace, and divide the corresponding subspace into two new subspaces according to the new segmentation point;

[0143] Repeat the subspace division operation until all dimensions of the four-dimensional point cloud data are used for the division operation or the number of points in the divided subspaces is less than the preset threshold.

[0144] In some embodiments, the vegetation area density and volume estimation module 330 is specifically further used for:

[0145] The performing nearest neighbor search on each subspace and estimating the vegetation area density and volume includes:

[0146] Perform nearest neighbor search on each subspace according to the K-D tree algorithm, and obtain the number of points in each searched subspace;

[0147] Calculate the vegetation area density and volume according to the number of points in each searched subspace and the volume of each subspace obtained according to the canopy height model.

[0148] The leaf area index inversion module 340 is used to construct a leaf area index inversion model according to the vegetation area density and volume by using the leaf area index formula, and obtain the vegetation leaf area index.

[0149] In some embodiments, the leaf area index inversion module 340 is specifically configured to:

[0150] The leaf area index formula includes:

[0151]

[0152] Where LAI represents the vegetation leaf area index, V, ρ, and A respectively represent the volume, density, and ground area of the vegetation area, and k1 represents the regression coefficient.

[0153] In some embodiments, the device 300 is further specifically configured to:

[0154] Perform a correlation analysis on the obtained vegetation leaf area index and the actually measured vegetation leaf area index of the same target acquisition area in the same period, and calculate the inversion accuracy of the leaf area index inversion model.

[0155] It can be understood that Figure 3 Each module / unit in the shown device 300 has the function of implementing each step in the detection method 100 provided in the embodiments of the present disclosure, and can achieve its corresponding technical effects. For the sake of brevity, they will not be described in detail here.

[0156] Figure 4 The structure diagram of an exemplary electronic device capable of implementing the embodiments of the present disclosure is shown. The electronic device 400 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device 400 can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0157] As Figure 4 shown, the electronic device 400 includes a computing unit 401, which can execute various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 402 or the computer program loaded from the storage unit 408 into the random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the electronic device 400 can also be stored. The computing unit 401, ROM 402, and RAM 403 are connected to each other through a bus 404. The I / O interface 405 is also connected to the bus 404.

[0158] Multiple components in the electronic device 400 are connected to the I / O interface 405, including: an input unit 406, such as a keyboard, a mouse, etc.; an output unit 407, such as various types of displays, speakers, etc.; a storage unit 408, such as a disk, an optical disc, etc.; and a communication unit 409, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 409 allows the electronic device 400 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0159] The computing unit 401 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include but are not limited to a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 executes the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage unit 408. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 400 via the ROM 402 and / or the communication unit 409. When the computer program is loaded into the RAM 403 and executed by the computing unit 401, one or more steps of method 100 described above can be executed. Alternatively, in other embodiments, the computing unit 401 can be configured to execute method 100 in any other suitable manner (e.g., by means of firmware).

[0160] The various embodiments of the systems and technologies described above in this article can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGA), application-specific integrated circuits (ASIC), application-specific standard products (ASSP), system-on-a-chip (SOC), complex programmable logic devices (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, where the programmable processor can be a special or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0161] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing devices, such that when the program codes are executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program codes can be executed entirely on the machine, partially on the machine, executed partially on the machine as an independent software package and partially on a remote machine, or executed entirely on a remote machine or server.

[0162] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0163] It should be noted that the present disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute Method 100 and achieve the corresponding technical effects achieved by the method of the embodiments of the present disclosure. For the sake of brevity of description, it will not be elaborated herein.

[0164] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0165] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by digital data communication in any form or medium (e.g., a communication network). Examples of communication networks include: local area network (LAN), wide area network (WAN), and the Internet.

[0166] A computer system can include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, or a server of a distributed system, or a server incorporating a blockchain.

[0167] It should be understood that various forms of the processes shown above can be used, steps can be reordered, added, or deleted. For example, the steps recited in this disclosure can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this is not limited herein.

[0168] The above specific embodiments do not constitute a limitation on the protection scope of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the protection scope of this disclosure.

Claims

1. A method for retrieving vegetation leaf area index based on unmanned aerial vehicle lidar data, characterized in that, Including: Preprocess the point cloud data of the target acquisition area, input the preprocessed point cloud data into a smooth constraint Markov random field model, and obtain the point cloud data of the ground point cloud and each vegetation point; Based on the point cloud data of the ground point cloud and each vegetation point, construct a canopy height model, and construct the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model; According to the K-D tree algorithm, divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point, perform nearest neighbor search on each subspace, and estimate the density and volume of the vegetation area; According to the density and volume of the vegetation area, use the leaf area index formula to construct a leaf area index inversion model and obtain the vegetation leaf area index.

2. The method according to claim 1, characterized in that, The preprocessing of the point cloud data of the target acquisition area includes: Perform adaptive filtering downsampling on the point cloud data of the target acquisition area to obtain sampled point cloud data, and perform singular value rejection and spatial median filtering operations on the sampled point cloud data.

3. The method according to claim 1, characterized in that The input of the preprocessed point cloud data into the smooth constraint Markov random field model to obtain the point cloud data of the ground point cloud and each vegetation point includes: Minimize the energy function through the smooth constraint Markov random field model, mark each input point, and obtain the point cloud data marked as the ground and the point cloud data marked as vegetation; where, The energy function includes a data term and a penalty term; where, The data term represents the fitting degree between the input point and the vegetation marked point; The penalty term represents the penalty for adjacent input points with different marks.

4. The method according to claim 1, characterized in that, The construction of the canopy height model based on the point cloud data of the ground point cloud and each vegetation point includes: Construct a digital elevation model according to the ground point cloud data; Construct a digital surface model according to the point cloud data of each vegetation point; Construct a canopy height model according to the digital surface model and the digital elevation model.

5. The method according to claim 4, wherein The construction of the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model includes: Extract the coordinates of each vegetation point on the x-axis, y-axis, and z-axis from the canopy height model; Extract the echo intensity of each vegetation point from the preprocessed point cloud data; Construct the four-dimensional point cloud data of each vegetation point according to the coordinates of each vegetation point on the x, y, and z axes and the echo intensity.

6. The method according to claim 5, wherein The division of the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point includes: Select a dimension from the dimensions of the four-dimensional point cloud data of each vegetation point, use the median of the four-dimensional point cloud data on the selected dimension as the division point, and divide the vegetation area composed of each vegetation point into two subspaces according to the division point; For each subspace, reselect a dimension from the remaining dimensions of the four-dimensional point cloud data, use the median of the four-dimensional point cloud data of each subspace on the reselected dimension as the new division point of the corresponding subspace, and divide the corresponding subspace into two new subspaces according to the new division point; Repeat the subspace division operation until all dimensions of the four-dimensional point cloud data are used for the division operation or the number of points in the divided subspaces is less than the preset threshold.

7. The method according to claim 6, wherein The nearest neighbor search for each subspace to estimate the density and volume of the vegetation area includes: Perform nearest neighbor search on each subspace according to the K-D tree algorithm to obtain the number of points in each searched subspace; Calculate the vegetation area density and volume based on the number of points in each searched subspace and the volume of each subspace obtained according to the canopy height model.

8. The method according to claim 1, characterized in that The leaf area index formula includes: Among them, LAI represents the vegetation leaf area index, V, ρ, and A represent the vegetation area volume, density, and ground area respectively, and k1 represents the regression coefficient.

9. The method according to claim 1, characterized in that, The method further includes: Perform correlation analysis on the obtained vegetation leaf area index and the vegetation leaf area index actually measured in the same target acquisition area at the same time, and calculate the inversion accuracy of the leaf area index inversion model.

10. A device for retrieving the leaf area index of vegetation based on unmanned aerial vehicle lidar data, characterized in that, It includes: A vegetation recognition module, which is used to preprocess the point cloud data of the target acquisition area, input the preprocessed point cloud data into the smooth constraint Markov random field model, and obtain the ground point cloud data and the point cloud data of each vegetation point; A four-dimensional point cloud data construction module, which is used to construct a canopy height model based on the ground point cloud data and the point cloud data of each vegetation point, and construct the four-dimensional point cloud data of each vegetation point according to the preprocessed point cloud data and the canopy height model; A vegetation area density and volume estimation module, which is used to divide the vegetation area composed of each vegetation point into multiple subspaces according to the dimensions of the four-dimensional point cloud data of each vegetation point according to the K-D tree algorithm, perform nearest neighbor search on each subspace, and estimate the vegetation area density and volume; A leaf area index inversion module, which is used to construct a leaf area index inversion model according to the vegetation area density and volume, use the leaf area index formula, and obtain the vegetation leaf area index.

Images