CN113780144A - Crop plant number and stem width automatic extraction method based on 3D point cloud - Google Patents
Crop plant number and stem width automatic extraction method based on 3D point cloud Download PDFInfo
- Publication number
- CN113780144A CN113780144A CN202111036169.9A CN202111036169A CN113780144A CN 113780144 A CN113780144 A CN 113780144A CN 202111036169 A CN202111036169 A CN 202111036169A CN 113780144 A CN113780144 A CN 113780144A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- stem
- crop
- point
- ground
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000605 extraction Methods 0.000 title claims abstract description 15
- 244000038559 crop plants Species 0.000 title claims abstract description 8
- 238000000034 method Methods 0.000 claims abstract description 38
- 241000196324 Embryophyta Species 0.000 claims abstract description 36
- 238000005070 sampling Methods 0.000 claims abstract description 20
- 230000011218 segmentation Effects 0.000 claims abstract description 5
- 238000007781 pre-processing Methods 0.000 claims abstract description 4
- 238000000926 separation method Methods 0.000 claims description 2
- 230000001488 breeding effect Effects 0.000 abstract description 9
- 238000009395 breeding Methods 0.000 abstract description 8
- 238000004364 calculation method Methods 0.000 abstract description 7
- 239000000463 material Substances 0.000 abstract description 3
- 238000005259 measurement Methods 0.000 description 14
- 238000012545 processing Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 7
- 239000000284 extract Substances 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 241000350052 Daniellia ogea Species 0.000 description 2
- 244000068988 Glycine max Species 0.000 description 2
- 235000010469 Glycine max Nutrition 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- 240000008067 Cucumis sativus Species 0.000 description 1
- 235000010799 Cucumis sativus var sativus Nutrition 0.000 description 1
- 241000238631 Hexapoda Species 0.000 description 1
- 230000009418 agronomic effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000004720 fertilization Effects 0.000 description 1
- 244000037666 field crops Species 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000003973 irrigation Methods 0.000 description 1
- 230000002262 irrigation Effects 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000004576 sand Substances 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000009333 weeding Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
- G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30242—Counting objects in image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a crop plant number and stem width automatic extraction method based on 3D point cloud, which comprises the following steps: s1, collecting 3D point cloud data of crops; s2, preprocessing the 3D point cloud data; s3, performing ground segmentation based on the preprocessed 3D point cloud data to obtain a crop 3D point cloud and a ground 3D point cloud; s4, setting stem section intercepting height based on the crop 3D point cloud and the ground 3D point cloud, and extracting the crop stem section point cloud with fixed height; s5, separating crop stems based on the extraction of the crop stem point cloud with fixed height to obtain single plant stem segments, and calculating the number of stems; and S6, calculating the diameter of the stem of the single plant stem. According to the method, the field crop 3D point cloud data is used for calculation, ground search and stem diameter fitting are performed in a sampling mode, the method can be well adapted to field environments, phenotype data can be automatically extracted, relevant information is provided for crop breeding and management, manpower and material resources are saved, and the method is convenient and rapid.
Description
Technical Field
The invention relates to the field of three-dimensional point cloud information analysis and processing, in particular to a method for automatically extracting the plant number and stem width of crops based on 3D point cloud.
Background
The phenotypic characteristics of the crops can directly reflect the yield of the crops, play an important role in screening superior varieties and guide field management, such as fertilization and irrigation, insect killing, weeding and the like. However, the traditional measurement is usually based on a large amount of manual measurement methods, the land sampling is estimated by using a ruler, a large amount of manpower is consumed, the traditional measurement is labor-intensive, the throughput is low, and errors are easily caused in the aspects of sampling, ruler adjustment, reading and data recording. Can obstruct the process of crop genotype research and cause a bottleneck in the breeding program.
The phenotypic researchers need to realize automatic extraction of phenotypic parameters of crops urgently, and researchers are convenient to liberate from huge field data, so that the phenotypic researchers develop a new way, various more convenient and fast instruments are adopted to extract the phenotypic parameters of the crops, the Xiaoodan Ma and the like construct a data acquisition system based on a near-end KinectV2 sensor platform, the plant height and the canopy width of a soybean group are obtained, and the optimal color index is extracted to represent the characteristics of the soybean canopy; yang et al calculated the plant height of individual seedlings of the cucumber seedlings by using a KinectV2 camera to realize automatic nondestructive measurement. The Kinect series cameras are favored by a plurality of researchers with low cost, portability, real-time performance and no damage, but have the defects of large influence of illumination, low matching precision, complex calculation process, rapid reduction of the precision of a single measurement plant in a group environment and the like.
Therefore, a nondestructive high-resolution three-dimensional measurement means is needed to realize the three-dimensional reconstruction of the real scene of the crop and extract the phenotypic characteristics of the crop.
Disclosure of Invention
The invention aims to provide a method for automatically extracting the number of plants and the stem width of crops based on 3D point cloud, which aims to solve the problems in the prior art, calculates by using field crop 3D point cloud data, performs ground search and stem diameter fitting in a sampling mode, can well adapt to the field environment, automatically extracts phenotype data, provides related information for crop breeding and management, saves manpower and material resources, and is convenient and quick.
In order to achieve the purpose, the invention provides the following scheme: the invention provides a crop plant number and stem width automatic extraction method based on 3D point cloud, which comprises the following steps:
s1, collecting 3D point cloud data of crops;
s2, preprocessing the 3D point cloud data;
s3, performing ground segmentation based on the preprocessed 3D point cloud data to obtain a crop 3D point cloud and a ground 3D point cloud;
s4, setting stem section intercepting height based on the crop 3D point cloud and the ground 3D point cloud, and extracting a crop stem section point cloud with fixed height;
s5, separating crop stems based on the extraction of the crop stem point cloud with fixed height to obtain single plant stem segments, and calculating the number of stems;
and S6, calculating the diameter of the stem of the single plant stem.
Preferably, the S2 includes:
s2.1, extracting a horizontal region of interest in the 3D point cloud data;
s2.2, carrying out point cloud registration based on the horizontal region of interest;
and S2.3, removing artifacts and noise in the 3D point cloud data after registration.
Preferably, in S3, a plane fitting is performed by using a random sampling consistency method to obtain a plane model of the crop planting ground.
Preferably, obtaining the crop planting ground plane model comprises:
s301, calculating the minimum sampling number M according to the given epsilon, P and M:
P=1-(1-(1-ε)m)Mwherein m is the minimum data size required for calculating the model parameters, P is the probability of a benign sampling subset, and epsilon is the proportion of error points in the sample;
s302, randomly extracting m points from the point cloud data of the horizontal region of interest, and calculating an initial value of a plane model parameter: ax + by + cz ═ d, wherein x, y and z are three-dimensional coordinates of point cloud points, and a, b and c are plane normal directions;
s303, calculating tolerance values delta of all points of the horizontal interested region based on the initial values of the plane model parameters, and setting a threshold range delta0If delta is at delta0If the model is internal, the model is classified as a model local internal point, otherwise, the model is classified as a model local external point;
s304, repeating S302 and S303M times, counting the number of the local points calculated each time, selecting the local point with the maximum number, and fitting the characteristic value of the local point with the maximum number to obtain the parameters of the crop planting ground plane model.
Preferably, the stem segment in S4 is cut at a height of 50 ± 5cm from the ground.
Preferably, in S5, a clustering algorithm DBSCAN based on spatial density is used to perform crop stem separation and calculate the number of stems.
Preferably, the S6 includes:
s6.1, intercepting stem slices of the single plant stem segments, and extracting point clouds of the stem slices;
s6.2, projecting the point cloud of the stem slice to a plane which is perpendicular to the growth direction of crop stems at the slice position to obtain a two-dimensional projection point;
s6.2, estimating optimal crop stem diameter parameters based on the two-dimensional projection points, and calculating the stem diameter of the single plant stem segment.
Preferably, the 3D point cloud data adopts FocusS70 scanner scan acquisition.
The invention discloses the following technical effects:
according to the automatic extraction method of the crop plant number and the stem width based on the 3D point cloud, provided by the invention, the 3D point cloud data of field crops are calculated, a high-flux automatic phenotype extraction process is constructed, ground search and stem diameter fitting are carried out in a sampling mode, the method can be well adapted to the field environment, single plant stem segments can be automatically and accurately separated out through space density clustering based on the Euclidean distance, the number of effective stems in a high-density planting cluster can be accurately counted, then the phenotype data can be automatically obtained in a high-flux mode, relevant information is provided for breeding variety selection and field crop management of the crops, manpower and material resources are saved, and the method is convenient and fast.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a schematic flow chart of a 3D point cloud-based method for automatically extracting plant number and stem width of crops according to an embodiment of the present invention;
FIG. 2 is a flow chart of field three-dimensional point cloud processing in an embodiment of the invention;
FIG. 3 illustrates the accessibility of DBSCAN points in the embodiment of the invention;
FIG. 4 is a crop stem point cloud projection in an embodiment of the invention;
FIG. 5 is the result of the distribution calculation of the stem segments of the crops by different clustering algorithms in the embodiment of the present invention, wherein a) is the result of the spatial distribution of the 3D point cloud stem segment data under MeanShift clustering; (b) the method comprises the following steps of (1) obtaining a spatial distribution result of 3D point cloud stem segment data under AHC clustering; (c) the space distribution result of the 3D point cloud stem segment data under the DBSCAN clustering is obtained.
FIG. 6 shows the results of linear regression of plant stem diameter point clouds and manual measurements in the examples of the invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
The invention provides a crop plant number and stem width automatic extraction method based on 3D point cloud, which refers to the following steps of 1-4:
s100, collecting 3D point cloud data of crops.
In the present embodiment, Focus is adoptedSAnd scanning by a 70-degree scanner, wherein the maximum measurement range of the scanner is 70m, the horizontal scanning field of view is 360 degrees, an effective scanning area formed by the scanner is a horizontal annular field of view with the outer diameter of 70m and the inner diameter of 0.6m, and 3D point cloud data of crops are obtained in the horizontal annular field of view. In addition, the 3D point cloud data in this embodiment is obtained in the field, and a dense multi-view effective three-dimensional point should be selected to ensure that the scanner scans as complete as possible three-dimensional stem point cloud data in an effective scanning range.
S200, preprocessing the collected 3D point cloud data of the crops.
S201, extracting a horizontal region of interest in the 3D point cloud data, wherein the horizontal region of interest selects point cloud data in an experimental scene from the horizontal direction (XOY plane) of the point cloud. This region is the focus of interest for data analysis, delineating the region for further processing.
In the later stage, point cloud registration is required, irrelevant areas in the horizontal view field are required to be removed for registration, and the identification interference of the target ball in the area and the environmental noise are reduced.
S202, acquiring point cloud data of a unit area based on the horizontal region of interest;
after the horizontal interesting area is obtained, point cloud data of a unit area is extracted from the horizontal interesting area, in this embodiment, the unit area is based on a single scanned field cluster, that is, the point cloud data of the unit area is point cloud data of the single cluster, and the single cluster in this embodiment represents a minimum planting unit.
And S203, removing artifacts and interference in the point cloud data of the unit area.
After the horizontal region of interest is extracted, the multi-view scheme is adopted to solve the self-occlusion problem and increase the point cloud density. Specifically, fast target ball-based registration is performed on a point cloud image in a horizontal region of interest by using FARO scene5.0 software, the FARO scene5.0 software automatically searches for a target ball in the point cloud image, and a plurality of point cloud images are registered according to the sphere center of the target ball.
After the registration is completed, more artifacts and noises caused by shooting or reflection exist in the point cloud image. Artifacts, optical phenomena in projection and imaging systems, and decoding of depth maps result in point clouds where these anomalies occur. The corresponding functions in the software are used to remove artifacts and noise that would interfere strongly with the subsequent processes.
S300, carrying out ground segmentation on the preprocessed 3D point cloud data to obtain a crop 3D point cloud and a ground 3D point cloud.
In this embodiment, the ground is segmented using random sample consensus (RANSAC). The random sampling consistency algorithm (RANSAC) is an effective robust estimation method, because the sampling idea of the algorithm can effectively avoid the interference problem under the complex environment, and an ideal processing result can be obtained for the data with the error rate of more than 50 percent, stable and reasonable ground point cloud is extracted from the crop point cloud data by using the algorithm according to the point cloud sampling information, and the RANSAC algorithm has the basic idea that: in the parameter estimation, in order to prevent data interference, all data are not utilized indiscriminately, but a ground model is estimated through partial sampling data, sampling data which are inconsistent with the estimated parameters are iteratively eliminated through a point number maximization criterion in the model, and then the parameters are estimated through correct sampling data. The specific process is as follows:
s301, given epsilon, P and M, calculating the minimum sampling number M;
under a certain confidence probability, the minimum sampling number M of the basic subset and the probability P (P > epsilon) of at least one benign sampling subset are satisfied with the relation shown in the formula (1):
P=1-(1-(1-ε)m)M (1),
where m is the minimum amount of data required to compute the model parameters,
the minimum number of samples M is calculated based on equation (1) using given ε, P, and M.
S302, randomly extracting m points from the point cloud data of the horizontal region of interest, and calculating the initial value of the plane model parameter.
In this embodiment, m is selected to be 3, that is, 3 points are randomly extracted from the point cloud data of the horizontal region of interest, the ground point cloud is described by using a normal equation of the spatial plane, as shown in formula (2), and the initial value of the plane model parameter is calculated by using the relationship between formula (2) and the parameter.
ax+by+cz=d (2)
Wherein,
[xi yi zi -1][a b c d]T=0 (3)
from the formula (3), the basic matrix [ a b c d]There are 3 degrees of freedom, i.e. at least 3 data points are needed to compute the fundamental matrix. The method specifically comprises the steps of calculating initial values of parameters by utilizing 3 randomly selected local interior points, and then searching other interior points of a point set according to the initial values. At this time, a judgment criterion needs to be established to determine whether the point is an intra-office point, and the method for judging whether the point is an intra-office point in this embodiment is to calculate a point (x)i,yi,zi) The distance to the plane, see equation (4).
S304, calculating a point (x) of the horizontal region of interest according to the initial value of the plane model parameteri,yi,zi) Setting a threshold range delta for the distance from the plane model, namely the tolerance value delta corresponding to each point0If delta is at delta0The interior points are classified as local interior points, otherwise, the interior points are classified as local exterior points.
The tolerance value delta is calculated according to the formula (4):
δi=|axi+byi+czi-d| (4)
s305, repeating S302 and S303M times, counting the number of the local points calculated each time, selecting the local point with the maximum number, and fitting the characteristic value of the local point with the maximum number to obtain the parameters of the crop planting ground plane model (shown in the formula 2).
And performing ground segmentation according to a crop ground plane fitting equation to obtain a crop 3D point cloud and a ground 3D point cloud.
S400, setting stem section intercepting height according to the crop 3D point cloud and the ground 3D point cloud, and extracting the crop stem section point cloud with fixed height.
In the fixed height position parallel to the ground, a vertical interested area is selected to obtain a crop stem section, stem diameter and stem number research is carried out, theoretically, the more complete stem section can be selected to carry out subsequent parameter extraction calculation, but due to surrounding environment interference of ground weeds, low split stems, canopy area blades and the like, the vertical interested area is properly selected in an area which is a certain distance away from the ground, meanwhile, the vertical interested area cannot be too high, canopy interference is prevented, in the embodiment, aiming at a jointing stage crop with the net height of about 2.5m as a process processing object, an area with the height of 50 +/-5 cm away from the ground is selected as the vertical interested area, the stem section in the high area is intercepted, and the point cloud of the crop stem section is extracted.
S500, separating crop stems according to the crop stem section point cloud with the fixed height to obtain single plant stem sections, and calculating the number of the stems.
After extracting the point cloud of the ROI crop stem section with fixed height, separating the crop stem by adopting a clustering mode and calculating the number of plants. Since some clustering algorithms need to specify the number of clusters, it is obvious that such algorithms are not suitable for automatically calculating the plant tree. Therefore, a clustering algorithm DBSCAN based on space density is selected for spatial clustering, and the crop stem number is automatically calculated.
Several definitions in DBSACN:
e, neighborhood: the area with the given object radius within E is called the E neighborhood of the object;
core object: if the number of sample points in the neighborhood of the given object E is more than or equal to MinPts, the object is called a core object;
the direct density can reach: for sample set D, if sample point q is within E-neighborhood of p, and p is a core object, then object q is directly density reachable from object p.
The density can reach: for a sample set D, a string of sample points p is given1,p2,…,pn,p=p1,q=pnIf the object p isiFrom pi-1The direct density is reachable, then object q is density reachable from object p.
Density connection: there is a point o in the sample set D, and if object o to object p and object q are density reachable, then p and q are density linked.
Density reachability is a transitive closure that is directly density reachable, and this relationship is asymmetric. The density connection is a symmetrical relationship. The purpose of DBSCAN is to find the largest set of density connected objects. If p and q belong to different clusters due to over-segmentation, the two small clusters are merged into one large cluster. And clustering the disordered point clouds into different single plant stem sections, and counting the number of stems. Examples of the definition of three reachability types are given as in fig. 2.
The specific process is as follows: 1. randomly selecting a point from the extracted crop stem section point cloud with fixed height as an iteration starting point; 2. searching points in a given neighborhood (E) from the position of the starting point, determining whether the central point is a core object according to MinPts, and finding out all objects connected with the point in density to form a cluster; 3. if the iteration point is an edge point (non-core object), jumping out of the loop, and searching a next iteration starting point; 4. and (5) sequentially iterating all unprocessed points, and finishing single-stem clustering.
The calculation of the spatial distribution of the crop stem segments according to the spatial information strongly influences the accuracy of the single-stem-segment character calculation. In particular, clustering a dense hollow cylindrical point set in a manner of calculating distances from all points to a cluster center according to a general clustering algorithm (such as KMeans) obviously suffers from obvious calculation interference due to point cloud distribution on the surface of the cylindrical body and three-dimensional stem segment length selection, so that two density-based algorithms, namely DBSCAN and MeanShift, and one hierarchical clustering (AHC) are selected. In addition, in order to calculate the number of clusters, the parameter of the number of clusters cannot be used as an input parameter at the beginning, so all the selected clustering algorithms are unsupervised methods.
Taking stem segment data (table 1) obtained by parallelly intercepting 4 clusters of crop stems in the field as an example, and classifying the three-dimensional point cloud data by using different unsupervised clustering algorithms to obtain a result of stem segment spatial distribution information.
TABLE 1
The original spatial distribution of the crop stem segments can be represented using the original stem segment image (as shown in fig. 5). The stem section clustering is carried out through the spatial distance information, the original disordered stem section point cloud can be clustered into single stem sections, the clustering categories are distinguished by using different colors in the graph 5, and the DBSCAN and AHC algorithms can separate all the stem sections from the dense planting area in the field. Comparing three clustering results of the four-cluster stem segment three-dimensional point cloud data according to the running time and the clustering quantity: DBSCAN < MeanShift < AHC at run time; compared with the actual stem segment quantity, the clustering quantity is 1 difference between the DBSCAN and the AHC on the cluster 4, the classification precision is more accurate, and the difference between the MeanShift and the actual stem segment quantity on the clusters 1, 2, 3 and 4 is 0, 11, 12 and 8, which shows that the clustering stability of the MeanShift on the dense space hollow cylindrical body point set is not as good as that of the DBSCAN and the AHC. And comprehensively comparing, and ordering the performance of the unsupervised algorithm for the dense space hollow cylindrical body point set to DBSCAN > AHC > MeanShift. Compared with other two algorithms, the DBSCAN algorithm can extract complete single stem sections from the point cloud of the crop stem sections more accurately, stably and quickly.
The crop stem is effectively extracted, and the accuracy of the automatic measurement of subsequent parameters is very important. Meanwhile, the automatic extraction of the stem number can provide the breeding workers and the field management personnel with the change information of crop tillering characteristics, growth conditions and effective stem number.
TABLE 2
S600, calculating the diameter of the stem of the single plant stem.
The stem diameter of a single plant can reflect the growth and development of the plant. In the field of agronomic research, the diameter of the stem of a crop plant is typically measured with a straight ruler or vernier caliper. However, these manual measurement methods are time consuming and laborious and are therefore not suitable for high throughput measurements. In this embodiment, the stem diameter of the individual stem segment is calculated by using a vertical plane projection and a circle fitting method. The specific process is as follows:
because the processing on the two-dimensional image is relatively easier and more mature, in the embodiment, the stem slice and the stem slice point cloud are collected, and the stem slice point cloud is projected to the plane which is perpendicular to the growth direction of the crop stem at the slice, so that the error caused by the inclination of the crop stem to the estimation of the stem diameter is avoided. The growth direction at the slice can be obtained by cross multiplication of all point clouds in the slice, as shown in formula (4):
wherein N iscIs the number of points of the crop stem slice point cloud, viIs the normal vector for point i. sgn (Z) is a sign function, ZijIs a vector vi×vjNor (v) represents the normalization of vector v. v. ofsIs the estimated growth direction of the stem section of the obtained crop. Then, a projection plane equation is obtained by a normal equation of the plane shown in the formula 2, and a crop stem projection along the growth direction is shown in fig. 3.
In order to carry out the circle fitting, the circle fitting is only carried out on an XY plane, and the projection point projected on the vertical surface of the crop stem is converted into the XY plane. And performing circle fitting on the projection points converted to the XY plane by using least squares based on the RANSAC algorithm to estimate the optimal crop stem diameter parameters. Computing post-projection point cloud set { x) by using least squarei,yi}(i=1,2,···,np) Three parameters at the minimum sum of squares of residuals achieved by the circular model, the three parameters specifically referring toxc,ycAnd R is represented by formula (5):
by devitalizing the parameters for the f objective function, as shown in equation (6):
solving the formula (6) to obtain xc,ycAnd R, obtaining an extreme point of the function f, and further calculating to obtain the crop stem diameter R.
The accuracy of the crop stem diameter calculated in this example was verified, and the results of comparing the 3D point cloud measurement of the plant stem diameter of cluster 2 with the ground truth were shown in table 3 for a total of 27 sets of data.
TABLE 3
As shown in Table 3, the measurements are highly correlated to manual measurements for a spatial circle fit (R)20.7585), the Root Mean Square Error (RMSE) is 5.73mm as shown in fig. 6. The parameter precision obtained by the algorithm can meet the precision required by crop measurement such as breeding and the like.
For agricultural breeding, field planting management and the like, the automatic extraction of the crop stem diameter can provide effective parameters for workers regularly, and the growth condition, yield estimation and plant species characteristic understanding of the workers are facilitated.
The scheme provides a high-throughput and automatic general framework for extracting the crop phenotype parameters of the plant number and stem width of the crop based on the 3D point cloud by using a high-resolution three-dimensional measurement means and a point cloud processing technology, can meet the high-throughput data processing requirements of the regular monitoring of the crop length conditions of agricultural breeding and field planting management, saves labor cost while regularly providing effective parameters for workers, and promotes breeding and planting management to realize modernization. The above-described embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solutions of the present invention can be made by those skilled in the art without departing from the spirit of the present invention, and the technical solutions of the present invention are within the scope of the present invention defined by the claims.
Claims (8)
1. A crop plant number and stem width automatic extraction method based on 3D point cloud is characterized in that: the method comprises the following steps:
s1, collecting 3D point cloud data of crops;
s2, preprocessing the 3D point cloud data;
s3, performing ground segmentation based on the preprocessed 3D point cloud data to obtain a crop 3D point cloud and a ground 3D point cloud;
s4, setting stem section intercepting height based on the crop 3D point cloud and the ground 3D point cloud, and extracting a crop stem section point cloud with fixed height;
s5, separating crop stems based on the extraction of the crop stem point cloud with fixed height to obtain single plant stem segments, and calculating the number of stems;
and S6, calculating the diameter of the stem of the single plant stem.
2. The method for automatically extracting the plant number and the stem width of the crop based on the 3D point cloud according to claim 1, wherein the method comprises the following steps: the S2 includes:
s2.1, extracting a horizontal region of interest in the 3D point cloud data;
s2.2, carrying out point cloud registration based on the horizontal region of interest;
and S2.3, removing artifacts and noise in the 3D point cloud data after registration.
3. The method for automatically extracting the plant number and the stem width of the crop based on the 3D point cloud as claimed in claim 2, wherein the method comprises the following steps: and in the S3, performing plane fitting by adopting a random sampling consistency method to obtain a plane model of the crop planting ground.
4. The method for automatically extracting the plant number and the stem width of the crop based on the 3D point cloud as claimed in claim 3, wherein the method comprises the following steps: obtaining the crop planting ground plane model comprises:
s301, calculating the minimum sampling number M according to the given epsilon, P and M:
P=1-(1-(1-ε)m)Mwherein m is the minimum data size required for calculating the model parameters, P is the probability of a benign sampling subset, and epsilon is the proportion of error points in the sample;
s302, randomly extracting m points from the point cloud data of the horizontal region of interest, and calculating an initial value of a plane model parameter: ax + by + cz ═ d, wherein x, y and z are three-dimensional coordinates of point cloud points, and a, b and c are plane normal directions;
s303, calculating tolerance values delta of all points of the horizontal interested region based on the initial values of the plane model parameters, and setting a threshold range delta0If delta is at delta0If the model is internal, the model is classified as a model local internal point, otherwise, the model is classified as a model local external point;
s304, repeating S302 and S303M times, counting the number of the local points calculated each time, selecting the local point with the maximum number, and fitting the characteristic value of the local point with the maximum number to obtain the parameters of the crop planting ground plane model.
5. The method for automatically extracting the plant number and the stem width of the crop based on the 3D point cloud according to claim 1, wherein the method comprises the following steps: the intercepting height of the stem segment in the S4 is 50 +/-5 cm from the ground.
6. The method for automatically extracting the plant number and the stem width of the crop based on the 3D point cloud according to claim 1, wherein the method comprises the following steps: and in the step S5, crop stem separation is carried out by adopting a clustering algorithm DBSCAN based on space density, and the stem number is calculated.
7. The method for automatically extracting the plant number and the stem width of the crop based on the 3D point cloud according to claim 1, wherein the method comprises the following steps: the S6 includes:
s6.1, intercepting stem slices of the single plant stem segments, and extracting point clouds of the stem slices;
s6.2, projecting the point cloud of the stem slice to a plane which is perpendicular to the growth direction of crop stems at the slice position to obtain a two-dimensional projection point;
s6.2, estimating optimal crop stem diameter parameters based on the two-dimensional projection points, and calculating the stem diameter of the single plant stem segment.
8. The method for automatically extracting the plant number and the stem width of the crop based on the 3D point cloud according to claim 1, wherein the method comprises the following steps: the 3D point cloud data adopts FocusS70 scanner scan acquisition.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111036169.9A CN113780144A (en) | 2021-09-06 | 2021-09-06 | Crop plant number and stem width automatic extraction method based on 3D point cloud |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111036169.9A CN113780144A (en) | 2021-09-06 | 2021-09-06 | Crop plant number and stem width automatic extraction method based on 3D point cloud |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113780144A true CN113780144A (en) | 2021-12-10 |
Family
ID=78841139
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111036169.9A Pending CN113780144A (en) | 2021-09-06 | 2021-09-06 | Crop plant number and stem width automatic extraction method based on 3D point cloud |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113780144A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114548277A (en) * | 2022-02-22 | 2022-05-27 | 电子科技大学 | Method and system for fitting ground points and extracting crop height based on point cloud data |
CN115995010A (en) * | 2023-03-22 | 2023-04-21 | 华南农业大学 | Plant height extraction method and system based on plant population point cloud |
WO2024009352A1 (en) * | 2022-07-04 | 2024-01-11 | 日本電信電話株式会社 | Location estimation device, location estimation method, and location estimation program |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110363861A (en) * | 2019-07-14 | 2019-10-22 | 南京林业大学 | Field crop three-dimensional reconstruction method based on laser radar point cloud |
CN111487646A (en) * | 2020-03-31 | 2020-08-04 | 安徽农业大学 | Online detection method for corn plant morphology |
CN111696122A (en) * | 2020-06-12 | 2020-09-22 | 北京数字绿土科技有限公司 | Crop phenotype parameter extraction method and device |
-
2021
- 2021-09-06 CN CN202111036169.9A patent/CN113780144A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110363861A (en) * | 2019-07-14 | 2019-10-22 | 南京林业大学 | Field crop three-dimensional reconstruction method based on laser radar point cloud |
CN111487646A (en) * | 2020-03-31 | 2020-08-04 | 安徽农业大学 | Online detection method for corn plant morphology |
CN111696122A (en) * | 2020-06-12 | 2020-09-22 | 北京数字绿土科技有限公司 | Crop phenotype parameter extraction method and device |
Non-Patent Citations (3)
Title |
---|
吴荣仁: "基于深度学习的点云树木分割及其胸径估计", 《中国优秀硕士学位论文全文数据库 农业科技辑》, 15 July 2020 (2020-07-15), pages 41 - 42 * |
周三章: "基于三维点云的农林作物三维重建方法与特征提取研究", 《中国优秀硕士学位论文全文数据库 农业科技辑》, pages 4 * |
李孟迪 等: "基于随机抽样一致性算法的稳健点云平面拟合方法", 《测绘科学》, vol. 40, no. 1, pages 2 - 3 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114548277A (en) * | 2022-02-22 | 2022-05-27 | 电子科技大学 | Method and system for fitting ground points and extracting crop height based on point cloud data |
CN114548277B (en) * | 2022-02-22 | 2023-09-08 | 电子科技大学 | Method and system for ground point fitting and crop height extraction based on point cloud data |
WO2024009352A1 (en) * | 2022-07-04 | 2024-01-11 | 日本電信電話株式会社 | Location estimation device, location estimation method, and location estimation program |
CN115995010A (en) * | 2023-03-22 | 2023-04-21 | 华南农业大学 | Plant height extraction method and system based on plant population point cloud |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Guo et al. | Crop 3D—a LiDAR based platform for 3D high-throughput crop phenotyping | |
Jin et al. | Stem–leaf segmentation and phenotypic trait extraction of individual maize using terrestrial LiDAR data | |
US11275941B2 (en) | Crop models and biometrics | |
Xu et al. | Aerial images and convolutional neural network for cotton bloom detection | |
Malambo et al. | Automated detection and measurement of individual sorghum panicles using density-based clustering of terrestrial lidar data | |
CN113780144A (en) | Crop plant number and stem width automatic extraction method based on 3D point cloud | |
CN109146948B (en) | Crop growth phenotype parameter quantification and yield correlation analysis method based on vision | |
CN111709981A (en) | Registration method of laser point cloud and analog image with characteristic line fusion | |
Miao et al. | Automatic stem-leaf segmentation of maize shoots using three-dimensional point cloud | |
Li et al. | A leaf segmentation and phenotypic feature extraction framework for multiview stereo plant point clouds | |
Zhou et al. | An integrated skeleton extraction and pruning method for spatial recognition of maize seedlings in MGV and UAV remote images | |
Zhang et al. | 3D monitoring for plant growth parameters in field with a single camera by multi-view approach | |
Király et al. | Tree height estimation methods for terrestrial laser scanning in a forest reserve | |
CN109961470B (en) | Living standing tree She Shuxing accurate estimation method based on laser point cloud | |
CN113920106B (en) | Corn growth vigor three-dimensional reconstruction and stem thickness measurement method based on RGB-D camera | |
CN110969654A (en) | Corn high-throughput phenotype measurement method and device based on harvester and harvester | |
Sun et al. | Remote estimation of grafted apple tree trunk diameter in modern orchard with RGB and point cloud based on SOLOv2 | |
Xiao et al. | High-throughput calculation of organ-scale traits with reconstructed accurate 3D canopy structures using a UAV RGB camera with an advanced cross-circling oblique route | |
Harandi et al. | How to make sense of 3D representations for plant phenotyping: a compendium of processing and analysis techniques | |
Lei et al. | Extraction of maize leaf base and inclination angles using terrestrial laser scanning (TLS) data | |
Li et al. | Multi-source data fusion improves time-series phenotype accuracy in maize under a field high-throughput phenotyping platform | |
Ma et al. | A method for calculating and simulating phenotype of soybean based on 3D reconstruction | |
Gatziolis et al. | Evaluation of terrestrial LiDAR technology for the development of local tree volume equations | |
CN111814666B (en) | Single tree parameter extraction method, system, medium and equipment under complex forest stand | |
Xiao et al. | 3D reconstruction and characterization of cotton bolls in situ based on UVA technology |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |