CN112308839A - Single-tree segmentation method and device for natural forest, computer equipment and storage medium - Google Patents

Single-tree segmentation method and device for natural forest, computer equipment and storage medium Download PDF

Info

Publication number
CN112308839A
CN112308839A CN202011198583.5A CN202011198583A CN112308839A CN 112308839 A CN112308839 A CN 112308839A CN 202011198583 A CN202011198583 A CN 202011198583A CN 112308839 A CN112308839 A CN 112308839A
Authority
CN
China
Prior art keywords
points
trunk
branch
seed
point
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.)
Granted
Application number
CN202011198583.5A
Other languages
Chinese (zh)
Other versions
CN112308839B (en
Inventor
刘钱威
王金亮
麻卫峰
张建鹏
刘一成
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Yunnan Normal University
Original Assignee
Yunnan Normal University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Yunnan Normal University filed Critical Yunnan Normal University
Priority to CN202011198583.5A priority Critical patent/CN112308839B/en
Publication of CN112308839A publication Critical patent/CN112308839A/en
Application granted granted Critical
Publication of CN112308839B publication Critical patent/CN112308839B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/0002—Inspection of images, e.g. flaw detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • 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
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • 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
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing
    • G06T2207/10044—Radar image
    • 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
    • G06T2207/20—Special algorithmic details
    • G06T2207/20112—Image segmentation details
    • G06T2207/20156—Automatic seed setting
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Quality & Reliability (AREA)
  • Image Processing (AREA)

Abstract

The invention is suitable for the technical field of computers, and provides a method and a device for cutting single trees of a natural forest, computer equipment and a storage medium, wherein the method comprises the following steps: acquiring point cloud data and determining normal vectors of all points and unit module values on height components; determining trunk seed points, and performing growth treatment on the trunk seed points by using a region growing algorithm; determining branch seed points, performing growth treatment on the branch seed points by using a region growing algorithm, and dividing an under-forest vegetation layer; distributing branch and leaf points according to a nearest neighbor method; and determining the single wood according to the trunk seed points, the branch seed points and the branch leaf points. According to the single-tree segmentation method provided by the invention, the growth of the trunk seed points is realized firstly based on the region growing algorithm to obtain a plurality of trunk seed points of the same type, then the extraction of the branches and the leaves are sequentially carried out, and the trunk, the branches and the leaves are sequentially segmented, so that the single-tree segmentation is realized, and the single-tree segmentation effect on natural forests with complex conditions is excellent.

Description

Single-tree segmentation method and device for natural forest, computer equipment and storage medium
Technical Field
The invention belongs to the technical field of computers, and particularly relates to a method and a device for cutting single trees of a natural forest, computer equipment and a storage medium.
Background
The forest resource inventory is an important component of forest management, plays an important role in maintaining forest sustainable development, monitoring forest plant diseases and insect pests, protecting biodiversity, evaluating fire hazard and the like, and the laser radar is a new remote sensing technology, can acquire information such as target position, height and the like, has strong characterization capability on vegetation space structures, and shows huge potential in describing and monitoring forest structures and functions.
However, in the prior art, there are many problems in using the laser radar to segment single trees in a forest, for example, point coordinates are projected onto a two-dimensional plane, a thiessen polygon is constructed according to the detected position of the single tree to segment the single tree, but in a natural forest, due to the influence of factors such as sunlight and moisture, the forest is dense and the single tree body often grows irregularly, when branches of the forest are intersected and canopy layers are overlapped, the separation of canopy layers of the single tree is difficult to achieve by using a cylindrical or linear cutting method, so that the method can obtain a better result in a forest land with sparse trees; in order to solve the problem of extracting irregular trees, the prior art provides point cloud voxel and hierarchical clustering to realize single tree segmentation, however, the method is suitable for artificial forests, the single trees are still difficult to segment when the trunks of the trees are inclined, and the robustness is poor; similarly, the shortest path comparison method adopts density clustering to identify the singletree, combines the graph theory, and belongs the point to the trunk with the shortest path, and finally realizes the distribution of the canopy to finish the splitting of the singletree, but the method identifies the serious over-splitting phenomenon of the singletree in the shrub and the herbaceous bush in the natural forest.
Therefore, the existing single-tree segmentation method is difficult to be suitable for natural forest lands with complex forest stand conditions, and the single-tree segmentation method is low in precision and poor in robustness due to factors such as shrub herbaceous clusters and forest tree trunk inclination existing in the natural forest lands.
Disclosure of Invention
The embodiment of the invention aims to provide a single tree segmentation method for a natural forest, and aims to solve the technical problems that the existing single tree segmentation method is difficult to be applied to natural forest lands with complex forest stand conditions, and the single tree segmentation method is low in precision and poor in robustness due to factors such as shrub herbaceous clusters and forest main trunk inclination existing in the natural forest lands.
The embodiment of the invention is realized in such a way that a method for cutting single trees of a natural forest comprises the following steps:
acquiring point cloud data of the natural forest;
determining a normal vector of each point in the point cloud data, and determining a unit module value of the normal vector on the height component;
determining points with unit module values smaller than a preset seed point threshold value as trunk seed points, and performing growth processing on the trunk seed points according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed points;
determining nearest neighbor points of the trunk seed points as branch seed points, and performing growth treatment on the branch seed points according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed points;
distributing the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and leaf seed points;
and determining the single tree according to the multiple same-category trunk points, the multiple same-category branch trunk points and the multiple same-category branch leaf points.
Another object of an embodiment of the present invention is to provide a device for cutting a single wood of a natural forest, including:
the point cloud data acquisition unit is used for acquiring point cloud data of the natural forest;
the normal vector calculation unit is used for determining the normal vector of each point in the point cloud data and determining the unit module value of the normal vector on the height component;
a trunk seed point determining unit, configured to determine, as a trunk seed point, a point where the unit modulus value is smaller than a preset seed point threshold, and perform growth processing on the trunk seed point according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed point;
the branch seed point determining unit is used for determining the nearest neighbor point of the trunk seed point as a branch seed point, and performing growth treatment on the branch seed point according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed point;
the branch and leaf point distribution unit is used for distributing the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and seed points;
and the single tree segmentation unit is used for determining the single tree according to the plurality of trunk points of the same category, the plurality of branch and trunk points of the same category and the plurality of branch and leaf points of the same category.
It is a further object of embodiments of the present invention to provide a computer apparatus, comprising a memory and a processor, the memory having stored therein a computer program, which, when executed by the processor, causes the processor to perform the steps of the method for cutting logs of native forest as described above.
It is a further object of embodiments of the present invention to provide a computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, causes the processor to perform the steps of the method for cutting logs of natural forest as described above.
According to the single tree segmentation method for the natural forest provided by the embodiment of the invention, after point cloud data of the natural forest is obtained, determining normal vectors of all points and unit module values on the height components by using the morphological principle of botany, determining trunk seed points based on the unit module values, and realizes the growth of the trunk seed points based on a region growing algorithm to obtain a plurality of trunk seed points of the same category, realizes the extraction of the single tree trunk, further generates the associated branch seed points by the trunk seed points, and the growth of the branch seed points is realized by using the region growing algorithm to obtain a plurality of branch seed points associated with the trunk seed points, thereby realizing the extraction of the branches, finally utilizing the nearest neighbor method to distribute the branch and leaf points, sequentially segmenting the main trunk, the branches and the leaves, therefore, the single-tree segmentation is completely realized, and the single-tree segmentation effect on the natural forest with complex conditions is excellent.
Drawings
FIG. 1 is a flow chart illustrating the steps of a method for cutting a single log of a natural forest according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating a step of determining normal vectors and unit norm values of points according to an embodiment of the present invention;
FIG. 3 is a flowchart illustrating a step of growing a stem seed point according to an embodiment of the present invention;
FIG. 4 is a flowchart illustrating another step of growing a stem seed point according to an embodiment of the present invention;
FIG. 5 is a flowchart illustrating a step of growing a branch seed point according to an embodiment of the present invention;
FIG. 6 is a flowchart illustrating another step of growing a branch seed point according to an embodiment of the present invention;
FIG. 7 is a flow chart illustrating steps of another method for cutting a single log of a natural forest according to an embodiment of the present invention;
FIG. 8 is a diagram illustrating the segmentation effect of various algorithms for the method for segmenting single trees of natural forest according to the embodiment of the present invention;
FIG. 9 is a schematic structural diagram of a single-wood splitting device for a natural forest according to an embodiment of the present invention;
fig. 10 is an internal structural diagram of a computer device for performing a method for cutting a single log of a natural forest according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
As shown in fig. 1, a flow chart of steps of a method for cutting a single wood of a natural forest according to an embodiment of the present invention specifically includes the following steps:
and S102, acquiring point cloud data of the natural forest.
In the embodiment of the invention, considering that a conventional Airborne Laser radar (ALS) adopts a top-down Scanning mode, only a crown part can be scanned, but trunk and branch points are difficult to obtain, and the single-tree segmentation effect of a natural forest is poor, therefore, the point cloud data of the natural forest is obtained by adopting a bottom-up Scanning mode based on a ground-based Laser radar (TLS), the under-forest trunk, branch and leaf points can be obtained in a high-density and complete manner through the TLS, and the obtained point cloud data of the natural forest is more accurate.
And step S104, determining normal vectors of all points in the point cloud data, and determining unit module values of the normal vectors on the height components.
In the embodiment of the invention, as the main trunk of the forest is often tall, big and stout in the natural forest, which is obviously different from young trees, shrubs, herbaceous plants and the like, the extraction of the single-tree main trunk is a prerequisite condition for identifying single trees and realizing single-tree segmentation in the natural forest. Consider the unit modulus value Z of the normal vector of each point in the point cloud data over the height componentnThe smaller the value of the feature value of a point on the vertical plane, the closer the point is to the vertical plane, so based on the above idea, the segmentation of the single tree trunk can be realized, wherein, please refer to fig. 2 and its explanation for the specific steps and calculation formulas for calculating the normal vector of each point in the point cloud data and determining the unit module value of the normal vector on the height component.
And S106, determining points with unit modulus values smaller than a preset seed point threshold value as trunk seed points, and performing growth processing on the trunk seed points according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed points.
In the embodiment of the invention, theUnit module value Z of normal vector of each point in point cloud data on height componentnThe characteristic value of a point on the vertical plane can be taken as a characteristic value, and the smaller the value, the closer the point is to the vertical plane. Therefore, the method determines the points with the unit modulus value smaller than the preset seed point threshold value as the trunk seed points (a plurality of trunk seed points can exist), and based on the trunk seed points, the trunk seed points are subjected to growth processing by using a preset region growing algorithm to obtain a plurality of trunk points of the same type associated with the trunk seed points, so that the trunk seed points are grown, and the trunk seed points and the trunk points of the same type are combined, thereby realizing the division of the single wood trunk. Specifically, a step of performing growth processing on the trunk seed points by using a region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed points is described with reference to subsequent fig. 3 and fig. 4 and an explanation thereof.
In the embodiment of the present invention, the seed point threshold is preferably 0.05.
And S108, determining nearest neighbor points of the trunk seed points as branch seed points, and performing growth treatment on the branch seed points according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed points.
In the embodiment of the invention, in consideration of the plant morphology, the branch is connected with the trunk and the branches and leaves and is a bridge of the trunk and the branches and leaves, and the point cloud data also meets the characteristic, so that the nearest neighbor point of the trunk seed point is determined as the branch seed point by searching the neighboring points around the trunk, and the branch seed point is subjected to growth processing by further utilizing a region growing algorithm, thereby realizing the regrowth of the branch and completing the segmentation of the branch. Specifically, a step of performing growth processing on the branch seed points by using a region growing algorithm to obtain a plurality of trunk points of the same category associated with the branch seed points is described with reference to subsequent fig. 5 and fig. 6 and an explanation thereof.
And step S110, distributing the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and seed points.
In the embodiment of the invention, considering that a natural forest is usually vertically layered and sequentially comprises a herbal layer, an shrub layer and a tree layer from low to high, wherein the tree layer and the shrub layer have obvious height difference, the method is mainly used for segmenting a single arbor, so that branch and leaf points can be determined by setting a certain height value as a boundary between the tree layer and the shrub layer, the distance from the branch and leaf points to peripheral known points is calculated based on a nearest algorithm, and then the nearest category is selected for distribution, so that the distribution of the branch and leaf points is realized, and a plurality of branch and leaf points of the same category associated with the branch and leaf seed points are obtained.
And S112, determining the single tree according to the multiple same-category trunk points, the multiple same-category branch trunk points and the multiple same-category branch leaf points.
In the embodiment of the invention, after the trunk points, the branch points and the branch and leaf points are divided, the single trees in the natural forest are finally determined according to the types of the single trees, namely the single trees belong to different single trees.
According to the single tree segmentation method for the natural forest provided by the embodiment of the invention, after point cloud data of the natural forest is obtained, determining normal vectors of all points and unit module values on the height components by using the morphological principle of botany, determining trunk seed points based on the unit module values, and realizes the growth of the trunk seed points based on a region growing algorithm to obtain a plurality of trunk seed points of the same category, realizes the extraction of the single tree trunk, further generates the associated branch seed points by the trunk seed points, and the growth of the branch seed points is realized by using the region growing algorithm to obtain a plurality of branch seed points associated with the trunk seed points, thereby realizing the extraction of the branches, finally utilizing the nearest neighbor method to distribute the branch and leaf points, sequentially segmenting the main trunk, the branches and the leaves, therefore, the single-tree segmentation is completely realized, and the single-tree segmentation effect on the natural forest with complex conditions is excellent.
As shown in fig. 2, a flowchart of a step of determining a normal vector and a unit module value of a point provided in an embodiment of the present invention specifically includes the following steps:
step S202, determining points of normal vectors to be calculated in the point cloud data and determining a plurality of adjacent points of the normal vectors to be calculated.
In the embodiment of the present invention, it is actually necessary to perform normal vector calculation on each point in the point cloud data in sequence, that is, it is necessary to determine each point in the point cloud data as a point of a normal vector to be calculated in sequence.
Step S204, determining the normal vector of the point of the normal vector to be calculated according to the covariance matrix of the point of the normal vector to be calculated and the plurality of adjacent points.
In the embodiment of the invention, the point P of the normal vector to be calculated in the point cloud data is determinediThen, the point P of the normal vector to be calculatediThe covariance matrix with k neighbors is as follows:
Figure BDA0002754716140000081
at this time, the covariance matrix is subjected to feature decomposition, specifically including:
Figure BDA0002754716140000082
where j is 1,2, and 3 in the formula, representing the number of eigenvectors and eigenvalues, e is the corresponding eigenvector, and λ is the corresponding eigenvalue (λ)1≤λ2≤λ3) Wherein the eigenvector corresponding to the minimum eigenvalue is the normal vector e of the point1。
Step S206, calculating a unit module value of the normal vector on the height component according to the normal vector of the point of the normal vector to be calculated.
In the embodiment of the present invention, n is made to [0,0,1 ═ 0]Then the unit module value Z of the normal vector on the height componentn＝|e1n|。
As shown in fig. 3, a flowchart of steps for performing growth processing on a trunk seed point provided in an embodiment of the present invention specifically includes the following steps:
step S302, a plurality of trunk neighboring points of the trunk seed points are determined, and the normal vector included angle between the trunk neighboring points and the trunk seed points is calculated.
In the embodiment of the present invention, the formula for calculating the normal vector included angle belongs to the common general knowledge of those skilled in the art, and is not specifically described herein, the stem points of the same category associated with the stem seed points are obtained by calculating the normal vector included angle between the adjacent points of the stem and the stem seed points,
step S304, determining neighboring trunk points whose normal vector included angles with the trunk seed points are smaller than a preset first threshold as a plurality of trunk points of the same category associated with the trunk seed points.
In the embodiment of the present invention, when an included angle between a neighboring point and a seed point normal vector is smaller than a certain threshold, the neighboring point is regarded as a point with a trunk, that is, the neighboring point can be determined as a plurality of trunk points of the same category associated with the trunk seed point.
In the embodiment of the present invention, the first threshold value is preferably set to 10 ° to 40 °, and more preferably to 20 °.
Fig. 4 is a flowchart of another step of growing a stem seed point according to an embodiment of the present invention, which is described in detail below.
In the embodiment of the present invention, the difference from the flowchart of the step of growing the trunk seed point shown in fig. 3 is that after the step S304, the method further includes:
step S402, the trunk points of the same category, the difference value of which with the unit module value of the trunk seed point is smaller than a preset second threshold value, are determined as the trunk seed points.
In the embodiment of the present invention, after the trunk points of the same type are determined, the trunk points of the same type whose difference value between the unit modulus value of the trunk seed point and the unit modulus value of the trunk seed point is smaller than the second threshold value may be determined again as the trunk seed points, at this time, the step S302 may be returned to, and the newly determined trunk seed points are continuously processed, and this is repeated, so that the growth processing of the trunk seed points is realized.
In the embodiment of the present invention, the second threshold is preferably 0.05.
As shown in fig. 5, a flowchart of steps for performing growth treatment on branch seed points provided in an embodiment of the present invention specifically includes the following steps:
step S502, determining a plurality of branch adjacent points of the branch seed points, and calculating the normal vector included angle between the branch adjacent points and the branch seed points.
In the embodiment of the present invention, similar to the trunk seed point, it is necessary to determine a normal vector angle between a branch neighboring point of the branch seed point and the branch seed point, and determine the branch points of the same category associated with the branch seed point through the normal vector angle.
Step S504, determining the adjacent points of the branch, the included angles of which with the normal vectors of the branch seed points are smaller than a preset third threshold value, as a plurality of branch points of the same category associated with the branch seed points.
In the embodiment of the present invention, when an included angle between a neighboring point and a normal vector of a seed point is smaller than a certain threshold, the neighboring point is regarded as a point with the trunk, that is, the neighboring point can be determined as a plurality of branches of the same category associated with the branch seed point.
In the embodiment of the present invention, the third threshold value is preferably set to 10 ° to 40 °, and more preferably to 20 °.
Fig. 6 is a flowchart of another step of growing a branch seed according to an embodiment of the present invention, which is described in detail below.
In the embodiment of the present invention, the difference from the flowchart of the step of growing the branch seed point shown in fig. 5 is that after step S504, the method further includes:
step S602, determining the branch points of the same category whose distance difference from the branch seed point is smaller than a preset fourth threshold as the branch seed points.
In the embodiment of the present invention, similarly, after the branch points of the same category are determined, the branch points of the same category whose distance difference from the branch seed point is smaller than the fourth threshold may be determined again as the branch seed points. At this time, the process may return to the step S502, and continue to process the newly determined branch seed points, and the process is repeated in this way, thereby implementing the growth process on the branch seed points.
Fig. 7 is a flowchart illustrating steps of another method for cutting single trees from natural forests according to an embodiment of the present invention, which is described in detail below.
In the embodiment of the present invention, the difference from the step flow chart of the method for splitting single trees of a natural forest shown in fig. 1 is that after step S102, the method further includes:
step S702, ground points are identified based on a progressive triangulation network filtering algorithm, elevation normalization processing is carried out on the point cloud data of the natural forest, and point cloud data after normalization processing are generated.
In the embodiment of the invention, in consideration of the fact that natural forests may grow in regions with fluctuating soil in the actual test process, the acquired point cloud data cannot be directly used, and a pretreatment step is required. Recognizing ground points by adopting improved progressive triangulation network filtering, and performing elevation normalization processing, namely subtracting elevation values of forest sites from the recognized ground points to convert point cloud data onto a plane; and cutting the point cloud data to the size of the sample plot, completing preprocessing, and obtaining point cloud data after normalization processing so as to facilitate subsequent single wood identification.
In order to understand the difference between the present invention and the conventional technical solution, the following will specifically illustrate the experimental effect of the algorithm provided by the present invention relative to the conventional algorithm when applied to the single-wood segmentation of the natural forest.
In the embodiment of the present invention, for convenience of description, the method for splitting single trees of a natural forest provided by the present invention is implemented by using a Trunk-growing algorithm (TG), which is abbreviated as TG in the subsequent experimental process.
In the embodiment of the present invention, the conventional algorithm for comparing with the method for splitting single trees of natural forest proposed by the present invention mainly comprises:
1) a single tree Segmentation method (PCS) combining region growing with a threshold value is provided based on ALS, the method firstly assumes that the highest Point in a Point set is a tree high Point, realizes the Segmentation of one tree through region growing, then compares other points with the distance of the known single tree to gradually realize the single tree Segmentation, is suitable for the TLS single tree Segmentation to a certain extent besides ALS, and is abbreviated as PCS in the subsequent test process;
2) the Shortest Path comparison method (CSP) is characterized in that Density-Based Spatial Clustering of Applications with Noise, DBSCAN (binary Clustering of spread with Noise) is adopted to identify single trees, points are assigned to the trunk with the Shortest Path by combining a graph theory, and finally, canopy distribution is realized to finish single tree segmentation, and the single trees are abbreviated as CSP in the subsequent test process.
In order to conveniently represent the recognition effect of the single wood recognition of each algorithm, the following precision verification and method comparison are provided:
when the single tree is successfully segmented, the single tree is marked as True segmentation (TP); when the single wood is not successfully divided, the single wood is marked as error division (FN); when the single wood does not exist but is segmented, the single wood is marked as False segmentation (FP); recall (Recall, R), Precision (Precision, P) and F-score were used as evaluation criteria for evaluating the segmentation of the singlewood: the calculation formulas of the recall rate R, the precision P and the F score are respectively as follows:
Figure BDA0002754716140000121
at this time, the three algorithms TG, PCS and CSP were used to identify single trees from the same natural forest (forest land one) and shrub (forest land one), and evaluation indexes of the respective algorithms were calculated, and specific calculation results are shown in table 1 below:
table 1:
Figure BDA0002754716140000122
it can be readily seen from the table that TG performs well in both plots; in a same sample, the R value reaches 1.00, each single tree is well divided, and F reaches 0.96; in the second plot, the R value is 0.95, one single tree is poor in growth vigor and is distributed to the shrub layer in the crushed combination, and the F fraction also reaches 0.95; the F-score of TG was highest in all three algorithms; obviously, although the CSP can well divide the single trees, the single trees are easily identified by shrubs, herbaceous plants and the like, so that the P value is low, and the F score is influenced finally; the PCS under-segments more severely in both sample plots; when two single-tree canopies are intersected, PCS is difficult to separate the two single-tree canopies due to the lack of constraint conditions, the R value is low, the single-tree segmentation condition is poor, and TG and CSP have prior identification conditions for the single trees, so that the single trees can be segmented more accurately.
As shown in fig. 8, a schematic diagram of the recognition result of using the three algorithms TG, PCS and CSP to respectively perform single-tree recognition on the same forest land and shrub, wherein the recognized single tree is marked with different colors, which is specifically shown in fig. 8. The data shown in the figures also better illustrate the superiority of TG and CSP algorithms over PCS algorithms in identifying singles.
For TG and CSP with high R value, the research compares the TG and CSP from the aspects of canopy segmentation, single-wood segmentation of irregular trunks and the like.
(1) And (4) dividing the canopy. Both TG and CSP show over-segmentation in the canopy. The growing places of naturally formed natural forest single trees are not regular usually, the distribution distance of the two single trees is not uniform, crowns are staggered during growing, point cloud data obtained is represented as point-point mixing, the canopy layers are still difficult to separate perfectly by nearest distribution or graph theory distribution points in the current algorithm, and the point-point mixing method is a key point and a difficulty point in follow-up research.
(2) Irregular trunks and crowns. Under the influence of factors such as sunlight irradiation angle and moisture, the same tree species in the forest stand is often in a competitive relationship, and the more nutrient substances are obtained by a single tree, the more divergent branches and leaves grow; under the condition of limited living space, the single-wood making with less nutrient substances has slower and poorer development, and the original development direction of the tree species can be changed to a certain extent for the growth, so that the S-shaped trunk grows. The TG and the CSP can better realize the identification and the segmentation of the single tree branches, the single tree branches are distributed with crown points according to the nearest trunk, and the precision is better under proper parameters; the latter realizes the crown point distribution by a method of comparing graph theory with the shortest path.
(3) Single-wood splitting of the oblique trunk. TG identifies the single tree trunk firstly, and then continuously supplements trunk and branch points in the subsequent growth, so that the algorithm can better identify the single tree with a larger inclination angle, but the condition of insufficient segmentation of the canopy occurs; compared with the CSP, the CSP is worse, wrong segmentation occurs, the original single wood is divided into two single wood, the limitation is that the algorithm preferentially identifies the single wood at the breast diameter, and when the inclination angle of the single wood is large or the inverted wood is formed, the trunk height is lower than 1.3m, wrong identification or no identification occurs. TG is more dominant in the face of such monoliths.
(4) Separating shrub and herb. Shrubs and herbs are competitors of single trees in natural forests, in previous researches, the single tree parameters (positions, breast diameters and the like) are usually interfered by the shrubs and the herbs, and how to accurately acquire effective information from the interference is a big problem. In the CSP cutting of single trees, the separation of the tree trunk from the herbaceous vegetation and shrub points can not be realized, because the algorithm considers that the points in the sample plot are all arbor points, the complexity of the natural forest is neglected, the non-main points with the close main trunk are difficult to separate by adopting the shortest path method, and the two are often mixed into one kind; TG takes the above point into consideration, and takes a certain height as the layering boundary of shrubs, herbs and trees, so that the under-forest environment can be effectively separated from trees.
In combination with the above, it can be known that the TG algorithm has better effect on the single-wood segmentation of the natural forest than the conventional PCS and CSP algorithms used in the field of single-wood segmentation.
Fig. 9 is a schematic structural diagram of a single-wood splitting device for natural forest according to an embodiment of the present invention, which is described in detail below.
In an embodiment of the present invention, the apparatus for splitting a single tree of a natural forest includes:
a point cloud data obtaining unit 910, configured to obtain point cloud data of the natural forest.
In the embodiment of the invention, considering that a conventional Airborne Laser radar (ALS) adopts a top-down Scanning mode, only a crown part can be scanned, but trunk and branch points are difficult to obtain, and the single-tree segmentation effect of a natural forest is poor, therefore, the point cloud data of the natural forest is obtained by adopting a bottom-up Scanning mode based on a ground-based Laser radar (TLS), the under-forest trunk, branch and leaf points can be obtained in a high-density and complete manner through the TLS, and the obtained point cloud data of the natural forest is more accurate.
And a normal vector calculation unit 920, configured to determine a normal vector of each point in the point cloud data, and determine a unit module value of the normal vector on the height component.
In the embodiment of the invention, as the main trunk of the forest is often tall, big and stout in the natural forest, which is obviously different from young trees, shrubs, herbaceous plants and the like, the extraction of the single-tree main trunk is a prerequisite condition for identifying single trees and realizing single-tree segmentation in the natural forest. Consider the unit modulus value Z of the normal vector of each point in the point cloud data over the height componentnThe smaller the value of the feature value of a point on the vertical plane, the closer the point is to the vertical plane, so based on the above idea, the segmentation of the single tree trunk can be realized, wherein, please refer to fig. 2 and its explanation for the specific steps and calculation formulas for calculating the normal vector of each point in the point cloud data and determining the unit module value of the normal vector on the height component.
A trunk seed point determining unit 930, configured to determine, as a trunk seed point, a point whose unit module value is smaller than a preset seed point threshold, and perform growth processing on the trunk seed point according to a preset region growing algorithm, so as to obtain a plurality of trunk points of the same category associated with the trunk seed point.
In the embodiment of the invention, the unit module value Z of the normal vector of each point in the point cloud data on the height componentnThe characteristic value of a point on the vertical plane can be taken as a characteristic value, and the smaller the value, the closer the point is to the vertical plane. Therefore, the present invention determines the point where the unit modulus value is smaller than the preset seed point threshold value as the trunk seed point (there may be a plurality of trunk seed points), and bases on the trunk seed pointAnd the trunk seed points are subjected to growth treatment by utilizing a preset region growing algorithm to obtain a plurality of trunk points of the same type associated with the trunk seed points, so that the growth of the trunk seed points is realized, and the trunk seed points and the trunk points of the same type are combined, thereby realizing the segmentation of the single-wood trunk. Specifically, please refer to fig. 3 and 4 and the explanation thereof, wherein the step of obtaining a plurality of trunk points of the same category associated with the trunk seed points is to perform growth processing on the trunk seed points by using a region growing algorithm.
The branch seed point determining unit 940 is configured to determine a nearest neighbor point of the trunk seed point as a branch seed point, and perform growth processing on the branch seed point according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed point.
In the embodiment of the invention, in consideration of the plant morphology, the branch is connected with the trunk and the branches and leaves and is a bridge of the trunk and the branches and leaves, and the point cloud data also meets the characteristic, so that the nearest neighbor point of the trunk seed point is determined as the branch seed point by searching the neighboring points around the trunk, and the branch seed point is subjected to growth processing by further utilizing a region growing algorithm, thereby realizing the regrowth of the branch and completing the segmentation of the branch. Specifically, a step of performing growth processing on the branch seed points by using a region growing algorithm to obtain a plurality of trunk points of the same category associated with the branch seed points is described with reference to fig. 5 and 6 and an explanation thereof.
And a branch and leaf point distribution unit 950, configured to distribute the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and leaf seed points.
In the embodiment of the invention, considering that a natural forest is usually vertically layered and sequentially comprises a herbal layer, an shrub layer and a tree layer from low to high, wherein the tree layer and the shrub layer have obvious height difference, the method is mainly used for segmenting a single arbor, so that branch and leaf points can be determined by setting a certain height value as a boundary between the tree layer and the shrub layer, the distance from the branch and leaf points to peripheral known points is calculated based on a nearest algorithm, and then the nearest category is selected for distribution, so that the distribution of the branch and leaf points is realized, and a plurality of branch and leaf points of the same category associated with the branch and leaf seed points are obtained.
The single-tree segmentation unit 960 is configured to determine a single tree according to the plurality of trunk points of the same category, the plurality of branches and trunk points of the same category, and the plurality of branches and leaves of the same category.
In the embodiment of the invention, after the trunk points, the branch points and the branch and leaf points are divided, the single trees in the natural forest are finally determined according to the types of the single trees, namely the single trees belong to different single trees.
According to the single-tree segmentation device for the natural forest provided by the embodiment of the invention, after point cloud data of the natural forest is obtained, determining normal vectors of all points and unit module values on the height components by using the morphological principle of botany, determining trunk seed points based on the unit module values, and realizes the growth of the trunk seed points based on a region growing algorithm to obtain a plurality of trunk seed points of the same category, realizes the extraction of the single tree trunk, further generates the associated branch seed points by the trunk seed points, and the growth of the branch seed points is realized by using the region growing algorithm to obtain a plurality of branch seed points associated with the trunk seed points, thereby realizing the extraction of the branches, finally utilizing the nearest neighbor method to distribute the branch and leaf points, sequentially segmenting the main trunk, the branches and the leaves, therefore, the single-tree segmentation is completely realized, and the single-tree segmentation effect on the natural forest with complex conditions is excellent.
FIG. 10 is a diagram illustrating an internal structure of a computer device in one embodiment. As shown in fig. 10, the computer apparatus includes a processor, a memory, a network interface, an input device, and a display screen connected through a system bus. Wherein the memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program that, when executed by the processor, causes the processor to implement a method of singulating trees in a natural forest. The internal memory may also have a computer program stored therein, which when executed by the processor, causes the processor to perform a method of singulating trees from natural forests. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.
Those skilled in the art will appreciate that the architecture shown in fig. 10 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, the cutting device for the single trees of the natural forest provided by the application can be realized in the form of a computer program, and the computer program can be run on a computer device as shown in fig. 10. The memory of the computer device may store various program modules constituting the individual tree splitting apparatus of the natural forest, such as a point cloud data acquisition unit 910, a normal vector calculation unit 920, and a trunk seed point determination unit 930 shown in fig. 9, and the like. The computer program constituted by the respective program modules causes the processor to execute the steps in the method for cutting single trees of natural forests according to the respective embodiments of the present application described in the present specification.
For example, the computer apparatus shown in fig. 10 may perform step S102 by the point cloud data acquisition unit 910 in the single wood division apparatus of the natural forest as shown in fig. 9; the computer device may perform step S104 by the normal vector calculation unit 920; the computer device may perform step S106 by the stem seed point determination unit 930.
In one embodiment, a computer device is proposed, the computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program:
acquiring point cloud data of the natural forest;
determining a normal vector of each point in the point cloud data, and determining a unit module value of the normal vector on the height component;
determining points with unit module values smaller than a preset seed point threshold value as trunk seed points, and performing growth processing on the trunk seed points according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed points;
determining nearest neighbor points of the trunk seed points as branch seed points, and performing growth treatment on the branch seed points according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed points;
distributing the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and leaf seed points;
and determining the single tree according to the multiple same-category trunk points, the multiple same-category branch trunk points and the multiple same-category branch leaf points.
In one embodiment, a computer readable storage medium is provided, having a computer program stored thereon, which, when executed by a processor, causes the processor to perform the steps of:
acquiring point cloud data of the natural forest;
determining a normal vector of each point in the point cloud data, and determining a unit module value of the normal vector on the height component;
determining points with unit module values smaller than a preset seed point threshold value as trunk seed points, and performing growth processing on the trunk seed points according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed points;
determining nearest neighbor points of the trunk seed points as branch seed points, and performing growth treatment on the branch seed points according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed points;
distributing the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and leaf seed points;
and determining the single tree according to the multiple same-category trunk points, the multiple same-category branch trunk points and the multiple same-category branch leaf points.
It should be understood that, although the steps in the flowcharts of the embodiments of the present invention are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in various embodiments may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the program is executed. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the embodiments described above may be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the embodiments described above are not described, but should be considered as being within the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention. Therefore, the protection scope of the present patent shall be subject to the appended claims.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.

Claims (10)

1. A method for cutting single trees of a natural forest is characterized by comprising the following steps:
acquiring point cloud data of the natural forest;
determining a normal vector of each point in the point cloud data, and determining a unit module value of the normal vector on the height component;
determining points with unit module values smaller than a preset seed point threshold value as trunk seed points, and performing growth processing on the trunk seed points according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed points;
determining nearest neighbor points of the trunk seed points as branch seed points, and performing growth treatment on the branch seed points according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed points;
distributing the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and leaf seed points;
and determining the single tree according to the multiple same-category trunk points, the multiple same-category branch trunk points and the multiple same-category branch leaf points.
2. The method for splitting single trees according to claim 1, wherein the step of determining the normal vector of each point in the point cloud data and determining the unit module value of the normal vector on the height component specifically comprises:
determining a point of a normal vector to be calculated in the point cloud data and determining a plurality of adjacent points of the point of the normal vector to be calculated;
determining the normal vector of the point of the normal vector to be calculated according to the covariance matrix of the point of the normal vector to be calculated and the plurality of adjacent points;
and calculating the unit module value of the normal vector on the height component according to the normal vector of the point of the normal vector to be calculated.
3. The veneer splitting method according to claim 1, wherein the step of growing the trunk seed points according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed points specifically comprises:
determining a plurality of trunk adjacent points of the trunk seed points, and calculating a normal vector included angle between the trunk adjacent points and the trunk seed points;
and determining the adjacent points of the trunk, of which the included angles with the normal vectors of the trunk seed points are smaller than a preset first threshold value, as a plurality of trunk points of the same category associated with the trunk seed points.
4. The method for splitting single trees according to claim 3, wherein after said step of determining the trunk neighboring points whose normal vector included angle with said trunk seed point is smaller than a preset first threshold as a plurality of homogeneous trunk points associated with said trunk seed point, further comprising:
and determining the trunk points of the same category, of which the difference value with the unit module value of the trunk seed points is smaller than a preset second threshold value, as the trunk seed points.
5. The method for cutting single trees according to claim 1, wherein the step of growing the branch seed points according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed points specifically comprises:
determining a plurality of branch adjacent points of the branch seed points, and calculating normal vector included angles between the branch adjacent points and the branch seed points;
and determining the adjacent points of the branches, of which the normal vector included angles with the branch seed points are smaller than a preset third threshold value, as a plurality of branch points of the same category associated with the branch seed points.
6. The method for single tree segmentation according to claim 5, wherein after the step of determining, as the plurality of stem points of the same category associated with the stem seed points, stem neighboring points whose normal vector included angle with the stem seed points is smaller than a preset third threshold value, the method further comprises:
and determining the branch points of the same category, the distance difference between which and the branch seed point is smaller than a preset fourth threshold value, as the branch seed points.
7. The method of singletree segmentation according to claim 1, further comprising, after the step of obtaining point cloud data of a native forest:
and identifying ground points based on a progressive triangulation filtering algorithm, and performing elevation normalization processing on the point cloud data of the natural forest to generate point cloud data after normalization processing.
8. A device for cutting single trees of natural forests, comprising:
the point cloud data acquisition unit is used for acquiring point cloud data of the natural forest;
the normal vector calculation unit is used for determining the normal vector of each point in the point cloud data and determining the unit module value of the normal vector on the height component;
a trunk seed point determining unit, configured to determine, as a trunk seed point, a point where the unit modulus value is smaller than a preset seed point threshold, and perform growth processing on the trunk seed point according to a preset region growing algorithm to obtain a plurality of trunk points of the same category associated with the trunk seed point;
the branch seed point determining unit is used for determining the nearest neighbor point of the trunk seed point as a branch seed point, and performing growth treatment on the branch seed point according to a preset region growing algorithm to obtain a plurality of branch points of the same category associated with the branch seed point;
the branch and leaf point distribution unit is used for distributing the branch and leaf points according to a preset nearest neighbor method to obtain a plurality of branch and leaf points of the same category associated with the branch and seed points;
and the single tree segmentation unit is used for determining the single tree according to the plurality of trunk points of the same category, the plurality of branch and trunk points of the same category and the plurality of branch and leaf points of the same category.
9. A computer arrangement, characterized by comprising a memory and a processor, the memory having stored therein a computer program which, when executed by the processor, causes the processor to carry out the steps of the method of singulation of native forests according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that a computer program is stored thereon, which, when being executed by a processor, causes the processor to carry out the steps of the method of singulation of native forests according to any one of claims 1 to 7.
CN202011198583.5A 2020-10-31 2020-10-31 Single-tree segmentation method and device for natural forest, computer equipment and storage medium Active CN112308839B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011198583.5A CN112308839B (en) 2020-10-31 2020-10-31 Single-tree segmentation method and device for natural forest, computer equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011198583.5A CN112308839B (en) 2020-10-31 2020-10-31 Single-tree segmentation method and device for natural forest, computer equipment and storage medium

Publications (2)

Publication Number Publication Date
CN112308839A true CN112308839A (en) 2021-02-02
CN112308839B CN112308839B (en) 2022-08-02

Family

ID=74332333

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011198583.5A Active CN112308839B (en) 2020-10-31 2020-10-31 Single-tree segmentation method and device for natural forest, computer equipment and storage medium

Country Status (1)

Country Link
CN (1) CN112308839B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101839701A (en) * 2010-05-27 2010-09-22 中国科学院自动化研究所 Method for automatically partitioning tree point cloud data
US20180082424A1 (en) * 2016-09-22 2018-03-22 Beijing Greenvalley Technology Co., Ltd. Single-tree segmentation method and apparatus
CN111462134A (en) * 2020-03-31 2020-07-28 武汉大学 Single-tree segmentation method and system for fusing high-resolution remote sensing image and laser radar point cloud
CN111598915A (en) * 2020-05-19 2020-08-28 北京数字绿土科技有限公司 Point cloud single wood segmentation method, device, equipment and computer readable medium

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101839701A (en) * 2010-05-27 2010-09-22 中国科学院自动化研究所 Method for automatically partitioning tree point cloud data
US20180082424A1 (en) * 2016-09-22 2018-03-22 Beijing Greenvalley Technology Co., Ltd. Single-tree segmentation method and apparatus
CN111462134A (en) * 2020-03-31 2020-07-28 武汉大学 Single-tree segmentation method and system for fusing high-resolution remote sensing image and laser radar point cloud
CN111598915A (en) * 2020-05-19 2020-08-28 北京数字绿土科技有限公司 Point cloud single wood segmentation method, device, equipment and computer readable medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
杨玉泽等: "基于激光点云数据的树木枝叶分割和三维重建", 《西北林学院学报》 *

Also Published As

Publication number Publication date
CN112308839B (en) 2022-08-02

Similar Documents

Publication Publication Date Title
Mia et al. Mango leaf disease recognition using neural network and support vector machine
Chan et al. Evaluation of Random Forest and Adaboost tree-based ensemble classification and spectral band selection for ecotope mapping using airborne hyperspectral imagery
US9639755B2 (en) Automated compound structure characterization in overhead imagery
Sandika et al. Random forest based classification of diseases in grapes from images captured in uncontrolled environments
CN110223314A (en) A kind of single wooden dividing method based on the distribution of tree crown three-dimensional point cloud
Bienert et al. A voxel-based technique to estimate the volume of trees from terrestrial laser scanner data
CN109446986B (en) Effective feature extraction and tree species identification method for tree laser point cloud
Estornell et al. Tree extraction and estimation of walnut structure parameters using airborne LiDAR data
Yao et al. Identifying standing dead trees in forest areas based on 3D single tree detection from full waveform lidar data
CN106845399A (en) A kind of method that use hierarchical cluster mode extracts individual tree information from LiDAR point cloud
Belton et al. Processing tree point clouds using Gaussian Mixture Models
Liu et al. Point-cloud segmentation of individual trees in complex natural forest scenes based on a trunk-growth method
Linhui et al. Extracting the forest type from remote sensing images by random forest
Ghosh et al. Calcrop21: A georeferenced multi-spectral dataset of satellite imagery and crop labels
Manfrini et al. Innovative approaches to orchard management: Assessing the variability in yield and maturity in a ‘Gala’apple orchard using a simple management unit modeling approach
Ferrer-Ferrer et al. Simultaneous fruit detection and size estimation using multitask deep neural networks
CN112308839B (en) Single-tree segmentation method and device for natural forest, computer equipment and storage medium
Räty et al. Fusing diameter distributions predicted by an area-based approach and individual-tree detection in coniferous-dominated forests
Zhang et al. Classification of landscape types based on land cover, successional stages and plant functional groups in a species-rich forest in Hainan Island, China
Madi et al. Univariate models to represent the diametric distribution of thinned stand of Tectona grandis Linn. F
Dahiya et al. An Effective Detection of Litchi Disease using Deep Learning
Zhu et al. Exploring soybean flower and pod variation patterns during reproductive period based on fusion deep learning
Sosa et al. An application using canny edge detection and multilayer perceptron for recognizing leaves of tropical plants
Ameur et al. A fusion system for tree species recognition through leaves and barks
Kamo A new approach to classify growth patterns based on growth function selection and k-means method

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
GR01 Patent grant
GR01 Patent grant