CN117315488B - Urban street tree extraction method based on point cloud features and morphological features - Google Patents
Urban street tree extraction method based on point cloud features and morphological features Download PDFInfo
- Publication number
- CN117315488B CN117315488B CN202311454780.2A CN202311454780A CN117315488B CN 117315488 B CN117315488 B CN 117315488B CN 202311454780 A CN202311454780 A CN 202311454780A CN 117315488 B CN117315488 B CN 117315488B
- Authority
- CN
- China
- Prior art keywords
- point
- point cloud
- tree
- clouds
- points
- 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.)
- Active
Links
- 230000000877 morphologic effect Effects 0.000 title claims abstract description 24
- 238000000605 extraction Methods 0.000 title claims description 16
- 238000000034 method Methods 0.000 claims abstract description 49
- 239000002023 wood Substances 0.000 claims abstract description 32
- 230000011218 segmentation Effects 0.000 claims abstract description 17
- 238000001914 filtration Methods 0.000 claims abstract description 15
- 210000000481 breast Anatomy 0.000 claims abstract description 14
- 238000001514 detection method Methods 0.000 claims abstract description 7
- 238000007781 pre-processing Methods 0.000 claims abstract description 7
- 238000004364 calculation method Methods 0.000 claims description 24
- 239000011159 matrix material Substances 0.000 claims description 15
- 230000035772 mutation Effects 0.000 claims description 13
- 238000010606 normalization Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 230000008030 elimination Effects 0.000 claims description 6
- 238000003379 elimination reaction Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 210000000038 chest Anatomy 0.000 claims description 5
- 102220605052 Histone H4-like protein type G_S61A_mutation Human genes 0.000 claims description 3
- 102220479869 Protein FAM180A_S62A_mutation Human genes 0.000 claims description 3
- 102220630896 Protein KRI1 homolog_S63A_mutation Human genes 0.000 claims description 3
- 239000000284 extract Substances 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 4
- 238000011835 investigation Methods 0.000 description 4
- 230000004927 fusion Effects 0.000 description 3
- 229910052760 oxygen Inorganic materials 0.000 description 3
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 2
- 229910052799 carbon Inorganic materials 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000003915 air pollution Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- QVGXLLKOCUKJST-UHFFFAOYSA-N atomic oxygen Chemical compound [O] QVGXLLKOCUKJST-UHFFFAOYSA-N 0.000 description 1
- 238000009412 basement excavation Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 239000007789 gas Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 229910052757 nitrogen Inorganic materials 0.000 description 1
- 239000001301 oxygen Substances 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/176—Urban or other man-made structures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A30/00—Adapting or protecting infrastructure or their operation
- Y02A30/60—Planning or developing urban green infrastructure
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Databases & Information Systems (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a method for extracting urban street trees based on point cloud features and morphological features, which comprises the following steps: s1, preprocessing LiDAR original point cloud data; s2, filtering the point clouds of the motor vehicle and the non-motor vehicle by adopting a breast diameter circle fitting method based on morphological characteristics of the street tree, and extracting the main point clouds; s3, filtering building point clouds based on the point cloud field characteristics, and extracting canopy point clouds; s4, fusing the trunk point cloud and the canopy point cloud of the street tree to obtain the whole point cloud of the street tree, and performing single tree segmentation on the street tree; s5, filtering the street lamp and the indication board point cloud by adopting a similar detection method to obtain an urban street tree point cloud; the invention fully considers the morphological characteristics of the pavement tree and the pavement tree point cloud characteristics in the LiDAR, divides the pavement tree into two parts of a single wood trunk and a crown layer, respectively extracts, and then fuses, thereby effectively eliminating the noise point cloud in the real scene and efficiently and accurately extracting the urban pavement tree.
Description
Technical Field
The invention relates to the technical field of urban street tree extraction, in particular to a method for extracting an urban street tree based on point cloud characteristics and morphological characteristics.
Background
The pavement tree is used as an important component of urban green space, absorbs tail gas discharged by motor vehicles on roads, shields the shade, adjusts the temperature, fixes nitrogen and releases oxygen, reduces air pollution and noise, and can effectively regulate and improve urban microclimate. The street tree resource investigation is an important precondition for urban ecological environment research, and is a key link for digital urban construction. The current investigation method mainly comprises manual sampling investigation, and the investigation has certain subjectivity and consumes a great amount of manpower and material resources.
As an active remote sensing technology, a laser radar (light detection AND RANGING, light radar) can acquire three-dimensional information of each element in urban space in real time, especially for ground features of a street tree which are particularly critical to main information, and the light radar has an advantage incomparable with other detection means, so that the light radar is a main means for monitoring forestry information. However, the laser radar sensor has non-target directivity when acquiring point cloud information, and in a real urban space, various ground feature elements coexist, and acquired data contains all ground feature point clouds, so that the difficulty of extracting a pure street tree from a complex street scene is increased.
The current street tree extraction method comprises algorithms by means of traditional machine learning, such as algorithms of a support vector machine, K nearest neighbor, decision tree, random forest and the like, and certainly comprises deep learning, such as PoinNet ++, BP neural network and the like. However, the methods are mostly based on a single classifier, a large number of training labels are needed, the underlying principle is complex and difficult to understand, the requirements on data quality are high, visual expression is difficult to be visually realized in the processing process, excavation on morphological characteristics of the pavement tree and surrounding environment is lacked, the processing result is greatly influenced by different environments, and the extraction precision still has a large improvement space. Therefore, the method for efficiently and accurately extracting and quantifying the urban street tree information is scientific guiding significance and practical value for urban green space planning and urban green carbon sink accounting.
Disclosure of Invention
In order to solve the problems, the invention provides an urban street tree extraction method based on point cloud features and morphological features, the morphological features of the street tree and the point cloud features of the street tree in LiDAR are fully considered, the street tree is divided into a single-tree trunk and a canopy, the two parts are respectively extracted, then fusion is carried out, noise point clouds such as buildings, motor vehicles, non-motor vehicles, street lamps and indication boards in a real scene are effectively eliminated, the complete street tree point clouds are completely extracted, then single-tree segmentation is carried out on the street tree through a region growing segmentation algorithm, and inversion of street tree parameters is completed by fusion point cloud features and street tree morphological features, so that the urban street tree can be efficiently and accurately extracted.
The invention adopts the following technical scheme:
a method for extracting urban street trees based on point cloud features and morphological features comprises the following steps:
S1, preprocessing LiDAR original point cloud data;
s2, filtering the point clouds of the motor vehicle and the non-motor vehicle by adopting a breast diameter circle fitting method based on morphological characteristics of the street tree, and extracting the main point clouds;
S3, filtering building point clouds based on the point cloud field characteristics, and extracting canopy point clouds;
S4, fusing the trunk point cloud and the canopy point cloud of the street tree to obtain the whole point cloud of the street tree, and performing single tree segmentation on the street tree;
and S5, filtering the street lamp and the indication board point cloud by adopting a similar detection method to obtain the urban street tree point cloud.
Preferably, the preprocessing in the step S1 includes denoising processing and elevation normalization processing on LiDAR original point cloud data; the specific process of denoising is to calculate the average distance from each point cloud to the neighborhood point cloud, and mark the point cloud with the average distance exceeding the set threshold as noise point elimination; the specific process of the elevation normalization processing is to separate the ground points from the non-ground points through filtering, search the nearest neighboring ground points corresponding to the non-ground points, and make differences on elevation attributes, so as to obtain elevation normalization point cloud data.
Preferably, the specific process of step S2 is:
s21, slicing the height position of the breast diameter of the normalized street tree point cloud;
S22, performing density clustering segmentation on the slice point cloud data;
S23, performing circle fitting on the point cloud data after clustering segmentation to enable the square sum of delta i to be minimum, and performing iterative calculation to solve a fitting circle, wherein a fitting algorithm is shown in a formula (1):
wherein δ i represents the difference between the distance of the point (x i,yi) to the edge of the fitted circle and the square of the radius of the circle, Representing the distance from the point (x i,yi) to the edge of the fitted circle, r representing the radius of the fitted circle, x i representing the abscissa of the point (x i,yi), y i representing the ordinate of the point (x i,yi), a representing the abscissa of the center of the fitted circle, B representing the ordinate of the center of the fitted circle, a, B and c representing the parameters to be solved;
S24, judging the number of the fitting round point clouds, when the number of the round point clouds is smaller than the number of the point clouds in the circle, eliminating the fitting circle, and reserving other fitting circles until all fitting circles are judged to be over, wherein the reserved fitting circle is the position of the real single wood, the circle center coordinate is the single wood coordinate, and the diameter of the fitting circle is the chest diameter of the single wood;
S25, taking all reserved fitting circle centers as center points, taking the average breast diameter of the street tree as an outward expansion distance, establishing a cube, screening point clouds, wherein the screened point clouds are trunk point clouds, and thus, the extraction of trunk point clouds of the street tree is completed.
Preferably, the chest height position in step S21 is 1.3m±0.05m; the average breast diameter of the pavement tree in the step S25 is 0.2m.
Preferably, the specific process of step S3 is:
S31, calculating plane features and discrete features of a point cloud R neighborhood by feature values extracted through a covariance matrix, extracting building point clouds by using the plane features and combining Euclidean distance clustering, and extracting basic vegetation point clouds by using the discrete features;
the covariance matrix is shown in formula (1) and formula (3):
the planar characteristics are shown in formula (4):
The discrete features are shown in equation (5):
Wherein, the point p i (i=1, 2,3 … n) is a point in the point cloud set C, and a covariance matrix formed by p i and a point in the vicinity of the point cloud R is shown in formula (2); n represents the number of points in the point cloud R neighborhood of the point p i, p represents the geometric center of the point set in the neighborhood, j represents the number of eigenvalues and eigenvectors, j= 3,e j and λ j represent the corresponding eigenvectors and eigenvalues, respectively, wherein λ 0>λ1>λ2 and T represent the transpose in matrix calculation;
s32, taking the extracted building point cloud as a constraint condition, searching the nearest neighbor distance in the vegetation point cloud extracted in the step S31 to find a building boundary point, and extracting an accurate vegetation point cloud through Euclidean distance clustering;
S33, setting a certain height threshold, and acquiring vegetation point clouds above the height threshold, so that the vegetation point clouds not only have the point clouds of a complete canopy but also have a small number of trunk point clouds;
And S34, searching points in a certain range in the vegetation point cloud extracted in the step S32, and complementing the vegetation point cloud to obtain a complete canopy vegetation point cloud.
Preferably, the height threshold in step S33 is 1.5m.
Preferably, the specific process of step S4 is:
s41, merging the trunk point cloud and the canopy point cloud of the street tree extracted in the step S2 and the step S3 to obtain a complete point cloud of the street tree;
s42, calculating the normal vector of the whole point cloud of the pavement tree, estimating the Z-axis component of the normal vector, selecting a trunk seed point according to the threshold value of the Z-axis component, and calculating the main seed point according to a formula (6) and a formula (7):
Zn＝|e1×n| (7)
Wherein, the point p i (i=1, 2,3 … n) is a point in the point cloud set C, and a covariance matrix formed by p i and a point in the vicinity of the point cloud R is shown in formula (2); n represents the number of points within the point cloud R neighborhood of point p i, Representing the geometric center of the point set in the neighborhood, j represents the number of eigenvalues and eigenvectors, j= 3,e j and λ j represent the corresponding eigenvectors and eigenvalues, respectively, and Z n represents the component of the normal vector in the Z axis;
s43, carrying out regional growth through trunk seed points to find all trunk points;
s44, searching a trunk point adjacent point as a branch seed point, and finding all the branch points through regional growth;
s45, performing nearest neighbor search to distribute branch and leaf points to the branches, and completing the pavement tree single-tree segmentation.
Preferably, the specific process of step S5 is:
s51, carrying out multi-layer slicing on the segmented point cloud trunk part at different heights, wherein the number of layers is more than or equal to 3;
s52, performing circle fitting on each slice of each single wood layer;
s52, judging the radius and the circle center position of each fitting circle of each single wood, and eliminating the category of each fitting circle if the radius and the circle center positions are completely consistent, so as to realize the elimination of the point cloud of the street lamp and the indication board.
Preferably, step S5 is followed by inversion of the parameters of the pavement tree; the single-tree parameter inversion of the pavement tree calculates the first underrun height, the canopy volume and the tree height respectively;
the calculation process of the first branch height is as follows:
S61A, establishing a single-wood Z-axis point cloud quantity frequency histogram;
S62A, detecting first mutation points of point cloud quantity in a histogram, wherein a calculation formula of the first mutation points is shown in a formula (8):
Countn-Countn-1>threshold (8)
Wherein, count n represents the number of slice point clouds of the nth layer, count n-1 represents the number of slice point clouds of the n-1 th layer, and threshold represents the mutation threshold;
S63A, feeding back the Z value of the mutation point to a real point cloud, namely finding a trunk bifurcation position, and obtaining a height, namely Shan Mudi a branch height;
The calculation of the canopy volume adopts a method of constructing a surface by a plurality of local convex hulls to calculate the volume of the polygonal cube, and the calculation process is as follows:
S61B, screening point clouds according to the calculated Shan Mudi-branch height, and only reserving the point clouds above the branch height, namely, the single-wood canopy point clouds;
S62B, arbitrarily selecting a point Q 1 from the reserved canopy point cloud set Q, forming a new point cloud set Q 1 by the point Q 1 and a point with the distance smaller than a threshold value 2d, and obtaining the sphere centers O and O' of the spheres with the radius d passing through Q 1、q2、q3 by taking the points Q 2 and Q 3 from Q 1;
S63B, traversing the point cloud set Q 1, respectively solving a set l and a set l ' of distances from the rest points to the sphere centers o and o ', judging Q 1、q2、q3 as a contour point if the distances from one set of the set l and the set l ' are more than d, connecting to form a boundary triangle, otherwise stopping traversing, and executing the step S64B;
S64B, selecting the next group of points in the point cloud set Q 1, and repeating the steps S62B and S63B until all the points in the point cloud set Q 1 are judged to be over, outputting a patch set A, wherein the exposed triangular patches in the set form a local convex hull;
S65B, selecting the next point in the point cloud set Q, executing the steps S62B-S64B until all points in the point cloud set Q are finished, constructing all local convex hulls into crown surfaces, and calculating the volume;
The calculation process of the tree height is as follows:
obtaining Shan Mudian a difference between the maximum value and the minimum value of the cloud Z axis to obtain the tree height, wherein a calculation formula is shown in a formula (9):
Height＝max(Z)-min(Z) (9)
Wherein Height represents the tree Height, max (Z) represents the single-wood point cloud Z-axis maximum value, and min (Z) represents the single-wood point cloud Z-axis minimum value.
After the technical scheme is adopted, compared with the background technology, the invention has the following advantages: aiming at urban street scene point cloud data, a relatively complete street tree extraction flow and a single wood parameter inversion method are constructed. The invention fully utilizes the normal vector characteristics of the laser point cloud to finish the extraction of the pavement tree canopy; according to the method, the special morphological characteristics of the street trees are utilized, the positions of the street trees are rapidly determined through a circle fitting method, and the number of fitted dot clouds is used for judging and eliminating complex non-motor vehicle point clouds and motor vehicle point clouds in urban scenes; according to the method, a region growing segmentation algorithm is constructed by combining the point cloud characteristics and the street tree morphological characteristics to segment single wood, so that a good result is obtained; the method constructed by the invention is completely based on the point cloud characteristics and the target morphological characteristics, can be suitable for vehicle-mounted laser point clouds and foundation laser point clouds, can provide technical support for rapid extraction of urban street trees, and provides scientific basis for accurate estimation of urban carbon reserves.
Drawings
FIG. 1 is a technical flow chart of the present invention;
FIG. 2 is a graph of pre-processed point cloud data of the present invention;
FIG. 3 is a graph of clustering partitions of the slice point cloud density according to the present invention;
FIG. 4 is a circle fit of the present invention;
FIG. 5 is a false detection fitted circle culling diagram of the present invention;
FIG. 6 is an extraction diagram of a street tree trunk of the present invention;
FIG. 7 is a cloud of building points of the present invention;
FIG. 8 is a rough vegetation point cloud of the present invention;
FIG. 9 is a precise vegetation point cloud of the present invention;
FIG. 10 is a full vegetation canopy point cloud of the present invention;
FIG. 11 is a full roadway tree point cloud map at two angles after the trunk-canopy fusion of the present invention;
FIG. 12 is a single-log segmentation diagram of the present invention;
FIG. 13 is a frequency histogram of the number of single Z-axis point clouds and a mutation position map according to the present invention;
FIG. 14 is a schematic view of an off-branch high point cloud of the present invention;
FIG. 15 is a single wood canopy point cloud of the present invention;
Fig. 16 is a crown surface convex hull diagram of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Examples
As shown in fig. 1 to 16, a method for extracting a city street tree based on point cloud features and morphological features includes the following steps:
S1, preprocessing LiDAR original point cloud data;
The preprocessing in the step S1 comprises denoising processing and elevation normalization processing on LiDAR original point cloud data; the specific process of denoising is to calculate the average distance from each point cloud to the neighborhood point cloud, and mark the point cloud with the average distance exceeding the set threshold as noise point elimination; the specific process of the elevation normalization processing is to separate the ground points from the non-ground points through filtering, find the nearest neighboring ground points corresponding to the non-ground points, and make differences on elevation attributes to obtain elevation normalization point cloud data;
s2, filtering the point clouds of the motor vehicle and the non-motor vehicle by adopting a breast diameter circle fitting method based on morphological characteristics of the street tree, and extracting the main point clouds;
the specific process of step S2 is as follows:
s21, slicing the height position of the breast diameter of the normalized street tree point cloud;
S22, performing density clustering segmentation on the slice point cloud data;
S23, performing circle fitting on the point cloud data after clustering segmentation to enable the square sum of delta i to be minimum, and performing iterative calculation to solve a fitting circle, wherein a fitting algorithm is shown in a formula (1):
wherein δ i represents the difference between the distance of the point (x i,yi) to the edge of the fitted circle and the square of the radius of the circle, Representing the distance from the point (x i,yi) to the edge of the fitted circle, r representing the radius of the fitted circle, x i representing the abscissa of the point (x i,yi), y i representing the ordinate of the point (x i,yi), a representing the abscissa of the center of the fitted circle, B representing the ordinate of the center of the fitted circle, a, B and c representing the parameters to be solved;
S24, judging the number of the fitting round point clouds, when the number of the round point clouds is smaller than the number of the point clouds in the circle, eliminating the fitting circle, and reserving other fitting circles until all fitting circles are judged to be over, wherein the reserved fitting circle is the position of the real single wood, the circle center coordinate is the single wood coordinate, and the diameter of the fitting circle is the chest diameter of the single wood;
s25, taking all reserved fitting circle centers as central points, taking the average breast diameter of the street tree as an outward expansion distance, establishing a cube, screening point clouds, wherein the screened point clouds are trunk point clouds, and thus, the trunk point clouds of the street tree are extracted;
The height position of the chest diameter in the step S21 is 1.3m plus or minus 0.05m; the average breast diameter of the pavement tree in the step S25 is 0.2m;
S3, filtering building point clouds based on the point cloud field characteristics, and extracting canopy point clouds;
The specific process of step S3 is as follows:
S31, calculating plane features and discrete features of a point cloud R neighborhood by feature values extracted through a covariance matrix, extracting building point clouds by using the plane features and combining Euclidean distance clustering, and extracting basic vegetation point clouds by using the discrete features;
the covariance matrix is shown in formula (1) and formula (3):
the planar characteristics are shown in formula (4):
The discrete features are shown in equation (5):
Wherein, the point p i (i=1, 2,3 … n) is a point in the point cloud set C, and a covariance matrix formed by p i and a point in the vicinity of the point cloud R is shown in formula (2); n represents the number of points within the point cloud R neighborhood of point p i, Representing the geometric center of the point set in the neighborhood, j represents the number of eigenvalues and eigenvectors, j= 3,e j and λ j represent the corresponding eigenvectors and eigenvalues, respectively, where λ 0>λ1>λ2, T represent the transpose in the matrix calculation;
s32, taking the extracted building point cloud as a constraint condition, searching the nearest neighbor distance in the vegetation point cloud extracted in the step S31 to find a building boundary point, and extracting an accurate vegetation point cloud through Euclidean distance clustering;
S33, setting a certain height threshold, and acquiring vegetation point clouds above the height threshold, so that the vegetation point clouds not only have the point clouds of a complete canopy but also have a small number of trunk point clouds;
the height threshold in step S33 is 1.5m;
s34, searching points in a certain range in the vegetation point cloud extracted in the step S32, and complementing the vegetation point cloud to obtain a complete canopy vegetation point cloud;
S4, fusing the trunk point cloud and the canopy point cloud of the street tree to obtain the whole point cloud of the street tree, and performing single tree segmentation on the street tree;
the specific process of step S4 is:
s41, merging the trunk point cloud and the canopy point cloud of the street tree extracted in the step S2 and the step S3 to obtain a complete point cloud of the street tree;
s42, calculating the normal vector of the whole point cloud of the pavement tree, estimating the Z-axis component of the normal vector, selecting a trunk seed point according to the threshold value of the Z-axis component, and calculating the main seed point according to a formula (6) and a formula (7):
Zn＝|e1×n| (7)
Wherein, the point p i (i=1, 2,3 … n) is a point in the point cloud set C, and a covariance matrix formed by p i and a point in the vicinity of the point cloud R is shown in formula (2); n represents the number of points within the point cloud R neighborhood of point p i, Representing the geometric center of the point set in the neighborhood, j represents the number of eigenvalues and eigenvectors, j= 3,e j and λ j represent the corresponding eigenvectors and eigenvalues, respectively, and Z n represents the component of the normal vector in the Z axis;
s43, carrying out regional growth through trunk seed points to find all trunk points;
s44, searching a trunk point adjacent point as a branch seed point, and finding all the branch points through regional growth;
S45, performing nearest neighbor search to distribute branch and leaf points to the branches, so as to finish the single tree segmentation of the street tree;
S5, filtering the street lamp and the indication board point cloud by adopting a similar detection method to obtain an urban street tree point cloud;
the specific process of step S5 is:
s51, carrying out multi-layer slicing on the segmented point cloud trunk part at different heights, wherein the number of layers is more than or equal to 3;
s52, performing circle fitting on each slice of each single wood layer;
S52, judging the radius and the circle center position of each fitting circle of each single wood, and eliminating the category of each fitting circle if the radius and the circle center positions are completely consistent, so as to realize the elimination of the point cloud of the street lamp and the indication board;
step S5, performing pavement tree single wood parameter inversion; the single-tree parameter inversion of the pavement tree calculates the first underrun height, the canopy volume and the tree height respectively;
the calculation process of the first branch height is as follows:
S61A, establishing a single-wood Z-axis point cloud quantity frequency histogram;
S62A, detecting first mutation points of point cloud quantity in a histogram, wherein a calculation formula of the first mutation points is shown in a formula (8):
Countn-Countn-1>threshold (8)
Wherein, count n represents the number of slice point clouds of the nth layer, count n-1 represents the number of slice point clouds of the n-1 th layer, and threshold represents the mutation threshold;
S63A, feeding back the Z value of the mutation point to a real point cloud, namely finding a trunk bifurcation position, and obtaining a height, namely Shan Mudi a branch height;
The calculation of the canopy volume adopts a method of constructing a surface by a plurality of local convex hulls to calculate the volume of the polygonal cube, and the calculation process is as follows:
S61B, screening point clouds according to the calculated Shan Mudi-branch height, and only reserving the point clouds above the branch height, namely, the single-wood canopy point clouds;
S62B, arbitrarily selecting a point Q 1 from the reserved canopy point cloud set Q, forming a new point cloud set Q 1 by the point Q 1 and a point with the distance smaller than a threshold value 2d, and obtaining the sphere centers o and o' of the sphere with the radius d passing through Q 1、q2、q3 by taking the points Q 2 and Q 3 from Q 1;
S63B, traversing the point cloud set Q 1, respectively solving a set l and a set l ' of distances from the rest points to the sphere centers o and o ', judging Q 1、q2、q3 as a contour point if the distances from one set of the set l and the set l ' are more than d, connecting to form a boundary triangle, otherwise stopping traversing, and executing the step S64B;
S64B, selecting the next group of points in the point cloud set Q 1, and repeating the steps S62B and S63B until all the points in the point cloud set Q 1 are judged to be over, outputting a patch set A, wherein the exposed triangular patches in the set form a local convex hull;
S65B, selecting the next point in the point cloud set Q, executing the steps S62B-S64B until all points in the point cloud set Q are finished, constructing all local convex hulls into crown surfaces, and calculating the volume;
The calculation process of the tree height is as follows:
obtaining Shan Mudian a difference between the maximum value and the minimum value of the cloud Z axis to obtain the tree height, wherein a calculation formula is shown in a formula (9):
Height＝max(Z)-min(Z) (9)
Wherein Height represents the tree Height, max (Z) represents the single-wood point cloud Z-axis maximum value, and min (Z) represents the single-wood point cloud Z-axis minimum value.
The present invention is not limited to the above-mentioned embodiments, and any changes or substitutions that can be easily understood by those skilled in the art within the technical scope of the present invention are intended to be included in the scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.
Claims (6)
1. A method for extracting urban street trees based on point cloud features and morphological features is characterized by comprising the following steps: the method comprises the following steps:
S1, preprocessing LiDAR original point cloud data;
s2, filtering the point clouds of the motor vehicle and the non-motor vehicle by adopting a breast diameter circle fitting method based on morphological characteristics of the street tree, and extracting the main point clouds;
the specific process of step S2 is as follows:
s21, slicing the height position of the breast diameter of the normalized street tree point cloud;
S22, performing density clustering segmentation on the slice point cloud data;
S23, performing circle fitting on the point cloud data after clustering segmentation to enable the square sum of delta i to be minimum, and performing iterative calculation to solve a fitting circle, wherein a fitting algorithm is shown in a formula (1):
wherein δ i represents the difference between the distance of the point (x i,yi) to the edge of the fitted circle and the square of the radius of the circle, Representing the distance from the point (x i,yi) to the edge of the fitted circle, r representing the radius of the fitted circle, x i representing the abscissa of the point (x i,yi), y i representing the ordinate of the point (x i,yi), a representing the abscissa of the center of the fitted circle, B representing the ordinate of the center of the fitted circle, a, B and c representing the parameters to be solved;
S24, judging the number of the fitting round point clouds, when the number of the round point clouds is smaller than the number of the point clouds in the circle, eliminating the fitting circle, and reserving other fitting circles until all fitting circles are judged to be over, wherein the reserved fitting circle is the position of the real single wood, the circle center coordinate is the single wood coordinate, and the diameter of the fitting circle is the chest diameter of the single wood;
s25, taking all reserved fitting circle centers as central points, taking the average breast diameter of the street tree as an outward expansion distance, establishing a cube, screening point clouds, wherein the screened point clouds are trunk point clouds, and thus, the trunk point clouds of the street tree are extracted;
S3, filtering building point clouds based on the point cloud field characteristics, and extracting canopy point clouds;
The specific process of step S3 is as follows:
S31, calculating plane features and discrete features of a point cloud R neighborhood by feature values extracted through a covariance matrix, extracting building point clouds by using the plane features and combining Euclidean distance clustering, and extracting basic vegetation point clouds by using the discrete features;
the covariance matrix is shown in formula (1) and formula (3):
the planar characteristics are shown in formula (4):
The discrete features are shown in equation (5):
Wherein, the point p i (i=1, 2,3 … n) is a point in the point cloud set C, and a covariance matrix formed by p i and a point in the vicinity of the point cloud R is shown in formula (2); n represents the number of points within the point cloud R neighborhood of point p i, Representing the geometric center of the point set in the neighborhood, j represents the number of eigenvalues and eigenvectors, j= 3,e j and λ j represent the corresponding eigenvectors and eigenvalues, respectively, where λ 0>λ1>λ2, T represent the transpose in the matrix calculation;
s32, taking the extracted building point cloud as a constraint condition, searching the nearest neighbor distance in the vegetation point cloud extracted in the step S31 to find a building boundary point, and extracting an accurate vegetation point cloud through Euclidean distance clustering;
S33, setting a certain height threshold, and acquiring vegetation point clouds above the height threshold, so that the vegetation point clouds not only have the point clouds of a complete canopy but also have a small number of trunk point clouds;
s34, searching points in a certain range in the vegetation point cloud extracted in the step S32, and complementing the vegetation point cloud to obtain a complete canopy vegetation point cloud;
S4, fusing the trunk point cloud and the canopy point cloud of the street tree to obtain the whole point cloud of the street tree, and performing single tree segmentation on the street tree;
the specific process of step S4 is:
s41, merging the trunk point cloud and the canopy point cloud of the street tree extracted in the step S2 and the step S3 to obtain a complete point cloud of the street tree;
s42, calculating the normal vector of the whole point cloud of the pavement tree, estimating the Z-axis component of the normal vector, selecting a trunk seed point according to the threshold value of the Z-axis component, and calculating the main seed point according to a formula (6) and a formula (7):
Zn＝|e1×n| (7)
Wherein, the point p i (i=1, 2,3 … n) is a point in the point cloud set C, and a covariance matrix formed by p i and a point in the vicinity of the point cloud R is shown in formula (2); n represents the number of points within the point cloud R neighborhood of point p i, Representing the geometric center of the point set in the neighborhood, j represents the number of eigenvalues and eigenvectors, j= 3,e j and λ j represent the corresponding eigenvectors and eigenvalues, respectively, and Z n represents the component of the normal vector in the Z axis;
s43, carrying out regional growth through trunk seed points to find all trunk points;
s44, searching a trunk point adjacent point as a branch seed point, and finding all the branch points through regional growth;
S45, performing nearest neighbor search to distribute branch and leaf points to the branches, so as to finish the single tree segmentation of the street tree;
and S5, filtering the street lamp and the indication board point cloud by adopting a similar detection method to obtain the urban street tree point cloud.
2. The urban street tree extraction method based on point cloud features and morphological features as claimed in claim 1, wherein the preprocessing in step S1 comprises denoising and elevation normalization of LiDAR original point cloud data; the specific process of denoising is to calculate the average distance from each point cloud to the neighborhood point cloud, and mark the point cloud with the average distance exceeding the set threshold as noise point elimination; the specific process of the elevation normalization processing is to separate the ground points from the non-ground points through filtering, search the nearest neighboring ground points corresponding to the non-ground points, and make differences on elevation attributes, so as to obtain elevation normalization point cloud data.
3. The urban street tree extraction method based on point cloud features and morphological features as set forth in claim 1, wherein the height position of the breast diameter in step S21 is 1.3m±0.05m; the average breast diameter of the pavement tree in the step S25 is 0.2m.
4. The urban street tree extraction method based on point cloud features and morphological features according to claim 1, wherein the height threshold in step S33 is 1.5m.
5. The urban street tree extraction method based on the point cloud characteristics and the morphological characteristics as set forth in claim 1, wherein the specific process of step S5 is as follows:
s51, carrying out multi-layer slicing on the segmented point cloud trunk part at different heights, wherein the number of layers is more than or equal to 3;
s52, performing circle fitting on each slice of each single wood layer;
s52, judging the radius and the circle center position of each fitting circle of each single wood, and eliminating the category of each fitting circle if the radius and the circle center positions are completely consistent, so as to realize the elimination of the point cloud of the street lamp and the indication board.
6. The urban street tree extraction method based on the point cloud characteristics and the morphological characteristics according to claim 5, wherein step S5 is followed by street tree single-tree parameter inversion; the single-tree parameter inversion of the pavement tree calculates the first underrun height, the canopy volume and the tree height respectively;
the calculation process of the first branch height is as follows:
S61A, establishing a single-wood Z-axis point cloud quantity frequency histogram;
S62A, detecting first mutation points of point cloud quantity in a histogram, wherein a calculation formula of the first mutation points is shown in a formula (8):
Countn-Countn-1＞threshold (8)
Wherein, count n represents the number of slice point clouds of the nth layer, count n-1 represents the number of slice point clouds of the n-1 th layer, and threshold represents the mutation threshold;
S63A, feeding back the Z value of the mutation point to a real point cloud, namely finding a trunk bifurcation position, and obtaining a height, namely Shan Mudi a branch height;
The calculation of the canopy volume adopts a method of constructing a surface by a plurality of local convex hulls to calculate the volume of the polygonal cube, and the calculation process is as follows:
S61B, screening point clouds according to the calculated Shan Mudi-branch height, and only reserving the point clouds above the branch height, namely, the single-wood canopy point clouds;
S62B, arbitrarily selecting a point Q 1 from the reserved canopy point cloud set Q, forming a new point cloud set Q 1 by the point Q 1 and a point with the distance smaller than a threshold value 2d, and obtaining the sphere centers o and o' of the sphere with the radius d passing through Q 1、q2、q3 by taking the points Q 2 and Q 3 from Q 1;
S63B, traversing the point cloud set Q 1, respectively solving a set l and a set l ' of distances from the rest points to the sphere centers o and o ', judging Q 1、q2、q3 as a contour point if the distances from one set of the set l and the set l ' are more than d, connecting to form a boundary triangle, otherwise stopping traversing, and executing the step S64B;
S64B, selecting the next group of points in the point cloud set Q 1, and repeating the steps S62B and S63B until all the points in the point cloud set Q 1 are judged to be over, outputting a patch set A, wherein the exposed triangular patches in the set form a local convex hull;
S65B, selecting the next point in the point cloud set Q, executing the steps S62B-S64B until all points in the point cloud set Q are finished, constructing all local convex hulls into crown surfaces, and calculating the volume;
The calculation process of the tree height is as follows:
obtaining Shan Mudian a difference between the maximum value and the minimum value of the cloud Z axis to obtain the tree height, wherein a calculation formula is shown in a formula (9):
Height＝max(Z)-min(Z) (9)
Wherein Height represents the tree Height, max (Z) represents the single-wood point cloud Z-axis maximum value, and min (Z) represents the single-wood point cloud Z-axis minimum value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311454780.2A CN117315488B (en) | 2023-11-03 | 2023-11-03 | Urban street tree extraction method based on point cloud features and morphological features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311454780.2A CN117315488B (en) | 2023-11-03 | 2023-11-03 | Urban street tree extraction method based on point cloud features and morphological features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117315488A CN117315488A (en) | 2023-12-29 |
| CN117315488B true CN117315488B (en) | 2024-07-05 |
Family
ID=89281230
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311454780.2A Active CN117315488B (en) | 2023-11-03 | 2023-11-03 | Urban street tree extraction method based on point cloud features and morphological features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117315488B (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110910407A (en) * | 2019-11-25 | 2020-03-24 | 南京林业大学 | Pavement tree trunk extraction method based on mobile laser scanning point cloud data |
| CN113657216A (en) * | 2021-07-30 | 2021-11-16 | 西安理工大学 | Method for separating tree crown and wood point of tree in point cloud scene based on shape characteristics |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105184852B (en) * | 2015-08-04 | 2018-01-30 | 百度在线网络技术（北京）有限公司 | A kind of urban road recognition methods and device based on laser point cloud |
| CN107833244A (en) * | 2017-11-02 | 2018-03-23 | 南京市测绘勘察研究院股份有限公司 | A kind of shade tree attribute automatic identifying method based on mobile lidar data |
| CN108226894A (en) * | 2017-11-29 | 2018-06-29 | 北京数字绿土科技有限公司 | A kind of Processing Method of Point-clouds and device |
| CA3239810A1 (en) * | 2019-03-08 | 2020-09-17 | Leddartech Inc. | Method, system and computer readable medium for evaluating influence of an action performed by an external entity |
| CN112308839B (en) * | 2020-10-31 | 2022-08-02 | 云南师范大学 | Single-tree segmentation method and device for natural forest, computer equipment and storage medium |
| CN116129391B (en) * | 2023-04-18 | 2023-07-14 | 山东省国土测绘院 | Method and system for extracting pavement tree from vehicle-mounted laser point cloud |
| CN116958808A (en) * | 2023-06-14 | 2023-10-27 | 西北农林科技大学 | Forest parameter estimation method based on real-time target detection network |
-
2023
- 2023-11-03 CN CN202311454780.2A patent/CN117315488B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110910407A (en) * | 2019-11-25 | 2020-03-24 | 南京林业大学 | Pavement tree trunk extraction method based on mobile laser scanning point cloud data |
| CN113657216A (en) * | 2021-07-30 | 2021-11-16 | 西安理工大学 | Method for separating tree crown and wood point of tree in point cloud scene based on shape characteristics |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117315488A (en) | 2023-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107784283B (en) | Object-oriented unmanned aerial vehicle high-resolution image coal fire area land cover classification method | |
| CN106780524A (en) | A kind of three-dimensional point cloud road boundary extraction method | |
| CN106097311A (en) | The building three-dimensional rebuilding method of airborne laser radar data | |
| CN111325138B (en) | Road boundary real-time detection method based on point cloud local concave-convex characteristics | |
| CN106971156A (en) | Rare earth mining area remote sensing information extraction method based on object-oriented classification | |
| CN108764012A (en) | The urban road shaft recognizer of mobile lidar data based on multi-frame joint | |
| CN115063555B (en) | Vehicle-mounted LiDAR point cloud street tree extraction method for Gaussian distribution area growth | |
| CN112529044B (en) | Method for extracting and classifying railway contact network based on vehicle-mounted LiDAR | |
| Kim et al. | 3D classification of power-line scene from airborne laser scanning data using random forests | |
| CN107944383A (en) | Building roof patch division method based on three-dimensional Voronoi diagram | |
| CN110532963B (en) | Vehicle-mounted laser radar point cloud driven road marking accurate extraction method | |
| CN110348478B (en) | Method for extracting trees in outdoor point cloud scene based on shape classification and combination | |
| CN109829423A (en) | A kind of icing lake infrared imaging detection method | |
| CN109492606A (en) | Multispectral vector picture capturing method and system, three dimensional monolithic method and system | |
| CN110363299B (en) | Spatial case reasoning method for outcrop rock stratum layering | |
| CN114170149A (en) | Road geometric information extraction method based on laser point cloud | |
| CN114119863A (en) | Method for automatically extracting street tree target and forest attribute thereof based on vehicle-mounted laser radar data | |
| CN115100232A (en) | Single-tree segmentation method based on fusion of LiDAR point cloud data | |
| CN113781431A (en) | Green vision rate calculation method based on urban point cloud data | |
| CN117197677A (en) | Tropical rain forest arbor-shrub separation method based on laser radar point cloud data | |
| CN113963259A (en) | Street view ground object multi-dimensional extraction method and system based on point cloud data | |
| CN113468982A (en) | Method, device and storage medium for classifying urban functional areas | |
| CN115761382A (en) | ALS point cloud classification method based on random forest | |
| CN111860359A (en) | Point cloud classification method based on improved random forest algorithm | |
| CN117315488B (en) | Urban street tree extraction method based on point cloud features and morphological features |
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 |