CN109212553A - The method that unmanned plane LiDAR and random forest extract ginkgo biological physical characteristic - Google Patents
The method that unmanned plane LiDAR and random forest extract ginkgo biological physical characteristic Download PDFInfo
- Publication number
- CN109212553A CN109212553A CN201811074556.XA CN201811074556A CN109212553A CN 109212553 A CN109212553 A CN 109212553A CN 201811074556 A CN201811074556 A CN 201811074556A CN 109212553 A CN109212553 A CN 109212553A
- Authority
- CN
- China
- Prior art keywords
- random forest
- canopy
- unmanned plane
- ground
- lidar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000007637 random forest analysis Methods 0.000 title claims abstract description 40
- 238000000034 method Methods 0.000 title claims abstract description 39
- 235000008100 Ginkgo biloba Nutrition 0.000 title claims abstract description 11
- 239000000284 extract Substances 0.000 title claims abstract description 11
- 235000011201 Ginkgo Nutrition 0.000 title claims description 10
- 241000218628 Ginkgo Species 0.000 title claims description 10
- 238000005259 measurement Methods 0.000 claims abstract description 12
- 238000009826 distribution Methods 0.000 claims description 10
- 238000003066 decision tree Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 8
- 239000002028 Biomass Substances 0.000 claims description 6
- 238000002790 cross-validation Methods 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- 238000009825 accumulation Methods 0.000 claims description 4
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 4
- 238000013210 evaluation model Methods 0.000 claims description 4
- 239000002023 wood Substances 0.000 claims description 3
- 239000004744 fabric Substances 0.000 claims description 2
- 230000007246 mechanism Effects 0.000 abstract description 2
- 244000194101 Ginkgo biloba Species 0.000 abstract 1
- 238000005516 engineering process Methods 0.000 description 6
- 230000008901 benefit Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000011835 investigation Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 235000018783 Dacrycarpus dacrydioides Nutrition 0.000 description 1
- 244000288671 Dacrycarpus dacrydioides Species 0.000 description 1
- 235000005016 Pinus halepensis Nutrition 0.000 description 1
- 241001236235 Pinus halepensis Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 229910052799 carbon Inorganic materials 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000010924 continuous production Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000000116 mitigating effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000012856 packing Methods 0.000 description 1
- 231100000289 photo-effect Toxicity 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S17/00—Systems using the reflection or reradiation of electromagnetic waves other than radio waves, e.g. lidar systems
- G01S17/88—Lidar systems specially adapted for specific applications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/4802—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
- G01S7/483—Details of pulse systems
- G01S7/486—Receivers
- G01S7/487—Extracting wanted echo signals, e.g. pulse detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Networks & Wireless Communication (AREA)
- General Physics & Mathematics (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Electromagnetism (AREA)
- Image Processing (AREA)
Abstract
本发明公开一种无人机LiDAR和随机森林提取银杏生物物理特性的方法，主要通过无人机搭载的LiDAR传感器采集数据，通过数据处理后，分别提取三组冠层结构特征变量，并与实测的林分生物物理特性一起构建随机森林模型预测提取生物物理特性。本方法采用随机森林模型方法去反演林分生物物理特性；相比较传统的参数方法变量之间需要满足一系列的先验假设，而通过随机森林模型可以将所有提取的特征变量输入模型进行运算，并且随机森林模型可以评估每个特征变量的重要性，从而高质量地提取了人工林林分生物物理特性。同时，通过上述数据获得的随机森林不仅利于特征变量的机理解释，也易于进行方法移植，即在天然林和次生林中也可以进行应用。
The invention discloses a method for extracting biophysical characteristics of ginkgo biloba by UAV LiDAR and random forest. The method mainly collects data through LiDAR sensors carried by the UAV, and after data processing, extracts three groups of canopy structure characteristic variables respectively, and compares them with actual measurements. The biophysical properties of the stand were constructed together with a random forest model to predict the extracted biophysical properties. This method uses the random forest model method to invert the biophysical characteristics of the stand; compared with the traditional parameter method, a series of a priori assumptions need to be satisfied between the variables, and through the random forest model, all the extracted characteristic variables can be input into the model for operation. , and the random forest model can evaluate the importance of each feature variable, thereby extracting the plantation stand biophysical properties with high quality. At the same time, the random forest obtained from the above data is not only conducive to the mechanism interpretation of characteristic variables, but also easy to transfer the method, that is, it can also be applied in natural forests and secondary forests.
Description
Technical field
The present invention relates to the fields such as forest resource monitoring and environmental factor investigation, specifically a kind of unmanned plane that passes through is generated
The method for extracting biophysical properties Random Forest model.
Background technique
The extraction of accurate artificial forest biophysical properties, for forest inventory investigation, Forest Productivity estimation and biology
Study on Diversity is of great significance.These information, which have, simultaneously can be used for exploring the forest space regularity of distribution, and can to forest
Continuous production operation, raising Forestry Carbon Mitigation is horizontal and local area ecological balance is maintained to provide data support and practical advice.
Conventional artificial forest biophysical properties extraction depends on ground investigation method, and time-consuming effort only can
The data on " point " are obtained, precision is not often high, and is not easy to be promoted over a large area.Laser radar (LiDAR) technology
It is that a kind of active remote sensing technology of its return signal to body surface and is analyzed by transmitting laser pulse.It can be obtained by LiDAR
The information such as high-precision Forest Canopy structure and hayashishita landform.
In recent years, the research extracted based on LiDAR data biophysical properties are as follows: Montealegre etc. 2016 exists
" the Use of low point density ALS data to estimate stand- delivered on " Forestry " volume 89
Level structural variables in Mediterranean Aleppo pine forest ", the research is by having
Man-machine platform carries laser radar and obtains low-density point cloud data, and extraction height characteristic variable is assessing these variable importances
On the basis of combined ground measured data be extracted the biophysical characteristics of Mediterranean Region kahikatea.
Bottalico etc. 2017 in " International Journal of Applied Earth
Observations and Geoinformation " " the Modeling Mediterranean forest that delivers on volume 57
Structure using airborne laser scanning data ", the research use airborne laser radar data, pass through
LiDAR altitude feature variable is extracted, in conjunction with sample measured data, is extracted Italian needle artificial forest with linear model approach
Multiple biophysical properties.However, above method is all based on the biophysical properties of the LiDAR data of someone's machine platform
It extracting, point of use cloud packing density is all very low, and one group of characteristic variable is used only to extract biophysical properties, meanwhile,
More have no that the method for imparametrization is used for the extraction of plantation stand biophysical properties.
Summary of the invention
Goal of the invention: in view of the deficiency of the prior art, a kind of generated by unmanned plane is provided and extracts biological object
The method for managing characteristic Random Forest model, can effectively improve the extraction accuracy of biophysical properties on woodland.
Technical solution: in order to achieve the above-mentioned object of the invention, The technical solution adopted by the invention is as follows: one kind passes through unmanned plane
Generate the method for extracting biophysical properties Random Forest model, comprising the following steps:
1) LiDAR data acquisition is carried out by the LiDAR sensor that multi-rotor unmanned aerial vehicle is carried；And it is arranged simultaneously on ground
Sample, record sample interior ground trees strain tree, while measuring the diameter of a cross-section of a tree trunk 1.3 meters above the ground of every wood and setting high；And actual measurement chest is combined by Binary formula
Diameter and the high estimation accumulation of tree；The actual measurement diameter of a cross-section of a tree trunk 1.3 meters above the ground and tree height is combined to calculate ground biomass by different rate growth formula；
2) noise point for removing unmanned plane LiDAR original point cloud data removes non-ground points based on IDW filtering algorithm, so
Afterwards by calculating the average value of laser point height in each rubber member, digital elevation model is generated；Pass through the digital elevation of the generation
A cloud is normalized in model, the unmanned plane LiDAR point cloud data after being normalized；
3) three groups of canopy structure characteristic variables are extracted respectively；This three groups of characteristic variables include percentile altitude feature variable,
Each layer coverage characteristic variable and canopy volume and profile features variable；
4) the standing forest biophysical properties and unmanned plane LiDAR point cloud feature construction based on the actual measurement of existing ground are gloomy at random
Woods model；M variable is randomly generated for the binary tree on node in mtry value specified first, mtry value is for modeling
Point cloud feature quantity；Secondly, randomly selecting k sample set using bootstrap method with putting back to, k decision tree is formed, this
K decision tree forms Random Forest model, finally, predicts biophysical properties to the Random Forest model.
Wherein, the spatial resolution of the step 2) digital elevation model is 0.5m.
Wherein, the percentile altitude feature variable in the step 3) includes: that the canopy at H25, H50, H75 and H95 is high
Degree distribution percentile；Canopy point cloud is distributed the coverage CCmean of average height or more and the variation lines of canopy point cloud distribution
Number Hcv.
Wherein, each layer coverage characteristic variable described in the step 3) includes: cloud quantity in each percentage height
30th, 50th, 70th, 90th or more point account for the percentage of all the points cloud.
Wherein, the canopy volume in the step 3) and profile features variable refer to Weibull function to canopy height point
Cloth section is fitted to obtain 2 profile features amount Weibull α and Weibull β and each structured sort volume accounting of canopy,
Each structured sort volume accounting of the canopy refers to this four canopy structure classifications of open tier, photic zone, low photosphere and confining bed
Volume percentage.
This method uses leave one cross validation；Coefficient of determination R is set2, root-mean-square error RMSE, opposite root-mean-square error
The effect and estimation precision of rRMSE evaluation model fitting:
X in formulaiFor certain standing forest biophysical properties measured value；Average value is surveyed for certain standing forest biophysical properties；
For the model estimated value of certain standing forest biophysical properties；N for sample ground quantity；I is for some sample.
The utility model has the advantages that compared with prior art, the present invention having the advantage that
(1) present invention is all based on someone's machine platform compared to previous LiDAR, and the LiDAR point cloud density of acquisition is all very
It is low, and the characteristic variable type extracted is less, and this method can obtain highdensity LiDAR point cloud based on unmanned aerial vehicle platform, it is right
The filtering of unmanned plane LiDAR discrete point cloud data, interpolation generate digital elevation model, point cloud data normalized；Based on normalizing
Change unmanned plane LiDAR discrete point cloud and extracts multiple groups artificial forest canopy characteristic variable；Based on the multiple groups unmanned plane LiDAR all extracted
Artificial forest canopy characteristic variable and ground measured data, construct Random Forest model, and debugging model parameter assesses characteristic variable
Importance, final each standing forest biophysical properties of inverting；The inversion accuracy of artificial forest biophysical properties is helped to improve, and is had
Effect inhibits biophysical properties inverting " saturation " problem of forest cover degree height, the high standing forest of biomass；
(2) this method removes inverting standing forest biophysical properties using Random Forest model method；Compared to more traditional parameter
It needs to meet a series of a priori assumption between method variable, and can be become the feature of all extractions by Random Forest model
It measures input model and carries out operation, and Random Forest model can assess the importance of each characteristic variable, thus in high quality
It is extracted plantation stand biophysical properties.Meanwhile characteristic variable is not only conducive to by the random forest that above-mentioned data obtain
Mechanism explain, be also easy to carry out method transplanting, i.e., also may be employed in wildwood and scondary forest.
Detailed description of the invention
Fig. 1 is different densities ginkgo artificial forest true color photo effect picture；
Fig. 2 is different densities ginkgo artificial forest unmanned plane laser radar point cloud atlas；
Fig. 3 is the characteristic variable important relationship figure based on Random Forest model Different forest stands biophysical properties.
Specific embodiment
In the following with reference to the drawings and specific embodiments, the present invention is furture elucidated, and the present embodiment is with technical solution of the present invention
Premised under implemented, it should be understood that these examples are only for illustrating the present invention and are not intended to limit the scope of the present invention.
1) LiDAR data acquisition is carried out by the LiDAR sensor that multi-rotor unmanned aerial vehicle is carried.In ground setting sample,
Sample interior ground trees strain tree is recorded, while measuring the diameter of a cross-section of a tree trunk 1.3 meters above the ground of every wood and setting high.Accumulation combines actual measurement chest according to Binary formula
Diameter and tree height are estimated that ground biomass is calculated by the different rate growth formula combination diameter of a cross-section of a tree trunk 1.3 meters above the ground and tree height；
2) noise point for removing unmanned plane LiDAR original point cloud data first, is gone based on IDW filtering algorithm unless ground
Point generates digital elevation model (DEM), and space point then by calculating the average value of laser point height in each rubber member
Resolution is 0.5m.A cloud is normalized by the digital elevation model of generation, the unmanned plane after being normalized
LiDAR point cloud data；
3) three groups of canopy structure characteristic variables are extracted, i.e. percentile altitude feature variable, each layer coverage feature
Variable and canopy volume and profile features variable；
Wherein, percentile altitude feature variable includes: canopy height distribution percentile (H25, H50, H75, H95), hat
Coverage (CCmean) more than layer point cloud distribution average height, the coefficient of variation (Hcv) of canopy point cloud distribution；
Wherein, each layer coverage characteristic variable include: cloud quantity each percentage height (30th, 50th, 70th,
90th, i.e. D3, D5, D7, D9) more than point account for the percentage of all the points cloud；
Wherein, canopy volume includes: that Weibull function is fitted canopy height profile with profile features variable
Obtain 2 profile features amounts α, β (i.e. Weibull α and Weibull β)；Each structured sort volume accounting of canopy, including open tier,
Photic zone, four canopy structure classifications of low photosphere and confining bed, the volume percentage of each canopy structure classification is (i.e.
OpenGap, Oligophotic, Euphotic, ClosedGap).
4) based on existing ground actual measurement standing forest biophysical properties and unmanned plane LiDAR point cloud feature construction for extracting
Predict the Random Forest model of biophysical properties；Building process is as follows: m variable is randomly generated in mtry value specified first
Binary tree on node, mtry value are the point cloud feature quantity for modeling；Secondly, being put using bootstrap method
It randomly selects k sample set with returning, forms k decision tree；Finally, according to the random forest that k decision tree finally forms, and it is right
Random Forest model carries out prediction and extracts biophysical properties.
This method uses leave one cross validation, the coefficient of determination (R2), root-mean-square error (RMSE), opposite root-mean-square error
(rRMSE) effect of evaluation model fitting and estimation precision:
X in formulaiFor certain standing forest biophysical properties measured value；Average value is surveyed for certain standing forest biophysical properties；
For the model estimated value of certain standing forest biophysical properties；N for sample ground quantity；I is for some sample.
Embodiment 1
Test block in the present embodiment is located at Jiangsu Province northern territory Pizhou City town Tie Fu, and geographical location is 34 ° of north latitude
33 ' 49 " -34 ° 34 ' 23 ", 118 ° 05 ' 1 of east longitude " -118 ° 06 ' 06 ".
The trial zone belongs to subhumid and temperate zones monsoon climate, and annual rainfall is about 903mm, and maximum rainfall concentrates on 7,8
Month plum rain season, year-round average temperature is about 13.9 DEG C, frost-free period 211 days, In The Soils be smolmitza earth, in acidity.
LiDAR data acquisition is carried out by the LiDAR sensor that multi-rotor unmanned aerial vehicle is carried first.It is provided according to history forest
The satellite remote-sensing image data that source survey data and early period obtain have chosen 5 pieces of 1 × 1km in ginkgo artificial forest Core distribution area
Square big plot, then in 5 sample ground according to the method for typical sampling the round sample that 9 pieces of radiuses are 15m is set,
The center on sample ground is positioned by Trimble GeoXH6000 Handhelds handhold GPS (in conjunction with JSCROS GPS wide area differential GPS system
System) it is positioned, precision is better than 0.5m.
During sample-plot survey, the Dan Mu of 5cm is greater than for the diameter of a cross-section of a tree trunk 1.3 meters above the ground, it is high to measure its diameter of a cross-section of a tree trunk 1.3 meters above the ground (diameter of a cross-section of a tree trunk 1.3 meters above the ground ruler measurement), tree one by one
(i.e. projector distance in two principal directions, uses tape measure with clear bole height (Vertex V ultrasonic wave and laser altimeter) and hat width
Measurement).By the individual tree information of actual measurement, summarize to obtain sample ground level stand characteristics, i.e. mean DBH increment, basal area, Lorey ' s
Mean stand height (using the basal area of each tree as weight, mean stand height that weighted sum obtains), the density of crop, accumulation
And ground biomass；Above-mentioned data see the table below 1.
Survey to 1 sample of table standing forest biophysical properties information summary sheet
When data prediction, the noise point of unmanned plane LiDAR original point cloud data is removed first, is based on IDW filtering algorithm
Non-ground points are removed, then by calculating the average value of laser point height in each pixel, are generated digital elevation model (DEM)
(spatial resolution 0.5m).And a cloud is normalized in the digital elevation model by generating, after obtaining normalization
Unmanned plane LiDAR point cloud data；The unmanned plane LiDAR point cloud effect data is shown in Fig. 1.
Canopy structure characteristic variable is extracted, and three groups of characteristic variables, i.e. percentile altitude feature variable, each layer coverage are extracted
Characteristic variable and canopy volume and profile features variable.
Percentile altitude feature variable includes: canopy height distribution percentile (H25, H50, H75, H95), canopy point cloud
It is distributed the coverage (CCmean) of average height or more, the coefficient of variation (Hcv) of canopy point cloud distribution；
Each layer coverage characteristic variable include: cloud quantity each percentage height (30th, 50th, 70th, 90th, i.e.,
D3, D5, D7, D9) more than point account for the percentage of all the points cloud；
Canopy volume and profile features variable include: that Weibull function is fitted to obtain 2 to canopy height profile
A profile features amount α, β (i.e. Weibull α and Weibull β)；Each structured sort volume accounting of canopy, including open tier, light transmission
Layer, four canopy structure classifications of low photosphere and confining bed, each canopy structure classification volume percentage (i.e. OpenGap,
Oligophotic, Euphotic, ClosedGap).
Standing forest biophysical properties and unmanned plane LiDAR point cloud feature construction random forest mould are surveyed based on existing ground
Type.M variable is randomly generated for the binary tree on node in mtry value specified first, mtry value is the point cloud for modeling
Feature quantity is 16；Secondly, randomly selecting k sample set using bootstrap method with putting back to, k decision tree is formed,
In, k 1000；It is predicted finally, treating sample according to the random forest that k decision tree forms.
This method uses leave one cross validation, the coefficient of determination (R2), root-mean-square error (RMSE), opposite root-mean-square error
(rRMSE) effect of evaluation model fitting and estimation precision:
X in formulaiFor certain standing forest biophysical properties measured value；Average value is surveyed for certain standing forest biophysical properties；For
The model estimated value of certain standing forest biophysical properties；N for sample ground quantity；I is for some sample.The Random Forest model middle forest
Decomposing biological physical characteristic estimation cross validation results see the table below 2,
Table 2 predicts cross validation results summary sheet based on each standing forest biophysical properties of Random Forest model
And three typical samples unmanned plane LiDAR point cloud atlas see Fig. 1 and Fig. 2, the difference based on Random Forest model prediction
The characteristic variable importance of standing forest biophysical properties is shown in Fig. 3.
Laser radar (LiDAR) technology is to body surface and to analyze one kind of its return signal by transmitting laser pulse
Active remote sensing technology.The information such as high-precision Forest Canopy structure and hayashishita landform can be obtained by LiDAR.Based on unmanned aerial vehicle platform
LiDAR with more flight cost it is low, data acquisition convenient and efficient can obtain the advantages such as high density point cloud, it will help improve
The inversion accuracy of artificial forest biophysical properties, and effectively inhibit the biophysics of forest cover degree height, the high standing forest of biomass special
Property inverting " saturation " problem.
Previous LiDAR is all based on someone's machine platform, and the LiDAR point cloud density of acquisition is all very low, and the feature extracted
Types of variables is less, and this method can obtain highdensity LiDAR point cloud data based on unmanned aerial vehicle platform, and in depth mention comprehensively
The unmanned plane LiDAR point cloud feature of the artificial storey of multiple groups is taken, finally, going inverting standing forest raw using Random Forest model method
Object physical characteristic enhances the ability and precision of inverting.
By the data comparison of table 2 and table 1, verification result shows through the invention to Plain ginkgo artificial forest biology object
Reason characteristic extracts compared with using other close remote sensing techniques to carry out the inverting of standing forest biophysical properties, opposite root mean square
Error reduces 5% or more, and inversion accuracy significantly improves.
Random forest in the present invention is a kind of prediction model, and the model is more superior than traditional Gradual regression analysis model,
Being embodied in it is a kind of machine learning method, belongs to non-parametric model, and usual precision of prediction is higher；And it is pre- by prediction model
The physical characteristic for surveying woods biology then belongs to a kind of existing technology, but the spy that the application is extracted by unmanned plane LiDAR data
It levies variable and ground measured data constructs random forest, and predict artificial forest biophysical properties based on the random forest of building
Will have very big advantage compared with traditional technology.
Specific embodiment is a preferred embodiment of the present invention, is not for limiting implementation and power of the invention
Sharp claimed range, the equivalence changes and modification that content described in all ranges of patent protection according to the present invention is made should all
It is included in the scope of the patent application of the present invention.
Claims (6)
1. a kind of method that unmanned plane LiDAR and random forest extract ginkgo biological physical characteristic, it is characterised in that: including following
Step:
1) LiDAR data acquisition is carried out by the LiDAR sensor that multi-rotor unmanned aerial vehicle is carried；And sample is set on ground simultaneously
Ground, record sample interior ground trees strain tree, while measuring the diameter of a cross-section of a tree trunk 1.3 meters above the ground of every wood and setting high；And the actual measurement diameter of a cross-section of a tree trunk 1.3 meters above the ground is combined by Binary formula
With the high estimation accumulation of tree；The actual measurement diameter of a cross-section of a tree trunk 1.3 meters above the ground and tree height is combined to calculate ground biomass by different rate growth formula；
2) noise point for removing unmanned plane LiDAR original point cloud data removes non-ground points based on IDW filtering algorithm, then leads to
The average value for calculating laser point height in each rubber member is crossed, digital elevation model is generated；Pass through the digital elevation model of the generation
A cloud is normalized, the unmanned plane LiDAR point cloud data after being normalized；
3) three groups of canopy structure characteristic variables are extracted respectively；This three groups of characteristic variables include percentile altitude feature variable, each layer
Coverage characteristic variable and canopy volume and profile features variable；
4) standing forest biophysical properties and unmanned plane LiDAR point cloud feature construction random forest mould based on the actual measurement of existing ground
Type；M variable is randomly generated for the binary tree on node in mtry value specified first, mtry value is the point cloud for modeling
Feature quantity；Secondly, randomly select k sample set using bootstrap method with putting back to, k decision tree is formed, this k
Decision tree forms Random Forest model, finally, based on the raw biophysical properties of Random Forest model prediction artificial forest.
2. the method that a kind of unmanned plane LiDAR according to claim 1 and random forest extract ginkgo biological physical characteristic,
It is characterized by: the spatial resolution of the step 2) digital elevation model is 0.5m.
3. the method that a kind of unmanned plane LiDAR according to claim 1 and random forest extract ginkgo biological physical characteristic,
It is characterized by: the percentile altitude feature variable in the step 3) includes: the canopy height at H25, H50, H75 and H95
It is distributed percentile；Canopy point cloud is distributed the coverage CCmean of average height or more and the coefficient of variation of canopy point cloud distribution
Hcv。
4. the method that a kind of unmanned plane LiDAR according to claim 1 and random forest extract ginkgo biological physical characteristic,
It is characterized by: each layer coverage characteristic variable described in the step 3) includes: cloud quantity in each percentage height
30th, 50th, 70th, 90th and D3, D5, D7, D9 or more point account for the percentage of all the points cloud.
5. a kind of side for generating extraction biophysical properties Random Forest model by unmanned plane according to claim 1
Method, it is characterised in that: canopy volume and profile features variable in the step 3) refer to Weibull function to canopy height point
Cloth section is fitted to obtain 2 profile features amount Weibull α and Weibull β and each structured sort volume accounting of canopy,
Each structured sort volume accounting of the canopy refers to this four canopy structure classifications of open tier, photic zone, low photosphere and confining bed
Volume percentage.
6. a kind of side for generating extraction biophysical properties Random Forest model by unmanned plane according to claim 1
Method, it is characterised in that: this method uses leave one cross validation；Coefficient of determination R is set2, root-mean-square error RMSE, it is relatively square
The effect and estimation precision of root error rRMSE evaluation model fitting:
X in formulaiFor certain standing forest biophysical properties measured value；Average value is surveyed for certain standing forest biophysical properties；For certain woods
The model estimated value of decomposing biological physical characteristic；N for sample ground quantity；I is for some sample.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811074556.XA CN109212553A (en) | 2018-09-14 | 2018-09-14 | The method that unmanned plane LiDAR and random forest extract ginkgo biological physical characteristic |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811074556.XA CN109212553A (en) | 2018-09-14 | 2018-09-14 | The method that unmanned plane LiDAR and random forest extract ginkgo biological physical characteristic |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109212553A true CN109212553A (en) | 2019-01-15 |
Family
ID=64984441
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811074556.XA Pending CN109212553A (en) | 2018-09-14 | 2018-09-14 | The method that unmanned plane LiDAR and random forest extract ginkgo biological physical characteristic |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109212553A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110133643A (en) * | 2019-05-22 | 2019-08-16 | 北京林业大学 | Plant root detection method and device |
| CN111489388A (en) * | 2020-04-20 | 2020-08-04 | 黑龙江工程学院 | Biomass inversion method based on effective crown information |
| CN113156394A (en) * | 2021-03-31 | 2021-07-23 | 国家林业和草原局华东调查规划设计院 | Forest resource monitoring method and device based on laser radar and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020489A (en) * | 2013-01-04 | 2013-04-03 | 吉林大学 | Novel method for forecasting siRNA interference efficiency based on ARM (Advanced RISC Machines) microprocessor |
| CN104849722A (en) * | 2015-05-08 | 2015-08-19 | 南京林业大学 | Light detection and ranging (LiDAR) waveform comprehensive feature-based individual tree identification method |
| CN104867180A (en) * | 2015-05-28 | 2015-08-26 | 南京林业大学 | UAV and LiDAR integrated forest stand characteristic inversion method |
-
2018
- 2018-09-14 CN CN201811074556.XA patent/CN109212553A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020489A (en) * | 2013-01-04 | 2013-04-03 | 吉林大学 | Novel method for forecasting siRNA interference efficiency based on ARM (Advanced RISC Machines) microprocessor |
| CN104849722A (en) * | 2015-05-08 | 2015-08-19 | 南京林业大学 | Light detection and ranging (LiDAR) waveform comprehensive feature-based individual tree identification method |
| CN104867180A (en) * | 2015-05-28 | 2015-08-26 | 南京林业大学 | UAV and LiDAR integrated forest stand characteristic inversion method |
Non-Patent Citations (4)
| Title |
|---|
| XIN SHEN等: "Tree-Species Classification in Subtropical Forests Using Airborne Hyperspectral and LiDAR Data", 《REMOTE SENSING》 * |
| ZHENGNAN ZHANG等: "Estimating Forest Structural Parameters Using Canopy Metrics Derived from Airborne LiDAR Data in Subtropical Forests", 《REMOTE SENSING》 * |
| 刘坤等: "银杏生物量分配格局及异速生长模型", 《北京林业大学学报》 * |
| 翟明普: "《现代森林培育理论与技术》", 30 June 2011 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110133643A (en) * | 2019-05-22 | 2019-08-16 | 北京林业大学 | Plant root detection method and device |
| CN110133643B (en) * | 2019-05-22 | 2021-08-20 | 北京林业大学 | Plant root detection method and device |
| CN111489388A (en) * | 2020-04-20 | 2020-08-04 | 黑龙江工程学院 | Biomass inversion method based on effective crown information |
| CN113156394A (en) * | 2021-03-31 | 2021-07-23 | 国家林业和草原局华东调查规划设计院 | Forest resource monitoring method and device based on laser radar and storage medium |
| CN113156394B (en) * | 2021-03-31 | 2024-04-12 | 国家林业和草原局华东调查规划设计院 | Forest resource monitoring method and device based on laser radar and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Hermosilla et al. | Estimation of forest structure and canopy fuel parameters from small-footprint full-waveform LiDAR data | |
| Hopkinson et al. | The uncertainty in conifer plantation growth prediction from multi-temporal lidar datasets | |
| Næsset et al. | Laser scanning of forest resources: the Nordic experience | |
| Moeslund et al. | Light detection and ranging explains diversity of plants, fungi, lichens, and bryophytes across multiple habitats and large geographic extent | |
| Parker et al. | A portable LIDAR system for rapid determination of forest canopy structure | |
| CN103529189B (en) | A kind of soil organism space distribution Forecasting Methodology based on quantitative and qualitative analysis auxiliary variable | |
| Laamrani et al. | Effects of topography and thickness of organic layer on productivity of black spruce boreal forests of the Canadian Clay Belt region | |
| Mandl et al. | Spaceborne LiDAR for characterizing forest structure across scales in the European Alps | |
| CN109061601A (en) | A method of based on unmanned plane laser radar inverting artificial forest forest structural variable | |
| CN104020274B (en) | The method of a kind of forest land site quality remote sensing and quantizing estimation | |
| CN108921885A (en) | A kind of method of comprehensive three classes data source joint inversion forest ground biomass | |
| CN109212505A (en) | A kind of forest stand characteristics inversion method based on the multispectral high degree of overlapping image of unmanned plane | |
| CN116452023B (en) | Kilometer-level carbon reserve evaluation method based on VOD data of low-frequency microwave radar | |
| CN108896021B (en) | A method for extracting structural parameters of plantation forest stand based on aerial photogrammetry point cloud | |
| CN109142679A (en) | The space predicting method of forest soil nutrient based on artificial neural network Kriging regression | |
| Cao et al. | Estimation of forest structural parameters using UAV-LiDAR data and a process-based model in ginkgo planted forests | |
| CN108981616B (en) | A method for inversion of effective leaf area index of plantation by UAV lidar | |
| Vastaranta et al. | Aboveground forest biomass derived using multiple dates of WorldView-2 stereo-imagery: quantifying the improvement in estimation accuracy | |
| CN109212553A (en) | The method that unmanned plane LiDAR and random forest extract ginkgo biological physical characteristic | |
| Hernández‐Stefanoni et al. | Estimating species richness and biomass of tropical dry forests using LIDAR during leaf‐on and leaf‐off canopy conditions | |
| CN109146951A (en) | A method of ginkgo artificial forest leaf area index is estimated based on unmanned plane laser radar porosity model | |
| Teixeira et al. | Spatial variability of soil CO2 emission in a sugarcane area characterized by secondary information | |
| Rosette et al. | Forestry applications for satellite lidar remote sensing | |
| Šrollerů et al. | Evaluating the applicability of high-density UAV LiDAR data for monitoring tundra grassland vegetation | |
| Ediriweera et al. | LiDAR remote sensing of structural properties of subtropical rainforest and eucalypt forest in complex terrain in North-eastern Australia |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190115 |
|
| RJ01 | Rejection of invention patent application after publication |