CN117576583A - Unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method - Google Patents
Unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method Download PDFInfo
- Publication number
- CN117576583A CN117576583A CN202311220937.5A CN202311220937A CN117576583A CN 117576583 A CN117576583 A CN 117576583A CN 202311220937 A CN202311220937 A CN 202311220937A CN 117576583 A CN117576583 A CN 117576583A
- Authority
- CN
- China
- Prior art keywords
- biomass
- unmanned aerial
- aerial vehicle
- winter wheat
- image
- 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
- 239000002028 Biomass Substances 0.000 title claims abstract description 115
- 241000209140 Triticum Species 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 title claims abstract description 53
- 235000021307 Triticum Nutrition 0.000 title claims abstract description 49
- 241000196324 Embryophyta Species 0.000 claims abstract description 44
- 238000011160 research Methods 0.000 claims abstract description 21
- 238000013528 artificial neural network Methods 0.000 claims abstract description 19
- 238000007781 pre-processing Methods 0.000 claims abstract description 11
- 238000004364 calculation method Methods 0.000 claims abstract description 7
- 238000006243 chemical reaction Methods 0.000 claims abstract description 7
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 claims description 47
- 238000011282 treatment Methods 0.000 claims description 30
- 238000012360 testing method Methods 0.000 claims description 28
- 229910052757 nitrogen Inorganic materials 0.000 claims description 24
- 210000004027 cell Anatomy 0.000 claims description 11
- 238000012937 correction Methods 0.000 claims description 11
- 230000005855 radiation Effects 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 5
- 239000003337 fertilizer Substances 0.000 claims description 5
- 238000013461 design Methods 0.000 claims description 4
- 238000002310 reflectometry Methods 0.000 claims description 4
- 230000017260 vegetative to reproductive phase transition of meristem Effects 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 210000002569 neuron Anatomy 0.000 claims description 3
- 238000005192 partition Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000005259 measurement Methods 0.000 claims description 2
- 238000009331 sowing Methods 0.000 claims description 2
- 238000004458 analytical method Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 3
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 229910052799 carbon Inorganic materials 0.000 description 2
- 230000000875 corresponding effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 240000006394 Sorghum bicolor Species 0.000 description 1
- 235000011684 Sorghum saccharatum Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 235000013339 cereals Nutrition 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013508 migration Methods 0.000 description 1
- 230000005012 migration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- QJGQUHMNIGDVPM-UHFFFAOYSA-N nitrogen group Chemical group [N] QJGQUHMNIGDVPM-UHFFFAOYSA-N 0.000 description 1
- 238000012805 post-processing Methods 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
- G06V20/17—Terrestrial scenes taken from planes or by drones
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4038—Image mosaicing, e.g. composing plane images from plane sub-images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- 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
- G06V10/766—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using regression, e.g. by projecting features on hyperplanes
-
- 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
- G06V10/778—Active pattern-learning, e.g. online learning of image or video features
- G06V10/7784—Active pattern-learning, e.g. online learning of image or video features based on feedback from supervisors
- G06V10/7792—Active pattern-learning, e.g. online learning of image or video features based on feedback from supervisors the supervisor being an automated module, e.g. "intelligent oracle"
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/194—Terrestrial scenes using hyperspectral data, i.e. more or other wavelengths than RGB
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
- G06V20/647—Three-dimensional objects by matching two-dimensional images to three-dimensional objects
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Remote Sensing (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method for rapidly and accurately estimating winter wheat biomass based on an unmanned aerial vehicle, which comprises the following steps: acquiring an unmanned aerial vehicle image of a research area by using an unmanned aerial vehicle multispectral sensor, and preprocessing the unmanned aerial vehicle image to obtain data of an orthographic image and a digital surface model; extracting the preprocessed orthographic image and the winter wheat plant height in the digital surface model based on grid calculation; establishing a biomass estimation model by adopting a BP neural network regression method according to the plant height of winter wheat; and (5) evaluating the accuracy of the biomass estimation model before and after improvement by using the model decision coefficients. The method has the characteristics of simplicity, rapidness and high precision, and can realize rapid and high-precision biomass estimation by fully utilizing the relation between the biomass of unit plant height and the plant height through analysis and conversion of image data, the precision is improved by 51.72 percent compared with a method for directly utilizing the plant height to carry out biomass estimation, and the low-cost and high-precision estimation of the biomass of winter wheat is realized.
Description
Technical Field
The invention relates to the technical field of agricultural engineering, in particular to a method for rapidly and accurately estimating winter wheat biomass based on an unmanned aerial vehicle.
Background
Biomass (AGB) has an important impact on light energy utilization, dry matter production, and accurate estimation has been one of the key problems in terrestrial ecological research because: on one hand, the crop biomass plays an important role in global carbon circulation, and accurate estimation of the crop biomass is helpful for grasping the feedback effect between climate change and farmland ecosystem; biomass, on the other hand, is closely related to the formation of the final yield per crop and yield. Therefore, the crop biomass estimation is timely and accurately carried out, the method has extremely important guiding significance for scientifically managing and reasonably utilizing farmlands and protecting and enhancing the carbon sink function of the farmlands, and provides important data reference for grain safety problems.
In recent years, the research of estimating crop biomass by using remote sensing technology at home and abroad is increasing. At present, unmanned aerial vehicle remote sensing is used for biomass estimation, spectral parameters are extracted by mainly utilizing RGB and multispectral digital orthographic images, vegetation indexes are calculated and the like for model construction, and if Yang Jun and the like (2019), ma and the like (2019) and Lu and the like (2019) are used for estimating wheat biomass by utilizing unmanned aerial vehicle digital images, the combination of image color indexes and texture features can improve the estimation precision of the wheat biomass, and the correlation coefficient of the biomass and the spectral indexes is 0.91 at most. However, as the canopy information acquired by the unmanned aerial vehicle is interacted by various factors such as crop planting areas, crop types, crop growth periods and the like, the vegetation index has saturation phenomenon, and the accuracy of the biomass estimation model and the migration capability of the model are affected.
The digital surface model (Digital surface model, DSM) is another key information of ultra-high resolution unmanned aerial vehicle images, highly correlated to crops. For example, chang et al (2017), niu Qinglin et al (2018) and Watanabe et al (2017) acquire digital remote sensing images of sorghum by using unmanned aerial vehicles, plant heights of crops are obtained based on digital ground models (Digital Terrain Model, DTM) and DSM, and RMSE (root mean square error) of the estimated plant heights and the actually measured plant heights is 0.33-0.88 m. In recent years, RTK (real-time kinematic) and PPK (dynamic post-processing) technologies are raised, so that unmanned aerial vehicles have outstanding advantages in terms of quickly and accurately acquiring plant heights, and research of Forlani et al (2018), wang et al (2022) shows that CSM (crop surface model) obtained by adopting PPK and RTK correction can improve plant height estimation accuracy, but research of directly constructing a biomass estimation model through DSM information is lacking at present, and model precision is insufficient.
Disclosure of Invention
Aiming at the technical problems that the existing winter wheat biomass estimation model is low in precision and poor in model applicability, the invention provides a winter wheat biomass rapid and high-precision estimation method based on an unmanned aerial vehicle.
In order to achieve the above purpose, the technical scheme of the invention is realized as follows: a fast and high-precision winter wheat biomass estimation method based on unmanned aerial vehicle comprises the following steps:
step one: acquiring an unmanned aerial vehicle image of a research area by using an unmanned aerial vehicle multispectral sensor, and preprocessing the unmanned aerial vehicle image to obtain data of an orthographic image and a digital surface model;
step two: extracting the preprocessed orthographic image and the winter wheat plant height in the digital surface model based on grid calculation;
step three: establishing a biomass estimation model by adopting a BP neural network regression method according to the plant height of winter wheat;
step four: and performing precision evaluation on the biomass estimation model by using the model decision coefficients.
Preferably, the unmanned aerial vehicle image acquisition method comprises the following steps: acquiring unmanned aerial vehicle images of a test area by adopting an unmanned aerial vehicle to carry a K6 multispectral sensor between 10:00 and 14:00 before sowing winter wheat in 2020 and during the booting period, the flowering period and the grouting period of winter wheat in 2021; the flying height of the unmanned aerial vehicle is 50m, and the spatial resolution of the acquired unmanned aerial vehicle image is 0.02m.
Preferably, the K6 multispectral sensor comprises a blue light wave band with a central wavelength of 450nm, a green light wave band with a central wavelength of 550nm, a red light wave band with a central wavelength of 685nm, a red side wave band with a central wavelength of 725nm, and a near infrared wave band with a central wavelength of 780 nm.
Preferably, ground control points are arranged at 4 vertexes and the middle position of the test area and used for accurately correcting the flying position and the altitude; 5 nitrogen treatment areas are designed in the test area, each nitrogen treatment is repeated for 3 times, 50% of nitrogen is applied as base fertilizer, and 50% of nitrogen is applied in the winter wheat jointing period; the test area adopts a random block test design, and each nitrogen treatment area is provided with 3 varieties of winter wheat.
Preferably, the unmanned aerial vehicle image preprocessing method comprises the following steps:
(1) Image format conversion: adopting K6 multispectral sensor self-provided MAPIR Camera Control software to convert MAPIR format unmanned aerial vehicle images into TIF format images in batches;
(2) Image screening: automatically deleting images outside the research area according to the geographic coordinate information by utilizing KML data of the research area;
(3) And (3) image stitching: adopting Agisoft PhotoScan Professional 12.0.0 software to splice images based on a 3Dmodel mode to obtain an orthographic image I of a research area and a digital surface model;
(4) Orthographic correction: based on RTK geographic information coordinates and ground control points, accurately correcting the geographic position of the research area orthographic image I and the digital surface model obtained in the step (3);
(5) Radiation calibration: based on DN values of the radiation calibration plates, a linear conversion formula is established by combining reflection characteristics of the calibration plates, DN values of the orthographic images I after orthographic correction are converted into reflectivity, and an orthographic image II directly used for biomass inversion is obtained.
Preferably, the method for extracting the plant height of the winter wheat in the second step comprises the following steps:
1) When the image stitching is carried out based on the 3Dmodel mode in the preprocessing, all images are aligned, camera depth information corresponding to each image is calculated according to RTK geographic information coordinates and information contained in the images, the depth information is combined and built into dense point clouds, grids and textures are generated based on the built dense point clouds, and an orthographic image I and a Crop Surface Model (CSM) are obtained;
2) Based on the fact that an orthographic image II obtained after orthographic correction in pretreatment meets the condition that the deviation between the value of CSM at a ground control point and the value of a GPS actual measurement digital elevation model DEM is smaller than 0.01m, outputting a CSM layer after reaching accuracy;
3) Based on the data of the region of interest of the cell, the orthographic image II and CSM layer data, outputting a plant height data layer in ArcGIS10.6 software through a grid calculator and an extraction value to obtain a plant height H dsm 。
Preferably, the image stitching is used for importing a CSM image layer into an ArcGIS10.6 software ArcTooLbox tool, obtaining a standardized winter wheat CSM by subtracting grid images of a digital elevation model of a bare land period of a test area from grid images of CSM of each growth period, recording the standardized winter wheat CSM as CSMs, and drawing a rectangular region of interest according to the range of each cell in the test area; obtaining the plant height H of each cell by using a calculation average tool on normalized CSMS partitions dsm Plant height H dsm In the range of 43.0 to 96.9cm.
Preferably, the steps 1) to 3) generate dense point clouds, generate grids, generate textures, construct a digital elevation model and obtain an orthographic image I in software Agisoft PhotoScan Professional 12.0.0; acquiring 5-day bare land unmanned aerial vehicle images of 10 months 2020, and establishing a digital elevation model DEM of the terrain of the test area; when drawing a rectangular region of interest, each edge and the cell edge are reserved for 1m.
Preferably, in the third step, the method for establishing the biomass estimation model by adopting the BP neural network regression method comprises the following steps: setting biomass as y, wherein the unit is kg/ha, and the plant height H is utilized dsm The model for estimating biomass is: y=f (H dsm ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein f is a functional expression;
the linear biomass estimation model is established by adopting a BP neural network regression method as follows: y= 419.23H dsm －15804；
Biomass y and plant height H dsm The ratio of (2) is denoted as y b The method comprises the following steps:
re-using BP neural network regression methodConstruction of biomass y and y b Improved biomass estimation model of relationship y=f (y b ) The improved biomass estimation model was obtained as:
wherein H is max Is H dsm Is a maximum value of (a).
Preferably, the BP neural network adopted by the BP neural network regression method is of a 3-layer network structure, a quasi-Newton method family optimizer is adopted to improve the running speed, and the number of hidden neurons is 1;
the resulting data is taken into an improved biomass estimation model y=f (y b ) Obtaining the biomass y and the ratio y b The model formula between the two is as follows: y=ky b -4004.8; wherein K is a coefficient;
intercepting biomass in different growth periods for multiple times to form different data sets, constructing a biomass estimation model by adopting a BP neural network model, and constructing a coefficient K and a plant height H dsm The relation between them is k=h max +0.3；
The improved biomass estimation model is converted into: y= (H) max +0.3)×y b －4000.8；
Will be the ratio y b Carrying out solving on the biomass y by taking the formula to obtain an estimated model of the biomass y:
the invention has the beneficial effects that: based on the images acquired by the low-altitude unmanned aerial vehicle, biomass estimation is performed by improving the plant height extracted by directly utilizing the unmanned aerial vehicle images, so that the accuracy of winter wheat biomass estimation is improved, and high-accuracy and rapid biomass estimation is realized. Compared with the prior art, the technical scheme provided by the invention has the characteristics of simplicity, rapidness and high precision, and on the basis of the existing image acquisition, the rapid and high-precision biomass estimation can be realized by fully utilizing the relation between the biomass of unit plant height and the plant height through analyzing and converting the data, the precision is improved by 51.72 percent compared with a method for directly utilizing the plant height to carry out biomass estimation, and the low-cost and high-precision estimation of the winter wheat biomass is realized.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of the present invention.
FIG. 2 is a schematic diagram of a test zone according to the present invention.
FIG. 3 is a schematic flow chart of the method for extracting winter wheat plant height.
FIG. 4 shows the relationship between measured and estimated biomass, wherein (a) is a linear biomass estimation model and (b) is a modified model.
FIG. 5 is a graph showing the relationship between different nitrogen treatments and biomass per plant height according to the present invention.
FIG. 6 is a graph showing the results of inversion of winter wheat biomass according to the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without any inventive effort, are intended to be within the scope of the invention.
As shown in fig. 1, a method for rapidly and accurately estimating the biomass of winter wheat based on an unmanned aerial vehicle comprises the following steps:
step one: and acquiring an unmanned aerial vehicle image of the research area by using an unmanned aerial vehicle multispectral sensor, and preprocessing the unmanned aerial vehicle image to obtain data of an orthographic image and a digital surface model.
Before the winter wheat in 2020 and during the booting period, the flowering period and the grouting period of the winter wheat in 2021, a K6 multispectral sensor is carried by a Dajiang unmanned aerial vehicle M600, unmanned aerial vehicle images of a test area are acquired at a ratio of 10:00-14:00, and the data format is MAPIR and is stored on an SD card. The K6 multispectral sensor mainly comprises 5 wave bands of blue light wave bands (450 nm of central wavelength, B), green light wave bands (550 nm of central wavelength, G), red light wave bands (685 nm of central wavelength, R), red side wave bands (725 nm of central wavelength, red edge) and near infrared wave bands (780 of central wave bands, nir). The unmanned aerial vehicle flight height is 50m, and the image space resolution who acquires is 0.02m, sets up ground control point (Ground control point, GCP) at 4 summit of test area and intermediate position for flight position and altitude are accurate to be rectified, reduce position error.
The test design of the test area is shown in figure 2. The test area designs 5 nitrogen treatment areas, which are respectively: n0 (0 kg/hm) 2 )、N6(90kg/hm 2 )、N12(180kg/hm 2 )、N18(270kg/hm 2 )、N24(360kg/hm 2 ) Each treatment was repeated 3 times, 50% nitrogen was applied as base fertilizer and 50% nitrogen was applied during winter wheat jointing. The test was designed as a randomized block test, with 3 varieties of winter wheat per treatment.
The preprocessing of the unmanned aerial vehicle image comprises the following steps:
(1) Image format conversion: firstly, copying images on an SD card of an unmanned aerial vehicle to a computer, and converting MAPIR format images into TIF format images in batches by adopting MAPIR Camera Control software of a K6 multispectral sensor;
(2) Image screening: in order to save time, the KML data of the research area is utilized, and images outside the research area are automatically deleted according to geographic coordinate information;
(3) And (3) image stitching: image stitching based on the 3Dmodel model using Agisoft PhotoScan Professional 12.0.0 software, an orthographic image I of the region of interest and a digital surface model (Digital Surface Model, DSM) were obtained. The advantage of the 3Dmodel mode is that height data can be better processed.
(4) Orthographic correction: and (3) accurately correcting the geographical positions of the orthographic images I and DSM of the research area obtained in the step (3) based on RTK geographical information coordinates and Ground Control Points (GCP), and eliminating position deviation caused by high flying speed, unstable posture and image distortion. The method can enable the positions of the unmanned aerial vehicle images of the research areas obtained in different periods to be accurately matched.
(5) Radiation calibration: based on DN (Digital Number DN) values of the radiation calibration plate, a linear conversion formula is established by combining reflection characteristics of the calibration plate, and an orthographic image DN value after orthographic correction is converted into reflectivity so as to eliminate influences caused by factors such as different time, solar altitude angle, weather and the like. The radiation calibration plate is paved in a flat place near the test area, and the DN value is converted into the reflectivity, so that the influence of factors such as different time, solar altitude angle, weather and the like can be eliminated. The orthographic image II of the unmanned aerial vehicle obtained by the step is mainly used for biomass inversion of a research area.
Step two: and extracting the preprocessed orthographic image and winter wheat plant height data in the digital surface model based on grid calculation.
Unmanned aerial vehicle winter wheat plant height H dsm The extraction steps (unit: cm) are shown in figure 3, and the specific implementation steps are as follows:
1) When the 3Dmodel mode is adopted to carry out image stitching in the preprocessing step (3), all images are aligned, camera depth information corresponding to each image is calculated according to RTK geographic information coordinates and information contained in the images, the information is combined and built into dense point clouds, grids and textures are generated based on the built dense point clouds, and an orthographic image I and a crop surface model (Crop Surface Model, CSM) are obtained.
2) Based on the orthographic image II obtained after orthographic correction in the step (4), the deviation between the CSM value at the ground control point and the actual measured DEM value of the GPS is smaller than 0.01m, and the CSM image layer is output after reaching the precision.
3) Based on the region ROI data, the orthographic image II and CSM layer data, outputting plant height H in ArcGIS10.6 software through a grid calculator, extraction values and other tools dsm And a data layer.
Step 1) to 3) are completed by several functional modules of generating dense point cloud, generating grid, generating texture, constructing DEM and orthographic image in software Agisoft PhotoScan Professional 12.0.
Specifically, the bare land unmanned aerial vehicle image is acquired at 5 months of 2020 for establishing a digital elevation model (Digital Elevation Model, DEM) of the terrain of the test area.
In the second image stitching step 2), the CSM image layer is imported into an ArcGIS10.6 software ArcTooLbox tool, the grid image of the test area bare land period DEM is subtracted from the grid image of each growth period CSM, the normalized winter wheat CSM is obtained and is recorded as CSMs, and a rectangular region of interest (Region of Interest, ROI) is drawn according to the range of each cell in the test area. When the ROI is drawn, each edge and the edge of the cell are reserved for 1m so as to eliminate the interference of the edge. Finally, obtaining the plant height H of each cell by using a calculation average value tool for the normalized CSMS partition dsm Plant height H dsm In the range of 43.0 to 96.9cm.
Step three: and establishing a biomass estimation model by adopting a BP neural network regression method according to winter wheat plant height data.
Firstly, biomass is set as y, the unit is kg/ha, and plant height H is utilized dsm The biomass estimation model is as follows:
y＝f(H dsm ) (1)
where f is a functional expression.
Specifically, a BP neural network regression method is adopted to establish a linear biomass estimation model, and the obtained biomass estimation model is as follows:
y＝419.23 H dsm －15804 (2)
model determination coefficient R 2 =0.58, and the relationship between the measured biomass and the estimated biomass is shown in fig. 4 (a).
The method for calculating the decision coefficient of the model by using the BP neural network regression method comprises the following steps:
wherein y is i 、Measured and predicted values of the ith sample biomass, +.>The average value of biomass is represented, and n is the number of samples.
The BP neural network regression method can automatically extract the reasonable rules between output and output data through the learning of biomass data, and has the capability of high self-learning and self-adaption. The BP neural network adopted in the invention is of a 3-layer network structure, the quasi-Newton method family optimizer is adopted to improve the running speed, and the number of hidden neurons is 1.
The error of the model is larger, and the comparison analysis shows that different nitrogen treatments show the biomass per plant height and the average plant height H dsm There is a good positive correlation. Therefore, a winter wheat biomass estimation model based on unit plant height biomass is provided, so that the model is improved, and biomass y and plant height H are obtained dsm The ratio of (2) is denoted as y b Biomass y, average plant height H dsm Sum ratio y b There is a relationship between:
there is a gradual trend between specific different nitrogen treatments and biomass per plant height ratios, as shown in figure 5. Different nitrogen treatments, biomass y and plant height H dsm The difference interval between adjacent nitrogen treatments is 30.95-40.09. The difference between the treatment N6 and the treatment N0 is 40.09, the difference between the treatment N12 and the treatment N6 is 34.86, the difference between the treatment N18 and the treatment N16 is 37.09, the difference between the treatment N24 and the treatment N18 is 30.95, and the difference between the treatment N6 and the treatment N0 is the largest among the treatments with different nitrogen, mainly due to the fertilizer effect of crops between nitrogen application and nitrogen application; the difference between the N24 treatment and the N18 treatment is small, mainly because when the nitrogen application amount reaches a certain level,the biomass of the wheat is increased to be stable, and the fertilizer effect is reduced.
Similarly, the BP neural network regression method is adopted to reconstruct the biomass y and the ratio y b An improved biomass estimation model of the relationship, noted:
y＝f(y b ) (5)
after the data is brought in, biomass y and ratio y are obtained b The specific model formula between the two is as follows: y=ky b －4004.8。
Taking into account the objective and practical fact that biomass gradually increases in the wheat booting stage, the flowering stage and the grouting stage, the coefficient K of the model is optimized. Specifically, biomass in different growth periods is intercepted for multiple times to form different data sets, a BP neural network model is adopted to construct a biomass estimation model, and coefficients K and H are obtained dsm There is a relationship of k=h max +0.3，H max For each data set H dsm The model formula can be converted to:
y＝(H max +0.3)×y b －4000.8 (6)
and (3) carrying out solution on y by using the formula (4) into the formula (6), and finally obtaining a biomass y estimation model as follows:
in the plant height H dsm Maximum value H of (2) max ＝96.9cm。
Model determination coefficient R' 2 =0.88. Through the improved biomass estimation model, the relation between the measured value and the estimated value is shown in fig. 4 (b), and it can be seen that the improved model is simple and quick and achieves higher estimation accuracy.
Step four: accuracy evaluation of biomass estimation model
The accuracy (accuracy) was evaluated using formula (7).
The accuracy of the improved biomass estimation model y' is improved by 51.72% compared with that of the original model y.
Based on the improved biomass estimation model, taking grouting period as an example, the winter wheat biomass is inverted, and the result is shown in fig. 6. From fig. 6 it can be seen that the higher the nitrogen level, the greater the biomass for the different nitrogen treatments.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (10)
1. A fast and high-precision winter wheat biomass estimation method based on unmanned aerial vehicle is characterized by comprising the following steps:
step one: acquiring an unmanned aerial vehicle image of a research area by using an unmanned aerial vehicle multispectral sensor, and preprocessing the unmanned aerial vehicle image to obtain data of an orthographic image and a digital surface model;
step two: extracting the preprocessed orthographic image and the winter wheat plant height in the digital surface model based on grid calculation;
step three: establishing a biomass estimation model by adopting a BP neural network regression method according to the plant height of winter wheat;
step four: and performing precision evaluation on the biomass estimation model by using the model decision coefficients.
2. The unmanned aerial vehicle-based rapid and high-precision estimation method for winter wheat biomass, which is characterized in that the unmanned aerial vehicle image acquisition method comprises the following steps: acquiring unmanned aerial vehicle images of a test area by adopting an unmanned aerial vehicle to carry a K6 multispectral sensor between 10:00 and 14:00 before sowing winter wheat in 2020 and during the booting period, the flowering period and the grouting period of winter wheat in 2021; the flying height of the unmanned aerial vehicle is 50m, and the spatial resolution of the acquired unmanned aerial vehicle image is 0.02m.
3. The unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method according to claim 2, wherein the K6 multispectral sensor comprises a blue light wave band with a central wavelength of 450nm, a green light wave band with a central wavelength of 550nm, a red light wave band with a central wavelength of 685nm, a red side wave band with a central wavelength of 725nm and a near infrared wave band with a central wavelength of 780 nm.
4. The unmanned aerial vehicle-based rapid and high-precision estimation method for winter wheat biomass according to claim 2 or 3, wherein ground control points are arranged at 4 vertexes and intermediate positions of a test area and used for precise correction of flying positions and altitudes; 5 nitrogen treatment areas are designed in the test area, each nitrogen treatment is repeated for 3 times, 50% of nitrogen is applied as base fertilizer, and 50% of nitrogen is applied in the winter wheat jointing period; the test area adopts a random block test design, and each nitrogen treatment area is provided with 3 varieties of winter wheat.
5. The unmanned aerial vehicle-based rapid and high-precision estimation method for winter wheat biomass according to claim 4, wherein the unmanned aerial vehicle image preprocessing method comprises the following steps:
(1) Image format conversion: adopting K6 multispectral sensor self-provided MAPIR Camera Control software to convert MAPIR format unmanned aerial vehicle images into TIF format images in batches;
(2) Image screening: automatically deleting images outside the research area according to the geographic coordinate information by utilizing KML data of the research area;
(3) And (3) image stitching: adopting Agisoft PhotoScan Professional 12.0.0 software to splice images based on a 3Dmodel mode to obtain an orthographic image I of a research area and a digital surface model;
(4) Orthographic correction: based on RTK geographic information coordinates and ground control points, accurately correcting the geographic position of the research area orthographic image I and the digital surface model obtained in the step (3);
(5) Radiation calibration: based on DN values of the radiation calibration plates, a linear conversion formula is established by combining reflection characteristics of the calibration plates, DN values of the orthographic images I after orthographic correction are converted into reflectivity, and an orthographic image II directly used for biomass inversion is obtained.
6. The unmanned aerial vehicle-based rapid and high-precision estimation method for winter wheat biomass, which is characterized in that the extraction method for winter wheat plant height in the second step is as follows:
1) When the image splicing is carried out based on the 3Dmodel mode in the preprocessing, all images are aligned, camera depth information corresponding to each image is calculated according to RTK geographic information coordinates and information contained in the images, the depth information is combined and built into dense point clouds, grids and textures are generated based on the built dense point clouds, and an orthographic image I and a crop surface model are obtained;
2) Based on the orthographic image II obtained after orthographic correction in the preprocessing, the deviation between the value of CSM at a ground control point and the value of a GPS actual measurement digital elevation model is less than 0.01m, and outputting a CSM layer after reaching the precision;
3) Based on the data of the region of interest of the cell, the orthographic image II and CSM layer data, outputting a plant height data layer in ArcGIS10.6 software through a grid calculator and an extraction value to obtain a plant height H dsm 。
7. The unmanned aerial vehicle-based rapid and high-precision estimation method for winter wheat biomass, which is characterized in that the image stitching is used for importing a CSM image layer into an ArcGIS10.6 software ArcTooLbox tool, obtaining a standardized winter wheat CSM by subtracting grid images of a digital elevation model of a bare land period of a test area from grid images of CSM of each growth period, recording the standardized winter wheat CSM as CSMs, and drawing a rectangular region of interest according to the range of each cell in the test area; obtaining the plant height H of each cell by using a calculation average tool on normalized CSMS partitions dsm Plant height H dsm In the range of 43.0 to 96.9cm.
8. The unmanned aerial vehicle-based rapid and high-precision estimation method for winter wheat biomass according to claim 7, wherein the steps 1) to 3) generate dense point clouds, generate grids, generate textures, construct a digital elevation model and obtain an orthographic image i in software Agisoft PhotoScan Professional 12.0.0; acquiring 5-day bare land unmanned aerial vehicle images of 10 months 2020, and establishing a digital elevation model DEM of the terrain of the test area; when drawing a rectangular region of interest, each edge and the cell edge are reserved for 1m.
9. The unmanned aerial vehicle-based rapid and high-precision estimation method for biomass of winter wheat according to any one of claims 1 to 3 and 5 to 8, wherein the method for establishing the biomass estimation model by adopting the BP neural network regression method in the third step is as follows: setting biomass as y, wherein the unit is kg/ha, and the plant height H is utilized dsm The model for estimating biomass is: y=f (H dsm ) The method comprises the steps of carrying out a first treatment on the surface of the Wherein f is a functional expression;
the linear biomass estimation model is established by adopting a BP neural network regression method as follows: y= 419.23H dsm －15804；
Biomass y and plant height H dsm The ratio of (2) is denoted as y b The method comprises the following steps:
reconstructing biomass y and ratio y by BP neural network regression method b Improved biomass estimation model of relationship y=f (y b ) The improved biomass estimation model was obtained as:
wherein H is max For plant height H dsm Is a maximum value of (a).
10. The unmanned aerial vehicle-based rapid and high-precision estimation method for winter wheat biomass, which is characterized in that the BP neural network adopted by the BP neural network regression method is of a 3-layer network structure, a quasi-Newton method family optimizer is adopted to improve the running speed, and the number of hidden neurons is 1;
the resulting data is taken into an improved biomass estimation model y=f (y b ) Obtaining the biomass y and the ratio y b The model formula between the two is as follows: y=ky b -4004.8; wherein K is a coefficient;
intercepting biomass in different growth periods for multiple times to form different data sets, constructing a biomass estimation model by adopting a BP neural network model, and constructing a coefficient K and a plant height H dsm The relation between them is k=h max +0.3；
The improved biomass estimation model is converted into: y= (H) max +0.3)×y b －4000.8；
Will be the ratio y b Carrying out solving on the biomass y by taking the formula to obtain an estimated model of the biomass y:
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211603974.XA CN116229289A (en) | 2022-12-13 | 2022-12-13 | Unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method |
| CN202211603974X | 2022-12-13 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117576583A true CN117576583A (en) | 2024-02-20 |
Family
ID=86583249
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211603974.XA Pending CN116229289A (en) | 2022-12-13 | 2022-12-13 | Unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method |
| CN202311220937.5A Pending CN117576583A (en) | 2022-12-13 | 2023-09-21 | Unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211603974.XA Pending CN116229289A (en) | 2022-12-13 | 2022-12-13 | Unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method |
Country Status (2)
| Country | Link |
|---|---|
| CN (2) | CN116229289A (en) |
| BE (1) | BE1030485B1 (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117274359B (en) * | 2023-10-13 | 2024-04-19 | 扬州大学 | Plant height measuring and calculating method and system for crop group |
| CN117253163A (en) * | 2023-11-14 | 2023-12-19 | 山东科技大学 | Unmanned plane laser radar-based field peanut plant height estimation method |
| CN117592604A (en) * | 2023-11-22 | 2024-02-23 | 河北省农林科学院旱作农业研究所 | Unmanned aerial vehicle-mounted remote sensing identification method for water-efficient wheat varieties |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102565061B (en) * | 2012-01-16 | 2014-03-12 | 沈阳农业大学 | Crop biomass nondestructive testing image acquisition and processing device and testing method |
| CN102967354B (en) * | 2012-10-31 | 2015-02-11 | 北京农业信息技术研究中心 | Detecting device and detecting method for crop biomass |
| CN105513096A (en) * | 2015-11-18 | 2016-04-20 | 青岛农业大学 | Method for estimating biomass of winter wheat |
| CN108007438B (en) * | 2017-12-01 | 2020-07-24 | 云南大学 | Estimation method for plant biomass of unmanned aerial vehicle aerial photography remote sensing wetland |
| CN109459392B (en) * | 2018-11-06 | 2019-06-14 | 南京农业大学 | A kind of rice the upperground part biomass estimating and measuring method based on unmanned plane multispectral image |
| US20230316555A1 (en) * | 2020-08-14 | 2023-10-05 | Agriculture Victoria Services Pty Ltd | System and Method for Image-Based Remote Sensing of Crop Plants |
| CN112215169B (en) * | 2020-10-10 | 2022-11-08 | 华中农业大学 | Crop plant height and biomass self-adaptive resolving method based on unmanned aerial vehicle passive remote sensing |
| CN112345458A (en) * | 2020-10-22 | 2021-02-09 | 南京农业大学 | Wheat yield estimation method based on multispectral image of unmanned aerial vehicle |
| CN112861435B (en) * | 2021-02-09 | 2023-07-18 | 深圳大学 | Mangrove quality remote sensing inversion method and intelligent terminal |
-
2022
- 2022-12-13 CN CN202211603974.XA patent/CN116229289A/en active Pending
-
2023
- 2023-07-19 BE BE20235601A patent/BE1030485B1/en active IP Right Grant
- 2023-09-21 CN CN202311220937.5A patent/CN117576583A/en active Pending
Also Published As
| Publication number | Publication date |
|---|---|
| BE1030485B1 (en) | 2024-05-21 |
| CN116229289A (en) | 2023-06-06 |
| BE1030485A1 (en) | 2023-11-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zheng et al. | Improved estimation of rice aboveground biomass combining textural and spectral analysis of UAV imagery | |
| CN117576583A (en) | Unmanned aerial vehicle-based winter wheat biomass rapid and high-precision estimation method | |
| CN108764255B (en) | Method for extracting winter wheat planting information | |
| CN108662991B (en) | Land block scale winter wheat leaf area index estimation method based on remote sensing satellite data | |
| CN110222475B (en) | Method for inverting moisture content of winter wheat plants based on multispectral remote sensing of unmanned aerial vehicle | |
| CN111798327A (en) | Construction method and application of wheat yield calculation model based on hyperspectral image | |
| CN111368817B (en) | Method and system for quantitatively evaluating thermal effect based on earth surface type | |
| CN110987821B (en) | Hyperspectral rapid atmospheric correction parameterization method | |
| CN105404873B (en) | A kind of winter wheat recognition methods based on the conversion of NDVI time series coordinates | |
| Yuan et al. | Unmanned aerial vehicle–based assessment of cover crop biomass and nitrogen uptake variability | |
| CN110610054A (en) | Method and system for constructing cuboid inversion model of soil humidity | |
| CN112836575A (en) | Multi-time-sequence image rice yield estimation method based on crop phenological period | |
| CN111044516B (en) | Remote sensing estimation method for chlorophyll content of rice | |
| Chen et al. | Global datasets of leaf photosynthetic capacity for ecological and earth system research | |
| CN109784300A (en) | A kind of crops science survey production method and system | |
| CN112348812A (en) | Forest stand age information measuring method and device | |
| CN109816542A (en) | A kind of crop production reduction Claims Resolution method and system | |
| CN115292616B (en) | Vegetation blue sky albedo estimation method and device based on spectrum invariant theory | |
| CN114511170A (en) | Soil nutrient estimation method and device based on multispectral image | |
| CN112836725A (en) | Weak supervision LSTM recurrent neural network rice field identification method based on time sequence remote sensing data | |
| CN116485757A (en) | Winter wheat total nitrogen content prediction method | |
| CN114140695B (en) | Prediction method and system for tea tree nitrogen diagnosis and quality index determination based on unmanned aerial vehicle multispectral remote sensing | |
| CN113887493A (en) | Black and odorous water body remote sensing image identification method based on ID3 algorithm | |
| Zhu et al. | UAV flight height impacts on wheat biomass estimation via machine and deep learning | |
| CN113534083A (en) | Corn stubble mode identification method, device and medium based on SAR |
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 |