CN109613022A - A kind of method, apparatus and system of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen - Google Patents
A kind of method, apparatus and system of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen Download PDFInfo
- Publication number
- CN109613022A CN109613022A CN201910075649.2A CN201910075649A CN109613022A CN 109613022 A CN109613022 A CN 109613022A CN 201910075649 A CN201910075649 A CN 201910075649A CN 109613022 A CN109613022 A CN 109613022A
- Authority
- CN
- China
- Prior art keywords
- citrus
- spectrum
- remote sensing
- plant
- spectrum remote
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/95—Investigating the presence of flaws or contamination characterised by the material or shape of the object to be examined
- G01N21/956—Inspecting patterns on the surface of objects
- G01N21/95607—Inspecting patterns on the surface of objects using a comparative method
Abstract
The present invention relates to the method, apparatus and system of a kind of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen, this method comprises: obtaining the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images include the distributed intelligence of citrus plant and the geography information of citrus plant；The high-spectrum remote sensing is pre-processed；By the low latitude high-spectrum remote sensing after pretreatment, it is input to Citrus Huanglongbing pathogen BP neural network model；Suffer from the distribution results of yellow twig in output citrus distributed areas.Detection of the detection method for the citrus orchard yellow twig of large area, can reduce the workload of manpower and material resources and detection；It can quickly detect, be accurately positioned, efficient lossless；And detection process is low-cost, simple laborsaving；As a result accurate and reliable.
Description
Technical field
The present invention relates to high-spectrum remote-sensing research field, in particular to a kind of low latitude high-spectrum remote-sensing detects Citrus Huanglongbing pathogen
Method, apparatus and system.
Background technique
Citrus is one of important fruit crop in China and one of the maximum fruit of world's output, in agricultural economy
In account for extremely important ingredient.Citrus Huanglongbing pathogen (citrus green disease or huanglongbing) is 20 generation
Early stage record in the citrus crop diseases of southern china place discovery, later even worldwide citrus planting site at home
Gradually it is found.The citrus diseased plant of infection yellow twig shows yellowing leaf, mottled in symptom, and plant growing way is weak, fruit
For " coppernose fruit " either Chinese olive, annesl etc. nowadays temporarily not can be radically cured yellow twig without active drug, at present to infection
The processing mode of yellow twig plant is that whole strain is excavated, and kills the agronomy processing of wood louse.The rate of propagation of yellow twig is fast, destroys
Power is big, heavy then fast between orchard if the citrus fruit tree for infecting yellow twig gently influences the growing way and yield of plant without timely processing
Speed sprawling, causes fruit-tree orchard large area withered, has seriously affected the development of Citrus Industry and compromised the income of orchard worker,
Therefore, Citrus Huanglongbing pathogen is considered as the destructive disease of Citrus Industry.
The symptom of Citrus Huanglongbing pathogen is more complicated, has various researchers just to carry out and grinds to Citrus Huanglongbing pathogen
Study carefully work, but most effective treatment means are that even root is excavated for plant that yellow twig has been infected in discovery early and whole tree, it can be big
Bring economic loss after big reduction citrus fruit tree infection yellow twig.How to detect citrus plant infection yellow twig is that many is ground
Study carefully the research contents of worker.
Currently, the method for detection Citrus Huanglongbing pathogen includes field diagnosis and laboratory biochemical analysis two major classes, field diagnosis
Be diagnosis the fastest method of yellow twig, it is simple and easy and without equipment assist, but knowledge and experience needed for the method deposit it is higher,
Subjective, accuracy rate is not high.Laboratory biochemical analysis includes cause of disease microscopic observation, biochemical indicator detection method, nucleic acid
Probe assay, PCR amplification detection method, LAMP fast detection method and serum detection method etc., these detection method detection process
It is complex, it is higher to testing staff's professional knowledge reserve requirements, detection cycle is long etc., it is unfavorable for being generalized to well agriculture real
In the production of border.Cause of disease microscopic observation is observed and then pays a home visit to cause of disease by electron microscope and ultramicrotomy
It is disconnected.Biochemical indicator detection method is detected using fluorescent indicators or foreign preteins etc..Molecules detection method is mainly nucleic acid
Probe in detecting, PCR amplification detection and LAMP are quickly detected, by the limit for the factors such as process is complicated, time-consuming, DNA dosage is big
The application of system, the technology is constantly subjected to constrain.These types of method is can accurately to detect very much Citrus Huanglongbing pathogen, but expense is high
It is expensive, instrumentation is complicated, professional must also detect under long period and particular surroundings, can not promote and answer well
It uses in production.Serum detection method is diagnosed using with the serum in conjunction with yellow twig cause of disease observe under Electronic Speculum, but this legal system
The disadvantages of standby technology is complicated and detection range is narrow is not solved.In recent years, many research workers carry out to citrus Huang
The EO-1 hyperion of imperial disease detects work, obtains good progress, illustrate EO-1 hyperion to the detection of Citrus Huanglongbing pathogen there are it is higher can
Row.
Therefore, for the citrus orchard of large area, for the workload for reducing manpower and material resources and detection, being badly in need of development can be quick
Detection, accurate positionin, efficient lossless, the reliable yellow twig detection method of low-cost, simple laborsaving and result.
Summary of the invention
In view of the above problems, the invention proposes a kind of method, apparatus of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen
And system, detection of the detection method for the citrus orchard yellow twig of large area can reduce the work of manpower and material resources and detection
Amount；It can quickly detect, be accurately positioned, efficient lossless；And detection process is low-cost, simple laborsaving；As a result accurate and reliable.
In order to solve the above-mentioned technical problem, in a first aspect, the embodiment of the present invention provides one kind based on low latitude high-spectrum remote-sensing
Citrus Huanglongbing pathogen detection method, comprising:
Obtain the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images include citrus plant
The geography information of distributed intelligence and citrus plant；
The high-spectrum remote sensing is pre-processed；
By the low latitude high-spectrum remote sensing after pretreatment, it is input to Citrus Huanglongbing pathogen BP neural network model；
Suffer from the distribution results of yellow twig in output citrus distributed areas.
In one embodiment, the generation step of the Citrus Huanglongbing pathogen BP neural network model, comprising:
The high-spectrum remote sensing for acquiring a large amount of citrus influences of plant crown, pre-processes the high-spectrum remote sensing；
The spectral information of citrus fruit tree healthy plant and infection yellow twig influences of plant crown in high-spectrum remote sensing is extracted, it is right
The spectral information is analyzed and processed, and obtains spectroscopic data；
By the spectroscopic data according to preset algorithm, characteristic wave bands are extracted；The characteristic wave bands include: maximum fault information,
Maximal projection or maximum distance characteristic wave bands；The characteristic wave bands are substituted into preset vegetation index model, obtain vegetation index；
Citrus Huanglongbing pathogen BP nerve net is generated by the parameter of debugging model using the vegetation index as training sample
Network model；The training sample is determining illness blade and healthy leaves data.
In one embodiment, the high-spectrum remote sensing is pre-processed, comprising:
Radiant correction, geometric correction and Panoramagram montage are carried out to the high-spectrum remote sensing.
In one embodiment, citrus fruit tree healthy plant and infection yellow twig plant hat in high-spectrum remote sensing are extracted
The spectral information of layer, is analyzed and processed the spectral information, obtains spectroscopic data, comprising:
Area-of-interest is drawn, the spectral information of each area-of-interest is calculated；
The spectral information of each region of interest is subjected to random combine, is averaging, obtains multiple spectrum samples；
The spectrum samples are used into mahalanobis distance method rejecting abnormalities sample, and are carried out using Savitzky-Golay algorithm
Smoothing denoising obtains spectroscopic data.
In one embodiment, the spectroscopic data is extracted into characteristic wave bands according to preset algorithm, comprising:
By the spectroscopic data according to SPA selection algorithm, the redundancy in original spectrum matrix is eliminated, feature is filtered out
Wave band；Steps are as follows by SPA:
(1) sample set sample number and number of wavelengths K are determined, spectrum matrix X is formedM×K；
(2) it initializes: n=1, in iteration for the first time, the optional column vector x in spectrum matrixj, it is denoted as xK(0), i.e. (K
(0)=j)；
(3) set S is defined as:Calculate separately xjTo in S to
The projection vector Px of amountj,
(4) maximum projection serial number is recorded, the projection vector by maximum projection as lower whorl iteration；
(5) superiority and inferiority for carrying out judgment models using RMSEP selects the smallest RMSEP, correspondingAnd N*As screening
Band combination out.
In one embodiment, the characteristic wave bands are substituted into preset vegetation index model, obtains vegetation index, wrapped
It includes:
The characteristic wave bands are substituted into NDVI, NDGI, TVI, RVI, NLI and DVI vegetation index model calculates all infection
Yellow twig plant and the respective averaged spectrum of healthy plant spectrum samples；
The value of vegetation index is calculated using the averaged spectrum, and is replaced to calculate with the different-waveband in wave-length coverage and be planted
By index；
Compare same a band math result with vegetation index model in healthy plant and infection yellow twig influences of plant crown
Difference, preferred bands of the maximum wave band of selection differences as vegetation index obtain preferred vegetation index.
Second aspect, the present invention also provides a kind of Citrus Huanglongbing pathogen detection devices based on low latitude high-spectrum remote-sensing, comprising:
Module is obtained, for obtaining the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images
The geography information of distributed intelligence and citrus plant including citrus plant；
Processing module, for being pre-processed to the high-spectrum remote sensing；
Input module, for being input to Citrus Huanglongbing pathogen BP for the low latitude high-spectrum remote sensing after pretreatment
Neural network model；
Output module, the distribution results for suffering from yellow twig for exporting citrus distributed areas.
In one embodiment, the Citrus Huanglongbing pathogen BP neural network model in the input module, comprising:
Acquisition process submodule, for acquiring the high-spectrum remote sensing of a large amount of citrus influences of plant crown, to the EO-1 hyperion
Remote sensing images are pre-processed；
Extraction process submodule is planted for extracting citrus fruit tree healthy plant and infection yellow twig in high-spectrum remote sensing
The spectral information of canopy layer is analyzed and processed the spectral information, obtains spectroscopic data；
It extracts and substitutes into submodule, for the spectroscopic data according to preset algorithm, to be extracted characteristic wave bands；The characteristic wave
Section includes: maximum fault information, maximal projection or maximum distance characteristic wave bands；The characteristic wave bands are substituted into preset vegetation index
Model obtains vegetation index；
Training generates submodule, for being generated using the vegetation index as training sample by the parameter of debugging model
Citrus Huanglongbing pathogen BP neural network model；The training sample is determining illness blade and healthy leaves data.
The third aspect, the present invention provide a kind of Citrus Huanglongbing pathogen detection system based on low latitude high-spectrum remote-sensing, comprising: nothing
Man-machine flying platform, earth station and cloud platform；
The unmanned plane during flying platform carries EO-1 hyperion camera, and the low latitude EO-1 hyperion for acquiring citrus distributed areas to be detected is distant
Feel image；
The earth station is navigated by water for controlling the unmanned plane during flying platform according to pre-set flight paths；
The cloud platform is connect with the unmanned plane during flying Platform communication, and the cloud platform includes:
Module is obtained, for obtaining the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images
The geography information of distributed intelligence and citrus plant including citrus plant；
Processing module, for being pre-processed to the high-spectrum remote sensing；
Input module, for being input to Citrus Huanglongbing pathogen BP for the low latitude high-spectrum remote sensing after pretreatment
Neural network model；
Output module, the distribution results for suffering from yellow twig for exporting citrus distributed areas.
It is an advantage of the current invention that the invention proposes a kind of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen method,
Apparatus and system, detection of the detection method for the citrus orchard yellow twig of large area can reduce manpower and material resources and detection
Workload；It can quickly detect, be accurately positioned, efficient lossless；And detection process is low-cost, simple laborsaving；As a result accurate and reliable.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention can be by written explanation
Specifically noted structure is achieved and obtained in book, claims and attached drawing.
Below by drawings and examples, technical scheme of the present invention will be described in further detail.
Detailed description of the invention
Attached drawing is used to provide further understanding of the present invention, and constitutes part of specification, with reality of the invention
It applies example to be used to explain the present invention together, not be construed as limiting the invention.In the accompanying drawings:
Fig. 1 is the process of the Citrus Huanglongbing pathogen detection method provided in an embodiment of the present invention based on low latitude high-spectrum remote-sensing
Figure；
Fig. 2 is the signal that UAV flight's EO-1 hyperion camera provided in an embodiment of the present invention acquires high-spectrum remote sensing
Figure；
Fig. 3 is the generation step flow chart of Citrus Huanglongbing pathogen BP neural network model provided in an embodiment of the present invention；
Fig. 4 ABC is low-to-medium altitude high-spectrum remote sensing schematic diagram of the present invention；
Fig. 5 is the flow chart of step S32 provided in an embodiment of the present invention；
Fig. 6 is the interested extraction of citrus influences of plant crown provided in an embodiment of the present invention and curve of spectrum situation schematic diagram；
Fig. 7 is the preferred spectral band schematic diagram of SPA algorithm provided in an embodiment of the present invention；
Fig. 8 is variation schematic diagram of the RMSE provided in an embodiment of the present invention with wave band number；
Fig. 9 is high spectrum image healthy plant provided in an embodiment of the present invention and the average light for infecting yellow twig influences of plant crown
It sets a song to music line chart；
Figure 10 is diagnostic result schematic diagram provided in an embodiment of the present invention；
Figure 11 is the block diagram of the Citrus Huanglongbing pathogen detection device provided in an embodiment of the present invention based on low latitude high-spectrum remote-sensing；
Figure 12 is the block diagram of Citrus Huanglongbing pathogen BP neural network model provided in an embodiment of the present invention；
Figure 13 is the signal of the Citrus Huanglongbing pathogen detection system provided in an embodiment of the present invention based on low latitude high-spectrum remote-sensing
Figure.
Specific embodiment
Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
It is fully disclosed to those skilled in the art.
The embodiment of the invention provides a kind of Citrus Huanglongbing pathogen detection methods based on low latitude high-spectrum remote-sensing, referring to Fig.1
It is shown, comprising:
S1, the low latitude high-spectrum remote sensing for obtaining citrus distributed areas to be detected；The remote sensing images include that citrus is planted
The distributed intelligence of strain and the geography information of citrus plant；
S2, the high-spectrum remote sensing is pre-processed；
S3, by the low latitude high-spectrum remote sensing after pretreatment, be input to Citrus Huanglongbing pathogen BP neural network mould
Type；
S4, output citrus distributed areas suffer from the distribution results of yellow twig.
In above-mentioned steps S1, for example unmanned plane or other low flyers carrying EO-1 hyperion camera, reference can be used
Shown in Fig. 2, the low latitude high-spectrum remote sensing of acquisition testing citrus distributed areas, the remote sensing images include point of citrus plant
The geography information of cloth information and citrus plant.
For example select Guangdong Huizhou, Qingyuan City etc. regional Citrus shatangju (Sugar Orange) is object in research, selection is planted
The orchard of Citrus shatangju includes new tree and veteran in orchard, also includes the infection serious sick tree of yellow twig and lesser extent of catching an illness
Citrus plant, is acquired by data, the low latitude EO-1 hyperion of the citrus plant of each orchard different growing stage is obtained, using Fig. 2
Mode acquire high spectrum image, the convenient and efficient in such a way that unmanned aerial vehicle remote sensing acquires high spectrum image, and having every in orchard
The specific location of a citrus fruit tree.
Above-mentioned steps S2~S4 pre-processes above-mentioned high-spectrum remote sensing, and Citrus Huanglongbing pathogen is input to after pretreatment
The distribution results of yellow twig are suffered from BP neural network model, exportable citrus distributed areas.
In the present embodiment, by unmanned plane or other low flyer low latitude high-spectrum remote-sensings, high spatial point can be obtained
The image of resolution and high spectral resolution, and longitude and latitude and the relative position of each tree can be obtained by image, it can be quickly high
It imitates and determines which citrus fruit tree has infected yellow twig in orchard；Citrus Huanglongbing pathogen is detected by unmanned plane low latitude high-spectrum remote-sensing,
Infected plant can be further determined that by the degree and disease of disease and the specific distribution of canopy.Diagnostic result is available entire
The case where yellow twig, is infected in citrus orchard, and obtains yellow twig in the distribution situation of illness influences of plant crown, and covering according to disease
Lid ratio determines the degree of disease of illness plant.The distribution situation and geography of entire orchard citrus plant are had in remote sensing images
Information, both it is known that the geography information position of illness plant is also known that the plantation situation of plant around it.The detection side
Detection of the method for the citrus orchard yellow twig of large area, can reduce the workload of manpower and material resources and detection；It can quickly detect, is quasi-
Determine position, efficient lossless；And detection process is low-cost, simple laborsaving；As a result accurate and reliable.
In one embodiment, in above-mentioned steps S3, the generation step of Citrus Huanglongbing pathogen BP neural network model, referring to figure
Shown in 3, comprising:
The high-spectrum remote sensing of S31, a large amount of citrus influences of plant crown of acquisition, locate the high-spectrum remote sensing in advance
Reason；
S32, the spectrum letter for extracting citrus fruit tree healthy plant and infection yellow twig influences of plant crown in high-spectrum remote sensing
Breath, is analyzed and processed the spectral information, obtains spectroscopic data；
S33, by the spectroscopic data according to preset algorithm, extract characteristic wave bands；The characteristic wave bands include: maximum information
Amount, maximal projection or maximum distance characteristic wave bands；The characteristic wave bands are substituted into preset vegetation index model, show that vegetation refers to
Number；
S34, Citrus Huanglongbing pathogen BP mind is generated by the parameter of debugging model using the vegetation index as training sample
Through network model；The training sample is determining illness blade and healthy leaves data.
Wherein in step S31, the high-spectrum remote sensing of a large amount of citrus influences of plant crown is acquired, it equally can also be by nobody
Machine or the high-spectrum remote sensing of other low latitude equipment acquisition；Preprocessing process and upper is carried out to above-mentioned high-spectrum remote sensing
It is identical for stating step S2.Pretreated process includes: to carry out radiant correction, geometric correction and panorama to high-spectrum remote sensing
Figure splicing etc., obtains the panorama high spectrum image that can detecte region, as shown in Figure 4 A.
By pretreated high-spectrum remote sensing, extracts citrus fruit tree healthy plant and infect yellow twig influences of plant crown
Spectral information, and the spectral information is analyzed and processed, obtain spectroscopic data；
Spectroscopic data is extracted into characteristic wave bands according to preset algorithm again；This feature wave band includes: maximum fault information, maximum
The characteristic wave bands such as projection or maximum distance；This feature wave band is substituted into preset vegetation index model, obtains vegetation index；Finally
Citrus Huanglongbing pathogen BP neural network model is generated by the parameter of debugging model using vegetation index as training sample；Wherein instruct
Practicing sample is determining illness blade and healthy leaves data.
High spectrum image number is established by acquiring the low latitude high spectrum image of a large amount of citrus influences of plant crown in the training stage
According to library；High spectrum image extracts canopy spectrum information and pre-processes, and extracts characteristic spectrum and preferred vegetation index, will be above-mentioned excellent
The vegetation index of choosing is trained as characteristic value by BP neural network, obtains preferably low latitude high-spectrum remote-sensing to citrus
The discrimination model of yellow twig, i.e. Citrus Huanglongbing pathogen BP neural network model；Differentiate the stage, extracts high-spectrum remote-sensing to be identified
Image is input in discrimination model, and moving model is differentiated as a result, the diagnostic result of model can be obtained.
It, should be comprising being previously mentioned vegetation in the single sample of training data using vegetation index as characteristic value in the present embodiment
The characteristic values such as index；
1, data set is handled, and the data set that will acquire is randomly divided into training set in the ratio of 3:1:1, verifying collects and test
Collection.
2, training set is used to training pattern, and verifying collection is used to verify the effect of the hyper parameter of model selection；Test set is used to
Verify the effect of model.
In diagnostic phases:
The canopy of citrus plant is extracted, only the spectrum of canopy is put into model diagnoses, such as can will be in diagnostic result
The canopy region for being determined as infection yellow twig is shown in red, the canopy region of health will be determined as in diagnostic result, is shown as
Green.
In one embodiment, step S32, referring to Figure 5, comprising:
S321, area-of-interest is drawn, calculates the spectral information of each area-of-interest；
S322, the spectral information of each region of interest is subjected to random combine, is averaging, obtains multiple spectrum samples
This；
S323, the spectrum samples are used to mahalanobis distance method rejecting abnormalities sample, and is calculated using Savitzky-Golay
Method carries out smoothing denoising, obtains spectroscopic data.
After carrying out radiant correction and geometric correction to high spectrum image, by drawing region of interest to influences of plant crown, extract
The spectral information of citrus influences of plant crown uses mahalanobis distance method rejecting abnormalities sample to the canopy spectrum information of said extracted, adopts
With the processing such as the smooth curve of spectrum of Savitzky-Golay algorithm and removal noise.
Above extracted spectrum samples are influenced by inside external environment and equipment, and there are individual wave band datas are different
Often, noise etc. influences, and to the spectrum of above-mentioned individual wave band data exceptions, unified remove in sample on the whole ceases abnormal wave
Section；By the spectrum to differ greatly between mahalanobis distance Rejection of samples, the abnormal spectrum samples of removal are obtained；Above-mentioned removal is abnormal
Spectrum samples carry out smooth and denoising, obtain pretreated spectrum samples.
Further, in addition to may be used also using the region of interest spectral information for stating the drafting of citrus influences of plant crown as training sample
The sample of the region of interest of same plant is further carried out random combine, it is combined into a new sample.Assuming that at one
N region of interest is drawn on plant, that just hasA sample can be trained, and be further increased
The data volume of sample.
Individual plant and influences of plant crown growing way in panorama high spectrum image are indicated referring to Fig. 4 B and 4C, infect yellow twig
Citrus plant can be all spread in whole plant, and the plant for lesser extent of catching an illness may only have part canopy branches and leaves to show
Therefore symptom by drawing region of interesting extraction canopy spectrum information, draws not direct whole tree when canopy region of interest
Canopy extracts, and can draw 18 region of interest to each tree as shown in Figure 6, calculate the spectral information of each region of interest, obtain
The spectral information of 18 region of interest, then each region of interest is carried out random combine averaging, available 310762 light
Compose sample.
In addition, citrus fruit crown canopy spectrum used in training must carry out ground validation, it is ensured that training sample does not occur
Mistake, the canopy top for infecting the citrus plant of yellow twig are divided into healthy leaves, mistake occur to avoid extracting spectroscopic data, make
With mahalanobis distance method rejecting abnormalities data.
In one embodiment, SPA is a kind of forward variable selection algorithm for minimizing vector space synteny, can be mentioned
Full wave several characteristic wave bands are taken, the redundancy in original spectrum matrix is eliminated, filters out characteristic wave bands.
Steps are as follows by SPA:
(1) sample set sample number and number of wavelengths K are determined, spectrum matrix X is formedM×K；
(2) it initializes: n=1, in iteration for the first time, the optional column vector x in spectrum matrixj, it is denoted as xk(0), i.e. (k
(0)=j)；
(3) set S is defined as:Calculate separately xjTo in S to
The projection vector Px of amountj,
(4) maximum projection serial number is recorded, the projection vector by maximum projection as lower whorl iteration；
(5) superiority and inferiority for carrying out judgment models using RMSEP selects the smallest RMSEP, correspondingAnd N*As screening
Band combination out.
Wherein, vegetation index includes NDVI, NDGI, TVI, RVI, NLI, DVI etc.,
NDVI: normalized differential vegetation index: in remote sensing image, the reflected value of the reflected value of near infrared band and red spectral band it
Calculating of the difference than upper sum of the two or two wave band reflectivity.
NDGI: the green degree index of normalized difference can be used to test to different vigor vegetation forms.
TVI: conversion hysteria vegetation index is determined.
RVI: ratio vegetation index is also known as green degree, for the ratio between two channel reflectivity can preferably reflect vegetation coverage and
The difference of upgrowth situation.
NLI: non-linear vegetation index.
DVI: difference environmental vegetation index, the calculating of DVI=NIR-R or two wave band reflectivity.
Its specific calculation is as follows:
Further, if having multiple wave bands in Nir, Red, Green, Blue wave-length coverage by the wave band that SPA is screened
It is available, the different-waveband in same a wave-length coverage is updated to operation in vegetation index model, relatively with a vegetation index
Model is in the difference of healthy plant and same a band math result of infection yellow twig influences of plant crown, the preferably maximum wave band of difference
As the preferred bands of vegetation index, preferred vegetation index is obtained.
By the diagnostic result of the exportable citrus orchard remote sensing images of optimal BP neural network model, Huanglong is infected if it exists
The plant of disease, and can further obtain extent and the geographical location of above-mentioned infection yellow twig plant.
In the present embodiment, treated, and spectroscopic data uses SPA algorithms selection preferred bands；
Above-mentioned spectrum samples are carried out into class label, it is preferred to carry out wave band with SPA algorithm, as shown in fig. 7, will contain 151
The spectrum samples of wave band imported into preferred bands in SPA algorithm, and according to the different of preferred wave band number, the variation of RMSE is as schemed
Shown in 8, when preferred bands number is 54, RMSE reaches minimum value.
Improved vegetation index formula is wanted in determination, and the spectral value of preferred bands is substituted into formula and is calculated；
Such as: choose NDVI, NDGI, TVI, RVI, NLI, the vegetation indexs model such as DVI, above-mentioned vegetation index model its
Middle Red, Green, Nir respectively refer to the spectral reflectivity of some wave band in a wave-length coverage, respective wave-length coverage
Are as follows: Green:576~492nm, Red:622~760nm, Nir:761~1000nm.Preferred wave band is referred to vegetation respectively
Operation in exponential model, Fig. 9 are that all infection yellow twig plant and the respective averaged spectrum of healthy plant spectrum samples, use are above-mentioned
Averaged spectrum calculates the value of vegetation index, and replaces and calculate vegetation index with the different preferred bands in wave-length coverage, by phase
The vegetation index of the illness sample and healthy sample that calculate with wave band compares, preferably the maximum wave band of vegetation index value difference value
As the calculating wave band of the vegetation index, and it is applied in all spectrum samples.
Using above-mentioned preferred vegetation index as the characteristic value of BP neural network model training, pass through the ginseng of debugging model
Number obtains optimal BP neural network model, realizes the detection to Citrus Huanglongbing pathogen.
Radiant correction, geometric correction and the panorama high spectrum image spliced are imported into optimal BP neural network model,
It is diagnosed by model, obtains the diagnostic result such as Figure 10.The available yellow twig of diagnostic result is in illness influences of plant crown
The geography information of distribution situation and each illness plant, both it is known that the extent of infection yellow twig plant plant around it
Plant situation.
In the present embodiment, planted by SPA preferred feature wave band and healthier plant and infection yellow twig influences of plant crown
By the difference of index, preferably vegetation index is trained as the characteristic value of BP neural network, obtained discrimination model can quickly,
Accurately, the case where efficiently and nondestructively diagnosing citrus plant, reduces the monitoring cost to large area citrus orchard, reduces big
The manpower and material resources of amount.
Based on the same inventive concept, the embodiment of the invention also provides the Citrus Huanglongbing pathogen inspections based on low latitude high-spectrum remote-sensing
Device is surveyed, by the principle and the aforementioned Citrus Huanglongbing pathogen detection method based on low latitude high-spectrum remote-sensing of the solved problem of the device
It is similar, therefore the implementation of the device may refer to the implementation of preceding method, overlaps will not be repeated.
Second aspect, the present invention also provides a kind of Citrus Huanglongbing pathogen detection device based on low latitude high-spectrum remote-sensing, references
Figure 11 includes:
Module 100 is obtained, for obtaining the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing figure
Distributed intelligence and the geography information of citrus plant as including citrus plant；
Processing module 101, for being pre-processed to the high-spectrum remote sensing；
Input module 102, for being input to Citrus Huanglongbing pathogen for the low latitude high-spectrum remote sensing after pretreatment
BP neural network model；
Output module 103, the distribution results for suffering from yellow twig for exporting citrus distributed areas.
In one embodiment, the Citrus Huanglongbing pathogen BP neural network model in the input module 102, referring to Fig.1 2,
Include:
Acquisition process submodule 1021, for acquiring the high-spectrum remote sensing of a large amount of citrus influences of plant crown, to the height
Spectral remote sensing image is pre-processed；
Extraction process submodule 1022, for extracting citrus fruit tree healthy plant and infection Huanglong in high-spectrum remote sensing
The spectral information of sick influences of plant crown is analyzed and processed the spectral information, obtains spectroscopic data；
It extracts and substitutes into submodule 1023, for the spectroscopic data according to preset algorithm, to be extracted characteristic wave bands；The spy
Levying wave band includes: maximum fault information, maximal projection or maximum distance characteristic wave bands；The characteristic wave bands are substituted into preset vegetation
Exponential model obtains vegetation index；
Training generates submodule 1024, for using the vegetation index as training sample, by the parameter of debugging model,
Generate Citrus Huanglongbing pathogen BP neural network model；The training sample is determining illness blade and healthy leaves data.
The third aspect, the present invention provide a kind of Citrus Huanglongbing pathogen detection system based on low latitude high-spectrum remote-sensing, referring to figure
Described in 12, comprising: unmanned plane during flying platform 1, earth station 2 and cloud platform 3；
The unmanned plane during flying platform carries EO-1 hyperion camera, and the low latitude EO-1 hyperion for acquiring citrus distributed areas to be detected is distant
Feel image；
The earth station is navigated by water for controlling the unmanned plane during flying platform according to pre-set flight paths；
The cloud platform is connect with the unmanned plane during flying Platform communication, and the cloud platform includes:
Module is obtained, for obtaining the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images
The geography information of distributed intelligence and citrus plant including citrus plant；
Processing module, for being pre-processed to the high-spectrum remote sensing；
Input module, for being input to Citrus Huanglongbing pathogen BP for the low latitude high-spectrum remote sensing after pretreatment
Neural network model；
Output module, the distribution results for suffering from yellow twig for exporting citrus distributed areas.
It should be noted last that the above examples are only used to illustrate the technical scheme of the present invention and are not limiting.Although ginseng
It is described the invention in detail according to embodiment, those skilled in the art should understand that, to technical side of the invention
Case is modified or replaced equivalently, and without departure from the spirit and scope of technical solution of the present invention, should all be covered in the present invention
Scope of the claims in.
Claims (9)
1. a kind of Citrus Huanglongbing pathogen detection method based on low latitude high-spectrum remote-sensing characterized by comprising
Obtain the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images include the distribution of citrus plant
The geography information of information and citrus plant；
The high-spectrum remote sensing is pre-processed；
By the low latitude high-spectrum remote sensing after pretreatment, it is input to Citrus Huanglongbing pathogen BP neural network model；
Suffer from the distribution results of yellow twig in output citrus distributed areas.
2. the method as described in claim 1, which is characterized in that the generation of the Citrus Huanglongbing pathogen BP neural network model walks
Suddenly, comprising:
The high-spectrum remote sensing for acquiring a large amount of citrus influences of plant crown, pre-processes the high-spectrum remote sensing；
The spectral information for extracting citrus fruit tree healthy plant and infection yellow twig influences of plant crown in high-spectrum remote sensing, to described
Spectral information is analyzed and processed, and obtains spectroscopic data；
By the spectroscopic data according to preset algorithm, characteristic wave bands are extracted；The characteristic wave bands include: maximum fault information, maximum
Projection or maximum distance characteristic wave bands；The characteristic wave bands are substituted into preset vegetation index model, obtain vegetation index；
Citrus Huanglongbing pathogen BP neural network mould is generated by the parameter of debugging model using the vegetation index as training sample
Type；The training sample is determining illness blade and healthy leaves data.
3. method according to claim 1 or 2, which is characterized in that pre-process, wrap to the high-spectrum remote sensing
It includes:
Radiant correction, geometric correction and Panoramagram montage are carried out to the high-spectrum remote sensing.
4. method according to claim 2, which is characterized in that extract high-spectrum remote sensing in citrus fruit tree healthy plant with
The spectral information for infecting yellow twig influences of plant crown, is analyzed and processed the spectral information, obtains spectroscopic data, comprising:
Area-of-interest is drawn, the spectral information of each area-of-interest is calculated；
The spectral information of each region of interest is subjected to random combine, is averaging, obtains multiple spectrum samples；
The spectrum samples are used into mahalanobis distance method rejecting abnormalities sample, and are carried out smoothly using Savitzky-Golay algorithm
Denoising obtains spectroscopic data.
5. method according to claim 2, which is characterized in that by the spectroscopic data according to preset algorithm, extract characteristic wave
Section, comprising:
By the spectroscopic data according to SPA selection algorithm, the redundancy in original spectrum matrix is eliminated, characteristic wave is filtered out
Section；Steps are as follows by SPA:
(1) sample set sample number and number of wavelengths K are determined, spectrum matrix X is formedM×K；
(2) it initializes: n=1, in iteration for the first time, the optional column vector x in spectrum matrixj, it is denoted as xK(0), i.e., (K (0)=
j)；
(3) set S is defined as:Calculate separately xjTo vector in S
Projection vector Pxj,
(4) maximum projection serial number is recorded, the projection vector by maximum projection as lower whorl iteration；
(5) superiority and inferiority for carrying out judgment models using RMSEP selects the smallest RMSEP, correspondingAnd N*As what is filtered out
Band combination.
6. method according to claim 2, which is characterized in that the characteristic wave bands are substituted into preset vegetation index model,
Obtain vegetation index, comprising:
The characteristic wave bands are substituted into NDVI, NDGI, TVI, RVI, NLI and DVI vegetation index model calculates all infection Huanglong
Sick plant and the respective averaged spectrum of healthy plant spectrum samples；
The value of vegetation index is calculated using the averaged spectrum, and is replaced and referred to the different-waveband calculating vegetation in wave-length coverage
Number；
Compare the difference with vegetation index model in healthy plant and same a band math result of infection yellow twig influences of plant crown
Different, preferred bands of the maximum wave band of selection differences as vegetation index obtain preferred vegetation index.
7. a kind of Citrus Huanglongbing pathogen detection device based on low latitude high-spectrum remote-sensing characterized by comprising
Module is obtained, for obtaining the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images include
The distributed intelligence of citrus plant and the geography information of citrus plant；
Processing module, for being pre-processed to the high-spectrum remote sensing；
Input module, for being input to Citrus Huanglongbing pathogen BP nerve for the low latitude high-spectrum remote sensing after pretreatment
Network model；
Output module, the distribution results for suffering from yellow twig for exporting citrus distributed areas.
8. device as claimed in claim 7, which is characterized in that the Citrus Huanglongbing pathogen BP neural network mould in the input module
Type, comprising:
Acquisition process submodule, for acquiring the high-spectrum remote sensing of a large amount of citrus influences of plant crown, to the high-spectrum remote-sensing
Image is pre-processed；
Extraction process submodule, for extracting citrus fruit tree healthy plant and infection yellow twig plant hat in high-spectrum remote sensing
The spectral information of layer, is analyzed and processed the spectral information, obtains spectroscopic data；
It extracts and substitutes into submodule, for the spectroscopic data according to preset algorithm, to be extracted characteristic wave bands；The characteristic wave bands packet
It includes: maximum fault information, maximal projection or maximum distance characteristic wave bands；The characteristic wave bands are substituted into preset vegetation index mould
Type obtains vegetation index；
Training generates submodule, for generating citrus by the parameter of debugging model using the vegetation index as training sample
Yellow twig BP neural network model；The training sample is determining illness blade and healthy leaves data.
9. a kind of Citrus Huanglongbing pathogen detection system based on low latitude high-spectrum remote-sensing characterized by comprising unmanned plane during flying is flat
Platform, earth station and cloud platform；
The unmanned plane during flying platform carries EO-1 hyperion camera, acquires the low latitude high-spectrum remote-sensing figure of citrus distributed areas to be detected
Picture；
The earth station is navigated by water for controlling the unmanned plane during flying platform according to pre-set flight paths；
The cloud platform is connect with the unmanned plane during flying Platform communication, and the cloud platform includes:
Module is obtained, for obtaining the low latitude high-spectrum remote sensing of citrus distributed areas to be detected；The remote sensing images include
The distributed intelligence of citrus plant and the geography information of citrus plant；
Processing module, for being pre-processed to the high-spectrum remote sensing；
Input module, for being input to Citrus Huanglongbing pathogen BP nerve for the low latitude high-spectrum remote sensing after pretreatment
Network model；
Output module, the distribution results for suffering from yellow twig for exporting citrus distributed areas.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910075649.2A CN109613022A (en) | 2019-01-25 | 2019-01-25 | A kind of method, apparatus and system of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910075649.2A CN109613022A (en) | 2019-01-25 | 2019-01-25 | A kind of method, apparatus and system of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109613022A true CN109613022A (en) | 2019-04-12 |
Family
ID=66018265
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910075649.2A Pending CN109613022A (en) | 2019-01-25 | 2019-01-25 | A kind of method, apparatus and system of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109613022A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110763698A (en) * | 2019-10-12 | 2020-02-07 | 仲恺农业工程学院 | Hyperspectral citrus leaf disease identification method based on characteristic wavelength |
| CN110940636A (en) * | 2019-12-05 | 2020-03-31 | 华南农业大学 | Intelligent identification and forest information monitoring system for citrus forest diseases and insect pests |
| CN112836616A (en) * | 2021-01-28 | 2021-05-25 | 广东技术师范大学 | Huanglongbing early warning method and device based on environmental parameter sequence and computer equipment |
| CN113218898A (en) * | 2021-05-06 | 2021-08-06 | 塔里木大学 | Plant disease remote sensing feature extraction method based on spectral analysis |
| CN113948220A (en) * | 2021-10-29 | 2022-01-18 | 广东技术师范大学 | Anthrax pathogen infection stage detection method with pre-analysis capability |
| CN114255262A (en) * | 2021-12-15 | 2022-03-29 | 赣南师范大学 | Pest killing system and method, electronic equipment and storage medium |
| CN114720436A (en) * | 2022-01-24 | 2022-07-08 | 四川农业大学 | Agricultural product quality parameter detection method and equipment based on fluorescence hyperspectral imaging |
| CN115115955A (en) * | 2022-07-08 | 2022-09-27 | 宁波大学 | Pine wood nematode disease tree monitoring method and device based on unmanned aerial vehicle hyperspectral remote sensing |
| CN116087118A (en) * | 2023-04-06 | 2023-05-09 | 黑龙江省农业科学院农业遥感与信息研究所 | Device and system for identifying corn northern leaf blight by hyperspectral remote sensing |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014078858A1 (en) * | 2012-11-19 | 2014-05-22 | Altria Client Services Inc. | Hyperspectral imaging system for monitoring agricultural products during processing and manufacturing |
| CN104036257A (en) * | 2014-06-25 | 2014-09-10 | 华南农业大学 | Citrus vein phloem degeneration detection and classification method based on D-S theory through multi-source data fusion |
| CN104050649A (en) * | 2014-06-13 | 2014-09-17 | 北京农业信息技术研究中心 | Agricultural remote sensing system |
| CN106778888A (en) * | 2016-12-27 | 2017-05-31 | 浙江大学 | A kind of orchard pest and disease damage survey system and method based on unmanned aerial vehicle remote sensing |
| CN106915462A (en) * | 2017-02-14 | 2017-07-04 | 福建兴宇信息科技有限公司 | Forestry pests ＆ diseases intelligent identifying system based on multi-source image information |
| CN108693119A (en) * | 2018-04-20 | 2018-10-23 | 北京麦飞科技有限公司 | Pest and disease damage based on unmanned plane high-spectrum remote-sensing intelligently examines the system of beating |
| CN109146508A (en) * | 2018-07-03 | 2019-01-04 | 深圳高科新农技术有限公司 | A kind of agricultural product traceability system based on unmanned plane |
-
2019
- 2019-01-25 CN CN201910075649.2A patent/CN109613022A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014078858A1 (en) * | 2012-11-19 | 2014-05-22 | Altria Client Services Inc. | Hyperspectral imaging system for monitoring agricultural products during processing and manufacturing |
| CN104050649A (en) * | 2014-06-13 | 2014-09-17 | 北京农业信息技术研究中心 | Agricultural remote sensing system |
| CN104036257A (en) * | 2014-06-25 | 2014-09-10 | 华南农业大学 | Citrus vein phloem degeneration detection and classification method based on D-S theory through multi-source data fusion |
| CN106778888A (en) * | 2016-12-27 | 2017-05-31 | 浙江大学 | A kind of orchard pest and disease damage survey system and method based on unmanned aerial vehicle remote sensing |
| CN106915462A (en) * | 2017-02-14 | 2017-07-04 | 福建兴宇信息科技有限公司 | Forestry pests ＆ diseases intelligent identifying system based on multi-source image information |
| CN108693119A (en) * | 2018-04-20 | 2018-10-23 | 北京麦飞科技有限公司 | Pest and disease damage based on unmanned plane high-spectrum remote-sensing intelligently examines the system of beating |
| CN109146508A (en) * | 2018-07-03 | 2019-01-04 | 深圳高科新农技术有限公司 | A kind of agricultural product traceability system based on unmanned plane |
Non-Patent Citations (8)
| Title |
|---|
| 刘燕德 等: "基于高光谱成像的柑橘黄龙病无损检测", 《农业机械学报》 * |
| 刘燕德 等: "柑桔叶片黄龙病光谱特征选择及检测模型", 《农业工程学报》 * |
| 杨燕: "基于高光谱成像技术的水稻稻瘟病诊断关键技术研究", 《中国博士学位论文全文数据库》 * |
| 王静 等: "冬小麦条锈病严重度不同估算方法对比研究", 《光谱学与光谱分析》 * |
| 邓小玲等: "基于主成分分析和BP神经网络的柑橘黄龙病诊断技术", 《光子学报》 * |
| 邓清: "柑橘黄龙病高光谱快速无损诊断方法", 《中国优秀硕士学位论文全文数据库》 * |
| 马淏 等: "基于高光谱成像技术应用光谱及纹理特征识别柑橘黄龙病", 《光谱学与光谱分析》 * |
| 马淏: "光谱及高光谱成像技术在作物特征信息提取中的应用研究", 《中国博士学位论文全文数据库》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110763698A (en) * | 2019-10-12 | 2020-02-07 | 仲恺农业工程学院 | Hyperspectral citrus leaf disease identification method based on characteristic wavelength |
| CN110940636A (en) * | 2019-12-05 | 2020-03-31 | 华南农业大学 | Intelligent identification and forest information monitoring system for citrus forest diseases and insect pests |
| CN110940636B (en) * | 2019-12-05 | 2020-12-08 | 华南农业大学 | Intelligent identification and forest information monitoring system for citrus forest diseases and insect pests |
| CN112836616A (en) * | 2021-01-28 | 2021-05-25 | 广东技术师范大学 | Huanglongbing early warning method and device based on environmental parameter sequence and computer equipment |
| CN112836616B (en) * | 2021-01-28 | 2022-04-22 | 广东技术师范大学 | Huanglongbing early warning method and device based on environmental parameter sequence and computer equipment |
| CN113218898A (en) * | 2021-05-06 | 2021-08-06 | 塔里木大学 | Plant disease remote sensing feature extraction method based on spectral analysis |
| CN113948220B (en) * | 2021-10-29 | 2022-12-09 | 广东技术师范大学 | Anthrax pathogen infection stage detection method with pre-analysis capability |
| CN113948220A (en) * | 2021-10-29 | 2022-01-18 | 广东技术师范大学 | Anthrax pathogen infection stage detection method with pre-analysis capability |
| CN114255262A (en) * | 2021-12-15 | 2022-03-29 | 赣南师范大学 | Pest killing system and method, electronic equipment and storage medium |
| CN114720436A (en) * | 2022-01-24 | 2022-07-08 | 四川农业大学 | Agricultural product quality parameter detection method and equipment based on fluorescence hyperspectral imaging |
| CN115115955A (en) * | 2022-07-08 | 2022-09-27 | 宁波大学 | Pine wood nematode disease tree monitoring method and device based on unmanned aerial vehicle hyperspectral remote sensing |
| CN116087118A (en) * | 2023-04-06 | 2023-05-09 | 黑龙江省农业科学院农业遥感与信息研究所 | Device and system for identifying corn northern leaf blight by hyperspectral remote sensing |
| CN116087118B (en) * | 2023-04-06 | 2023-06-09 | 黑龙江省农业科学院农业遥感与信息研究所 | Device and system for identifying corn northern leaf blight by hyperspectral remote sensing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109613022A (en) | A kind of method, apparatus and system of low latitude high-spectrum remote-sensing detection Citrus Huanglongbing pathogen | |
| Singh et al. | A review of imaging techniques for plant disease detection | |
| Zhou et al. | Strawberry maturity classification from UAV and near-ground imaging using deep learning | |
| CN107346434A (en) | A kind of plant pest detection method based on multiple features and SVMs | |
| de Oca et al. | The AgriQ: A low-cost unmanned aerial system for precision agriculture | |
| Wang et al. | Tomato diseases recognition based on faster RCNN | |
| CN113948220B (en) | Anthrax pathogen infection stage detection method with pre-analysis capability | |
| Gupta et al. | An image processing approach for measurement of chili plant height and width under field conditions | |
| CN112712038A (en) | Method and system for monitoring wheat lodging condition based on multispectral satellite image | |
| CN113657158A (en) | Google Earth Engine-based large-scale soybean planting region extraction algorithm | |
| Li et al. | An automatic approach for detecting seedlings per hill of machine-transplanted hybrid rice utilizing machine vision | |
| Lizarazo et al. | Identification of symptoms related to potato Verticillium wilt from UAV-based multispectral imagery using an ensemble of gradient boosting machines | |
| Cao et al. | Detecting asymptomatic infections of rice bacterial leaf blight using hyperspectral imaging and 3-dimensional convolutional neural network with spectral dilated convolution | |
| Liao et al. | A hybrid CNN-LSTM model for diagnosing rice nutrient levels at the rice panicle initiation stage | |
| Martins et al. | Metbots: metabolomics robots for precision viticulture | |
| Ariza-Sentís et al. | BBR: An open-source standard workflow based on biophysical crop parameters for automatic Botrytis cinerea assessment in vineyards | |
| Tang et al. | Chlorophyll detector development based on snapshot-mosaic multispectral image sensing and field wheat canopy processing | |
| Aleynikov et al. | Automated diseases detection of plant diseases in space greenhouses | |
| CN112164041A (en) | Automatic diagnosis and treatment system and method for huanglongbing based on multi-scale deep neural network | |
| Yang et al. | Feature extraction of cotton plant height based on DSM difference method | |
| Widiyanto et al. | Monitoring the growth of tomatoes in real time with deep learning-based image segmentation | |
| Nascimento et al. | A soybean seedlings dataset for soil condition and genotype classification | |
| James et al. | CitDet: A Benchmark Dataset for Citrus Fruit Detection | |
| Kifle | IoT and ML-Based precision agriculture: A case of Rwanda coffee | |
| Petrellis | A Windows Phone Application for Plant Disease Diagnosis. |
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: 20190412 |
|
| RJ01 | Rejection of invention patent application after publication |