CN111767863B - Remote sensing identification method for winter wheat scab based on near-earth hyperspectral technology - Google Patents
Remote sensing identification method for winter wheat scab based on near-earth hyperspectral technology Download PDFInfo
- Publication number
- CN111767863B CN111767863B CN202010616580.2A CN202010616580A CN111767863B CN 111767863 B CN111767863 B CN 111767863B CN 202010616580 A CN202010616580 A CN 202010616580A CN 111767863 B CN111767863 B CN 111767863B
- Authority
- CN
- China
- Prior art keywords
- winter wheat
- scab
- sample
- wavelet
- severity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 241000209140 Triticum Species 0.000 title claims abstract description 222
- 235000021307 Triticum Nutrition 0.000 title claims abstract description 221
- 206010039509 Scab Diseases 0.000 title claims abstract description 109
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000005516 engineering process Methods 0.000 title claims abstract description 17
- 238000001228 spectrum Methods 0.000 claims abstract description 63
- 238000012549 training Methods 0.000 claims abstract description 32
- 238000012216 screening Methods 0.000 claims abstract description 26
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims description 71
- 230000003595 spectral effect Effects 0.000 claims description 65
- 201000010099 disease Diseases 0.000 claims description 61
- 238000002310 reflectometry Methods 0.000 claims description 36
- 241000495841 Oenanthe oenanthe Species 0.000 claims description 28
- 241001024327 Oenanthe <Aves> Species 0.000 claims description 23
- 235000013339 cereals Nutrition 0.000 claims description 16
- 238000010219 correlation analysis Methods 0.000 claims description 14
- 230000009466 transformation Effects 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000004422 calculation algorithm Methods 0.000 claims description 10
- 230000005855 radiation Effects 0.000 claims description 8
- 230000036541 health Effects 0.000 claims description 7
- 230000003902 lesion Effects 0.000 claims description 5
- 238000000985 reflectance spectrum Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract description 2
- 239000000523 sample Substances 0.000 description 81
- 238000004458 analytical method Methods 0.000 description 6
- 238000005259 measurement Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 4
- 238000002474 experimental method Methods 0.000 description 4
- 230000008859 change Effects 0.000 description 3
- 210000005069 ears Anatomy 0.000 description 3
- 239000000284 extract Substances 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000000701 chemical imaging Methods 0.000 description 2
- 239000004744 fabric Substances 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000010832 independent-sample T-test Methods 0.000 description 2
- 230000001575 pathological effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000004611 spectroscopical analysis Methods 0.000 description 2
- 241000607479 Yersinia pestis Species 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 208000032625 disorder of ear Diseases 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 239000003337 fertilizer Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 208000021245 head disease Diseases 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 239000000049 pigment Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000011425 standardization method Methods 0.000 description 1
- 238000012795 verification 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
- G06V20/188—Vegetation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
- G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The invention relates to a method for identifying winter wheat scab based on a near-earth hyperspectral technology, which solves the defect of low accuracy of identifying the severity degree of winter wheat scab from an upright angle compared with the prior art. The invention comprises the following steps: obtaining hyperspectral data; calculating the severity of scab; screening original spectrum band characteristics; screening the optimal wavelet characteristics; constructing a winter wheat head blight identification model; training a winter wheat head blight recognition model; obtaining the identification result of the wheat head blight of winter wheat. The invention not only realizes the identification of the severity of the scab of the winter wheat under the vertical angle, but also greatly improves the accuracy of the identification of the severity of the scab of the winter wheat.
Description
Technical Field
The invention relates to the technical field of winter wheat head blight identification, in particular to a winter wheat head blight remote sensing identification method based on a near-field hyperspectral technology.
Background
The non-imaging hyperspectral technology has the characteristics of high spectral resolution, multiple wave bands, rich information and the like, and the non-imaging hyperspectral data collected near ground is less influenced by the atmosphere and the external environment, has high signal-to-noise ratio and is more similar to the real spectrum of the ground object. The distinction between healthy wheat and wheat infested by pests can be manifested by differences in their spectra. The spectral characteristics of wheat are determined by both its chemistry and morphology, and change as the growth conditions, health, and growth conditions of wheat plants change. Germ-infected wheat, its tissue pigment, moisture content, and internal structure are subject to changes that can be studied as a function, so spectroscopic analysis can be used to monitor the health condition of wheat.
In the prior art, liang et al (2015) identified scab of wheat grain by spectroscopic analysis and image processing using hyperspectral imaging techniques. The established linear discriminant analysis, support vector machine and BP (back propagation) neural network model have good identification effect on scab infected seeds, and the accuracy is above 90%. Ewa et al (2018) established classification models based on hyperspectral image texture parameters to identify infected kernels, classifying kernels located on the ventral side with 100% accuracy. Liu Shuang (2019) extracts the hyperspectral image information of the wheat grains and establishes an identification model, and the hyperspectral image technology is utilized to realize the efficient and accurate identification of the wheat scab grains. Jin et al (2018) classify hyperspectral pixels of healthy spikes and scab infected spikes in the field by using a deep neural network, and the classification accuracy reaches 74.3%. Whetton et al (2018) measured wheat scab in the field on-line using a hyperspectral imager and used RGB photographs collected from plots to assess crop disease incidence (ratio of infected to healthy spikes). This study achieved good accuracy (82%), but the severity of the disease was not determined. Previous studies on scab have focused on classifying diseased wheat kernels by hyperspectral imaging or identifying diseased ears under laboratory conditions, either as a classification of health and disease, and not achieving the ideal recognition effect, or as a classification of disease severity. Therefore, how to realize classification and identification of the severity of wheat scab from an upright angle by utilizing a hyperspectral technology, and improvement of identification accuracy has become an urgent technical problem to be solved.
Disclosure of Invention
The invention aims to solve the defect of low accuracy in identifying the severe degree of the scab of winter wheat from an upright angle in the prior art, and provides a remote sensing identification method for the scab of winter wheat based on a near-field hyperspectral technology.
In order to achieve the above object, the technical scheme of the present invention is as follows:
a remote sensing identification method for winter wheat scab based on a near-earth hyperspectral technology comprises the following steps:
acquisition of hyperspectral data: reflectance spectra were obtained for winter wheat single spike samples at an upright angle.
Calculation of the severity of scab: and (3) solving the disease severity of each winter wheat single spike sample by adopting a disease severity calculation formula, and classifying the winter wheat single spike samples according to the disease severity.
Screening of original spectral band characteristics: and carrying out correlation analysis on the reflectivity spectrum and the disease severity of the winter wheat single spike sample, and screening out the original spectrum band characteristics with high correlation.
Screening of optimal wavelet characteristics: and carrying out continuous wavelet transformation on the reflectivity spectrum of the winter wheat single spike sample to obtain corresponding wavelet characteristics, and screening the wavelet characteristics which have high correlation with the severity of the illness and obvious difference between severity classes from the corresponding wavelet characteristics as optimal wavelet characteristics.
Building a winter wheat head blight identification model: and constructing a winter wheat head blight identification model by utilizing an SVM algorithm.
Training of winter wheat head blight recognition model: and training the winter wheat head blight recognition model by using the screened original spectrum band characteristics and the optimal wavelet characteristics as input variables of the winter wheat head blight recognition model.
Obtaining a winter wheat head blight recognition result: and inputting the screened original spectrum band characteristics and optimal wavelet characteristics to a winter wheat head blight sensitive characteristic set, so as to obtain a winter wheat head blight identification result.
Further, the calculation of the severity of the scab disease comprises the following steps:
the severity of scab was calculated for each winter wheat single spike sample using the following formula:
y＝a/b*(1+c/a)
wherein a represents the number of diseased wheat grains in a single wheat ear, b represents the number of all wheat grains in the single wheat ear, c represents the number of wheat grains with serious shrinkage and lesion degree, and y represents the severity of scab disease of winter wheat ears.
Winter wheat Shan Suiyang is classified into healthy or infected type according to the disease severity of each winter wheat single spike sample, or into healthy, mild and severe type.
Further, the screening of the original spectrum band characteristics comprises the following steps:
according to the ratio of the spectral radiation value in the single-spike sample reflectivity spectrum of winter wheat to the corresponding wave Duan Baiban spectral radiation value, obtaining a winter wheat spike reflectivity spectrum curve, and obtaining the spectral reflectivity of each wave band between 350nm and 950 nm.
Carrying out correlation analysis on the spectral reflectance of each wave band of each winter wheat single spike sample between 350nm and 950nm and the disease severity corresponding to the spectral reflectance of each winter wheat single spike sample by adopting the following formula, and obtaining a correlation coefficient R1 between the spectral reflectance and the disease severity;
wherein x is i And y i Representing each wave band between 350nm and 950nm of winter wheat single spike sampleSpectral reflectance and scab severity of winter wheat single spike samples,represents the average value of the spectral reflectivities of all samples in the same band,/-)>The average value of the disease severity of the sample is represented by n, the number of samples is represented by n, the absolute value of R1 is between 0 and 1, the closer the value of R1 is to 1, the stronger the correlation degree between two variables is represented, and the closer to 0 is the weaker the correlation degree between the two variables is.
And selecting two characteristic wave bands 491nm and 699nm with the largest correlation with the disease severity as the original spectral band characteristics of the single spike sample reflectivity spectrum of winter wheat in two different wave bands of 400nm-500nm and 600nm-700 nm.
Further, the screening of the optimal wavelet characteristics includes the following steps:
adopting a Gaussian function as a mother wavelet function, carrying out continuous wavelet transformation on the reflectivity spectrum of a winter wheat single spike sample by using the following formula to obtain wavelet energy coefficients W of 601 different wavebands and 10 different scales between 350nm and 950nm f (a，b)：
Wherein ψ (λ) represents a mother wavelet function, f (λ) represents a reflectance spectrum of a winter wheat single spike sample, and a and b represent a decomposition scale and a time shift factor of a wavelet, respectively; 601 is obtained from each winter wheat single spike sample Wave band ×10 Dimension of Wavelet energy coefficients of different wavebands and scales.
The wavelet energy coefficient W is calculated by the following formula f Performing correlation analysis on the disease severity and the (a, b) to obtain a correlation coefficient R2, and screening out a wavelet feature set I with the scab correlation of more than 0.65 by using a threshold method;
wherein z is i And y i Respectively representing two variables of wavelet energy coefficient of a certain wave band scale and scab disease severity of winter wheat single spike sample,mean value of wavelet energy coefficients representing all samples of the same band and the same scale, +.>An average value of the severity of the disease of the sample, n represents the number of the samples, and the absolute value of R2 is between 0 and 1; the closer the absolute value of R2 is to 1, the stronger the correlation between the two variables, whereas the closer to 0, the weaker the correlation between the two variables.
The method comprises the steps of screening wavelet feature set two which shows significant differences in three types of combinations by adopting an independent T test function in spss software for the two-by-two combination of healthy, mild and severe winter wheat spike samples;
and intersecting the wavelet feature set I and the wavelet feature set II to obtain the optimal wavelet feature.
Further, a decision function adopted by the construction of the winter wheat head blight identification model by using the SVM algorithm is as follows:
wherein alpha is i I=1,..n is Lagrange multiplier corresponding to the training sample; e, e i The method comprises the steps of representing a winter wheat scab sensitive characteristic set obtained by combining original spectral band characteristics and wavelet characteristics of a winter wheat ear sample; f (f) i ∈{-1,1}；K(e i E) is a radial basis function and meets the Mercer condition; b is a threshold; the f (e) value was used to determine the severity of the condition predicted for winter wheat head blight.
Further, the training of the winter wheat head blight recognition model comprises the following steps:
training a two-class winter wheat head blight identification model:
all winter wheat single spike sample data are classified into two types of health and disease, 73 samples are randomly selected as a training sample set D, and D= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) -any sample e) i Is an n-dimensional vector, f i ∈{-1,1}；f i I.e. the corresponding classification result, f i 1 represents healthy winter wheat, f i A value equal to-1 indicates diseased winter wheat.
Combining the original spectral band characteristics and wavelet characteristics of a single spike sample of winter wheat to obtain a scab sensitive characteristic set of winter wheat, wherein the scab sensitive characteristic set of winter wheat is used as a sample variable e i 。
Training sample set d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) And) inputting the model into a winter wheat head blight recognition model to obtain a trained two-class winter wheat head scab recognition model.
Training a three-classification winter wheat head blight identification model:
dividing the severity of scab of winter wheat ears into three types of healthy, mild and severe samples, and respectively constructing a scab recognition model classifier svm for two types of winter wheat ears in each two types of the three types of samples i-j Obtaining SVM Health-mild 、SVM Health-severe And SVM Mild-severe 。
All winter wheat single spike sample data are classified into three categories, healthy, mild and severe, wherein 73 samples were randomly selected as training sample set D, d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) -any sample e) i Is an n-dimensional vector, f i ∈{-1,1}；f i I.e. the corresponding classification result, f i Equal to-1 represents class i, f i A value equal to 1 represents class j.
Winter wheatCombining the original spectral characteristics and the optimal wavelet characteristics of a single spike sample to obtain a winter wheat scab sensitive characteristic set as a sample variable e i 。
Training sample set d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) And (3) respectively inputting the samples into three classifiers, wherein each classifier counts the classification of the samples, the classification of the samples obtains the most votes in a certain category, and the samples are judged to be of the category, so that a trained three-classification winter wheat head blight recognition model is finally obtained.
Advantageous effects
According to the winter wheat scab identification method based on the near-to-ground hyperspectral technology, the original spectral band characteristics and the optimal wavelet characteristics are combined to obtain the winter wheat scab sensitive characteristic set, and the winter wheat scab sensitive characteristic set is utilized to construct a winter wheat scab identification model, so that the winter wheat scab severity is identified under the vertical angle, and the accuracy of winter wheat scab severity identification is greatly improved. The accuracy of the winter wheat scab identification model constructed by utilizing the winter wheat scab sensitive feature set is superior to that of a model constructed by independently using wavelet features or original spectral features, and the estimated severity of the affected wheat ears is more in line with the actual illness state of the wheat ears, wherein the overall accuracy of the health-disease two-classification is 91.11%, the overall accuracy of the three-classification is 75.56%, and Kappa coefficients are 0.82 and 0.63 respectively.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a graph of reflectance spectra of healthy, mild and severe wheat ears;
FIG. 3 is a plot of severity of disease versus spectral reflectance;
FIG. 4 is a flow chart of a method of screening for optimal wavelet features;
fig. 5 is a graph showing the overall accuracy and kappa coefficient of two and three classifications identified by the existing disease severity calculation formula (original severity) and the disease severity calculation formula (novel severity) in the present invention in an identification model constructed based on the original spectral features, wavelet features and the winter wheat scab sensitive feature set, respectively.
Detailed Description
For a further understanding and appreciation of the structural features and advantages achieved by the present invention, the following description is provided in connection with the accompanying drawings, which are presently preferred embodiments and are incorporated in the accompanying drawings, in which:
the remote sensing identification method for winter wheat scab based on the near-earth hyperspectral technology shown in fig. 1 comprises the following steps:
first, acquiring hyperspectral data: and obtaining the reflectivity spectrum of the winter wheat single spike sample under the vertical angle, wherein the reflectivity spectrum is hyperspectral data.
(1) The winter wheat hyperspectral data in the invention are collected by using a ASD Field Spec Pro FR (350-2500 nm) spectrometer. The spectrum resolution of the spectrometer is 3nm in the range of 350 nm-1000 nm, 10nm in the range of 1000-2500 nm, and the spectrum sampling interval is 1nm. The method is characterized in that the method is used for selecting the wheat in the grouting period and the mature period as objects, a piece of black cloth with the thickness of 1m multiplied by 1m is used for measuring, a small hole is reserved in the middle of the black cloth, wheat ears are penetrated from the lower part to the upper part of the small hole, a probe of a spectrometer is placed at the top of the wheat ears, and the spectrum at an upright angle is measured. The probe was placed directly above the ear for spectroscopic measurements. Each wheat ear was measured 10 times at each angle and the average of the 10 measurements was recorded. Spectral measurements of all samples were performed between 10:00 and 14:00 (Beijing local time), under sunny and cloudless breeze weather conditions. Calculating to obtain a wheat head reflectivity spectrum curve through the ratio of the spectrum radiation value of the wheat head to the spectrum radiation value of the white board, wherein the formula is as follows:
(2) The data acquisition time, the growing period, the geographical area and the wheat variety of all samples in the winter wheat head blight severity identification experiment are different. In order to eliminate spectral differences due to differences in the background of the data acquisition, the raw spectral data is normalized prior to data analysis. Spectra of all samples from zone AAs a standard spectrum. The spectral average of the healthy samples in the A region is divided by the spectral average of the healthy samples in the B and C regions, respectively, to obtain two ratio curves. These two ratio curves reflect the spectral differences between the regions. Calculating the Ratio of a certain wavelength by using the following method i :
Wherein: i denotes the wavelength, Σ denotes the sum of the reflectivities of all healthy samples in the region at wavelength i, ref denotes the reflectivities, A, B and C represent Guo Hezhen, white lake town and Shu Cheng county experimental regions, respectively; multiplying the original spectrum data of each sample acquired in the area B and the area C by the corresponding ratio curve to obtain normalized spectrum data, wherein a normalized calculation formula under the wavelength is as follows:
Ref'(B/C) i ＝Ref(B/C) i ×Ratio i
wherein:indicating the reflectivity of the sample spectrum in region B or C at the wavelength, ref' (B/C) i Representing the normalized reflectance.
The reliability of the spectrum identification result can be improved by adopting a spectrum standardization method. In this process, the reflectance of healthy samples of zone a is used as the standard reflectance, and the sample reflectances of both zone B and zone C are adjusted to the same level as the reflectance of zone a. Thus, the background effects of the spectral data of the three different sets of regions are largely eliminated. Since the same ratio curve was used for adjustment for each of the spectral data in the B and C regions, the relative differences between the healthy and affected sample spectra did not change.
Second, calculating the severity of scab: and (3) solving the disease severity of each winter wheat single spike sample by adopting a disease severity calculation formula, and classifying the winter wheat single spike samples according to the disease severity.
(1) According to the technical specification of wheat scab prediction (GB/T15796-2011) published and implemented in 2011, the existing calculation formula of the severity of the illness is as follows: severity = number of sick spikes/total number of spikes. Considering that disease kernels are slightly and heavily infected, the formula cannot fully reflect and distinguish the actual severity of the wheat head disease.
In order to embody the difference of pathological degrees among the infected spikelets and strengthen the actual disease of the pathological degrees of the wheat ears, the invention calculates the disease severity of scab of each winter wheat single spike sample by adopting the following formula:
y＝a/b*(1+c/a)
wherein a represents the number of diseased wheat grains in a single wheat ear, b represents the number of all wheat grains in the single wheat ear, c represents the number of wheat grains with serious shrinkage and lesion degree, and y represents the severity of scab disease of winter wheat ears.
The scab disease severity calculation formula provided by the invention improves the correlation between the disease severity of winter wheat ears and the spectral reflectivities of different disease severity of winter wheat ears, and improves the recognition accuracy of the winter wheat ear scab recognition model constructed by the invention to a certain extent.
(2) The winter wheat single ear samples are classified according to the disease severity of each winter wheat single ear sample, and the winter wheat Shan Suiyang is classified into healthy or infected type, or into healthy, mild and severe type.
During winter wheat grouting, 118 winter wheat ear samples are selected, each winter wheat ear sample is visually judged, and the number a of diseased wheat grains in a single wheat ear, the number b of all wheat grains in the single wheat ear and the number c of wheat grains with serious shrinkage and pathological changes are counted. The severity of scab y was calculated for each winter wheat ear using the formula y=a/b (1+c/a). 118 winter wheat scab is classified into a healthy (y < 1/5) and an infected (y.gtoreq.1/5) group, or into a healthy (y=0), a mild (0 < y < 1/2) group and a severe (y.gtoreq.1/2) group. Experiments were carried out at 5.3.2019 in the city of the co-fertilizer, lu Jiang county Guo Hezhen (31℃29'N,117℃13' E). In the experiment, the measurement of the vertical and horizontal hyperspectral data of wheat ears is selected in the wheat filling period, and the statistics of the data quantity of the wheat ear sample obtained by investigation and measurement on the disease severity of the wheat ears and the severity of grain diseases are shown in table 1. The experiment provides data guarantee for the establishment of a novel disease severity and scab identification model on the wheat ear scale.
TABLE 1 number of samples for wheat ear model construction
The spectrum curves of the upstanding head blight samples of the diseases with different degrees are obviously different from the spectrum curves of the wheat with different degrees as shown in figure 2, and the spectrum reflectivity is integrally increased along with the aggravation of the disease. The correlation coefficient curves between the existing disease severity and the disease severity in the invention and the spectral reflectivities of the winter wheat single spike samples are shown in figure 3, and the correlation between the spectral band reflectivities and the disease severity in the invention is higher than the correlation between the existing disease severity.
Thirdly, screening original spectrum band characteristics: and carrying out correlation analysis on the reflectivity spectrum and the disease severity of the winter wheat single spike sample, and screening out the original spectrum band characteristics with high correlation. The method comprises the following specific steps:
(1) According to the ratio of the spectral radiation value in the single-spike sample reflectivity spectrum of winter wheat to the corresponding wave Duan Baiban spectral radiation value, obtaining a winter wheat spike reflectivity spectrum curve, and obtaining the spectral reflectivity of each wave band between 350nm and 950 nm.
(2) Carrying out correlation analysis on the spectral reflectance of each band of each winter wheat single spike sample between 350nm and 950nm and the disease severity corresponding to the spectral reflectance of each winter wheat single spike sample by adopting the following formula, solving a linear relation (namely a correlation coefficient R1) between the spectral reflectance and the disease severity, and evaluating the identification capability of each original spectral band on scab;
wherein x is i And y i Respectively representing two variables of the spectral reflectance of each wave band between 350nm and 950nm of the winter wheat single spike sample and the scab severity of the winter wheat single spike sample,represents the average value of the spectral reflectivities of all samples in the same band,/-)>The average value of the disease severity of the sample is represented by n, the number of samples is represented by n, the absolute value of R1 is between 0 and 1, the closer the value of R1 is to 1, the stronger the correlation degree between two variables is represented, and the closer to 0 is the weaker the correlation degree between the two variables is.
The identification ability of the original spectral features to scab was evaluated using a correlation analysis method, the correlation analysis results of which are shown in Table 2 (R is a correlation coefficient, R 2 For determining coefficients), the novel severity and the original severity are compared with the spectral band correlation analysis results respectively, the extracted characteristic bands are found to be the same, and the correlation of the novel severity in 491nm band and 699nm band is slightly higher than the correlation coefficient of the original severity.
TABLE 2 original band characteristics
(3) In order to eliminate information redundancy of adjacent wave bands, two characteristic wave bands 491nm and 699nm with the greatest correlation with disease severity are selected as original spectrum wave band characteristics of the single spike sample reflectivity spectrum of winter wheat in two different wave band ranges of 400nm-500nm and 600nm-700 nm.
Fourth, screening the optimal wavelet characteristics: and carrying out continuous wavelet transformation on the reflectivity spectrum of the winter wheat single spike sample to obtain corresponding wavelet characteristics, and screening the wavelet characteristics which have high correlation with the severity of the illness and obvious difference between severity classes from the corresponding wavelet characteristics as optimal wavelet characteristics.
Wavelet transform is an effective time-frequency analysis method following fourier transform, and is divided into continuous wavelet transform and discrete wavelet transform (continuous wavelet transform, CWT), and has been widely used in various fields of scientific research such as image processing and pattern recognition. The remote sensing field mainly uses discrete wavelet transform to perform data filtering and denoising on image data, and continuous wavelet transform is mainly applied to signal analysis. In crop spectral signal analysis, continuous wavelet transformation correlates the original spectral curve with a gaussian function at different locations and scales to generate a series of continuous wavelet energy coefficients that can be extracted for weak information in different disease spectra. The continuous wavelet transformation is an important processing method capable of localizing the frequency domain and the time domain of a signal at the same time, utilizes a mother wavelet function (ψ (lambda)) to refine the signal in different scales and positions, converts the ear spectral data into wavelet energy coefficients composed of positions and scales, and can extract sensitive information in different disease spectrums.
The invention utilizes continuous wavelet transformation to extract detail information of the single spike spectrum of winter wheat, and is used for identifying scab of winter wheat spike. Firstly, continuous wavelet transformation is carried out on wheat ear spectral data at different positions and different scales to obtain wavelet energy coefficients, correlation analysis is utilized to obtain correlation of the wavelet coefficients and severity, wavelet characteristics (R > 0.65) sensitive to wheat scab are screened out, health is carried out on the screened characteristics, independent sample T test is carried out between the light and the severity, and finally, the wavelet characteristics which are sensitive to the scab and have obvious differences between different severity classes (P-value < 0.05) are obtained as optimal wavelet characteristics.
As shown in fig. 4, the specific steps of the optimal wavelet feature screening are:
(1) Performing continuous wavelet transformation on the reflectivity spectrum of each winter wheat single spike sample to obtain a wavelet coefficient diagram, adopting a Gaussian function as a mother wavelet function, and performing continuous wavelet transformation on the reflectivity spectrum of the winter wheat single spike sample by using the following formula to obtain 601 different wave bands and 10 different-scale wavelet energy coefficients W between 350nm and 950nm f (a，b)：
Wherein ψ (λ) represents a mother wavelet function, f (λ) represents a reflectance spectrum of a winter wheat single spike sample, and a and b represent a decomposition scale and a time shift factor of a wavelet, respectively; each winter wheat single spike sample can be obtained as 601 Wave band ×10 Dimension of Wavelet energy coefficients of different wavebands and scales.
(2) The wavelet energy coefficient W is calculated by the following formula f Performing correlation analysis on the disease severity and the (a, b) to obtain a correlation coefficient R2, and screening out a wavelet feature set I with the scab correlation of more than 0.65 by using a threshold method;
wherein z is i And y i Respectively representing two variables of wavelet energy coefficient of a certain wave band scale and scab disease severity of winter wheat single spike sample,mean value of wavelet energy coefficients representing all samples of the same band and the same scale, +.>An average value of the severity of the disease of the sample, n represents the number of the samples, and the absolute value of R2 is between 0 and 1; the closer the absolute value of R2 is to 1, the stronger the correlation between the two variables, whereas the closer to 0, the weaker the correlation between the two variables.
(3) And (3) screening wavelet feature sets II which show significant differences (P-value < 0.05) in three types of combinations by adopting an independent T test function in spss software for the two-by-two combinations of healthy, mild and severe winter wheat spike samples.
(4) And intersecting the wavelet feature set I and the wavelet feature set II to obtain the optimal wavelet feature. The optimal wavelet features are wavelet features that are sensitive to scab and have significant differences between the severity of three scab.
After the original spectrum data is subjected to continuous wavelet transformation, 6 sensitive characteristic variables at different scales and different wave band positions are finally obtained after correlation analysis and independent sample T test, namely the characteristics with high correlation to disease severity and obvious difference among different severity classes are obtained, as shown in a table 3 (R in the table 2 To determine coefficients). Comparing the wavelet characteristics extracted based on the novel severity and the original severity with the determination coefficients of the severity and the band position can know that the decomposition scale and the band position of the wavelet characteristics extracted based on the two severity and sensitive to diseases are consistent, and the determination coefficients of the novel severity and the wavelet characteristics are higher than the determination coefficients of the original severity and the wavelet characteristics.
TABLE 3 wavelet characteristics
Fifthly, constructing a winter wheat head blight recognition model: the method for constructing the winter wheat head blight identification model by utilizing the SVM algorithm comprises the following specific steps:
(1) The support vector machine decision function is:
wherein alpha is i Is Lagrange multiplier, e i Representing training sample variables, f i E { -1,1}, b is a threshold, e
Representing the test input variable.
(2) Because the selected winter wheat ear training sample is non-linearly separable by non-linearityLinear transformation v=θ (e i ) Mapping the samples to a high dimensional space, bringing in a new support vector machine decision function available in the above equation:
when the kernel function meets the Mercer condition, a proper inner product function is adopted to obtain a classification function of a high-dimensional space, so that the linear separable inner product function is realized, and the inner product function is:
(3) Substituting the model to replace to obtain a decision function of the winter wheat head blight recognition model:
wherein alpha is i I=1,..n is Lagrange multiplier corresponding to the training sample; e, e i The method comprises the steps of representing a winter wheat scab sensitive characteristic set obtained by combining original spectral band characteristics and wavelet characteristics of a winter wheat ear sample; f (f) i ∈{-1,1}；K(e i E) is a radial basis function and meets the Mercer condition; b is a threshold; the f (e) value was used to determine the predicted severity of winter wheat ear disease.
Sixth, training a winter wheat head blight recognition model: and training the model by using the screened original spectrum band characteristics and the optimal wavelet characteristics as input variables of the winter wheat head blight identification model. The method comprises the following specific steps:
(1) Training a two-class winter wheat head blight identification model:
dividing all winter wheat ear sample data into two types of health and disease, randomly selecting 73 samples as a training sample set D, wherein D= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) -x) where any sample x i Is an n-dimensional vector, y i ∈{-1,1}；y i Namely, the corresponding classification result is y i 1 represents healthy winter wheat, y i A value equal to-1 indicates diseased winter wheat.
Combining the original spectral band characteristics and wavelet characteristics of the screened single spike sample of winter wheat to obtain a scab sensitive characteristic set of winter wheat, wherein the scab sensitive characteristic set of winter wheat is used as a sample variable e i 。
Training sample set d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) And) inputting the model into a winter wheat head blight recognition model to obtain a trained two-class winter wheat head scab recognition model.
(2) Training a three-classification winter wheat head blight identification model:
dividing the severity of scab of winter wheat ears into three types of healthy, mild and severe samples, and respectively constructing scab recognition model classifiers svm of two types of winter wheat ears for each two types of the three types of winter wheat ear samples i-j A total of 3 classifiers, which are SVM Health-mild 、SVM Health-severe And SVM Mild-severe 。
All winter wheat ear sample data were classified into three categories, healthy, mild and severe, with 73 samples randomly selected as training sample set D, d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) -any sample e) i Is an n-dimensional vector, f i ∈{-1,1}；y i Namely, the corresponding classification result is y i Equal to-1 represents class i, y i A value equal to 1 represents class j.
Combining the original spectral characteristics of the winter wheat single spike sample and the optimal wavelet characteristics together to obtain a winter wheat scab sensitive characteristic set as a sample variable e i ；
Training sample set d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) Respectively inputting into three classifiers, each of which performs statistics on classification of the sample,and if the number of tickets obtained by classifying the sample in a certain class is the largest, judging the sample to the class, and finally obtaining the trained three-classification winter wheat head blight recognition model.
Screening out the original spectral band characteristics and wavelet characteristics of a single spike sample of winter wheat from a test winter wheat sample to form a winter wheat scab sensitive characteristic set; the method comprises the steps of inputting a winter wheat scab sensitive characteristic set into a trained winter wheat head scab recognition model, and outputting a scab severity recognition result by the trained winter wheat head scab recognition model, wherein the result is used for accurately judging whether the wheat scab disease is in the healthy, mild and severe states.
Seventh, obtaining a wheat head scab identification result of winter wheat: and inputting the winter wheat scab sensitive feature set composed of the original spectral band features and the optimal wavelet features into a winter wheat head scab recognition model to obtain a winter wheat head scab recognition result. The method comprises the following specific steps:
(1) And obtaining the original spectrum band characteristics and the optimal wavelet characteristics of the winter wheat sample to be identified by adopting the steps.
(2) And inputting the winter wheat scab sensitive feature set composed of the original spectral band features and the optimal wavelet features into a winter wheat head scab recognition model to obtain a winter wheat head scab recognition result.
As shown in fig. 5, an SVM algorithm is adopted to construct a winter wheat scab recognition model based on original spectral characteristics, the model classification precision based on novel severity is higher than the model precision of original severity, the precision of second classification is higher than the precision of third classification, the overall precision of second classification based on the novel severity model reaches 88.89%, the overall precision of third classification reaches 68.89%, and Kappa coefficients are respectively 0.53 and 0.78. The winter wheat scab recognition model based on the optimal wavelet features is constructed by adopting an SVM algorithm, the model classification precision based on the novel severity is higher than the model precision of the original severity, the precision of the second classification is higher than the precision of the third classification, the total precision of the second classification based on the novel severity model reaches 91.11%, the total precision of the third classification reaches 71.11%, and Kappa coefficients are respectively 0.82 and 0.57. And constructing a winter wheat scab recognition model based on a winter wheat scab sensitive feature set by adopting an SVM algorithm, wherein the recognition result, the overall precision and the Kappa coefficient of the model are shown in tables 4 and 5. It can be seen from the table that the model classification accuracy based on the novel severity is higher than the model accuracy of the original severity, the accuracy of the second classification is higher than the accuracy of the third classification, wherein the overall accuracy of the second classification based on the novel severity model reaches 91.11%, the overall accuracy of the third classification reaches 75.56%, and Kappa coefficients are respectively 0.82 and 0.63.
Table 4 recognition accuracy versus analysis of different models
Table 5 recognition accuracy versus analysis of different models
Under the condition that a model is built by the SVM algorithm of the same algorithm, wavelet features are superior to the original spectrum band features in the identification capability of scab, because the continuous wavelet transformation can extract hidden detail information in the spectrum, the information has higher correlation with severity than the original spectrum, and the scab features can be expressed more. The correlation between the novel severity and the spectral characteristics is higher than the original severity in the prior art, because the evaluation method of the original severity only counts the ratio of the number of diseased spikes to the total spike number, and does not consider the difference of the severity of single spikes. For example, if two wheat ears with the same number of ears and the same number of diseased ears, wherein the diseased ear particles of one wheat ear are severe lesions and the other wheat ear is mild lesions, the hyperspectral data of the two wheat ears have great difference. In consideration of the point, the invention adjusts the ratio of the number of severely diseased kernels to the total diseased kernels to the original severity, and enhances the spectrum difference of wheat ears with different severely diseased kernels. And tracking, comparing and observing the original severity category label of the verification sample and the classified result label to find that the classification result of the misclassified sample is the same as the novel severity category of the sample. The results show that the novel severity can more accurately express the actual illness state of the single wheat ears.
The foregoing has shown and described the basic principles, principal features and advantages of the invention. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, and that the above embodiments and descriptions are merely illustrative of the principles of the present invention, and various changes and modifications may be made therein without departing from the spirit and scope of the invention, which is defined by the appended claims. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (6)
1. The remote sensing identification method for the winter wheat scab based on the near-earth hyperspectral technology is characterized by comprising the following steps of:
11 Acquisition of hyperspectral data: obtaining a reflectivity spectrum of a winter wheat single spike sample under an upright angle;
12 Calculation of the severity of the scab: the disease severity of each winter wheat single spike sample is obtained by adopting a disease severity calculation formula, and the winter wheat single spike samples are classified according to the disease severity;
13 Screening of original spectral band features: carrying out correlation analysis on the reflectivity spectrum and the disease severity of a winter wheat single spike sample, and screening out the original spectrum band characteristics with high correlation;
14 Screening of optimal wavelet features): carrying out continuous wavelet transformation on the reflectivity spectrum of a winter wheat single spike sample to obtain corresponding wavelet characteristics, and screening out the wavelet characteristics which have high correlation with the severity of the illness and obvious difference between severity classes from the corresponding wavelet characteristics as optimal wavelet characteristics;
15 Construction of winter wheat head blight identification model: constructing a winter wheat head scab recognition model by using an SVM algorithm;
16 Training of winter wheat head blight recognition model: training the winter wheat head blight recognition model by using the screened original spectrum band characteristics and the optimal wavelet characteristics as input variables of the winter wheat head blight recognition model;
17 Obtaining the identification result of the wheat head blight of winter wheat: and inputting the screened original spectrum band characteristics and optimal wavelet characteristics to a winter wheat head blight sensitive characteristic set, so as to obtain a winter wheat head blight identification result.
2. The remote sensing identification method for winter wheat scab based on near-field hyperspectral technology as claimed in claim 1, wherein the calculation of the severity of the scab condition comprises the following steps:
21 Using the following formula to calculate the severity of scab for each winter wheat single spike sample:
y＝a/b*(1+c/a)
wherein a represents the number of diseased wheat grains in a single wheat ear, b represents the number of all wheat grains in the single wheat ear, c represents the number of wheat grains with serious shrinkage lesion degree, and y represents the severity of scab disease of winter wheat ears;
22 According to the severity of the illness of each single spike sample of winter wheat, the winter wheat Shan Suiyang is classified into healthy or infected type, or into healthy, mild and severe type.
3. The remote sensing identification method for winter wheat scab based on near-to-ground hyperspectral technology as claimed in claim 2, wherein the screening of the characteristics of the original spectrum band comprises the following steps:
31 According to the ratio of the spectral radiation value in the single-spike sample reflectivity spectrum of winter wheat to the corresponding wave Duan Baiban spectral radiation value, obtaining a spike reflectivity spectrum curve of winter wheat, and obtaining the spectral reflectivity of each wave band between 350nm and 950 nm;
32 Carrying out correlation analysis on the spectral reflectance of each wave band of each winter wheat single spike sample between 350nm and 950nm and the disease severity corresponding to the spectral reflectance of each winter wheat single spike sample by adopting the following formula to obtain a correlation coefficient R1 between the spectral reflectance and the disease severity;
wherein x is i And y i Respectively representing two variables of the spectral reflectance of each wave band between 350nm and 950nm of the winter wheat single spike sample and the scab disease severity of the winter wheat single spike sample,represents the average value of the spectral reflectivities of all samples in the same band,/-)>The average value of the disease severity of the sample is represented, n represents the number of samples, the absolute value of R1 is between 0 and 1, the closer the value of R1 is to 1, the stronger the correlation degree between two variables is represented, and the closer the value is to 0, the weaker the correlation degree between the two variables is;
33 In the two different wave bands of 400nm-500nm and 600nm-700nm, selecting the two characteristic wave bands 491nm and 699nm with the largest correlation with the disease severity as the original spectral wave band characteristics of the single spike sample reflectivity spectrum of winter wheat.
4. The remote sensing identification method for winter wheat scab based on near-earth hyperspectral technology as claimed in claim 3, wherein the screening of the optimal wavelet features comprises the following steps:
41 Using Gaussian function as mother wavelet function, performing continuous wavelet transformation on reflectivity spectrum of winter wheat single spike sample by using the following formula to obtain 601 different wave bands between 350-950nm and 10 different scale wavelet energy coefficients W f (a，b)：
Wherein ψ (λ) represents a mother wavelet function, f (λ) represents a reflectance spectrum of a winter wheat single spike sample, and a and b represent a decomposition scale and a time shift factor of a wavelet, respectively; 601 is obtained from each winter wheat single spike sample Wave band ×10 Dimension of Wavelet energy coefficients of different wave bands and scales;
42 Using the following formula for wavelet energy coefficient W f Performing correlation analysis on the disease severity and the (a, b) to obtain a correlation coefficient R2, and screening out a wavelet feature set I with the scab correlation of more than 0.65 by using a threshold method;
wherein z is i And y i Respectively representing two variables of wavelet energy coefficient of a certain wave band scale and scab disease severity of winter wheat single spike sample,mean value of wavelet energy coefficients representing all samples of the same band and the same scale, +.>An average value of the severity of the disease of the sample, n represents the number of the samples, and the absolute value of R2 is between 0 and 1; the closer the absolute value of R2 is to 1, the stronger the correlation degree between the two variables is, whereas the closer to 0, the weaker the correlation degree between the two variables is;
43 For the two-by-two combination of the healthy, mild and severe winter wheat ear samples, adopting an independent T test function in spss software to screen out a wavelet feature set II which shows significant difference in all three types of combination;
44 Intersecting the wavelet feature set one and the wavelet feature set two to obtain the optimal wavelet feature.
5. The remote sensing recognition method for winter wheat scab based on the near-to-ground hyperspectral technology as claimed in claim 1, wherein the decision function adopted for constructing the winter wheat head scab recognition model by using the SVM algorithm is as follows:
wherein alpha is i I=1,..n is Lagrange multiplier corresponding to the training sample; e, e i The method comprises the steps of representing a winter wheat scab sensitive characteristic set obtained by combining original spectral band characteristics and wavelet characteristics of a winter wheat ear sample; f (f) i ∈{-1,1}；K(e i E) is a radial basis function and meets the Mercer condition; b is a threshold; the f (e) value was used to determine the severity of the condition predicted for winter wheat head blight.
6. The remote sensing recognition method for winter wheat scab based on near-field hyperspectral technology as claimed in claim 1, wherein the training of the winter wheat head scab recognition model comprises the following steps:
61 Training a two-classification winter wheat head blight recognition model:
611 Dividing all winter wheat single spike sample data into two types of health and disease, randomly selecting 73 samples as a training sample set D, and D= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) -c), wherein, any sample e i Is an n-dimensional vector, f i ∈{-1,1}；f i F for its corresponding classification result i 1 represents healthy winter wheat, f i Equal to-1 represents diseased winter wheat;
612 Combining the original spectral band characteristics and wavelet characteristics of a single spike sample of winter wheat to obtain a scab sensitive characteristic set of winter wheat, wherein the scab sensitive characteristic set of winter wheat is used as a sample variable e i ；
613 Training sample set d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) Inputting the model into a winter wheat head blight recognition model to obtain a trained two-class winter wheat head scab recognition model;
62 Training a three-classification winter wheat head blight recognition model:
621 Dividing the scab severity of winter wheat ears into three types of healthy, mild and severe samples, and respectively constructing a scab recognition model classifier svm for two types of winter wheat ears in each two types of the three types of samples i-j Obtaining SVM Health-mild 、SVM Health-severe And SVM Mild-severe ；
622 All winter wheat single spike sample data are classified into three categories, healthy, mild and severe, with 73 samples randomly selected as training sample set D, d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) -c), wherein, any sample e i Is an n-dimensional vector, f i ∈{-1,1}；f i I.e. the corresponding classification result, f i Equal to-1 represents class i, f i A value equal to 1 represents class j;
623 Combining the original spectral characteristics and the optimal wavelet characteristics of a single spike sample of winter wheat to obtain a scab sensitive characteristic set of winter wheat as a sample variable e i ；
624 Training sample set d= { (e) 1 ,f 1 ),(e 2 ,f 2 ),...,((e m ,f m ) And (3) respectively inputting the samples into three classifiers, wherein each classifier counts the classification of the samples, the classification of the samples obtains the most votes in a certain category, and the samples are judged to be of the category, so that a trained three-classification winter wheat head blight recognition model is finally obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010616580.2A CN111767863B (en) | 2020-07-01 | 2020-07-01 | Remote sensing identification method for winter wheat scab based on near-earth hyperspectral technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010616580.2A CN111767863B (en) | 2020-07-01 | 2020-07-01 | Remote sensing identification method for winter wheat scab based on near-earth hyperspectral technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111767863A CN111767863A (en) | 2020-10-13 |
| CN111767863B true CN111767863B (en) | 2023-10-31 |
Family
ID=72723032
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010616580.2A Active CN111767863B (en) | 2020-07-01 | 2020-07-01 | Remote sensing identification method for winter wheat scab based on near-earth hyperspectral technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111767863B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113176214A (en) * | 2021-04-20 | 2021-07-27 | 安徽大学 | Wheat head gibberellic disease monitoring method by applying continuous wavelet analysis through non-imaging spectrum |
| CN113409292B (en) * | 2021-06-29 | 2022-12-30 | 中国科学院空天信息创新研究院 | Method and device for detecting wheat scab and electronic equipment |
| CN114199800B (en) * | 2021-12-08 | 2023-11-21 | 华智生物技术有限公司 | Rice sheath blight disease identification method, system, equipment and medium |
| CN114740004A (en) * | 2022-04-29 | 2022-07-12 | 吉林农业科技学院 | Method for detecting root rot and positioning based on collection of hyperspectral corn by unmanned aerial vehicle |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019174572A1 (en) * | 2018-03-12 | 2019-09-19 | 浙江大学 | Method for identifying frostbite status of cereal seeds using embryo spectrogram characteristic wavebands |
| CN110346312A (en) * | 2019-07-19 | 2019-10-18 | 安徽大学 | Winter wheat fringe head blight recognition methods based on Fei Shi linear discriminant and support vector machines technology |
| CN110596280A (en) * | 2019-10-29 | 2019-12-20 | 南京财经大学 | Rapid detection method for wheat vomitoxin pollution level based on hyperspectral image and spectrum information fusion |
-
2020
- 2020-07-01 CN CN202010616580.2A patent/CN111767863B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019174572A1 (en) * | 2018-03-12 | 2019-09-19 | 浙江大学 | Method for identifying frostbite status of cereal seeds using embryo spectrogram characteristic wavebands |
| CN110346312A (en) * | 2019-07-19 | 2019-10-18 | 安徽大学 | Winter wheat fringe head blight recognition methods based on Fei Shi linear discriminant and support vector machines technology |
| CN110596280A (en) * | 2019-10-29 | 2019-12-20 | 南京财经大学 | Rapid detection method for wheat vomitoxin pollution level based on hyperspectral image and spectrum information fusion |
Non-Patent Citations (2)
| Title |
|---|
| 基于小波特征的小麦白粉病与条锈病的定量识别;鲁军景;黄文江;张竞成;蒋金豹;;光谱学与光谱分析(第06期);全文 * |
| 基于温湿度与遥感植被指数的冬小麦赤霉病估测;李卫国;黄文江;董莹莹;陈华;王晶晶;单婕;;农业工程学报(第23期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111767863A (en) | 2020-10-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111767863B (en) | Remote sensing identification method for winter wheat scab based on near-earth hyperspectral technology | |
| Bhargava et al. | Fruits and vegetables quality evaluation using computer vision: A review | |
| CN109657653B (en) | Wheat grain gibberellic disease identification method based on imaging hyperspectral data | |
| Shi et al. | Detection and discrimination of pests and diseases in winter wheat based on spectral indices and kernel discriminant analysis | |
| CN110346312B (en) | Winter wheat head gibberellic disease identification method based on Fisher linear discrimination and support vector machine technology | |
| CN107103306B (en) | Winter wheat powdery mildew remote sensing monitoring method based on wavelet analysis and support vector machine | |
| CN101881726B (en) | Nondestructive detection method for comprehensive character living bodies of plant seedlings | |
| Zhang et al. | Identification of corn seeds with different freezing damage degree based on hyperspectral reflectance imaging and deep learning method | |
| Zhang et al. | Integrating spectral and image data to detect Fusarium head blight of wheat | |
| Liu et al. | Discrimination of rice panicles by hyperspectral reflectance data based on principal component analysis and support vector classification | |
| CN112557393B (en) | Wheat leaf layer nitrogen content estimation method based on hyperspectral image fusion map features | |
| Huang et al. | Hyperspectral image analysis based on BoSW model for rice panicle blast grading | |
| CN106951914B (en) | Method for identifying vinegar variety by electronic nose for optimizing fuzzy identification vector extraction | |
| Saccon et al. | Assessment of Fusarium and deoxynivalenol using optical methods | |
| CN109470648A (en) | A kind of single grain crop unsound grain quick nondestructive determination method | |
| CN110779875B (en) | Method for detecting moisture content of winter wheat ear based on hyperspectral technology | |
| Wang et al. | Variety identification of sweet maize seeds based on hyperspectral imaging combined with deep learning | |
| Vinod et al. | Hyperspectral remote sensing for agriculture: a review | |
| Zhao et al. | Study on the classification method of rice leaf blast levels based on fusion features and adaptive-weight immune particle swarm optimization extreme learning machine algorithm | |
| Cao et al. | Discrimination of tea plant variety using in-situ multispectral imaging system and multi-feature analysis | |
| CN105869161B (en) | Hyperspectral image band selection method based on image quality evaluation | |
| An et al. | Classification of wheat powdery mildew based on hyperspectral: From leaves to canopy | |
| Yang et al. | A quality detection method of corn based on spectral technology and deep learning model | |
| Jinzhu et al. | Discrimination of tomato yellow leaf curl disease using hyperspectral imaging | |
| CN117434006A (en) | Wheat stripe rust identification method based on spectral characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |