CN110376202A - Tea tree anthracnose scab recognition methods based on imaging hyperspectral technique - Google Patents
Tea tree anthracnose scab recognition methods based on imaging hyperspectral technique Download PDFInfo
- Publication number
- CN110376202A CN110376202A CN201910513049.XA CN201910513049A CN110376202A CN 110376202 A CN110376202 A CN 110376202A CN 201910513049 A CN201910513049 A CN 201910513049A CN 110376202 A CN110376202 A CN 110376202A
- Authority
- CN
- China
- Prior art keywords
- tea tree
- value
- ref
- scab
- reflectivity
- 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.)
- Granted
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
- G01N21/17—Systems in which incident light is modified in accordance with the properties of the material investigated
- G01N21/25—Colour; Spectral properties, i.e. comparison of effect of material on the light at two or more different wavelengths or wavelength bands
-
- 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
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
- G01N2021/8466—Investigation of vegetal material, e.g. leaves, plants, fruits
-
- 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
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
-
- 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
- G06V10/58—Extraction of image or video features relating to hyperspectral data
Abstract
The invention discloses a kind of tea tree anthracnose scab recognition methods based on imaging hyperspectral technique, include the following steps: data acquisition, blade goes background, the building of tea tree anthracnose set of preferred features, based on the automatic cluster of iteration self-organizing data analysis algorithm IsoData, the tea tree anthracnose scab of two-dimension spectrum feature space analysis is identified；The characteristics of present invention has scab recognition accuracy height, the strong robustness of tea tree anthracnose, and the automatic of tea tree anthracnose, accurate measurements may be implemented.
Description
Technical field
The present invention relates to agricultural engineering technology fields, and in particular to it is a kind of it is practical, propagable it is adaptive based on
The tea tree anthracnose scab recognition methods of hyperspectral technique is imaged.
Background technique
Anthracnose is a kind of important and typical foliage disease in tea tree, seriously threatens the yield and quality of tealeaves.Closely
Nian Lai, along with Global climate change, the trend aggravated is presented in the generation of anthracnose, proposes stern challenge to disease control.
And some unscientific prevention and control measures such as blindness pesticide abuse it will cause soil hardening, acidification the problems such as, seriously affect tea place
The quality of ecology and tealeaves；Importantly, for blade be using the tealeaves at position for, the excessive residual bring food of agriculture
Safety problem is attracted extensive attention in various countries.In recent years, high light spectrum image-forming technology spectral information combined with image information
(HSI) plant disease diagnosis, nondestructive measuring method of the farm product and in terms of show big advantage, be disease inspection
It surveys, a kind of valuable tool of identification and quantization.However, traditional blade disease recognition side based on Imaging Hyperspectral Data
Method need to get first it is a large amount of and it is typical health and disease sample area-of-interest (ROI), and with machine learning algorithm phase
In conjunction with.Due to the complexity of model and the otherness of background, for different leaves sample, these models are often unstable
's.
Summary of the invention
Goal of the invention of the invention is to overcome traditional blade based on Imaging Hyperspectral Data in the prior art
Disease recognition method is due to the complexity of model and the otherness of background, and for different leaves sample, model is unstable
Deficiency provides a kind of practical, propagable adaptive tea tree anthracnose scab knowledge based on imaging hyperspectral technique
Other method.
To achieve the goals above, the invention adopts the following technical scheme:
A kind of tea tree anthracnose scab recognition methods based on imaging hyperspectral technique, includes the following steps:
(1-1) data acquisition；
(1-2) blade goes background；
The building of (1-3) tea tree anthracnose set of preferred features；
The automatic cluster of (1-4) based on iteration self-organizing data analysis algorithm IsoData；
The tea tree anthracnose scab of (1-5) two-dimension spectrum feature space analysis identifies；
A classical near infrared band-infrared band NIR-Red two-dimensional space is constructed, by whole pixels of leaf samples
It is projected on two-dimensional space, these pixels have the tag along sort of IsoData；Red wave band Red is the reflection of 630-690nm wave band
Rate mean value, near infrared band NIR are 760-900nm wave band reflectivity mean values, calculate the center point coordinate of every one kind, pass through training
The method of sample threshold traversal, sets the threshold range of near infrared band and the threshold range of red wave band；If in a class
Heart point is located in the threshold range of near infrared band and the threshold range of red wave band, then such all pixels incorporate into as disease
Class；Otherwise, such all pixels incorporate into as healthy class.
It is divided especially by the central point and certain threshold value of every one kind.The center point coordinate of every one kind is calculated, is led to
The method for crossing training sample threshold value traversal sets the optimal threshold of near infrared band and red wave band；If the central point of a class
It falls in greater than red wave band threshold value and is less than near infrared band threshold range, then such all pixels incorporate into as disease class；
Otherwise, such all pixels incorporate into as healthy class.
Present invention spectrum abundant and on the basis of image information in taking full advantage of Imaging Hyperspectral Data, firstly,
Specific spectra feature based on withdrawing spectral information tea tree anthracnose；Then, by image analysis and unsupervised machine learning side
Method combines, and constructs adaptive algorithm；Training is not only it needs to be determined that set of preferred features and two-dimensional space classification thresholds, need complexity
Training modeling.It is combined in addition, pixel cluster is classified with two-dimensional space, reduces classification results to the sensibility of threshold value,
To ensure that stronger disease recognition ability.
The present invention combines IsoData classification method with Two Dimensional Thresholding method, realizes the automatic classification of tea tree anthracnose；
The validity of this method is evaluated in two scales of pixel and blade；Finally, it is examined to develop the remote sensing monitoring of disease for tea plant
Disconnected software and hardware system provides theoretical and model supports, facilitates the efficiency and the level of IT application of practical promotion tea place plant protection work.
The present invention is based on high light spectrum image-forming data, carry out feature construction and Optimization Analysis, by independent t test analysis and ratio
Analysis combines, and filters out the disease sensitive band positioned at 542nm, 686nm and 754nm；On this basis, ratio is respectively adopted
With normalization structure, two kinds of disease indexs are constructed, i.e. tea anthracnose Ratio index (TARI) and tea anthracnose normalizes index
(TANI)；Based on sensitive bloom spectrum signature, a kind of automatic identification algorithm is proposed and demonstrated, is realized to tea leaf anthrax
The adaptivity of scab detects.By verifying sample, model is reachable to the identification overall accuracy (OAA) of scab region picture
0.98, Kappa coefficient is up to 0.94.Compared with the classification results for being directly based upon pixel, this method has accuracy rate high and robust
Property is strong, can provide effective means for the identification of disease and degree analyzing.Therefore, it may be implemented using high light spectrum image-forming technology
Automatic, the accurate measurements of tea tree anthracnose.
Preferably, (1-1) includes the following steps:
Using halogen lamp provide light source, with 185 imaging spectrometer of UHD carry out high spectrum image acquisition, by tea leaf by
Piece is successively open and flat on testboard, is taken pictures using hyperspectral imager to every blade, obtains tea leaf in 450-
High spectrum image within the scope of 950nm, using high spectrum image as sample；
For high spectrum image, the R that the reference white plate that reflectivity is approximately 1 obtains is acquiredwhiteIt is approximately 0 with reflectivity
The R obtained with reference to blackboarddark, the high spectrum image reflectance value R after correction is calculated according to the following formula；
In formula: Rorigina1For the hyper spectral reflectance value of the raw video of acquisition；RdarkFor the reflectance value of blackboard；Rwhite
For the reflectance value of blank；R is the high spectrum image reflectance value after correction.
Preferably, (1-2) includes the following steps:
Setting of image segmentation threshold W1 extracts the albedo image Ref of the 750nm wave band of high spectrum image750；
Work as Ref750In pixel reflectivity >=W1, determine at the pixel for blade；
Work as Ref750In pixel reflectivity < W1, determine at the pixel for background；
It sets and is determined as the region pixel value of blade as 1, set and be determined as that the region pixel value of background as 0, obtains a width
Two-value mask image Maskleaf；Utilize the two-value mask image Maskleaf, exposure mask is carried out to high spectrum image, removes background,
Obtain leaf area image.
Preferably, (1-3) includes the following steps:
For the imaging high spectrum image of sample, the spectral reflectivity of tri- wave bands of 542nm, 686nm and 754nm is extracted
Ref542, Ref686And Ref754, to construct anthracnose sensitivity spectrum index: tea tree anthracnose Ratio index TARI and tea tree anthrax
Disease normalization index TANI, expression formula are as follows:
Extract the 686nm band spectrum reflectivity Ref of the imaging high spectrum image of sample686nm, extract 490-530nm wave band
Maximum first differential value Db the first differential value of all wave bands in the wavelength band is calculated according to following formula, by calculate tie
Fruit is depicted as a curve, wherein maximum value, that is, Db:
ρ′(λi)≈[ρ(λi+1)-ρ(λi-1)]/Δλ
In formula, ρ ' (λi) it is first differential, ρ (λi) be wave band i reflectivity；Δ λ is adjacent wavelength Xi+1And λi-1Between
Every.
Red side vegetation, which is calculated, using following formula coerces index RVSI:
Wherein, Ref712, Ref732, Ref752Respectively the imaging high spectrum image of sample is in 712nm, 732nm and 752nm
Band spectrum reflectivity.
Preferably, (1-4) includes the following steps:
(1-4) includes the following steps:
By TARI, TANI, Ref of each pixel of sample686nm, Db and the input of RVSI5 feature be based on iteration self-organizing
In data analysis algorithm, infima species number and maximum kind number based on iteration self-organizing data analysis algorithm are inputted；Certainly based on iteration
Tissue data analysis algorithm is trained, and is obtained the classification results of optimization, is generated the classification chart of sample.
Iteration self-organizing data analysis algorithm (IsoData) is to increase on the basis of k-means algorithm to cluster result
" merging " and " division " two operation, and a kind of clustering algorithm of set algorithm operational parameter control.Algorithm idea passes through
It sets initial parameter and introduces human-computer dialogue link, and using the mechanism of merger and division, when certain two class cluster centre distance is small
When a certain threshold value, they are merged into one kind, when certain quasi-standard deviation is greater than a certain threshold value or its number of samples more than a certain threshold
When value, it is classified as two classes.When certain class number of samples is less than certain threshold value, need to be cancelled.In this way, according to initial cluster center
With the parameter iterations such as the class number of setting, a more satisfactory classification results are finally obtained.
Preferably, the threshold range of near infrared band NIR is NIR < 0.6, the threshold range of red wave band Red is Red >
0.1, wherein Red is 630-690nm wave band reflectivity mean value, and NIR is 760-900nm wave band reflectivity mean value.Therefore, this hair
Bright scab recognition accuracy height, the strong robustness for having the following beneficial effects: tea tree anthracnose, is the identification and degree point of disease
Analysis provides effective means；Automatic, the accurate measurements of tea tree anthracnose may be implemented.
Detailed description of the invention
Fig. 1 is a kind of flow chart of the invention；
Fig. 2 is a kind of anthracnose infection blade sample of the invention and five features schematic diagram；
Fig. 3 is a kind of anthracnose scab extracted region process schematic of the invention；
Fig. 4 is the tea tree anthracnose scab recognition result of typical sample of the invention；
Fig. 5 is a kind of scab range identification schematic diagram of two-dimension spectrum feature space of the invention；
Fig. 6 is a kind of comparison diagram of the invention based on IsoData clustering method and based on traditional pixel classification method.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and detailed description.
Embodiment as shown in Figure 1 is a kind of tea tree anthracnose scab recognition methods based on imaging hyperspectral technique, packet
Include following steps:
(1-1) data acquisition；
Using halogen lamp provide light source, with 185 imaging spectrometer of UHD carry out high spectrum image acquisition, by tea leaf by
Piece is successively open and flat on testboard, is taken pictures using hyperspectral imager to every blade, obtains tea leaf in 450-
High spectrum image within the scope of 950nm, using high spectrum image as sample；
For high spectrum image, the R that the reference white plate that reflectivity is approximately 1 obtains is acquiredwhiteIt is approximately 0 with reflectivity
The R obtained with reference to blackboarddark, the high spectrum image reflectance value R after correction is calculated according to the following formula；
In formula: RoriginalFor the hyper spectral reflectance value of the raw video of acquisition；RdarkFor the reflectance value of blackboard；Rwhite
For the reflectance value of blank；R is the high spectrum image reflectance value after correction.
(1-2) blade goes background；
Setting of image segmentation threshold W1 extracts the albedo image Ref of the 750nm wave band of high spectrum image750；
Work as Ref750In pixel reflectivity >=W1, determine at the pixel for blade；
Work as Ref750In pixel reflectivity < W1, determine at the pixel for background；
It sets and is determined as the region pixel value of blade as 1, set and be determined as that the region pixel value of background as 0, obtains a width
Two-value mask image Maskleaf；Utilize the two-value mask image Maskleaf, exposure mask is carried out to high spectrum image, removes background,
Obtain leaf area image.
The building of (1-3) tea tree anthracnose set of preferred features；
For the imaging high spectrum image of sample, the spectral reflectivity of tri- wave bands of 542nm, 686nm and 754nm is extracted
Ref542, Ref686And Ref754, to construct anthracnose sensitivity spectrum index: tea tree anthracnose Ratio index TARI and tea tree anthrax
Disease normalization index TANI, expression formula are as follows:
Extract the 686nm band spectrum reflectivity Ref of the imaging high spectrum image of sample686nm, extract 490-530nm wave band
Maximum first differential value Db the first differential value of all wave bands in the wavelength band is calculated according to following formula, by calculate tie
Fruit is depicted as a curve, wherein maximum value, that is, Db:
ρ′(λi)≈[ρ(λi+1)-ρ(λi-1)]/Δλ
In formula, ρ ' (λi) it is first differential, ρ (λi) be wave band i reflectivity；Δ λ is adjacent wavelength Xi+1And λi-1Between
Every.
Red side vegetation, which is calculated, using following formula coerces index RVSI:
Wherein, Ref712, Ref732, Ref752Respectively the imaging high spectrum image of sample is in 712nm, 732nm and 752nm
Band spectrum reflectivity.
The automatic cluster of (1-4) based on iteration self-organizing data analysis algorithm IsoData；
As shown in Fig. 2,
By TARI, TANI, Ref of each pixel of sample686nm, Db and the input of RVSI5 feature be based on iteration self-organizing
In data analysis algorithm, infima species number and maximum kind number based on iteration self-organizing data analysis algorithm are inputted；Certainly based on iteration
Tissue data analysis algorithm is trained, and is obtained the classification results of optimization, is generated the classification chart of sample.
The tea tree anthracnose scab of (1-5) two-dimension spectrum feature space analysis identifies；
As shown in figure 3, building one classical near infrared band-infrared band NIR-Red two-dimensional space, by leaf samples
Whole pixels be projected on two-dimensional space, these pixels have the tag along sort of IsoData.In the present invention, with tradition
The classification that method carries out each pixel is different, and the class of IsoData is further incorporated into normal and two class of scab as a whole
In.It is divided especially by the central point and certain threshold value of every one kind.The center point coordinate for calculating every one kind, passes through training
The method of sample threshold traversal, sets the threshold range of near infrared band and the threshold range of red wave band；The threshold of near infrared band
Value range is NIR < 0.6, and the threshold range of red wave band is Red > 0.1, wherein Red is that 630-690nm wave band reflectivity is equal
Value, NIR is 760-900nm wave band reflectivity mean value；If the central point of a class be located near infrared band threshold range and
In the threshold range of red wave band, then such all pixels incorporate into as disease class；Otherwise, such all pixels incorporate into for
Healthy class.
Precision evaluation
Visual interpretation is carried out to leaf spot lesion region based on false color image, extracts scab region ROI as reference, to certainly
The precision of dynamic classification results is verified.And for healthy leaves sample, due to disease-free spot, integrated plate blade (i.e. vane region
Each pixel in domain) it is regarded as healthy area.The differentiation result of the above method and actual visual interpretation result are compared
Compared with being evaluated to the precision to this method.Model accuracy evaluation index includes overall accuracy (0A) and Kappa coefficient.Institute
Have statistical analysis and modeling using MATLAB software complete (MathWorks Inc., Natick, Massachusetts,
USA)。
In the present invention, the precision of method is evaluated from two scales of pixel and blade respectively.Following table is to two
The precision of a scale training and verifying sample is summarized.
Blade and grid cell size tea tree anthracnose accuracy of identification
Note: OA and Kappa on grid cell size are each sick leaf visual interpretation result institute's score compared with category of model result
The average value of class precision.
The result shows that training and verifying sample standard deviation reach degree of precision, i.e. the OA of training sample is on grid cell size
96%, Kappa 0.91, the OA for verifying sample is 94%, Kappa 0.87.
In addition, as a whole by blade, the precision of this method is investigated in blade level.For training sample
Speech, all samples correctly can be classified (OA 100%, Kappa 1)；For verifying sample, only one disease sample
Healthy leaves are divided by mistake, other blade sample standard deviations classify correct (OA 98%, Kappa 0.96).Fig. 4 illustrates some allusion quotations
Pattern is clustered based on Isodata and the tea tree anthracnose scab recognition result of two-dimension spectrum feature space analysis method, dark
Scab is represented, light color represents healthy area, and the effect of judging result can be intuitively found out from figure.
Fig. 5 is a kind of scab range identification schematic diagram of two-dimension spectrum feature space of the invention."+" represents in figure
The center position of every class in the optimal classification result obtained after IsoData automatic cluster；Grey rectangle frame represent the present invention to
The threshold range (NIR < 0.6, Red > 0.1) of near infrared band and red wave band out.If the central point of a class falls in this
In threshold range, then such all pixels incorporate into as disease class；Otherwise, such all pixels incorporate into as healthy class.
Further to verify IsoData and two-dimensional space threshold determination method to the recognition effect of anthracnose, based on identical
Spectral Properties collection and threshold value, compared traditional based on the classification method of pixel with the present invention.Fig. 6 shows difference
Classification results based on both methods.The result shows that the classification results based on IsoData method used herein are substantially better than
Using only the classification results of the method based on pixel of two waveband threshold value.Can in addition, although IsoData classification method is used only
To obtain the image clustering of sharpness of border, but it is difficult to determine which class belongs to disease region, which class belongs to healthy area.Pass through
Analysis to more multisample, health and disease pixel show obvious cluster feature and are divided in NIR-Red two-dimensional space
Cloth regularity.Therefore, it is broken in scab and healthy area edge based on the Threshold segmentation of pixel to efficiently solve tradition for this method
Broken and poor stability problem.Meanwhile IsoData clustering method is a kind of effective automatic judging method.
Traditional blade disease recognition method needs to get a large amount of and typical health and disease sample ROI first, and
It is combined with machine learning algorithm.Due to the complexity of model and the otherness of background, for different leaves sample, these
Model is often unstable.On the contrary, the present invention takes full advantage of spectrum abundant and image information in Imaging Hyperspectral Data.
Firstly, the specific spectra feature based on withdrawing spectral information tea tree anthracnose；Then, by image analysis and unsupervised engineering
Learning method combines, and constructs adaptive algorithm；Training is not only it needs to be determined that set of preferred features and two-dimensional space classification thresholds, need
Complicated training modeling.In addition, pixel cluster is combined with two-dimensional space classification, classification results are reduced to the sensitivity of threshold value
Property, to ensure that stronger disease recognition ability.
In general, method proposed in this paper can be generalized to the plant leaf blade with clear spectral response characteristics damage,
The automatic identification and diagnosis of pest and disease damage stress.Simultaneously as pest and disease damage usually occurs simultaneously, future may need to identify inhomogeneity
The pest and disease damage type of type, this will test to the specificity of this method.In addition, canopy scale also can be the one of the following application
A scene, or the camera of exploitation customization wave band can be considered.
It should be understood that this embodiment is only used to illustrate the invention but not to limit the scope of the invention.In addition, it should also be understood that,
After having read the content of the invention lectured, those skilled in the art can make various modifications or changes to the present invention, these etc.
Valence form is also fallen within the scope of the appended claims of the present application.
Claims (6)
1. a kind of tea tree anthracnose scab recognition methods based on imaging hyperspectral technique, characterized in that include the following steps:
(1-1) data acquisition；
(1-2) blade goes background；
The building of (1-3) tea tree anthracnose set of preferred features；
The automatic cluster of (1-4) based on iteration self-organizing data analysis algorithm IsoData；
The tea tree anthracnose scab of (1-5) two-dimension spectrum feature space analysis identifies；
A classical near infrared band-infrared band NIR-Red two-dimensional space is constructed, whole pixels of leaf samples are projected
On two-dimensional space, these pixels have the tag along sort of IsoData；Red wave band Red is that 630-690nm wave band reflectivity is equal
Value, near infrared band NIR is 760-900nm wave band reflectivity mean value, calculates the center point coordinate of every one kind, passes through training sample
The method of threshold value traversal, sets the threshold range of near infrared band and the threshold range of red wave band；If the central point of a class
In the threshold range of near infrared band and the threshold range of red wave band, then such all pixels incorporate into as disease class；
Otherwise, such all pixels incorporate into as healthy class.
2. the tea tree anthracnose scab recognition methods according to claim 1 based on imaging hyperspectral technique, characterized in that
(1-1) includes the following steps:
Using halogen lamp provide light source, with 185 imaging spectrometer of UHD carry out high spectrum image acquisition, by tea leaf piecewise according to
It is secondary open and flat on testboard, it is taken pictures using hyperspectral imager to every blade, obtains tea leaf in 450-950nm model
Interior high spectrum image is enclosed, using high spectrum image as sample；
For high spectrum image, the R that the reference white plate that reflectivity is approximately 1 obtains is acquiredwhiteIt is approximately 0 reference with reflectivity
The R that blackboard obtainsdark, the high spectrum image reflectance value R after correction is calculated according to the following formula；
In formula: RoriginalFor the hyper spectral reflectance value of the raw video of acquisition；RdarkFor the reflectance value of blackboard；RwhiteIt is white
The reflectance value of plate；R is the high spectrum image reflectance value after correction.
3. the tea tree anthracnose scab recognition methods according to claim 1 based on imaging hyperspectral technique, characterized in that
(1-2) includes the following steps:
Setting of image segmentation threshold W1 extracts the albedo image Ref of the 750nm wave band of high spectrum image750；
Work as Ref750In pixel reflectivity >=W1, determine at the pixel for blade；
Work as Ref750In pixel reflectivity < W1, determine at the pixel for background；
It sets and is determined as the region pixel value of blade as 1, set and be determined as that the region pixel value of background as 0, obtains a width two-value
Mask image Maskleaf；Utilize the two-value mask image Maskleaf, exposure mask is carried out to high spectrum image, background is removed, obtains
Leaf area image.
4. the tea tree anthracnose scab recognition methods according to claim 1 based on imaging hyperspectral technique, characterized in that
(1-3) includes the following steps:
For the imaging high spectrum image of sample, the spectral reflectivity Ref of tri- wave bands of 542nm, 686nm and 754nm is extracted542,
Ref686And Ref754, to construct anthracnose sensitivity spectrum index: tea tree anthracnose Ratio index TARI and tea tree anthracnose normalizing
Change index TANI, expression formula is as follows:
Extract the 686nm band spectrum reflectivity Ref of the imaging high spectrum image of sample686nm, extract 490-530nm wave band most
Big first differential value Db calculates the first differential value of all wave bands in the wavelength band, calculated result is drawn according to following formula
A curve is made, wherein maximum value, that is, Db:
ρ′(λi)≈[ρ(λi+1)-ρ(λi-1)]/Δλ
In formula, ρ ' (λi) it is first differential, ρ (λi) be wave band i reflectivity；Δ λ is adjacent wavelength Xi+1And λi-1Interval.
Red side vegetation, which is calculated, using following formula coerces index RVSI:
Wherein, Ref712, Ref732, Ref752Respectively the imaging high spectrum image of sample is in 712nm, 732nm and 752nm wave band
Spectral reflectivity.
5. the tea tree anthracnose scab recognition methods according to claim 1 based on imaging hyperspectral technique, characterized in that
(1-4) includes the following steps:
By TARI, TANI, Ref of each pixel of sample686nm, Db and the input of RVSI5 feature be based on iteration self-organizing data
In parser, infima species number and maximum kind number based on iteration self-organizing data analysis algorithm are inputted；Based on iteration self-organizing
Data analysis algorithm is trained, and obtains the classification results of optimization, generates the classification chart of sample.
6. the tea tree anthracnose scab recognition methods according to claim 1 based on imaging hyperspectral technique, characterized in that
The threshold range of near infrared band NIR is NIR < 0.6, and the threshold range of red wave band Red is Red > 0.1, wherein Red is
630-690nm wave band reflectivity mean value, NIR are 760-900nm wave band reflectivity mean values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910513049.XA CN110376202B (en) | 2019-06-13 | 2019-06-13 | Tea tree anthracnose lesion identification method based on imaging hyperspectral technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910513049.XA CN110376202B (en) | 2019-06-13 | 2019-06-13 | Tea tree anthracnose lesion identification method based on imaging hyperspectral technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110376202A true CN110376202A (en) | 2019-10-25 |
| CN110376202B CN110376202B (en) | 2021-11-19 |
Family
ID=68250250
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910513049.XA Active CN110376202B (en) | 2019-06-13 | 2019-06-13 | Tea tree anthracnose lesion identification method based on imaging hyperspectral technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110376202B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111562273A (en) * | 2020-06-05 | 2020-08-21 | 大连工业大学 | Hyperspectrum-based fish water jet descaling slight damage visualization method |
| CN112699756A (en) * | 2020-12-24 | 2021-04-23 | 中国农业科学院农业信息研究所 | Hyperspectral image-based tea origin identification method and system |
| CN112967233A (en) * | 2021-02-07 | 2021-06-15 | 海南大学 | Rubber tree anthracnose identification system |
| CN113610768A (en) * | 2021-07-14 | 2021-11-05 | 南方电网科学研究院有限责任公司 | Method and device for measuring and calculating coverage rate of algae on surface of insulator and storage medium |
| CN117115662A (en) * | 2023-09-25 | 2023-11-24 | 中国科学院空天信息创新研究院 | Jujube tree spider mite pest identification method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105021529A (en) * | 2015-06-11 | 2015-11-04 | 浙江水利水电学院 | Spectrum information and image information fusing crop plant disease and insect pest identifying and distinguishing method |
| CN108020511A (en) * | 2016-11-01 | 2018-05-11 | 中国科学院遥感与数字地球研究所 | A kind of shallow macrophytic lake water quality parameter remote-sensing monitoring method and device |
| CN109360117A (en) * | 2018-10-08 | 2019-02-19 | 西充恒河农牧业开发有限公司 | A kind of crop growing mode recognition methods |
-
2019
- 2019-06-13 CN CN201910513049.XA patent/CN110376202B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105021529A (en) * | 2015-06-11 | 2015-11-04 | 浙江水利水电学院 | Spectrum information and image information fusing crop plant disease and insect pest identifying and distinguishing method |
| CN108020511A (en) * | 2016-11-01 | 2018-05-11 | 中国科学院遥感与数字地球研究所 | A kind of shallow macrophytic lake water quality parameter remote-sensing monitoring method and device |
| CN109360117A (en) * | 2018-10-08 | 2019-02-19 | 西充恒河农牧业开发有限公司 | A kind of crop growing mode recognition methods |
Non-Patent Citations (6)
| Title |
|---|
| ASSESSING CROP DAMAGE FROM DICAMBA ON: "Jingcheng Zhang", 《RESEARCH ARTICLE》 * |
| MERTON.R: "EARLY SIMULATION RESULTS OF THE ARIES-1 SATELLITE SENSOR FOR MULTI-TEMPORAL VEGETATION RESEARCH DERIVED FROM AVIRIS", 《JPL PUBLICATION》 * |
| 朱林等: "基于图像处理的酿酒葡萄叶片病害分割与诊断", 《中国农机化学报》 * |
| 王志平等: "利用高光谱数据进行地物识别分类研究", 《光子学报》 * |
| 王晓庆等: "炭疽病胁迫下的茶树叶片高光谱特征分析", 《植物保护》 * |
| 陈艳华等: "基于分类知识利用神经网络反演叶面积指数", 《生态学报》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111562273A (en) * | 2020-06-05 | 2020-08-21 | 大连工业大学 | Hyperspectrum-based fish water jet descaling slight damage visualization method |
| CN112699756A (en) * | 2020-12-24 | 2021-04-23 | 中国农业科学院农业信息研究所 | Hyperspectral image-based tea origin identification method and system |
| CN112699756B (en) * | 2020-12-24 | 2023-08-25 | 中国农业科学院农业信息研究所 | Hyperspectral image-based tea origin identification method and system |
| CN112967233A (en) * | 2021-02-07 | 2021-06-15 | 海南大学 | Rubber tree anthracnose identification system |
| CN113610768A (en) * | 2021-07-14 | 2021-11-05 | 南方电网科学研究院有限责任公司 | Method and device for measuring and calculating coverage rate of algae on surface of insulator and storage medium |
| CN117115662A (en) * | 2023-09-25 | 2023-11-24 | 中国科学院空天信息创新研究院 | Jujube tree spider mite pest identification method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110376202B (en) | 2021-11-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110376202A (en) | Tea tree anthracnose scab recognition methods based on imaging hyperspectral technique | |
| Chen et al. | Mapping croplands, cropping patterns, and crop types using MODIS time-series data | |
| CN109308697B (en) | Leaf disease identification method based on machine learning algorithm | |
| Kurtulmus et al. | Green citrus detection using ‘eigenfruit’, color and circular Gabor texture features under natural outdoor conditions | |
| He et al. | A method of green litchi recognition in natural environment based on improved LDA classifier | |
| CN107103306B (en) | Winter wheat powdery mildew remote-sensing monitoring method based on wavelet analysis and support vector machines | |
| CN104374738B (en) | A kind of method for qualitative analysis improving identification result based on near-infrared | |
| CN109325431B (en) | Method and device for detecting vegetation coverage in feeding path of grassland grazing sheep | |
| Zhang et al. | Robust hyperspectral vision-based classification for multi-season weed mapping | |
| CN112699756B (en) | Hyperspectral image-based tea origin identification method and system | |
| CN110363125A (en) | Using the method for Model Transfer identification different cultivars Citrus Huanglongbing pathogen | |
| Zhang et al. | Extraction of tree crowns damaged by Dendrolimus tabulaeformis Tsai et Liu via spectral-spatial classification using UAV-based hyperspectral images | |
| CN107145831A (en) | Based on vector probabilistic diffusion and markov random file Hyperspectral Image Classification method | |
| Zhang et al. | Diagnosing the symptoms of sheath blight disease on rice stalk with an in-situ hyperspectral imaging technique | |
| CN105067532A (en) | Method for identifying early-stage disease spots of sclerotinia sclerotiorum and botrytis of rape | |
| CN108764284A (en) | The classification denoising method and system of a kind of high resolution image to pine tree Deceased wood | |
| CN109598284A (en) | A kind of hyperspectral image classification method based on large-spacing distribution and space characteristics | |
| CN115292334A (en) | Intelligent planting method and system based on vision, electronic equipment and storage medium | |
| CN108764285A (en) | A kind of recognition methods of pine tree Deceased wood and system based on high resolution image | |
| CN106568730B (en) | A kind of rice yin-yang leaf fringe recognition methods based on Hyperspectral imaging near the ground | |
| Cao et al. | Discrimination of tea plant variety using in-situ multispectral imaging system and multi-feature analysis | |
| Kuswidiyanto et al. | Airborne hyperspectral imaging for early diagnosis of kimchi cabbage downy mildew using 3D-ResNet and leaf segmentation | |
| Verma et al. | Recent advancements in image-based prediction models for diagnosis of plant diseases | |
| CN110096970A (en) | Pine forest discoloration standing tree single plant identification method based on WV3 satellite image | |
| CN110335249A (en) | Citrus Huanglongbing pathogen detection method based on high light spectrum image-forming technology |
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 |