CN106404689A - Identification method of components of exocarpium - Google Patents
Identification method of components of exocarpium Download PDFInfo
- Publication number
- CN106404689A CN106404689A CN201610846460.5A CN201610846460A CN106404689A CN 106404689 A CN106404689 A CN 106404689A CN 201610846460 A CN201610846460 A CN 201610846460A CN 106404689 A CN106404689 A CN 106404689A
- Authority
- CN
- China
- Prior art keywords
- sample
- spectral data
- modeling
- data
- modeling collection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/17—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
- G01N21/31—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry
Landscapes
- Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Immunology (AREA)
- Pathology (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The embodiment of the invention discloses an identification method of the components of exocarpium. The identification method comprises the steps of scanning samples of a sample set through at least one waveband, and collecting hyperspectral images of the samples of the sample set; obtaining hyperspectral data of the samples of the sample set according to the hyperspectral images; processing the hyperspectral data of the samples of the sample set, dividing the samples of the sample set into modeling set samples and testing set samples, and obtaining the hyperspectral data of the modeling set samples and the testing set samples; selecting characteristic wavelength in the hyperspectral data of the modeling set samples through a continuous projection algorithm; regarding the hyperspectral data of the testing set samples, the hyperspectral data of the modeling set samples and the hyperspectral data which corresponds to the characteristic wavelength as input variables of a discriminant analysis model, and obtaining the identification results of the components of the modeling set samples. The identification method of the components of the exocarpium is simple in operating step and can identify the components of the exocarpium precisely.
Description
Technical field
The present invention relates to Med Mat Appreciation technical field, more particularly, to a kind of Exocarpium Citri Grandis Components identification method.
Background technology
Exocarpium Citri Grandis also known as change skin, Huajuhong, are the outside rind of the immature fruit of rutaceae Citrus grandis.Change tangerine
Red not only have control coughing eat, stomach invigorating promoting the circulation of qi, anti-alcohol function, and or human body beauty treatment optimum feed stock, before having wide market
Scape.Research shows, volatile oil, flavone compound, and polysaccharide and coumarin kind compound etc. are mainly effectively becoming of Exocarpium Citri Grandis
Point.The content of different kind active ingredient is different, and effect is different, and also differs larger in price, with certified products skin
Best results.Therefore on market, the certified products fruit of presence many Exocarpium Citri Grandises, adulterant fruit, adulterant skin pretend to be certified products skin, compromise
Consumer's interests, have also impacted the interests of the peasants of plantation improved seeds.
At present to the conventional discrimination method major sexual shape identification of Exocarpium Citri Grandis composition, Microscopic Identification, high performance liquid chromatography.
Although these methods are each advantageous, presence subjectivity in varying degrees is strong, need pretreatment, experimentation complexity etc. to lack
Point is it is impossible to meet quick, the reliable needs detecting in market.
Content of the invention
In view of this, the embodiment of the present invention provides a kind of Exocarpium Citri Grandis Components identification method, and operating procedure is simple, can be accurate
The composition of identification Exocarpium Citri Grandis.
A kind of embodiment of the present invention Exocarpium Citri Grandis authentication method, including：
By at least one wave band, the sample in sample set is scanned, collecting sample concentrates the high-spectrum of sample
Picture；
According to described high spectrum image, obtain the high-spectral data of sample in sample set；
The high-spectral data of sample in described sample set is processed, the sample in sample set is divided into modeling collection sample
Basis and inspection set sample, and obtain the described modeling collection high-spectral data of sample and the EO-1 hyperion number of described inspection set sample
According to；
Characteristic wavelength is selected in the high-spectral data of described modeling collection sample by successive projection algorithm；
By the high-spectral data of described inspection set sample, the classification assignment of described inspection set sample and described modeling collection sample
This high-spectral data or the corresponding high-spectral data of described characteristic wavelength are built as the input variable of discriminant analysis model, acquisition
The composition recognition result of mould collection sample.
Further, described discriminant analysis model includes discrimination model or the extreme learning machine structure of PLS structure
The discrimination model built.
Further, at least one wave band described includes the wave band of 400nm-1000nm or the wave band of 1000nm-2500nm.
Further, in described sample set, sample includes certified products skin, adulterant skin, certified products fruit and the adulterant fruit of Exocarpium Citri Grandis.
Further, described high-spectral data includes spatial position data, wavelength data and spectral absorbance values.
Further, the described high-spectral data to sample in described sample set carries out processing and includes：
The high-spectral data of sample in described sample set is corrected based on formula below：
Wherein, RrefFor the high-spectral data after correction；DNrawHigh-spectral data for sample in sample set before correction；
DNwhiteFor blank correction data；DNdarkFor blackboard correction data.
Dry process is carried out to the high-spectral data after correction by SG smoothing algorithm.
Further, characteristic wavelength bag is selected in the high-spectral data of described modeling collection sample by successive projection algorithm
Include：
Classification assignment is carried out to the composition of described modeling collection sample；
The high-spectral data of described modeling collection sample and described classification assignment are become as the input of successive projection algorithm
Amount, selects characteristic wavelength in the high-spectral data of described modeling collection sample.
Methods described also includes：Accuracy of identification is calculated based on recognition result.
A kind of Exocarpium Citri Grandis Components identification method provided in an embodiment of the present invention, by by the EO-1 hyperion number of inspection set sample
High-spectral data or the corresponding high-spectral data of characteristic wavelength according to, the classification assignment of inspection set sample and modeling collection sample are made
For the input variable of discriminant analysis model, obtain the composition recognition result of modeling collection sample, operating procedure is simple, can accurately know
The composition of other Exocarpium Citri Grandis.
Brief description
By reading the detailed description that non-limiting example is made made with reference to the following drawings, other of the present invention
Feature, objects and advantages will become more apparent upon：
Fig. 1 a is a kind of Exocarpium Citri Grandis Components identification method flow diagram that the embodiment of the present invention one provides；
Fig. 1 b be the embodiment of the present invention one provide 400nm-1000nm scanning wavelength band in, sample in sample set
The spectrum comparison diagram that high-spectral data is formed；
Fig. 1 c is the embodiment of the present invention one offer in 1000nm-2500nm scanning wavelength band, sample in sample set
The spectrum comparison diagram that high-spectral data is formed；
Fig. 1 d is working as of the embodiment of the present invention one offer to collect the high-spectral data of sample as input variable with modeling, and
During 400nm-1000nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；
Fig. 1 e be the embodiment of the present invention one provide when with the corresponding high-spectral data of characteristic wavelength as input variable, and
During 400nm-1000nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；
Fig. 1 f is working as of the embodiment of the present invention one offer to collect the high-spectral data of sample as input variable with modeling, and
During 1000nm-2500nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；
Fig. 1 g be the embodiment of the present invention one provide when with the corresponding high-spectral data of characteristic wavelength as input variable, and
During 1000nm-2500nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；
Fig. 2 a works as with the high-spectral data modeling collection sample as input variable for what the embodiment of the present invention two provided, and
During 400nm-1000nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；
Fig. 2 b works as with the corresponding high-spectral data of characteristic wavelength as input variable for what the embodiment of the present invention two provided, and
During 400nm-1000nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；
Fig. 2 c works as with the high-spectral data modeling collection sample as input variable for what the embodiment of the present invention two provided, and
During 1000nm-2500nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；
What Fig. 2 d was that the embodiment of the present invention two provides is when with the corresponding high-spectral data of characteristic wavelength as input variable, and
In 1000nm-2500nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling.
Specific embodiment
The present invention is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining the present invention, rather than limitation of the invention.It also should be noted that, in order to just
Part related to the present invention rather than full content is illustrate only in description, accompanying drawing.
Embodiment one
Fig. 1 a is a kind of Exocarpium Citri Grandis Components identification method flow diagram that the embodiment of the present invention one provides, as shown in Figure 1a, institute
The method of stating includes：
S110：By at least one wave band, sample in sample set is scanned, collecting sample concentrates the EO-1 hyperion of sample
Image.
In the present embodiment, the sample in sample set includes certified products skin, adulterant skin, certified products fruit and the adulterant of Exocarpium Citri Grandis
Really.Wherein, certified products skin is 32, certified products skin 10, adulterant fruit 11, adulterant skin 7.Wherein, by the certified products skin in sample set,
Adulterant skin, certified products fruit and adulterant fruit powder broken uniformly after, respectively take on 5g culture dish holding, for gathering high spectrum image.
In the present embodiment, at least one wave band includes 400nm-1000nm wave band or 1000nm-2500nm wave band.Bloom
The collection of spectrogram picture adopt Sichuan Shuan Lihepu Science and Technology Ltd. GaiaSorter EO-1 hyperion sort instrument system (V10E,
N25E-SWIR).This system is mainly made up of hyperspectral imager, CCD camera, light source, camera bellows, computer.Table 1 is EO-1 hyperion
The parameter list of laboratory apparatus in sorting instrument system.
Table 1
| Sequence number | Relevant parameter | V10E | N25E-SWIR |
| 1 | Spectral region | 400-1000nm | 1000-2500nm |
| 2 | Spectral resolution | 2.8nm | 12nm |
| 3 | Image planes size | 6.15×14.2 | 7.6×14.2 |
| 4 | Line falling dispersion | 97.5nm/mm | 208nm/mm |
| 5 | Relative aperture | F/2.4 | F/2.0 |
| 6 | Veiling glare | <0.5% | <0.5% |
| 7 | Wave band number | 520 | 288 |
When carrying out high spectrum image collection, need to arrange the camera exposure time, carry the platform translational speed of sample with
And the distance between object lens and sample.This 3 parameters influence each other, and make the image size to fit of collection, clearly, indeformable mistake
Very.Through making repeated attempts, object lens are highly set to 31cm, and the time for exposure is set to 10ms, and platform translational speed is set to 46mm/
s.Image capture software is completed using the Hyperspectral imager acquisition software that Sichuan Shuan Lihepu Science and Technology Ltd. provides.When
When sample set sample being scanned using the wave band of 400nm-1000nm, obtain the high spectrum image of each sample；Work as employing
When the wave band of 1000nm-2500nm is scanned to modeling collection sample, obtain the high spectrum image of each sample.
S120：According to described high spectrum image, obtain the high-spectral data of sample in sample set.
In the present embodiment, the high spectrum image according to the modeling collection sample obtaining is it is possible to obtain modeling collection sample
High-spectral data.Wherein, high-spectral data includes spatial position data, wavelength data and spectral absorbance values.
S130：The high-spectral data of sample in described sample set is processed, and according to the described sample set after processing
Sample in sample set is divided into modeling collection sample and inspection set sample by the high-spectral data of middle sample, and builds described in obtaining
The high-spectral data of mould collection sample and the high-spectral data of described inspection set sample.
In the present embodiment, exemplary, the described high-spectral data to sample in sample set carries out processing inclusion：To institute
The high-spectral data stating sample in sample set is corrected based on formula below：Wherein,
RrefFor the high-spectral data after correction；DNrawHigh-spectral data for sample in sample set before correction；DNwhiteFor blank school
Correction data；DNdarkFor blackboard correction data；Dry process is carried out to the high-spectral data after correction by SG smoothing algorithm.
In the present embodiment, Fig. 1 b is the EO-1 hyperion of sample in sample set in 400nm-1000nm scanning wavelength band
The spectrum comparison diagram that data is formed；Fig. 1 c is to scan wavelength band, the EO-1 hyperion number of sample in sample set in 1000nm-2500nm
According to the spectrum comparison diagram being formed, as shown in fig. 1b and fig. lc, the curve of spectrum 11 of certified products fruit, the curve of spectrum 12 of adulterant fruit, just
The variation tendency of the curve of spectrum 13 of the curve of spectrum 14 of product skin and adulterant skin is generally identical, and the light of certified products skin is set a song to music
The spectral reflectance values of line 14 are less than the spectral reflectance values of the curve of spectrum of other three kinds of compositions, from the point of view of plots changes
Four kinds of heterogeneities do not have fairly obvious difference.In the present embodiment, using Kennard-Stone algorithm by sample set
Sample be divided into modeling collection sample and inspection set sample, and the high-spectral data of sample process from sample set in extract modeling
The collection high-spectral data of sample and the high-spectral data of inspection set sample.
S140：Characteristic wavelength is selected in the high-spectral data of described modeling collection sample by successive projection algorithm.
In the present embodiment, exemplary, by successive projection algorithm in the high-spectral data of described modeling collection sample
Characteristic wavelength is selected to include：Classification assignment is carried out to the sample in described modeling collection sample；Bloom by described modeling collection sample
Modal data and described classification assignment, as the input variable of successive projection algorithm, collect the high-spectral data of sample in described modeling
Middle selection characteristic wavelength.Wherein, to certified products skin, adulterant skin, certified products fruit, adulterant fruit be entered as 1,2,3,4 respectively.Table 2 is modeling
Collection sample and inspection set sample divide list；As shown in table 2, modeling collection sample totally 38 samples, inspection set sample totally 32 samples
This, in modeling collection sample, the quantity of certified products skin is 22, and in inspection set sample, the quantity of certified products skin is 20, therefore 10 certified products skins
Sample both collected the sample in sample as modeling, and as the sample in inspection set sample.
Table 2
| Certified products skin | Adulterant skin | Certified products fruit | Adulterant fruit | |
| Classification assignment | 1 | 2 | 3 | 4 |
| Modeling collection sample | 22 | 4 | 5 | 7 |
| Inspection set sample | 20 | 3 | 5 | 4 |
Table 3 is the list of the characteristic wavelength chosen, and as shown in table 3, selects in the sweep limits of 400nm-1000nm wave band
Characteristic wavelength be 15, in the range of 1000nm-2500nm select characteristic wavelength be 5.
Table 3
S150：By the high-spectral data of inspection set sample, the classification assignment of inspection set sample and described modeling collection sample
High-spectral data or the corresponding high-spectral data of described characteristic wavelength as discrimination model input variable, obtain modeling collection sample
This composition recognition result.
In the present embodiment, the discrimination model that discriminant analysis model builds for PLS.Wherein, by inspection set sample
The high-spectral data of this high-spectral data, the classification assignment of inspection set sample and described modeling collection sample is as discrimination model
Input variable, can obtain modeling collection sample composition recognition result；Or by the high-spectral data of inspection set sample, inspection
The classification assignment of collection sample and the corresponding high-spectral data of described characteristic wavelength, as the input variable of discrimination model, can obtain
Take the composition recognition result of modeling collection sample.Specifically, by high-spectral data and the inspection set sample of the inspection set sample of input
This classification assignment, can set up the relation between high-spectral data and classification assignment；By high-spectral data and the class of foundation
Relation between other assignment, the high-spectral data of input modeling collection sample or the corresponding high-spectral data of described characteristic wavelength, just
The classification assignment of corresponding modeling collection sample can be obtained.And the composition identification knot of modeling collection sample is obtained by classification assignment
Really.
On the basis of above-described embodiment, can also be such a way：Can be by each sample in inspection set sample
High-spectral data be input to PLS structure discrimination model in, make PLS build discrimination model defeated
The classification assignment going out is identical with the classification assignment of each sample of inspection set, by said method, PLS structure is sentenced
Other model is trained.Then, by the bloom of corresponding for the characteristic wavelength of each sample high-spectral data and modeling collection sample
Modal data is separately input in the discrimination model of PLS structure, exports the classification assignment of each sample respectively.
In the present embodiment, Fig. 1 d is to work as with the high-spectral data modeling collection sample as input variable, and in 400nm-
During 1000nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；Fig. 1 e is when with characteristic wave
Long corresponding high-spectral data is input variable, and in 400nm-1000nm wavelength band, the inspection set of acquisition is collected with modeling
The class prediction value schematic diagram of sample；Fig. 1 f is to work as with the high-spectral data modeling collection sample as input variable, and in 1000nm-
During 2500nm wavelength band, the inspection set of acquisition collects the class prediction value schematic diagram of sample with modeling；Fig. 1 g is when with characteristic wave
Long corresponding high-spectral data is input variable, and in 1000nm-2500nm wavelength band, the inspection set of acquisition is collected with modeling
The class prediction value schematic diagram of sample.As shown in Fig. 1 d-1g, PLS-DA is the bloom based on modeling collection sample for the PLS
The discrimination model that modal data (all band) builds, PLS-DA-SPA is that PLS feature based wavelength (characteristic wave bands) is right
The discrimination model that the high-spectral data answered builds, has one in modeling collection sample during composition identification using above two model
Fixed mistake, but error rate is relatively low.
On the basis of above-described embodiment, described method also includes：Accuracy of identification is calculated based on recognition result.Wherein,
Accuracy of identification is the ratio of the correct number of class prediction value and overall number in modeling collection sample.Table 4 is based on an inclined young waiter in a wineshop or an inn
The list of the accuracy of identification that the discrimination model that multiplication builds calculates.As shown in table 4, the composition of modeling collection sample is identified
During, overall recognition accuracy, certified products skin accuracy of identification highest are in 1000nm-2500nm wavelength band, partially minimum
The discrimination model that square law is built based on the high-spectral data of modeling collection sample, is 78% and 90% respectively, and the identification of certified products skin is wrong
What by mistake rate was minimum is then in the range of 1000nm-2500nm, the high-spectral data based on modeling collection sample for the PLS and
The discrimination model that the corresponding high-spectral data of characteristic wavelength builds, is 5%.
As shown in table 4, either in 400nm-1000nm or 1000nm-2500nm wavelength band, PLS
The discrimination model being built based on the high-spectral data of modeling collection sample, overall discrimination and certified products skin discrimination are above based on spy
Levy the discrimination model of wavelength corresponding high-spectral data structure.And the error recognition rate of certified products skin, either 400nm-1000nm
Or in 1000-2500nm wavelength band, the discrimination model that PLS is built based on the high-spectral data of modeling collection sample
And the corresponding high-spectral data of feature based wavelength builds discrimination model, the error recognition rate of certified products skin is identical.Wherein,
In 400n-1000nm wavelength band, the error recognition rate of certified products skin is 10%；In 1000nm-2500nm wavelength band, certified products
Skin error recognition rate is then 5%.
Table 4
A kind of Exocarpium Citri Grandis Components identification method provided in an embodiment of the present invention, by by the EO-1 hyperion number of inspection set sample
According to and the modeling collection high-spectral data of sample or the corresponding high-spectral data of characteristic wavelength build as PLS
The input variable of model, obtains the composition recognition result of modeling collection sample, and operating procedure is simple, can accurately identify Exocarpium Citri Grandis
Composition.
Embodiment two
This enforcement two is with the difference of the present embodiment one：Using discrimination model be sentencing that extreme learning machine builds
Other model.
In the present embodiment, as shown in Figure 2 a-2d, ELM is the high-spectral data based on modeling collection sample for the extreme learning machine
The discrimination model building, ELM-SPA is the discrimination model that the corresponding high-spectral data of extreme learning machine feature based wavelength builds.
During composition identification, there is certain mistake in modeling collection sample in the model being built by extreme learning machine, but mistake
Rate is relatively low.
Table 5 is the list of the accuracy of identification being calculated based on the discrimination model that extreme learning machine builds.As shown in table 5, totally
Accuracy of identification, certified products skin accuracy of identification highest are ELM model and ELM-SPA in 1000nm-2500nm wavelength band
Model, is 84% and 95% respectively.What certified products skin identification error rate was minimum is then in 1000nm-2500nm wavelength band
ELM model and ELM-SPA model, are 5%.In 400nm-1000nm wavelength band, it is overall that ELM-SPA model calculates
Discrimination and certified products skin discrimination are above ELM model.For the identification error rate of certified products skin, ELM-SPA model and ELM model
The error recognition rate calculating is identical.In 1000nm-2500nm wavelength band, either ELM-SPA model and ELM model,
The overall discrimination that calculates, certified products skin discrimination, certified products skin identification error rate all same, respectively 84%, 95% and 5%.
Table 5
As can be seen here, in the discrimination model of extreme learning machine structure and the discrimination model of PLS structure to modeling
Collection sample be identified during, in 1000nm-2500nm wavelength band, according to extreme learning machine build discrimination model and
The overall discrimination that the discrimination model that PLS builds calculates, certified products skin discrimination are above 400nm-1000nm ripple
In segment limit.And, the accuracy rate of the discrimination model that extreme learning machine builds is higher than the differentiation mould that PLS builds
Type.The accuracy rate of the discrimination model that the corresponding high-spectral data of PLS feature based wavelength builds is less than based on modeling
The discrimination model that the high-spectral data of collection sample builds.But, in 400nm-1000nm wavelength band, extreme learning machine is based on
The accuracy rate of the discrimination model that the corresponding high-spectral data of characteristic wavelength builds is higher than the high-spectral data based on modeling collection sample
The discrimination model building, in the range of 1000nm-25nm, the corresponding high-spectral data of extreme learning machine feature based wavelength builds
The accuracy rate of discrimination model with based on modeling collection sample high-spectral data build discrimination model identical.
A kind of Exocarpium Citri Grandis Components identification method provided in an embodiment of the present invention, by by the EO-1 hyperion number of inspection set sample
According to and the mould that builds as extreme learning machine of the modeling collection high-spectral data of sample or the corresponding high-spectral data of characteristic wavelength
The input variable of type, obtains the composition recognition result of modeling collection sample, and operating procedure is simple, can accurately identify the one-tenth of Exocarpium Citri Grandis
Point.
Note, above are only presently preferred embodiments of the present invention and institute's application technology principle.It will be appreciated by those skilled in the art that
The invention is not restricted to specific embodiment described here, can carry out for a person skilled in the art various obvious changes,
Readjust and substitute without departing from protection scope of the present invention.Therefore although being carried out to the present invention by above example
It is described in further detail, but the present invention is not limited only to above example, without departing from the inventive concept, also
Other Equivalent embodiments more can be included, and the scope of the present invention is determined by scope of the appended claims.
Claims (8)
1. a kind of Exocarpium Citri Grandis Components identification method is it is characterised in that include：
By at least one wave band, the sample in sample set is scanned, collecting sample concentrates the high spectrum image of sample；
According to described high spectrum image, obtain the high-spectral data of sample in sample set；
The high-spectral data of sample in described sample set is processed, by the sample in sample set be divided into modeling collection sample with
And inspection set sample, and obtain the described modeling collection high-spectral data of sample and the high-spectral data of described inspection set sample；
Characteristic wavelength is selected in the high-spectral data of described modeling collection sample by successive projection algorithm；
By the high-spectral data of described inspection set sample, the classification assignment of described inspection set sample and described modeling collection sample
High-spectral data or the corresponding high-spectral data of described characteristic wavelength, as the input variable of discriminant analysis model, obtain modeling collection
The composition recognition result of sample.
2. method according to claim 1 is it is characterised in that described discriminant analysis model includes PLS structure
Discrimination model or extreme learning machine build discrimination model.
3. method according to claim 1 is it is characterised in that at least one wave band described includes the ripple of 400nm-1000nm
Section or the wave band of 1000nm-2500nm.
4. method according to claim 1 it is characterised in that in described sample set sample include Exocarpium Citri Grandis certified products skin,
Adulterant skin, certified products fruit and adulterant fruit.
5. method according to claim 1 is it is characterised in that described high-spectral data includes spatial position data, wavelength
Data and spectral absorbance values.
6. method according to claim 1 is it is characterised in that the described high-spectral data to sample in described sample set enters
Row processes and includes：
The high-spectral data of sample in described sample set is corrected based on formula below：
Wherein, RrefFor the high-spectral data after correction；DNrawHigh-spectral data for sample in sample set before correction；DNwhite
For blank correction data；DNdarkFor blackboard correction data.
Dry process is carried out to the high-spectral data after correction by SG smoothing algorithm.
7. method according to claim 1 is it is characterised in that described collect sample by successive projection algorithm in described modeling
High-spectral data in select characteristic wavelength include：
Classification assignment is carried out to the sample in described modeling collection sample；
Using the described modeling collection high-spectral data of sample and described classification assignment as successive projection algorithm input variable,
Characteristic wavelength is selected in the high-spectral data of described modeling collection sample.
8. method according to claim 1 is it is characterised in that also include：Accuracy of identification is calculated based on recognition result.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610846460.5A CN106404689A (en) | 2016-09-23 | 2016-09-23 | Identification method of components of exocarpium |
| PCT/CN2017/086901 WO2018054091A1 (en) | 2016-09-23 | 2017-06-02 | Method for identifying components of exocarpium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610846460.5A CN106404689A (en) | 2016-09-23 | 2016-09-23 | Identification method of components of exocarpium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106404689A true CN106404689A (en) | 2017-02-15 |
Family
ID=57997942
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610846460.5A Pending CN106404689A (en) | 2016-09-23 | 2016-09-23 | Identification method of components of exocarpium |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN106404689A (en) |
| WO (1) | WO2018054091A1 (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018054091A1 (en) * | 2016-09-23 | 2018-03-29 | 广州道地南药技术研究有限公司 | Method for identifying components of exocarpium |
| CN110763698A (en) * | 2019-10-12 | 2020-02-07 | 仲恺农业工程学院 | Hyperspectral citrus leaf disease identification method based on characteristic wavelength |
| CN111122579A (en) * | 2020-01-17 | 2020-05-08 | 中国农业科学院都市农业研究所 | Method for measuring total content of flavone in lettuce leaves |
| CN113887543A (en) * | 2021-12-07 | 2022-01-04 | 深圳市海谱纳米光学科技有限公司 | Luggage counterfeit discrimination method based on hyperspectral characteristics and spectrum acquisition device |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109975217A (en) * | 2019-03-26 | 2019-07-05 | 贵阳学院 | Plum soluble solid content value detection method based on Hyperspectral imager |
| CN112129709A (en) * | 2020-09-16 | 2020-12-25 | 西北农林科技大学 | Apple tree canopy scale nitrogen content diagnosis method |
| CN113008815A (en) * | 2021-02-24 | 2021-06-22 | 浙江工业大学 | Hyperspectral image information-based method for nondestructive detection of total flavonoids in spina date seeds |
| CN113902673A (en) * | 2021-09-02 | 2022-01-07 | 北京市农林科学院信息技术研究中心 | Tomato gray mold degree identification method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1831516A (en) * | 2006-04-03 | 2006-09-13 | 浙江大学 | Method for nondistructive discriminating variety and true and false of cigarette using visible light and near-infrared spectrum technology |
| CN1995987A (en) * | 2007-02-08 | 2007-07-11 | 江苏大学 | Non-destructive detection method and device for agricultural and animal products based on hyperspectral image technology |
| CN102621077A (en) * | 2012-03-30 | 2012-08-01 | 江南大学 | Hyper-spectral reflection image collecting system and corn seed purity nondestructive detection method based on same |
| CN103942749A (en) * | 2014-02-24 | 2014-07-23 | 西安电子科技大学 | Hyperspectral ground feature classification method based on modified cluster hypothesis and semi-supervised extreme learning machine |
| CN105527241A (en) * | 2015-01-14 | 2016-04-27 | 青海春天药用资源科技利用有限公司 | Non-destructive method for detecting authenticity of raw cordyceps sinensis |
| CN105548037A (en) * | 2015-01-14 | 2016-05-04 | 青海春天药用资源科技利用有限公司 | Method for nondestructively detecting raw traditional Chinese medicinal materials |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103620381B (en) * | 2011-06-29 | 2015-09-30 | 富士通株式会社 | Floristics recognition device and method |
| CN104198398A (en) * | 2014-08-07 | 2014-12-10 | 浙江大学 | Method for distinguishing donkey-hide gelatin based on hyperspectral imaging |
| CN106404689A (en) * | 2016-09-23 | 2017-02-15 | 广东食品药品职业学院 | Identification method of components of exocarpium |
-
2016
- 2016-09-23 CN CN201610846460.5A patent/CN106404689A/en active Pending
-
2017
- 2017-06-02 WO PCT/CN2017/086901 patent/WO2018054091A1/en active Application Filing
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1831516A (en) * | 2006-04-03 | 2006-09-13 | 浙江大学 | Method for nondistructive discriminating variety and true and false of cigarette using visible light and near-infrared spectrum technology |
| CN1995987A (en) * | 2007-02-08 | 2007-07-11 | 江苏大学 | Non-destructive detection method and device for agricultural and animal products based on hyperspectral image technology |
| CN102621077A (en) * | 2012-03-30 | 2012-08-01 | 江南大学 | Hyper-spectral reflection image collecting system and corn seed purity nondestructive detection method based on same |
| CN103942749A (en) * | 2014-02-24 | 2014-07-23 | 西安电子科技大学 | Hyperspectral ground feature classification method based on modified cluster hypothesis and semi-supervised extreme learning machine |
| CN105527241A (en) * | 2015-01-14 | 2016-04-27 | 青海春天药用资源科技利用有限公司 | Non-destructive method for detecting authenticity of raw cordyceps sinensis |
| CN105548037A (en) * | 2015-01-14 | 2016-05-04 | 青海春天药用资源科技利用有限公司 | Method for nondestructively detecting raw traditional Chinese medicinal materials |
Non-Patent Citations (2)
| Title |
|---|
| 王铭海: ""猕猴桃、桃和梨品质特性的近红外光谱无损检测模型优化研究"", 《中国优秀学位论文全文数据库 工程科技I辑》 * |
| 高洪智 等: ""基于连续投影算法的土壤总氮近红外特征波长的选取"", 《光谱学与光谱分析》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018054091A1 (en) * | 2016-09-23 | 2018-03-29 | 广州道地南药技术研究有限公司 | Method for identifying components of exocarpium |
| CN110763698A (en) * | 2019-10-12 | 2020-02-07 | 仲恺农业工程学院 | Hyperspectral citrus leaf disease identification method based on characteristic wavelength |
| CN111122579A (en) * | 2020-01-17 | 2020-05-08 | 中国农业科学院都市农业研究所 | Method for measuring total content of flavone in lettuce leaves |
| CN113887543A (en) * | 2021-12-07 | 2022-01-04 | 深圳市海谱纳米光学科技有限公司 | Luggage counterfeit discrimination method based on hyperspectral characteristics and spectrum acquisition device |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2018054091A1 (en) | 2018-03-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106404689A (en) | Identification method of components of exocarpium | |
| Nie et al. | Classification of hybrid seeds using near-infrared hyperspectral imaging technology combined with deep learning | |
| Mahendran et al. | Application of computer vision technique on sorting and grading of fruits and vegetables | |
| CN104374738B (en) | A kind of method for qualitative analysis improving identification result based on near-infrared | |
| Lei et al. | Rapid classification of commercial Cheddar cheeses from different brands using PLSDA, LDA and SPA–LDA models built by hyperspectral data | |
| CN109540831B (en) | Wolfberry variety identification method based on hyperspectral imaging technology | |
| CN110163101B (en) | Method for rapidly distinguishing seeds of traditional Chinese medicinal materials and rapidly judging grades of seeds | |
| CN109858477A (en) | The Raman spectrum analysis method of object is identified in complex environment with depth forest | |
| Cao et al. | Identification of species and geographical strains of Sitophilus oryzae and Sitophilus zeamais using the visible/near‐infrared hyperspectral imaging technique | |
| Liu et al. | Rapid identification of chrysanthemum teas by computer vision and deep learning | |
| CN110108644A (en) | A kind of maize variety identification method based on depth cascade forest and high spectrum image | |
| Yu et al. | Accuracy and stability improvement in detecting Wuchang rice adulteration by piece-wise multiplicative scatter correction in the hyperspectral imaging system | |
| CN110006844A (en) | Near infrared spectrum feature extracting method and system based on functionality pivot analysis | |
| CN108898156A (en) | A kind of green green pepper recognition methods based on high spectrum image | |
| Shao et al. | Identification of adulterated cooked millet flour with Hyperspectral Imaging Analysis | |
| CN107576600B (en) | Quick detection method for matcha granularity grade | |
| Hao et al. | Combined hyperspectral imaging technology with 2D convolutional neural network for near geographical origins identification of wolfberry | |
| CN114295556A (en) | Hyperspectral physicochemical index fitting method and system based on one-dimensional convolution | |
| Xie et al. | Modeling for mung bean variety classification using visible and near-infrared hyperspectral imaging | |
| Gómez-Camperos et al. | Digital image processing techniques for detection of pests and diseases in crops: a review | |
| Yadav et al. | Citrus disease classification with convolution neural network generated features and machine learning classifiers on hyperspectral image data | |
| CN201041547Y (en) | Device for analyzing and recognizing different types of green teas based on multi-spectrum image texture | |
| CN117132843A (en) | Wild ginseng, under-forest mountain ginseng and garden ginseng in-situ identification method, system and related equipment | |
| CN107255624A (en) | A kind of fast and convenient method for determining protein content in fan's power mushroom entity | |
| Seo et al. | Detection of spinach juice residues on stainless steel surfaces using VNIR hyperspectral images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20170418 Address after: And Whampoa District of Guangzhou City, North Street Tianlu 510525 No. 123 of Guangdong Province Applicant after: Guangzhou Province Research Southern Technology Co. Ltd. Applicant after: Guangdong Food and Drug Vocational College Address before: 515000 Tianhe District Longdong Road, Guangdong, No. 321, North Road, No. Applicant before: Guangdong Food and Drug Vocational College |
|
| TA01 | Transfer of patent application right | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170215 |
|
| RJ01 | Rejection of invention patent application after publication |