CN109470648B - Rapid nondestructive determination method for imperfect grains of single-grain crops - Google Patents
Rapid nondestructive determination method for imperfect grains of single-grain crops Download PDFInfo
- Publication number
- CN109470648B CN109470648B CN201811390652.5A CN201811390652A CN109470648B CN 109470648 B CN109470648 B CN 109470648B CN 201811390652 A CN201811390652 A CN 201811390652A CN 109470648 B CN109470648 B CN 109470648B
- Authority
- CN
- China
- Prior art keywords
- grains
- grain
- imperfect
- spectrum
- crop
- 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
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000001514 detection method Methods 0.000 claims abstract description 29
- 239000011159 matrix material Substances 0.000 claims abstract description 24
- 238000002329 infrared spectrum Methods 0.000 claims abstract description 15
- 238000004458 analytical method Methods 0.000 claims abstract description 10
- 235000013339 cereals Nutrition 0.000 claims description 160
- 241000209140 Triticum Species 0.000 claims description 67
- 235000021307 Triticum Nutrition 0.000 claims description 67
- 238000001228 spectrum Methods 0.000 claims description 51
- 238000004422 calculation algorithm Methods 0.000 claims description 22
- 230000002950 deficient Effects 0.000 claims description 19
- 238000007781 pre-processing Methods 0.000 claims description 9
- 206010039509 Scab Diseases 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 240000008042 Zea mays Species 0.000 claims description 5
- 235000005824 Zea mays ssp. parviglumis Nutrition 0.000 claims description 5
- 235000002017 Zea mays subsp mays Nutrition 0.000 claims description 5
- 235000005822 corn Nutrition 0.000 claims description 5
- 235000007164 Oryza sativa Nutrition 0.000 claims description 4
- 235000009566 rice Nutrition 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 claims description 3
- 238000000411 transmission spectrum Methods 0.000 claims description 3
- 238000004497 NIR spectroscopy Methods 0.000 claims 1
- 240000007594 Oryza sativa Species 0.000 claims 1
- 239000008187 granular material Substances 0.000 claims 1
- 238000012937 correction Methods 0.000 description 8
- 238000005516 engineering process Methods 0.000 description 5
- 238000012795 verification Methods 0.000 description 5
- 230000001066 destructive effect Effects 0.000 description 4
- 210000001161 mammalian embryo Anatomy 0.000 description 4
- 239000002245 particle Substances 0.000 description 4
- 238000010200 validation analysis Methods 0.000 description 4
- 241000238631 Hexapoda Species 0.000 description 3
- 241000209094 Oryza Species 0.000 description 3
- 235000013305 food Nutrition 0.000 description 3
- 230000035784 germination Effects 0.000 description 3
- 238000003306 harvesting Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000003628 erosive effect Effects 0.000 description 2
- 229910052739 hydrogen Inorganic materials 0.000 description 2
- 230000003902 lesion Effects 0.000 description 2
- 230000003595 spectral effect Effects 0.000 description 2
- 238000010183 spectrum analysis Methods 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 108010068370 Glutens Proteins 0.000 description 1
- UFHFLCQGNIYNRP-UHFFFAOYSA-N Hydrogen Chemical compound [H][H] UFHFLCQGNIYNRP-UHFFFAOYSA-N 0.000 description 1
- 241000607479 Yersinia pestis Species 0.000 description 1
- 210000001015 abdomen Anatomy 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 238000012630 chemometric algorithm Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000003337 fertilizer Substances 0.000 description 1
- 230000002538 fungal effect Effects 0.000 description 1
- 235000021312 gluten Nutrition 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 239000001257 hydrogen Substances 0.000 description 1
- 208000015181 infectious disease Diseases 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 150000002894 organic compounds Chemical class 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Images
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
- G01N21/35—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light
- G01N21/359—Investigating relative effect of material at wavelengths characteristic of specific elements or molecules, e.g. atomic absorption spectrometry using infrared light using near infrared light
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)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
The invention discloses a rapid nondestructive judgment method for imperfect grains of single-grain crops, which comprises the following steps: the method comprises the following steps: collecting single-grain crop samples, detecting the imperfect grain condition of each single-grain crop, and establishing a category information matrix; step two: collecting the near infrared spectrum of single grains of the single-grain crop; step three: constructing a near-infrared discrimination analysis model of imperfect grains of single-grain crops; step four: distinguishing normal single-grain crops and imperfect single-grain crops by using the established model; the method has the advantages of objective and accurate detection result, no damage to the sample in the detection process, rapidness, simplicity and convenience, and can realize high-efficiency and quantitative judgment of imperfect grains of single-grain crops.
Description
Technical Field
The invention relates to a method for judging the quality of single-grain crops, in particular to a method for quickly and nondestructively judging imperfect grains of single-grain crops.
Background
The defective single-grain crop grains refer to damaged single-grain crop grains with use value, and include scab grains, wormhole grains, sprouting grains, damaged grains and mildewing grains. The imperfect grains comprehensively reflect the yield and quality conditions of the single-grain crops during growth, harvest, circulation and storage, and are an important index for quality detection of the single-grain crops. However, the detection of imperfect grains of single-grain crops in China still needs to completely depend on artificial sensory identification, the result is not objective and accurate enough, the repeatability is poor, and the time and the labor are consumed during the detection. The method is necessary for improving the quality inspection efficiency, facilitating the price per quality in the grain industry, guaranteeing the benefits of farmers in grain planting and the national grain safety and researching the rapid, accurate and simple imperfect grain detection technology of single-grain crops.
The near infrared spectrum is an electromagnetic wave with the wavelength within the range of 780 nm-2526 nm, can reflect the combined frequency of vibration of hydrogen-containing groups (C-H, N-H, O-H) in organic compounds and mixtures and the absorption of frequency doubling at all levels, is widely applied to the fields of agriculture, chemical industry, food, life science, environmental protection, quality supervision and the like in recent years, and has the advantages of rapidness, high efficiency, accuracy, no damage to samples, and capability of analyzing a plurality of components simultaneously. At present, the near infrared technology is applied to various indexes of single-kernel crops, such as moisture, hardness, protein content, wet gluten content and the like, but the near infrared technology is rarely reported on the detection of imperfect kernels of single-kernel crops. The essence of the phenomenon of defective single-grain crops is that the embryo or endosperm of the single-grain crops is damaged mechanically or physiologically and invaded by microorganisms, so that the quality of products and food is reduced. The germination single-seed crops and the mildewed single-seed crops are usually caused by overcast and rainy conditions in the later stage of single-seed crop grouting and overhigh moisture during harvesting or storage, and the single-seed crop seeds have the phenomena of germination and mildewing; the scab single-grain crops are typified by gibberellic disease grains and black embryo grains, the causes of the scab single-grain crops are also usually caused by rainy weather and fungal breeding during ear emergence, and the grains have the phenomena of shrinkage, white retention, pink mildew, black embryo and the like; insect erosion of single-grain crops is usually caused by insect pest infection and improper control during storage of the single-grain crops, and the phenomenon of insect erosion of embryo or endosperm parts of the grains exists; the damaged single-kernel crop is mainly caused by mechanical damage during harvesting, transporting and storing of the single-kernel crop, and kernels are incomplete and deformed to different degrees. Compared with normal single-grain crops, the imperfect grains have different changes in physical and chemical properties and can be reflected on the near infrared spectrum, so that the near infrared analysis of the imperfect grains of the single-grain crops has theoretical feasibility.
Disclosure of Invention
The invention aims to solve the technical problems that detection results of imperfect single-kernel crop detection in the prior art are basically all dependent on sense, low in accuracy, time-consuming and labor-consuming.
The invention solves the technical problems through the following technical scheme: a method for rapidly and nondestructively judging defective grains of single-grain crops comprises the following steps:
the method comprises the following steps: collecting single-grain crop samples, detecting the imperfect grain condition of each single-grain crop, and establishing a category information matrix;
step two: collecting the near infrared spectrum of single grains of the single-grain crop;
step three: constructing a near-infrared discrimination analysis model of imperfect grains of single-grain crops;
step four: and distinguishing normal single-grain crops and imperfect single-grain crops by using the established model.
Preferably, the first step, the second step, the third step and the fourth step are implemented on a platform of a near infrared spectrum analysis instrument with a single particle detection function.
Preferably, the single-grain crop samples collected in the step one are different varieties harvested at different times and different production places, the total number of the single-grain crop samples is 1000-3000, wherein the number of the imperfect grains of each type of single-grain crop accounts for 25% -75% of the total number of the samples.
Preferably, in the step one, the detected various imperfect single-grain crops are assigned with the value of 2, the detected normal single-grain crops are assigned with the value of 1, and the classification information matrix is established as a classification label.
Preferably, the near infrared spectrum in the second step is near infrared diffuse reflection spectrum or near infrared diffuse transmission spectrum.
Preferably, the third step specifically comprises:
spectrum preprocessing, selecting effective wavelength and establishing a spectrum matrix;
and b, constructing a relation model between the spectrum matrix and the single-grain crop category information matrix by adopting a chemometrics algorithm.
Preferably, the chemometric algorithm is discriminant partial least squares algorithm or stacked self-encoder algorithm.
Preferably, when the discriminative partial least square algorithm is adopted to construct the imperfect grain near-infrared discriminative analysis model of the single-grain crop, the corresponding spectrum preprocessing is as follows: cut at 9596.6-7992cm-1And 5600.6-4790cm-1Spectrum of (2), first derivative + vector normalization of the spectrumThe number of smoothing points 17; spectrum matrixes are obtained after spectrum preprocessing, and a discriminant partial least square algorithm is adopted to construct a relation model, wherein the relation model is a discriminant partial least square algorithm model, and the specific parameters are as follows: setting the main component number as 5, the classification threshold value as 1.57, judging that the crop is a normal single-grain crop when the predicted value is less than or equal to 1.57, and judging that the crop has the predicted value>1.57 is imperfect.
Preferably, when the stacked self-encoder algorithm is adopted to construct the imperfect grain near-infrared discrimination analysis model of the single-grain crop, the corresponding spectrum preprocessing is as follows: -1,1] normalization; spectrum matrixes are obtained after spectrum preprocessing, a stack type self-encoder algorithm is adopted to construct a relation model, the relation model is a stack type self-encoder algorithm model, and the specific parameters are as follows: setting the sparsity parameter to be 0.1, the weight attenuation coefficient to be 0.003, the weight of the sparse penalty term to be 3, training each layer of neural network parameter by using a minFunc gradient descent method, setting the training function to be l-bfgs, and setting the maximum iteration number to be 400 times; the size of the input layer is equal to the number of wavelength points of the acquired spectrum; 2 hidden layers are arranged, the number of the neuron nodes of the hidden layer 1 is set to be 100, and the number of the neuron nodes of the hidden layer 2 is set to be 200; the output layer size is 2, the defective kernels are determined when the output value is 2, and the normal single-kernel crops are determined when the output value is 1.
Preferably, the single-grain crop comprises rice, wheat or corn, and the imperfect rice grains comprise immature grains, scab grains, worm-eaten grains, germinated grains, damaged grains and mildewed grains; imperfect wheat grains include diseased speckles, worm-eaten grains, germinated grains, damaged grains and mildewed grains; the imperfect corn grains include scab grains, worm-eaten grains, germinated grains, damaged grains, mildewed grains and heat-damaged grains.
Compared with the prior art, the invention has the following advantages:
(1) the detection process of the invention is not influenced by subjective judgment of inspectors, and the detection result is objective and accurate; the detection process does not damage the sample, is rapid and simple, and can realize the high-efficiency and quantitative judgment of the imperfect grains of the single-grain crops.
(2) The method for rapidly and nondestructively judging the defective grains of the single-grain crops can be used in combination with other single-grain crop quality detection technologies, for example, a near infrared spectrum detection model combining the moisture and hardness of the single-grain crops can be used for simultaneously realizing rapid and nondestructive detection on multiple indexes of the moisture, the defective grains, the hardness and the like of the single-grain crops, realizing automation of multiple quality detection of the single-grain crops, and better serving the grain industry.
Drawings
In order to more clearly illustrate the technical solution of the present invention, the drawings needed to be used in the description of the present invention will be briefly introduced below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained based on these drawings without inventive labor.
FIG. 1 is a flow chart of a method for rapid non-destructive determination of defective single-kernel crops according to an embodiment of the present invention;
FIG. 2 is a near-infrared diffuse reflection average spectrum of a single kernel crop sample in a method for rapid non-destructive determination of imperfect kernels of a single kernel crop disclosed in an embodiment of the present invention;
FIG. 3 is a scatter diagram of a PLS-DA model for predicting values and class labels of a correction set sample in the method for rapid non-destructive determination of imperfect grains of single-grain crops disclosed in the embodiments of the present invention;
FIG. 4 is a scatter plot of predicted values of correction set samples with different numbers by an SAE model in the single-kernel crop imperfect kernel fast nondestructive determination method disclosed by the embodiment of the present invention;
FIG. 5 is a scatter plot of a PLS-DA model on a prediction value and a category label of a validation set sample in the method for rapid non-destructive determination of defective grains in single-grain crops disclosed in the embodiments of the present invention;
fig. 6 is a scatter diagram of predicted values of different numbers of validation set samples by an SAE model in the single-kernel crop imperfect kernel fast nondestructive determination method disclosed by the embodiment of the present invention.
Detailed Description
The following examples are given for the detailed implementation and specific operation of the present invention, but the scope of the present invention is not limited to the following examples.
The following examples illustrate the present invention in detail with respect to a single grain crop, wheat. As shown in fig. 1, a method for rapidly and nondestructively determining defective grains of single-grain crops comprises the following steps:
s1: and collecting single-grain crop samples, wherein the collected single-grain crop samples are different varieties harvested at different times and different production places, the total number of the single-grain crop samples is 3000, and the number of the imperfect grains of each type of single-grain crop accounts for 25% -75% of the total number of the samples. The imperfect single-grain crop grains comprise scab grains, wormhole grains, germination grains, damaged grains and mildew grains. And (3) assigning the value of each detected imperfect single-grain crop to be 2, assigning the value of the normal single-grain crop to be 1, and establishing a category information matrix as a category label.
As shown in Table 1, 6 wheat samples of different varieties were collected in different regions of China. And selecting normal grain wheat and various imperfect grain wheat for 2169 grains in total for each sample according to GB 1351-2008. Wherein 75% of samples, namely 1631 wheat, are randomly selected as a calibration set for modeling; the remaining 25%, 538 grains, were used as validation sets to validate the predicted performance of the model.
TABLE 1 wheat sample sources and number statistics
The category label of the detected normal wheat is marked as 1, the detected imperfect wheat comprises scab grains, worm-eaten grains, germinated grains, damaged grains and mildewed grains, the category labels of the detected normal wheat are all marked as 2, the size of a category information matrix is established to be 1631 rows and 1 columns, and each element in the matrix sequentially corresponds to the category label of each correction set wheat.
S2: and (3) acquiring the near infrared spectrum of the single grain crop, wherein the near infrared spectrum is a near infrared diffuse reflection spectrum or a near infrared diffuse transmission spectrum. Using a Bruker lightThe spectrometer collects the spectra of all wheat samples. When collecting samples, each wheat grain is horizontally placed at the center of a sample window, and the diffuse reflection scanning parameters of a sample cup are used for collecting the wheat grains, wherein the spectrum scanning range is 11987.2-3996.8cm-1Resolution of 8cm-1And the number of scanning times is 32, and in the scanning process, 1 spectrum is respectively collected on the abdomen and the back of the wheat and then averaged to be used as the spectrum of the sample.
The spectra of the collected normal wheat, lesion wheat, worm-eaten wheat, germinated wheat, damaged wheat and moldy wheat were averaged, and the averaged spectral images are shown in fig. 2. As can be seen from FIG. 2, the difference between normal wheat and various imperfect grains of wheat is significant, which shows that the near infrared spectrum can reflect the difference between the physicochemical components of normal wheat and imperfect grains of wheat, so that the two types of wheat can be distinguished by modeling.
S3: after acquiring the correction set spectrum and the corresponding category information, preprocessing the spectrum, selecting effective wavelengths and establishing a spectrum matrix; then 2 methods are respectively adopted: the wheat imperfect grain model is constructed by a Discriminant Partial least squares algorithm (PLS-DA) and a Stacked self-coding algorithm (SAE).
Constructing a PLS-DA model of imperfect wheat grains: using the optimization software of the Bruker spectrometer to optimize parameters, and obtaining the optimal spectrum pretreatment as follows: the spectral range is cut off at 9596.6-7992cm-1And 5600.6-4790cm-1Then, the first derivative + vector normalization is performed on the spectrum, and the number of points is smoothed 17. After the spectrum pretreatment is carried out on the spectrum range, a spectrum matrix is obtained, wherein the spectrum matrix is 1631 rows and 1037 columns, 1631 is the number of the calibration set samples, 1037 is the number of the spectrum wavelength points, each row corresponds to one pretreated calibration set sample spectrum, different rows correspond to different pretreated calibration set sample spectra, and the spectrum matrix and the class information matrix correspond to the same wheat sample in each row.
Using PLS-DA to construct a relation model between the spectrum matrix and the category matrix, setting the number of principal components as 5, setting a threshold value as 1.57, judging the wheat to be normal when the predicted value is less than or equal to 1.57, and judging the wheat to be incomplete when the predicted value is greater than 1.57Wheat, R of the model2The correlation index was 0.5309, and the RMSECV cross-validation root mean square was 0.298. A scatter plot of the PLS-DA model against the predicted values and class labels of the correction set samples is shown in FIG. 3. With 1.57 as the threshold in fig. 3, most samples with a normal wheat class label of 1 were correctly classified under the threshold, while most samples with an imperfect grain wheat class label of 2 were correctly classified above the threshold. Specifically, in this embodiment, 165 models are misjudged. Wherein, the normal wheat is wrongly judged to be 79 defective grains, the diseased speckled grains, the worsted grains, the germinated grains, the damaged grains and the mildewed grains are respectively wrongly judged to be 21, 35, 21, 6 and 3 normal grains, the respective correct recognition rates are respectively 80.96%, 94.70%, 84.58%, 92.88%, 97.65% and 93.02%, and the total average correct classification rate is 89.88%.
SAE model construction of imperfect wheat grains: and (4) carrying out [ -1,1] normalization pretreatment on the spectrum of the correction set to obtain a spectrum matrix. And then training the spectrum matrix and the category information matrix by using an SAE algorithm to construct a model. The parameters of the model are as follows: the sparsity parameter is 0.1, the weight attenuation coefficient is 0.003, the weight of the sparse penalty term is 3, the neural network parameter of each layer is trained by using a minFunc gradient descent method toolbox, the training function is l-bfgs, and the maximum iteration frequency is 400 times. The number of the input layers is equal to the number of wavelength points of a spectrum, and is 1037; 2 hidden layers are arranged, the neuron node number of the hidden layer 1 is set as 100, and the neuron node number of the hidden layer 2 is set as 200. The size number of the output layer is 2, wherein the defective wheat is judged when the output value is 2, and the normal wheat is judged when the output value is 1.
The training results of the model on the correction set samples are shown in fig. 4. 1631 in the correction set, the first 1216 grains are imperfect grains, and the last 415 grains are normal grains. As can be seen from fig. 4, the imperfect grains and normal wheat were largely correctly discriminated to be predicted as 2 and 1, respectively. Specifically, in this embodiment, the model misjudges 50 pieces in total. Wherein the normal wheat is wrongly judged as 19 imperfect grains, the diseased speckled grains, the worsted grains, the germinated grains, the damaged grains and the mildewed grains are wrongly judged as 12, 5, 10, 4 and 0 normal grains respectively, the correct recognition rates are 95.42%, 96.96%, 97.79%, 96.61%, 98.82% and 100%, and the total average correct recognition rate is 96.93%.
S4: and distinguishing normal single-grain crops and imperfect single-grain crops by using the established model. For the verification set wheat sample, near infrared spectrum data are collected under the same condition, the data are subjected to the same spectrum pretreatment, and then the established model is used for distinguishing normal wheat and imperfect wheat.
In this example, the wheat samples to be tested were 538 validation set wheat samples, wherein the first 404 were imperfect wheat grains and the last 134 were normal wheat grains. For different modeling modes, the results of judging imperfect grains of wheat are respectively as follows:
firstly, predicting result of PLS-DA model of imperfect wheat grains on verification set
And (3) predicting the verification set wheat by adopting a PLS-DA model of imperfect wheat grains, wherein the discrimination result is shown in a figure 5. Most of the imperfect grain wheat and the normal grain wheat in the graph are correctly judged by the model to be samples with output values respectively larger than 1.57 and smaller than 1.57, and are correctly judged by the model to be normal grains. Specifically, in this embodiment, the total number of false positives is 49. Wherein, the normal wheat is wrongly judged as 17 imperfect grains, the diseased grains, the worsted grains, the germinated grains, the damaged grains and the mildewed grains are wrongly judged as 7 normal grains, 14 normal grains, 7 normal grains, 2 normal grains and 2 normal grains, the respective correct recognition rates are respectively 87.31 percent, 94.70 percent, 81.33 percent, 92.86 percent, 97.65 percent and 85.71 percent, and the total average correct recognition rate is 90.89 percent.
② the prediction result of imperfect wheat SAE model to verification set
And (3) predicting the verification set wheat by adopting an SEA model of imperfect wheat grains, wherein the discrimination result is shown in figure 6. Most of the imperfect grain wheat and normal wheat in the graph are correctly judged by the model to have output values of 2 and 1, respectively. Specifically, in this example, the number of erroneous judgments was 36, wherein the normal wheat was erroneously judged as the defective grain 13, and the lesion grain, the worm-eaten grain, the germinated grain, the damaged grain, and the mildewed grain were erroneously judged as the normal grains 5, 7, 8, 2, and 1, and the respective correct recognition rates were 90.31%, 96.21%, 90.67%, 91.84%, 97.65%, and 92.86%, respectively, and the total average correct recognition rate was 93.31%.
Specifically, the first step, the second step, the third step and the fourth step are implemented on the basis of a near infrared spectrum analysis instrument platform with a single particle detection function.
Through the technical scheme, the detection process of the rapid nondestructive judgment method for the imperfect grains of the single-grain crops does not depend on subjective judgment influence of inspectors, and the detection result is objective and accurate; the detection process does not damage the sample, and is rapid, simple and convenient. The invention can be implemented not only on the basis of the spectral analysis instrument with the single-particle detection function, namely the Bruker MPA type Fourier transform near infrared spectrometer, but also on the basis of a single-particle detection and separation near infrared spectral analysis instrument platform, such as a QSC-1 type crop quality intelligent detection sorter developed by the institute of society for fertilizer and Material science of China, so as to realize the efficient and quantitative judgment of imperfect wheat grains. Besides the wheat indicated in the embodiment, the method can also be popularized to the imperfect grain discrimination of other single-grain crops, such as corn, rice and the like; meanwhile, the method can be used in combination with other quality detection technologies, such as a near infrared spectrum detection model combining wheat moisture and hardness, rapid and nondestructive detection of multiple indexes such as wheat moisture, imperfect grains and hardness can be realized simultaneously, automation of multiple quality detection of wheat is realized, and the food industry is better served.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (6)
1. A rapid nondestructive judgment method for imperfect grains of single-grain crops is characterized by comprising the following steps:
the method comprises the following steps: collecting single-grain crop samples, detecting the imperfect grain condition of each single-grain crop, and establishing a category information matrix;
step two: collecting the near infrared spectrum of single grains of the single-grain crop;
step three: constructing a near-infrared discrimination analysis model of imperfect grains of single-grain crops;
step four: distinguishing normal single-grain crops and imperfect single-grain crops by using the established model;
the third step specifically comprises:
spectrum preprocessing, selecting effective wavelength and establishing a spectrum matrix;
b, constructing a relation model between the spectrum matrix and the single-grain crop category information matrix by adopting a chemometrics algorithm; the chemometrics algorithm is a discriminant partial least square algorithm or a stacked self-encoder algorithm;
when a discriminative partial least square algorithm is adopted to construct a near-infrared discriminative analysis model of imperfect grains of single-grain crops, the corresponding spectrum pretreatment is as follows: cut at 9596.6-7992cm-1And 5600.6-4790cm-1The spectrum of (2), the first derivative + vector normalization is carried out on the spectrum, and the number of smooth points is 17; spectrum matrixes are obtained after spectrum preprocessing, and a discriminant partial least square algorithm is adopted to construct a relation model, wherein the relation model is a discriminant partial least square algorithm model, and the specific parameters are as follows: setting the main component number as 5, the classification threshold value as 1.57, judging that the crop is a normal single-grain crop when the predicted value is less than or equal to 1.57, and judging that the crop has the predicted value>Imperfect granules at 1.57;
when a stack type self-encoder algorithm is adopted to construct a near-infrared discrimination analysis model of imperfect grains of single-grain crops, the corresponding spectrum pretreatment is as follows: -1,1] normalization; spectrum matrixes are obtained after spectrum preprocessing, a stack type self-encoder algorithm is adopted to construct a relation model, the relation model is a stack type self-encoder algorithm model, and the specific parameters are as follows: setting the sparsity parameter to be 0.1, the weight attenuation coefficient to be 0.003, the weight of the sparse penalty term to be 3, training each layer of neural network parameter by using a minFunc gradient descent method, setting the training function to be l-bfgs, and setting the maximum iteration number to be 400 times; the size of the input layer is equal to the number of wavelength points of the acquired spectrum; 2 hidden layers are arranged, the number of the neuron nodes of the hidden layer 1 is set to be 100, and the number of the neuron nodes of the hidden layer 2 is set to be 200; the output layer size is 2, the defective kernels are determined when the output value is 2, and the normal single-kernel crops are determined when the output value is 1.
2. The method for rapidly and nondestructively judging the defective grains of the single-grain crop according to claim 1, wherein the first step, the second step, the third step and the fourth step are implemented on a platform based on a near infrared spectroscopy analyzer with a single-grain detection function.
3. The method as claimed in claim 1, wherein the collected single kernel crop samples in the first step are different varieties harvested at different times and different production places, and the total number of the single kernel crop samples is 1000-3000, wherein the number of the single kernel crop imperfect grains in each type accounts for 25% -75% of the total number of the samples.
4. The method for rapidly and nondestructively judging the defective single-grain crops according to claim 1, wherein in the step one, the detected various defective single-grain crops are assigned with the value of 2, the detected normal single-grain crops are assigned with the value of 1, and the classification information matrix is established as a classification label.
5. The method for rapidly and nondestructively judging the defective grains of the single-grain crops according to claim 1, wherein the near infrared spectrum in the second step is a near infrared diffuse reflection spectrum or a near infrared diffuse transmission spectrum.
6. The method for rapidly and nondestructively judging the defective grains of the single-grain crop according to any one of claims 1 to 5, wherein the single-grain crop comprises rice, wheat or corn, and the defective grains comprise immature grains, diseased speckled grains, wormhead grains, germinated grains, damaged grains and mildewed grains; imperfect wheat grains include diseased speckles, worm-eaten grains, germinated grains, damaged grains and mildewed grains; the imperfect corn grains include scab grains, worm-eaten grains, germinated grains, damaged grains, mildewed grains and heat-damaged grains.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811390652.5A CN109470648B (en) | 2018-11-21 | 2018-11-21 | Rapid nondestructive determination method for imperfect grains of single-grain crops |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811390652.5A CN109470648B (en) | 2018-11-21 | 2018-11-21 | Rapid nondestructive determination method for imperfect grains of single-grain crops |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109470648A CN109470648A (en) | 2019-03-15 |
| CN109470648B true CN109470648B (en) | 2021-03-16 |
Family
ID=65674223
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811390652.5A Active CN109470648B (en) | 2018-11-21 | 2018-11-21 | Rapid nondestructive determination method for imperfect grains of single-grain crops |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109470648B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110926905B (en) * | 2019-12-30 | 2022-05-06 | 中国农业科学院作物科学研究所 | Corn imperfect granule standard sample and preparation method thereof |
| CN112580540A (en) * | 2020-12-23 | 2021-03-30 | 安徽高哲信息技术有限公司 | Artificial intelligent crop processing system and method |
| CN112924412B (en) * | 2021-01-22 | 2022-11-04 | 中国科学院合肥物质科学研究院 | Single-grain rice variety authenticity distinguishing method and device based on near infrared spectrum |
| CN114062305B (en) * | 2021-10-15 | 2024-01-26 | 中国科学院合肥物质科学研究院 | Single grain variety identification method and system based on near infrared spectrum and 1D-In-Resnet network |
| WO2023200006A1 (en) * | 2022-04-15 | 2023-10-19 | 国立研究開発法人農業・食品産業技術総合研究機構 | Determination device, determination method, and control program |
| CN114813631B (en) * | 2022-06-23 | 2022-09-06 | 中化现代农业有限公司 | Wheat variety purity detection method and device and electronic equipment |
| CN116311232B (en) * | 2023-05-26 | 2023-09-19 | 安徽高哲信息技术有限公司 | Imperfect grain type identification method and device and computer equipment |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103091284B (en) * | 2013-02-28 | 2015-07-15 | 河南工业大学 | Near-infrared spectrum technology based method for rapidly identifying wheat grains with FHB (Fusarium Head Blight) |
| CN105115938A (en) * | 2015-07-29 | 2015-12-02 | 北京航空航天大学 | Quantitative determination method for melamine adulteration in feedstuff |
| CN106503734A (en) * | 2016-10-14 | 2017-03-15 | 哈尔滨工程大学 | Based on trilateral filter and the image classification method of the sparse autocoder of storehouse |
| CN107609579A (en) * | 2017-08-25 | 2018-01-19 | 西安电子科技大学 | Classification of radar targets method based on sane variation self-encoding encoder |
| CN108801971A (en) * | 2018-04-08 | 2018-11-13 | 浙江大学 | Mould based on high light spectrum image-forming technology infects the detection method of Chinese chestnut |
-
2018
- 2018-11-21 CN CN201811390652.5A patent/CN109470648B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103091284B (en) * | 2013-02-28 | 2015-07-15 | 河南工业大学 | Near-infrared spectrum technology based method for rapidly identifying wheat grains with FHB (Fusarium Head Blight) |
| CN105115938A (en) * | 2015-07-29 | 2015-12-02 | 北京航空航天大学 | Quantitative determination method for melamine adulteration in feedstuff |
| CN106503734A (en) * | 2016-10-14 | 2017-03-15 | 哈尔滨工程大学 | Based on trilateral filter and the image classification method of the sparse autocoder of storehouse |
| CN107609579A (en) * | 2017-08-25 | 2018-01-19 | 西安电子科技大学 | Classification of radar targets method based on sane variation self-encoding encoder |
| CN108801971A (en) * | 2018-04-08 | 2018-11-13 | 浙江大学 | Mould based on high light spectrum image-forming technology infects the detection method of Chinese chestnut |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109470648A (en) | 2019-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109470648B (en) | Rapid nondestructive determination method for imperfect grains of single-grain crops | |
| Caporaso et al. | Near-Infrared spectroscopy and hyperspectral imaging for non-destructive quality assessment of cereal grains | |
| Walsh et al. | The uses of near infra-red spectroscopy in postharvest decision support: A review | |
| Agelet et al. | Feasibility of near infrared spectroscopy for analyzing corn kernel damage and viability of soybean and corn kernels | |
| Huang et al. | Review of seed quality and safety tests using optical sensing technologies | |
| Theanjumpol et al. | Non-destructive identification and estimation of granulation in ‘Sai Num Pung’tangerine fruit using near infrared spectroscopy and chemometrics | |
| TU et al. | Selection for high quality pepper seeds by machine vision and classifiers | |
| Agelet et al. | Limitations and current applications of Near Infrared Spectroscopy for single seed analysis | |
| ElMasry et al. | Utilization of computer vision and multispectral imaging techniques for classification of cowpea (Vigna unguiculata) seeds | |
| Li et al. | Discriminating varieties of tea plant based on Vis/NIR spectral characteristics and using artificial neural networks | |
| Singh et al. | Fungal detection in wheat using near-infrared hyperspectral imaging | |
| CN113049530B (en) | Single-seed corn seed moisture content detection method based on near-infrared hyperspectrum | |
| Singh et al. | Detection of sprouted and midge‐damaged wheat kernels using near‐infrared hyperspectral imaging | |
| CN108680515B (en) | Single-grain rice amylose quantitative analysis model construction and detection method thereof | |
| CN112924412B (en) | Single-grain rice variety authenticity distinguishing method and device based on near infrared spectrum | |
| CN108613943B (en) | Near-infrared single-grain crop component detection method based on spectrum morphology transfer | |
| Hadiwijaya et al. | Multi-product calibration model for soluble solids and water content quantification in Cucurbitaceae family, using visible/near-infrared spectroscopy | |
| Wang et al. | Discrimination of cracked soybean seeds by near-infrared spectroscopy and random forest variable selection | |
| Pang et al. | Rapid seed viability prediction of Sophora japonica by improved successive projection algorithm and hyperspectral imaging | |
| CN111272668A (en) | Construction method of wheat variety identification model | |
| Rathna Priya et al. | Characterising corn grain using infrared imaging and spectroscopic techniques: a review | |
| Wu et al. | Deep convolution neural network with weighted loss to detect rice seeds vigor based on hyperspectral imaging under the sample-imbalanced condition | |
| Liu et al. | Identification of heat damage in imported soybeans based on hyperspectral imaging technology | |
| Jha et al. | Authentication of mango varieties using near-infrared spectroscopy | |
| Hong et al. | Nondestructive prediction of rice seed viability using spectral and spatial information modeling of visible–near infrared hyperspectral images |
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 |