CN110479636A - Method and device based on neural network automatic sorting tobacco leaf - Google Patents
Method and device based on neural network automatic sorting tobacco leaf Download PDFInfo
- Publication number
- CN110479636A CN110479636A CN201910660071.7A CN201910660071A CN110479636A CN 110479636 A CN110479636 A CN 110479636A CN 201910660071 A CN201910660071 A CN 201910660071A CN 110479636 A CN110479636 A CN 110479636A
- Authority
- CN
- China
- Prior art keywords
- tobacco leaf
- physical feature
- training
- information data
- sorted
- 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
-
- A—HUMAN NECESSITIES
- A24—TOBACCO; CIGARS; CIGARETTES; SIMULATED SMOKING DEVICES; SMOKERS' REQUISITES
- A24B—MANUFACTURE OR PREPARATION OF TOBACCO FOR SMOKING OR CHEWING; TOBACCO; SNUFF
- A24B3/00—Preparing tobacco in the factory
- A24B3/16—Classifying or aligning leaves
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/34—Sorting according to other particular properties
- B07C5/342—Sorting according to other particular properties according to optical properties, e.g. colour
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B07—SEPARATING SOLIDS FROM SOLIDS; SORTING
- B07C—POSTAL SORTING; SORTING INDIVIDUAL ARTICLES, OR BULK MATERIAL FIT TO BE SORTED PIECE-MEAL, e.g. BY PICKING
- B07C5/00—Sorting according to a characteristic or feature of the articles or material being sorted, e.g. by control effected by devices which detect or measure such characteristic or feature; Sorting by manually actuated devices, e.g. switches
- B07C5/36—Sorting apparatus characterised by the means used for distribution
- B07C5/361—Processing or control devices therefor, e.g. escort memory
Landscapes
- Manufacture Of Tobacco Products (AREA)
- Image Analysis (AREA)
Abstract
The application provides a kind of method and device based on neural network automatic sorting tobacco leaf, this method comprises: acquiring the natural light picture of training tobacco leaf, obtains training the physical feature information data of tobacco leaf by preset tobacco leaf characteristic dimension information analysis natural light picture;Training tobacco leaf is subjected to spectrum analysis, obtains the spectral signature information data for training tobacco leaf;The physical feature information data of training tobacco leaf and spectral signature information data is corresponding, it carries out neural metwork training and obtains the physical feature of tobacco leaf and the correlation model of spectral signature;The spectrum tobacco leaf grading strategy that tobacco leaf is established according to spectral signature information data obtains the Natural classification strategy of tobacco leaf based on correlation model;The natural image of tobacco leaf to be sorted is obtained, analysis obtains the physical feature information data of tobacco leaf to be sorted, compares to obtain the sorting rank of tobacco leaf to be sorted with Natural classification strategy, sorts tobacco leaf to be sorted according to sorting rank.The present invention realizes the tobacco leaf sorting of intelligence, efficient, standardization and low cost.
Description
Technical field
This application involves the technical fields of automation control more particularly to a kind of based on neural network automatic sorting tobacco leaf
Method and device.
Background technique
Tobacco leaf is the raw material of tobacco industry, is a kind of important industrial crops, and the cultivated area in China is also very wide
It is general, it occupies an important position in the agricultural production in China.The quality and production technology of tobacco leaf will have a direct impact on by tobacco leaf system
At tobacco product quality.During tobacco leaf planting, due to by different geographical weather, environment, soil and tobacco variety,
Inserted part, planting system and baking production and the influence for saving technique, so that the quality difference of tobacco leaf raw material is very big.
Therefore, it is necessary to the tobacco leaves to different qualities to be classified, and be directed to different brackets cigarette with different fabrication process conditions
Leaf makes to obtain the tobacco leaf product of different brackets economic value.The main foundation of tobacco leaf grading be growth site based on tobacco leaf,
The features such as color, maturity, blade construction, residual wound, length dimension.The existing stage, to the classification of tobacco leaf mainly using according to
According to national tobacco leaf grading standard, being sorted by artificial sense organ, the large labor intensity that needs to expend, manpower are more, low efficiency,
And influenced by artificial subjective factor, so that the precision of tobacco leaf sorting is low, sorting quality is not achieved classification and requires.Due to tobacco leaf shape
State is various, does not have specific textural characteristics, and color structure is also complex, only by artificially single from form, texture and color etc.
Feature carries out the case where sorting is easy to generation erroneous judgement and fails to judge to distinguish tobacco leaf.
It is also higher and higher to the quality pursuit of tobacco leaf product with the promotion of people's living standard, in quality of tobacco requirement
The method of the today stepped up, traditional artificial sorting tobacco leaf are neither able to satisfy the requirement of tobacco leaf grading quality, are also not achieved point
The requirement for picking efficiency also adds the cost of tobacco leaf sorting, is unfavorable for the development of tobacco leaf industry.
Therefore, how a kind of intelligent, efficient, standardization is provided and the inexpensive scheme sorted to tobacco leaf is this field
Technical problem urgently to be resolved.
Summary of the invention
The application's is designed to provide a kind of method and device based on neural network automatic sorting tobacco leaf, solves existing
Tobacco leaf sorting is without standardization in technology, sorts the technical issues of poor quality, sorting efficiency are low, at high cost.
In order to achieve the above objectives, the application provides a kind of method based on neural network automatic sorting tobacco leaf, comprising:
The natural light picture for acquiring training tobacco leaf, is obtained by natural light picture described in preset tobacco leaf characteristic dimension information analysis
To the physical feature information data of the trained tobacco leaf;
The trained tobacco leaf is subjected to spectrum analysis, obtains the spectral signature information data of the trained tobacco leaf;It will be described
The physical feature information data and spectral signature information data of training tobacco leaf are corresponding, carry out neural metwork training and obtain tobacco leaf
Physical feature and spectral signature correlation model;
The spectrum tobacco leaf grading strategy of the tobacco leaf is established according to the spectral signature information data, is based on the association mould
Type obtains the Natural classification strategy of the tobacco leaf;
The natural image of tobacco leaf to be sorted is obtained, analysis obtains the physical feature information data of the tobacco leaf to be sorted, with
The Natural classification strategy compares to obtain the sorting rank of the tobacco leaf to be sorted, described wait divide according to sorting rank sorting
Pick tobacco leaf.
Optionally, wherein the natural light picture for acquiring training tobacco leaf, by described in preset tobacco leaf characteristic dimension information analysis
Natural light picture obtains the physical feature information data of the trained tobacco leaf, are as follows:
With the natural light picture of the training tobacco leaf of pre-set velocity acquisition predetermined quantity;
The natural light picture is analyzed simultaneously by preset tobacco leaf characteristic dimension information, it is respectively right to obtain the trained tobacco leaf
The physical feature information data answered.
Optionally, wherein this method further include:
The physical feature information data of the tobacco leaf to be sorted and the Natural classification strategy compare, no related sorting rank
When, the tobacco leaf to be sorted is included into retraining tobacco classification;
The retraining tobacco leaf is extracted into nature characteristic information data and Spectral Properties reference as training tobacco leaf again again
Data are ceased, the correlation model that neural metwork training is updated is imported;
The correlation model based on update sorts tobacco leaf to be sorted.
Optionally, wherein the natural light picture for acquiring training tobacco leaf, by described in preset tobacco leaf characteristic dimension information analysis
Natural light picture obtains the physical feature information data of the trained tobacco leaf, are as follows:
Training tobacco leaf/tobacco leaf to be sorted is imported into detection darkroom, polishing is carried out by preset constant light source, with camera pair
Training Tobacco Leaves/tobacco leaf to be sorted described in every group is taken pictures, and the natural light picture of the trained tobacco leaf/tobacco leaf to be sorted is acquired;
By natural light picture described in preset tobacco leaf characteristic dimension information analysis, the trained tobacco leaf/cigarette to be sorted is obtained
The physical feature information data of leaf.
Optionally, wherein the trained tobacco leaf is subjected to spectrum analysis, obtains the spectral signature information of the trained tobacco leaf
Data, are as follows:
The trained tobacco leaf is imaged with preset spectral band using bloom spectrum sensor, obtains the bloom for training tobacco leaf
Spectrogram picture;
With high spectrum image described in preset Spectra feature extraction project analysis, the Spectral Properties reference for stating trained tobacco leaf is obtained
Cease data.
On the other hand, the present invention also provides a kind of devices based on neural network automatic sorting tobacco leaf, comprising: physical feature
Analysis processor, physical feature and spectral signature association processor, Natural classification policy handler and tobacco leaf sorting processor;Its
In,
The physical feature analysis processor, is connected with the physical feature with spectral signature association processor, acquisition
The natural light picture of training tobacco leaf, obtains the trained cigarette by natural light picture described in preset tobacco leaf characteristic dimension information analysis
The physical feature information data of leaf;
The physical feature and spectral signature association processor, with the physical feature analysis processor and Natural classification plan
Slightly processor is connected, and the trained tobacco leaf is carried out spectrum analysis, obtains the spectral signature information data of the trained tobacco leaf;
The physical feature information data and spectral signature information data of the trained tobacco leaf is corresponding, it carries out neural metwork training and obtains
To the physical feature of tobacco leaf and the correlation model of spectral signature;
The Natural classification policy handler, with the physical feature and spectral signature association processor and tobacco leaf sorting office
Reason device is connected, and the spectrum tobacco leaf grading strategy of the tobacco leaf is established according to the spectral signature information data, is based on the pass
Gang mould type obtains the Natural classification strategy of the tobacco leaf;
The tobacco leaf sorting processor is connected with the Natural classification policy handler, obtains tobacco leaf to be sorted oneself
Right image, analysis obtain the physical feature information data of the tobacco leaf to be sorted, compare to obtain institute with the Natural classification strategy
The sorting rank for stating tobacco leaf to be sorted sorts the tobacco leaf to be sorted according to the sorting rank.
Optionally, wherein the physical feature analysis processor, comprising: natural light training tobacco leaf picture collection device and from
Right feature information processing device;Wherein,
The natural light training tobacco leaf picture collection device, is connected with the physical feature message handler, with default speed
The natural light picture of the training tobacco leaf of degree acquisition predetermined quantity;
The physical feature message handler, with natural light training tobacco leaf picture collection device and physical feature and spectrum
Feature association processor is connected, and analyzes the natural light picture simultaneously by preset tobacco leaf characteristic dimension information, obtains described
The training corresponding physical feature information data of tobacco leaf.
Optionally, wherein the device further include: tobacco leaf feature more new processor, with the tobacco leaf sorting processor and certainly
Right signature analysis processor is connected,
It is compared in the physical feature information data of the tobacco leaf to be sorted and the Natural classification strategy, no related sorting grade
When other, the tobacco leaf to be sorted is included into retraining tobacco classification;
The retraining tobacco leaf is extracted into nature characteristic information data and Spectral Properties reference as training tobacco leaf again again
Data are ceased, the correlation model that neural metwork training is updated is imported;
The correlation model based on update sorts tobacco leaf to be sorted.
Optionally, wherein the physical feature analysis processor, comprising: tobacco natural light picture collection device and naturally spy
Levy message handler;Wherein,
The tobacco natural light picture collection device, is connected with the physical feature message handler, will training tobacco leaf/to
Sort tobacco leaf import detection darkroom, by preset constant light source carry out polishing, with camera described in every group training Tobacco Leaves/to
Sorting tobacco leaf is taken pictures, and the natural light picture of the trained tobacco leaf/tobacco leaf to be sorted is acquired;
The physical feature message handler, with the tobacco natural light picture collection device and physical feature and spectral signature
Association processor is connected, by natural light picture described in preset tobacco leaf characteristic dimension information analysis, obtain the trained tobacco leaf/
The physical feature information data of tobacco leaf to be sorted.
Optionally, wherein the physical feature and spectral signature association processor, comprising: high spectrum image collector, light
Spectrum signature extraction processor and correlation model create processor;Wherein,
The high spectrum image collector is connected with the Spectra feature extraction processor, utilizes bloom spectrum sensor
The trained tobacco leaf is imaged with preset spectral band, obtains the high spectrum image for training tobacco leaf;
The Spectra feature extraction processor is connected with the high spectrum image collector and correlation model creation processor
It connects, with high spectrum image described in preset Spectra feature extraction project analysis, obtains the spectral signature Information Number for stating trained tobacco leaf
According to;
The correlation model creates processor, with the physical feature analysis processor and Spectra feature extraction processor phase
Connection, the physical feature information data and spectral signature information data of the trained tobacco leaf is corresponding, carry out neural network
Training obtains the physical feature of tobacco leaf and the correlation model of spectral signature.
The method and device based on neural network automatic sorting tobacco leaf of the application, realization have the beneficial effect that:
(1) method and device based on neural network automatic sorting tobacco leaf of the application obtains cigarette according to the feature of tobacco leaf
The natural light image analysis of leaf obtains natural light characteristic, and the spectrum analysis for obtaining tobacco leaf obtains spectrum characteristic data, benefit
Being associated between natural light feature and spectral signature is established with neural network learning, point based on spectral signature setting tobacco leaf sorting
Grade strategy, the automatic sorting system of tobacco leaf is established with this, realizes the tobacco leaf sorting of intelligence, efficient, standardization and low cost.
(2) method and device based on neural network automatic sorting tobacco leaf of the application, is established using neural network learning
Being associated between tobacco natural light feature and spectral signature obtains tobacco leaf sorting control from various dimensions training study according to tobacco leaf feature
System, is learnt renewal learning model to unrecognized tobacco leaf again, is adapted to the tobacco leaf sorting of variety classes, region, is mentioned
The intelligence and accuracy of tobacco leaf sorting are risen.
(3) method and device based on neural network automatic sorting tobacco leaf of the application, is established using neural network learning
Being associated between tobacco natural light feature and spectral signature obtains tobacco leaf sorting control from various dimensions training study according to tobacco leaf feature
System, by adjusting or selecting the spectral signature of EO-1 hyperion can choose the tobacco leaf grading of different hierarchical policies, thus as needed
The tobacco leaf of various different sorting classifications is obtained, automatic sorting tobacco leaf widespread popularity is improved.
Detailed description of the invention
In order to illustrate the technical solutions in the embodiments of the present application or in the prior art more clearly, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
The some embodiments recorded in application can also be obtained according to these attached drawings other attached for those skilled in the art
Figure.
Fig. 1 is a kind of flow diagram of the method based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 2 is the flow diagram of second of method based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 3 is the flow diagram of the third method based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 4 is the flow diagram of the 4th kind of method based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 5 is the flow diagram of the 5th kind of method based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 6 is a kind of structural schematic diagram of the device based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 7 is the structural schematic diagram of second of device based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 8 is the structural schematic diagram of the third device based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Fig. 9 is the structural schematic diagram of the 4th kind of device based on neural network automatic sorting tobacco leaf in the embodiment of the present invention;
Figure 10 is the structural schematic diagram of the 5th kind of device based on neural network automatic sorting tobacco leaf in the embodiment of the present invention.
Specific embodiment
Below with reference to the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Ground description, it is clear that described embodiment is some embodiments of the present application, instead of all the embodiments.Based on the application
In embodiment, those skilled in the art's every other embodiment obtained without making creative work, all
Belong to the range of the application protection.
Embodiment 1
As shown in Figure 1, for a kind of flow diagram based on neural network automatic sorting tobacco leaf in the present embodiment, by bloom
Imaging region is counter shifts on nature visible light region for spectrum, establishes tobacco leaf EO-1 hyperion multidimensional characteristic library, then that EO-1 hyperion multidimensional is special
Machine learning is done in sign library and natural light multidimensional characteristic library, can be established tobacco leaf by the model learning method of lightgbm and be sorted
Correlation models in journey among natural light and spectral signature, and then carry out tobacco leaf automatic sorting.Specifically, this method includes such as
Lower step:
Step 101, the natural light picture for acquiring training tobacco leaf, by preset tobacco leaf characteristic dimension information analysis natural light figure
Piece obtains training the physical feature information data of tobacco leaf.
Optionally, tobacco leaf is taken off out according to the picture of tobacco leaf, color mean value and centerline length is found out, then gathered
Class becomes multiple parameters, and each flue-cured tobacco point can set parameters according to their own needs to do the preliminary sorting of tobacco leaf.It adopts
The mode of deep learning, the implicit network structure of 50~200 layers of building, for storing the low-dimensional and high dimensional feature of blade,
The characteristic dimension used can be red (Red), green (Green), indigo plant (Blue), tone (H), saturation degree (S), lightness (V).
Feature Engineering is established first since natural light, due to blade non-deployed condition, so will be first by the blade of front and back sides
Region is individually taken out, and construction feature engineering respectively, and natural light Feature Engineering essential characteristic parameter is RGB (RGB) value, in
Transparency, hue, saturation, intensity can be all added in Feature Engineering by the phase, and due to the Feature Engineering mistake of pixel
In wide in range, so different region area parameters can also be clustered out according to these foundation characteristics, it can also be with artificial constructed higher-dimension
Feature.
Training tobacco leaf is carried out spectrum analysis by step 102, obtains the spectral signature information data for training tobacco leaf;It will train
The physical feature information data and spectral signature information data of tobacco leaf are corresponding, carry out neural metwork training and obtain the natural special of tobacco leaf
The correlation model of sign and spectral signature.
Neural network, is a kind of method of machine learning, and the concept of deep learning is derived from the research of artificial neural network.Contain
The multilayer perceptron of more hidden layers is exactly a kind of deep learning structure.Deep learning is more abstracted by combination low-level feature formation
High level indicates attribute classification or feature, to find that the distributed nature of data indicates.
The concept of deep learning was proposed by Hinton et al. in 2006.Non- prison is proposed based on depth confidence network (DBN)
The layer-by-layer training algorithm of greed is superintended and directed, hope is brought to solve the relevant optimization problem of deep structure, then proposes multilayer autocoding
Device deep structure.Furthermore the convolutional neural networks that Lecun et al. is proposed are first real multilayered structure learning algorithms, it is utilized
Spatial correlation reduces number of parameters to improve training performance.
Deep learning is a kind of based on the method for carrying out representative learning to data in machine learning.Observation (such as a width
Image) various ways can be used to indicate, such as vector of each pixel intensity value, or be more abstractively expressed as a series of
Side, region of specific shape etc..And use certain specific representation methods be easier from example learning tasks (for example, face
Identification or human facial expression recognition).The benefit of deep learning is feature learning and the layered characteristic with non-supervisory formula or Semi-supervised
It extracts highly effective algorithm and obtains feature by hand to substitute.
Deep learning is a new field in machine learning research, and motivation is that foundation, simulation human brain are divided
The neural network of study is analysed, it imitates the mechanism of human brain to explain data, such as image, sound and text.
The same with machine learning method, also supervised learning and unsupervised learning divide different to depth machine learning method
Learning framework under the learning model very difference that establishes for example, convolutional neural networks (Convolutional neural
Networks, abbreviation CNNs) be exactly a kind of depth supervised learning under machine learning model, and depth confidence net (Deep
Belief Nets, abbreviation DBNs) it is exactly a kind of machine learning model under unsupervised learning.According at present in the acquisition of flue-cured tobacco point
Data from the point of view of, the scheme of Unsupervised clustering is not suitable for present case, can be using traditional number that have supervision to have mark
According to training neural network, it is preferable that the convolutional neural networks mode of learning of image recognition can be used.
Spectrum picture of the spectral resolution within the scope of the 10l order of magnitude is known as high spectrum image (Hyperspectral
Image).Remote sensing technology pass through the second half in 20th century development, no matter theoretically, technically and application it is upper have occurred it is great
Variation.Wherein, the appearance of hyper-spectral image technique and fast development are undoubtedly one aspect very outstanding in this variation.
By the bloom spectrum sensor being mounted on different spaces platform, i.e. imaging spectrometer, the ultraviolet of electromagnetic spectrum, visible light,
Near-infrared and mid infrared region are imaged target area with tens of to hundreds of continuous and subdivision spectral band simultaneously.
Spectrum analysis is done into tobacco leaf feeding laboratory while natural light construction feature engineering, it can be deduced that each tobacco leaf
Spectral signature, for recording the parameters such as protein content, likewise, spectral signature also can using region clustering method construct
High dimensional feature, last developing algorithm model analyze the relevance of natural light feature and spectral signature, can be with by spectral signature
The qualitative characteristics of tobacco leaf are well reflected out, the quality grading of tobacco leaf can be set in the qualitative characteristics based on tobacco leaf, thus, it is possible to
By the standardization of the quality grading of tobacco leaf, planningization, avoid the subjective judgement artificially sorted, improve tobacco leaf sorting quality and
Efficiency.
Step 103, the spectrum tobacco leaf grading strategy that tobacco leaf is established according to spectral signature information data, are obtained based on correlation model
To the Natural classification strategy of tobacco leaf.
After big data analysis, spectral signature is predicted with natural light, is established further according to tobacco leaf grading under spectrum certainly
Tobacco leaf grading under right light.It may select LightGBM model herein, in precision and xgboost is similar, but speed is very
Fastly, LightGBM can choose the decision Tree algorithms based on histogram, compared to the algorithm pre-sorted of another mainstream
(the exact algorithm in such as xgboost), histogram have many advantages, In on memory consumption and calculating cost
On histogram algorithm, further optimized by LightGBM, promptly obtaining a result can be with very convenient quick
Ground adjustment algorithm can save a large amount of time.
Step 104, the natural image for obtaining tobacco leaf to be sorted, analysis obtain the physical feature Information Number of tobacco leaf to be sorted
According to comparing to obtain the sorting rank of tobacco leaf to be sorted with Natural classification strategy, sort tobacco leaf to be sorted according to sorting rank.
Machine learning (Machine Learning, ML) is a multi-field cross discipline, be related to probability theory, statistics,
The multiple subjects such as Approximation Theory, convextiry analysis, algorithm complexity theory.Specialize in the study that the mankind were simulated or realized to computer how
Behavior reorganizes the existing structure of knowledge and is allowed to constantly improve the performance of itself to obtain new knowledge or skills.
It is the core of artificial intelligence, is the fundamental way for making computer have intelligence, application is throughout artificial intelligence
Every field, it is mainly using conclusion, comprehensive rather than deduction.
Machine learning herein and deep learning are not overlapped, although the classification deep learning identification of tobacco leaf also can be effective,
But the black box characteristic of deep learning rapidly can not observe and construct the feature being really classified, thus, it is possible to use deep learning
The classification of early period is done, the Feature Engineering or human intervention of machine learning are to facilitate subsequent tobacco leaf grading.
In some alternative embodiments, as shown in Fig. 2, being based on neural network automatic sorting for second in the present embodiment
The flow diagram of the method for tobacco leaf.Unlike Fig. 1, the natural light picture of training tobacco leaf is acquired, by preset tobacco leaf
Characteristic dimension information analysis natural light picture obtains training the physical feature information data of tobacco leaf, are as follows:
Step 201, with pre-set velocity acquisition predetermined quantity training tobacco leaf natural light picture.
The image recognition under natural light is done using deep learning method, due to being contours extract, so operand is than frame
That extracts is much larger, a picture can be optimized to 0.6 second, and has 4 tobacco leaves inside a picture, can analyze simultaneously,
So averagely arrive classification of flue-cured tobacco leaves be 0.15 second one, a hour 14400 opens, one flue-cured tobacco point one day classification tobacco leaf exist
120000 or so, if efficiency is inadequate, partition can also be increased, 5 tobacco leaves can be put inside a picture, detection time is same
It is 0.6 second.The number and shooting speed of any shooting tobacco leaf image of selection according to actual needs should be in the range of the present embodiment.
Step 202 analyzes natural light picture simultaneously by preset tobacco leaf characteristic dimension information, obtains training tobacco leaf respectively right
The physical feature information data answered.
In some alternative embodiments, as shown in figure 3, being based on neural network automatic sorting for the third in the present embodiment
The flow diagram of the method for tobacco leaf.Unlike Fig. 1, further includes:
Step 301, the physical feature information data of tobacco leaf to be sorted and Natural classification strategy compare, no related sorting rank
When, tobacco leaf to be sorted is included into retraining tobacco classification.
Retraining tobacco leaf is extracted nature characteristic information data and Spectral Properties as training tobacco leaf again by step 302 again
Information data is levied, the correlation model that neural metwork training is updated is imported.
Step 303 sorts tobacco leaf to be sorted based on the correlation model of update.
In some alternative embodiments, as shown in figure 4, being based on neural network automatic sorting for the 4th kind in the present embodiment
The flow diagram of the method for tobacco leaf.Unlike Fig. 1, the natural light picture of training tobacco leaf is acquired, by preset tobacco leaf
Characteristic dimension information analysis natural light picture obtains training the physical feature information data of tobacco leaf, are as follows:
Training tobacco leaf/tobacco leaf to be sorted is imported detection darkroom by step 401, carries out polishing by preset constant light source,
It is taken pictures with camera to every group of trained Tobacco Leaves/tobacco leaf to be sorted, acquires the natural light picture of training tobacco leaf/tobacco leaf to be sorted.
Step 402 presses preset tobacco leaf characteristic dimension information analysis natural light picture, obtains training tobacco leaf/cigarette to be sorted
The physical feature information data of leaf.
In some alternative embodiments, as shown in figure 5, being based on neural network automatic sorting for the 5th kind in the present embodiment
The flow diagram of the method for tobacco leaf.Unlike Fig. 1, training tobacco leaf is subjected to spectrum analysis, obtains training tobacco leaf
Spectral signature information data, are as follows:
Step 501 is imaged training tobacco leaf with preset spectral band using bloom spectrum sensor, obtains training tobacco leaf
High spectrum image.
Step 502, with preset Spectra feature extraction project analysis high spectrum image, obtain the Spectral Properties for stating trained tobacco leaf
Levy information data.
In some alternative embodiments, as shown in fig. 6, being based on neural network automatic sorting cigarette to be a kind of in the present embodiment
The structural schematic diagram of the device 600 of leaf, the device can be used for implementing the above-mentioned method based on neural network automatic sorting tobacco leaf.
Specifically, which includes: physical feature analysis processor 601, physical feature and spectral signature association processor 602, nature
Hierarchical policy processor 603 and tobacco leaf sorting processor 604.
Wherein, physical feature analysis processor 601 is connected with physical feature with spectral signature association processor 602, adopts
The natural light picture for collecting trained tobacco leaf obtains oneself of trained tobacco leaf by preset tobacco leaf characteristic dimension information analysis natural light picture
Right characteristic information data.
Physical feature and spectral signature association processor 602, with physical feature analysis processor 601 and Natural classification strategy
Processor 603 is connected, and training tobacco leaf is carried out spectrum analysis, obtains the spectral signature information data for training tobacco leaf;It will train
The physical feature information data and spectral signature information data of tobacco leaf are corresponding, carry out neural metwork training and obtain the natural special of tobacco leaf
The correlation model of sign and spectral signature.
Natural classification policy handler 603 is handled with physical feature and spectral signature association processor 602 and tobacco leaf sorting
Device 604 is connected, and the spectrum tobacco leaf grading strategy of tobacco leaf is established according to spectral signature information data, obtains cigarette based on correlation model
The Natural classification strategy of leaf.
Tobacco leaf sorting processor 604 is connected with Natural classification policy handler 603, obtains the nature of tobacco leaf to be sorted
Image, analysis obtain the physical feature information data of tobacco leaf to be sorted, compare to obtain tobacco leaf to be sorted with Natural classification strategy
Rank is sorted, sorts tobacco leaf to be sorted according to sorting rank.
In some alternative embodiments, as shown in fig. 7, being based on neural network automatic sorting for second in the present embodiment
The structural schematic diagram of the device 700 of tobacco leaf.Unlike Fig. 6, physical feature analysis processor 601, comprising: natural light instruction
Practice tobacco leaf picture collection device 701 and physical feature message handler 702.
Wherein, natural light training tobacco leaf picture collection device 701, is connected with physical feature message handler 702, with default
The natural light picture of the training tobacco leaf of speed acquisition predetermined quantity.
Physical feature message handler 702, with natural light training tobacco leaf picture collection device 701 and physical feature and Spectral Properties
Sign association processor 602 is connected, and analyzes natural light picture simultaneously by preset tobacco leaf characteristic dimension information, obtains training tobacco leaf
Corresponding physical feature information data.
In some alternative embodiments, as shown in figure 8, being based on neural network automatic sorting for the third in the present embodiment
The structural schematic diagram of the device 800 of tobacco leaf.Unlike Fig. 6, the device further include: tobacco leaf feature more new processor 801,
It is connected with tobacco leaf sorting processor 604 and physical feature analysis processor 601.In the physical feature Information Number of tobacco leaf to be sorted
It is compared according to Natural classification strategy, when no correlation sorts rank, tobacco leaf to be sorted is included into retraining tobacco classification.
Retraining tobacco leaf is extracted into nature characteristic information data and spectral signature Information Number as training tobacco leaf again again
According to the correlation model that importing neural metwork training is updated;Tobacco leaf to be sorted is divided based on the correlation model of update
It picks.
In some alternative embodiments, as shown in figure 9, being based on neural network automatic sorting for the 4th kind in the present embodiment
The structural schematic diagram of the device 900 of tobacco leaf.Unlike Fig. 6, physical feature analysis processor 601, comprising: tobacco natural
Light picture collection device 901 and physical feature message handler 902.
Wherein, tobacco natural light picture collection device 901, is connected with physical feature message handler 902, by training cigarette
Leaf/tobacco leaf to be sorted imports detection darkroom, carries out polishing by preset constant light source, with camera to every group of trained Tobacco Leaves/
Tobacco leaf to be sorted is taken pictures, and the natural light picture of training tobacco leaf/tobacco leaf to be sorted is acquired.
Physical feature message handler 902 is closed with tobacco natural light picture collection device 901 and physical feature and spectral signature
Connection processor 602 is connected, and by preset tobacco leaf characteristic dimension information analysis natural light picture, obtains trained tobacco leaf/to be sorted
The physical feature information data of tobacco leaf.
In some alternative embodiments, as shown in Figure 10, divided automatically for the 5th kind in the present embodiment based on neural network
Pick the structural schematic diagram of the device 1000 of tobacco leaf.Unlike Fig. 6, physical feature and spectral signature association processor 602,
It include: high spectrum image collector 1001, Spectra feature extraction processor 1002 and correlation model creation processor 1003.
Wherein, high spectrum image collector 1001 is connected with Spectra feature extraction processor 1002, is passed using EO-1 hyperion
Sensor, to training tobacco leaf imaging, obtains the high spectrum image for training tobacco leaf with preset spectral band.
Spectra feature extraction processor 1002 creates processor 1003 with high spectrum image collector 1001 and correlation model
It is connected, with preset Spectra feature extraction project analysis high spectrum image, obtains the spectral signature Information Number for stating trained tobacco leaf
According to.
Correlation model creates processor 1003, with physical feature analysis processor 601 and Spectra feature extraction processor
1002 are connected, and the physical feature information data of training tobacco leaf and spectral signature information data is corresponding, carry out neural network instruction
Get the physical feature of tobacco leaf and the correlation model of spectral signature.
Although the preferred embodiment of the application has been described, it is created once a person skilled in the art knows basic
Property concept, then additional changes and modifications may be made to these embodiments.So it includes excellent that the following claims are intended to be interpreted as
It selects embodiment and falls into all change and modification of the application range.Obviously, those skilled in the art can be to the application
Various modification and variations are carried out without departing from spirit and scope.If in this way, these modifications and variations of the application
Belong within the scope of the claim of this application and its equivalent technologies, then the application is also intended to encompass these modification and variations and exists
It is interior.
Claims (10)
1. a kind of method based on neural network automatic sorting tobacco leaf characterized by comprising
The natural light picture for acquiring training tobacco leaf, obtains institute by natural light picture described in preset tobacco leaf characteristic dimension information analysis
State the physical feature information data of trained tobacco leaf;
The trained tobacco leaf is subjected to spectrum analysis, obtains the spectral signature information data of the trained tobacco leaf;By the training
The physical feature information data and spectral signature the information data correspondence of tobacco leaf, progress neural metwork training obtain oneself of tobacco leaf
The correlation model of right feature and spectral signature;
The spectrum tobacco leaf grading strategy that the tobacco leaf is established according to the spectral signature information data, is obtained based on the correlation model
To the Natural classification strategy of the tobacco leaf;
Obtain the natural image of tobacco leaf to be sorted, analysis obtains the physical feature information data of the tobacco leaf to be sorted, and described
Natural classification strategy compares to obtain the sorting rank of the tobacco leaf to be sorted, sorts the cigarette to be sorted according to the sorting rank
Leaf.
2. the method according to claim 1 based on neural network automatic sorting tobacco leaf, which is characterized in that acquire training cigarette
The natural light picture of leaf obtains oneself of the trained tobacco leaf by natural light picture described in preset tobacco leaf characteristic dimension information analysis
Right characteristic information data, are as follows:
With the natural light picture of the training tobacco leaf of pre-set velocity acquisition predetermined quantity;
The natural light picture is analyzed simultaneously by preset tobacco leaf characteristic dimension information, it is corresponding to obtain the trained tobacco leaf
Physical feature information data.
3. the method according to claim 1 based on neural network automatic sorting tobacco leaf, which is characterized in that further include:
The physical feature information data of the tobacco leaf to be sorted and the Natural classification strategy compare, when no correlation sorts rank,
The tobacco leaf to be sorted is included into retraining tobacco classification;
The retraining tobacco leaf is extracted into nature characteristic information data and spectral signature Information Number as training tobacco leaf again again
According to the correlation model that importing neural metwork training is updated;
The correlation model based on update sorts tobacco leaf to be sorted.
4. the method according to claim 1 based on neural network automatic sorting tobacco leaf, which is characterized in that acquire training cigarette
The natural light picture of leaf obtains oneself of the trained tobacco leaf by natural light picture described in preset tobacco leaf characteristic dimension information analysis
Right characteristic information data, are as follows:
Training tobacco leaf/tobacco leaf to be sorted is imported into detection darkroom, polishing is carried out by preset constant light source, with camera to every group
The trained Tobacco Leaves/tobacco leaf to be sorted is taken pictures, and the natural light picture of the trained tobacco leaf/tobacco leaf to be sorted is acquired;
By natural light picture described in preset tobacco leaf characteristic dimension information analysis, the trained tobacco leaf/tobacco leaf to be sorted is obtained
Physical feature information data.
5. the method according to claim 1 based on neural network automatic sorting tobacco leaf, which is characterized in that by the training
Tobacco leaf carries out spectrum analysis, obtains the spectral signature information data of the trained tobacco leaf, are as follows:
The trained tobacco leaf is imaged with preset spectral band using bloom spectrum sensor, obtains the high-spectrum for training tobacco leaf
Picture;
With high spectrum image described in preset Spectra feature extraction project analysis, the spectral signature Information Number for stating trained tobacco leaf is obtained
According to.
6. a kind of device based on neural network automatic sorting tobacco leaf characterized by comprising physical feature analysis processor,
Physical feature and spectral signature association processor, Natural classification policy handler and tobacco leaf sorting processor;Wherein,
The physical feature analysis processor, is connected with the physical feature with spectral signature association processor, acquisition training
The natural light picture of tobacco leaf obtains the trained tobacco leaf by natural light picture described in preset tobacco leaf characteristic dimension information analysis
Physical feature information data;
At the physical feature and spectral signature association processor, with the physical feature analysis processor and Natural classification strategy
Reason device is connected, and the trained tobacco leaf is carried out spectrum analysis, obtains the spectral signature information data of the trained tobacco leaf;By institute
The physical feature information data and the spectral signature information data for stating trained tobacco leaf are corresponding, carry out neural metwork training and obtain cigarette
The physical feature of leaf and the correlation model of spectral signature;
The Natural classification policy handler, with the physical feature and spectral signature association processor and tobacco leaf sorting processor
It is connected, the spectrum tobacco leaf grading strategy of the tobacco leaf is established according to the spectral signature information data, is based on the association mould
Type obtains the Natural classification strategy of the tobacco leaf;
The tobacco leaf sorting processor is connected with the Natural classification policy handler, obtains the natural figure of tobacco leaf to be sorted
Picture, analysis obtain the physical feature information data of the tobacco leaf to be sorted, with the Natural classification strategy compare to obtain it is described to
The sorting rank for sorting tobacco leaf sorts the tobacco leaf to be sorted according to the sorting rank.
7. the device according to claim 6 based on neural network automatic sorting tobacco leaf, which is characterized in that described naturally special
Levy analysis processor, comprising: natural light training tobacco leaf picture collection device and physical feature message handler;Wherein,
The natural light training tobacco leaf picture collection device, is connected with the physical feature message handler, is adopted with pre-set velocity
Collect the natural light picture of the training tobacco leaf of predetermined quantity;
The physical feature message handler, with natural light training tobacco leaf picture collection device and physical feature and spectral signature
Association processor is connected, and analyzes the natural light picture simultaneously by preset tobacco leaf characteristic dimension information, obtains the training
The corresponding physical feature information data of tobacco leaf.
8. the device according to claim 6 based on neural network automatic sorting tobacco leaf, which is characterized in that further include: cigarette
Leaf feature more new processor is connected with the tobacco leaf sorting processor and physical feature analysis processor,
It is compared in the physical feature information data of the tobacco leaf to be sorted and the Natural classification strategy, no related sorting rank
When, the tobacco leaf to be sorted is included into retraining tobacco classification;
The retraining tobacco leaf is extracted into nature characteristic information data and spectral signature Information Number as training tobacco leaf again again
According to the correlation model that importing neural metwork training is updated;
The correlation model based on update sorts tobacco leaf to be sorted.
9. the device according to claim 6 based on neural network automatic sorting tobacco leaf, which is characterized in that described naturally special
Levy analysis processor, comprising: tobacco natural light picture collection device and physical feature message handler;Wherein,
The tobacco natural light picture collection device, is connected with the physical feature message handler, will training tobacco leaf/to be sorted
Tobacco leaf imports detection darkroom, carries out polishing by preset constant light source, with camera training Tobacco Leaves/to be sorted described in every group
Tobacco leaf is taken pictures, and the natural light picture of the trained tobacco leaf/tobacco leaf to be sorted is acquired;
The physical feature message handler is associated with the tobacco natural light picture collection device and physical feature with spectral signature
Processor is connected, by natural light picture described in preset tobacco leaf characteristic dimension information analysis, obtain the trained tobacco leaf/to point
Pick the physical feature information data of tobacco leaf.
10. the device according to claim 6 based on neural network automatic sorting tobacco leaf, which is characterized in that the nature
Feature and spectral signature association processor, comprising: high spectrum image collector, Spectra feature extraction processor and correlation model wound
Build processor;Wherein,
The high spectrum image collector is connected, using bloom spectrum sensor with pre- with the Spectra feature extraction processor
If spectral band the trained tobacco leaf is imaged, obtain the high spectrum image for training tobacco leaf;
The Spectra feature extraction processor is connected with the high spectrum image collector and correlation model creation processor,
With high spectrum image described in preset Spectra feature extraction project analysis, the spectral signature information data for stating trained tobacco leaf is obtained;
The correlation model creates processor, is connected with the physical feature analysis processor and Spectra feature extraction processor
It connects, the physical feature information data and spectral signature information data of the trained tobacco leaf is corresponding, carry out neural network instruction
Get the physical feature of tobacco leaf and the correlation model of spectral signature.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910660071.7A CN110479636B (en) | 2019-07-19 | 2019-07-19 | Method and device for automatically sorting tobacco leaves based on neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910660071.7A CN110479636B (en) | 2019-07-19 | 2019-07-19 | Method and device for automatically sorting tobacco leaves based on neural network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110479636A true CN110479636A (en) | 2019-11-22 |
CN110479636B CN110479636B (en) | 2021-11-09 |
Family
ID=68547811
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910660071.7A Active CN110479636B (en) | 2019-07-19 | 2019-07-19 | Method and device for automatically sorting tobacco leaves based on neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110479636B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111067131A (en) * | 2019-12-25 | 2020-04-28 | 福建武夷烟叶有限公司 | Automatic tobacco grade identification and sorting method |
CN111334901A (en) * | 2020-02-18 | 2020-06-26 | 常熟理工学院 | Goat raw wool automatic sorting device based on memristor neural network |
CN112246686A (en) * | 2020-10-30 | 2021-01-22 | 南通市怡天时纺织有限公司 | Textile production conveyer |
CN112704259A (en) * | 2020-12-11 | 2021-04-27 | 杭州安脉盛智能技术有限公司 | Threshing quality evaluation method and system for dense tobacco leaf recognition |
CN113240277A (en) * | 2021-05-13 | 2021-08-10 | 云南省烟草质量监督检测站 | Flue-cured tobacco purchase grade classification sample manufacturing method based on artificial intelligence grading |
CN114504116A (en) * | 2021-12-29 | 2022-05-17 | 中国烟草总公司郑州烟草研究院 | Tobacco leaf classification device and method based on laser-induced breakdown spectroscopy |
CN114521664A (en) * | 2022-01-13 | 2022-05-24 | 浙江大学 | Automatic tobacco leaf grading system and device based on multi-mode image data and deep neural network |
CN114766706A (en) * | 2022-05-09 | 2022-07-22 | 北京天地数联科技有限公司 | Tobacco leaf impurity removal and grading method |
Citations (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS5639791A (en) * | 1979-09-07 | 1981-04-15 | Japan Tobacco Inc | Preparation of 3-hydroxy-solavetivone |
JPH0360775A (en) * | 1989-07-31 | 1991-03-15 | Japan Tobacco Inc | Color detection type device for discriminating kind of leaf tobacco |
JPH0929185A (en) * | 1995-07-17 | 1997-02-04 | Takaaki Satake | Method for deciding grade of vegetables and fruits having net patterns intrinsic to pericarp surface |
JP2002154888A (en) * | 2000-09-07 | 2002-05-28 | Mitsubishi Chemicals Corp | Classifying method for granular fertilizer and classifying equipment for granular fertilizer |
US20020133627A1 (en) * | 2001-03-19 | 2002-09-19 | International Business Machines Corporation | Intelligent document filtering |
CN1482578A (en) * | 2002-06-21 | 2004-03-17 | �����ɷ� | Graphic identification method and system for process endpoint curve and computer readable media having instructions executing the same method |
CN102590211A (en) * | 2011-01-11 | 2012-07-18 | 郑州大学 | Method for utilizing spectral and image characteristics to grade tobacco leaves |
CN103743486A (en) * | 2014-01-02 | 2014-04-23 | 上海大学 | Automatic grading system and method based on mass tobacco leaf data |
CN104432452A (en) * | 2014-11-07 | 2015-03-25 | 山东中烟工业有限责任公司 | Method for controlling loosening and conditioning of tobacco leaves |
CN106250631A (en) * | 2016-08-03 | 2016-12-21 | 中国科学院西安光学精密机械研究所 | A kind of method for diagnosing faults based on fault test correlation matrix |
CN107101971A (en) * | 2017-05-22 | 2017-08-29 | 上海创和亿电子科技发展有限公司 | Processing method, system and the electronic equipment of tobacco leaf attribute |
CN107576618A (en) * | 2017-07-20 | 2018-01-12 | 华南理工大学 | Pyricularia Oryzae detection method and system based on depth convolutional neural networks |
CN107578067A (en) * | 2017-09-15 | 2018-01-12 | 国家烟草质量监督检验中心 | A kind of recognition methods of pipe tobacco component and device based on convolutional neural networks |
CN107958216A (en) * | 2017-11-27 | 2018-04-24 | 沈阳航空航天大学 | Based on semi-supervised multi-modal deep learning sorting technique |
CN108888281A (en) * | 2018-08-16 | 2018-11-27 | 华南理工大学 | State of mind appraisal procedure, equipment and system |
CN109344891A (en) * | 2018-09-21 | 2019-02-15 | 北京航空航天大学 | A kind of high-spectrum remote sensing data classification method based on deep neural network |
CN109711611A (en) * | 2018-12-17 | 2019-05-03 | 北京木业邦科技有限公司 | Timber cuts volume recovery recognition methods, device, electronic equipment and storage medium |
CN109719055A (en) * | 2017-10-30 | 2019-05-07 | 松下知识产权经营株式会社 | Resin determination method and resin decision maker |
CN110479635A (en) * | 2019-07-19 | 2019-11-22 | 深圳市微蓝智能科技有限公司 | Method and device based on neural network automatic sorting tobacco leaf |
US20200338599A1 (en) * | 2017-12-03 | 2020-10-29 | Seedx Technologies Inc. | Systems and methods for sorting of seeds |
-
2019
- 2019-07-19 CN CN201910660071.7A patent/CN110479636B/en active Active
Patent Citations (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS5639791A (en) * | 1979-09-07 | 1981-04-15 | Japan Tobacco Inc | Preparation of 3-hydroxy-solavetivone |
JPH0360775A (en) * | 1989-07-31 | 1991-03-15 | Japan Tobacco Inc | Color detection type device for discriminating kind of leaf tobacco |
JPH0929185A (en) * | 1995-07-17 | 1997-02-04 | Takaaki Satake | Method for deciding grade of vegetables and fruits having net patterns intrinsic to pericarp surface |
JP2002154888A (en) * | 2000-09-07 | 2002-05-28 | Mitsubishi Chemicals Corp | Classifying method for granular fertilizer and classifying equipment for granular fertilizer |
US20020133627A1 (en) * | 2001-03-19 | 2002-09-19 | International Business Machines Corporation | Intelligent document filtering |
CN1482578A (en) * | 2002-06-21 | 2004-03-17 | �����ɷ� | Graphic identification method and system for process endpoint curve and computer readable media having instructions executing the same method |
CN102590211A (en) * | 2011-01-11 | 2012-07-18 | 郑州大学 | Method for utilizing spectral and image characteristics to grade tobacco leaves |
CN103743486A (en) * | 2014-01-02 | 2014-04-23 | 上海大学 | Automatic grading system and method based on mass tobacco leaf data |
CN104432452A (en) * | 2014-11-07 | 2015-03-25 | 山东中烟工业有限责任公司 | Method for controlling loosening and conditioning of tobacco leaves |
CN106250631A (en) * | 2016-08-03 | 2016-12-21 | 中国科学院西安光学精密机械研究所 | A kind of method for diagnosing faults based on fault test correlation matrix |
CN107101971A (en) * | 2017-05-22 | 2017-08-29 | 上海创和亿电子科技发展有限公司 | Processing method, system and the electronic equipment of tobacco leaf attribute |
CN107576618A (en) * | 2017-07-20 | 2018-01-12 | 华南理工大学 | Pyricularia Oryzae detection method and system based on depth convolutional neural networks |
CN107578067A (en) * | 2017-09-15 | 2018-01-12 | 国家烟草质量监督检验中心 | A kind of recognition methods of pipe tobacco component and device based on convolutional neural networks |
CN109719055A (en) * | 2017-10-30 | 2019-05-07 | 松下知识产权经营株式会社 | Resin determination method and resin decision maker |
CN107958216A (en) * | 2017-11-27 | 2018-04-24 | 沈阳航空航天大学 | Based on semi-supervised multi-modal deep learning sorting technique |
US20200338599A1 (en) * | 2017-12-03 | 2020-10-29 | Seedx Technologies Inc. | Systems and methods for sorting of seeds |
CN108888281A (en) * | 2018-08-16 | 2018-11-27 | 华南理工大学 | State of mind appraisal procedure, equipment and system |
CN109344891A (en) * | 2018-09-21 | 2019-02-15 | 北京航空航天大学 | A kind of high-spectrum remote sensing data classification method based on deep neural network |
CN109711611A (en) * | 2018-12-17 | 2019-05-03 | 北京木业邦科技有限公司 | Timber cuts volume recovery recognition methods, device, electronic equipment and storage medium |
CN110479635A (en) * | 2019-07-19 | 2019-11-22 | 深圳市微蓝智能科技有限公司 | Method and device based on neural network automatic sorting tobacco leaf |
Non-Patent Citations (4)
Title |
---|
任德齐等: "《物联网技术概论》", 31 July 2013, 重庆大学出版社 * |
何宗耀: "《新型智慧城市建设现状 技术与研究》", 31 August 2018, 北京邮电大学出版社 * |
何月顺: "关联规则挖掘技术的研究及应用", 《中国优秀博硕士学位论文全文数据库(博士)信息科技辑》 * |
高小鹏等: "基于教学数据的课程关联挖掘研究", 《计算机教育》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111067131A (en) * | 2019-12-25 | 2020-04-28 | 福建武夷烟叶有限公司 | Automatic tobacco grade identification and sorting method |
CN111334901A (en) * | 2020-02-18 | 2020-06-26 | 常熟理工学院 | Goat raw wool automatic sorting device based on memristor neural network |
CN112246686A (en) * | 2020-10-30 | 2021-01-22 | 南通市怡天时纺织有限公司 | Textile production conveyer |
CN112704259A (en) * | 2020-12-11 | 2021-04-27 | 杭州安脉盛智能技术有限公司 | Threshing quality evaluation method and system for dense tobacco leaf recognition |
CN113240277A (en) * | 2021-05-13 | 2021-08-10 | 云南省烟草质量监督检测站 | Flue-cured tobacco purchase grade classification sample manufacturing method based on artificial intelligence grading |
CN114504116A (en) * | 2021-12-29 | 2022-05-17 | 中国烟草总公司郑州烟草研究院 | Tobacco leaf classification device and method based on laser-induced breakdown spectroscopy |
CN114521664A (en) * | 2022-01-13 | 2022-05-24 | 浙江大学 | Automatic tobacco leaf grading system and device based on multi-mode image data and deep neural network |
CN114766706A (en) * | 2022-05-09 | 2022-07-22 | 北京天地数联科技有限公司 | Tobacco leaf impurity removal and grading method |
CN114766706B (en) * | 2022-05-09 | 2023-09-12 | 北京天地数联科技有限公司 | Tobacco impurity removing and grading method |
Also Published As
Publication number | Publication date |
---|---|
CN110479636B (en) | 2021-11-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110479636A (en) | Method and device based on neural network automatic sorting tobacco leaf | |
CN107016405B (en) | A kind of pest image classification method based on classification prediction convolutional neural networks | |
CN107239514A (en) | A kind of plants identification method and system based on convolutional neural networks | |
CN109961024A (en) | Wheat weeds in field detection method based on deep learning | |
CN109478232A (en) | The identification of weeds in natural environment | |
Tetila et al. | A deep-learning approach for automatic counting of soybean insect pests | |
CN106971160A (en) | Winter jujube disease recognition method based on depth convolutional neural networks and disease geo-radar image | |
CN107330892A (en) | A kind of sunflower disease recognition method based on random forest method | |
CN110807760B (en) | Tobacco leaf grading method and system | |
CN110479635B (en) | Method and device for automatically sorting tobacco leaves based on neural network | |
CN109543630A (en) | Remote sensing image forest land extracting method and system, storage medium, electronic equipment based on deep learning | |
Lim et al. | Performance effect analysis for insect classification using convolutional neural network | |
CN116058195B (en) | Illumination regulation and control method, system and device for leaf vegetable growth environment | |
CN110084198A (en) | The airport CNN indoor scene recognition methods based on Fisher signature analysis | |
Treboux et al. | Decision tree ensemble vs. nn deep learning: efficiency comparison for a small image dataset | |
CN116129260A (en) | Forage grass image recognition method based on deep learning | |
Yuan et al. | Sensitivity examination of YOLOv4 regarding test image distortion and training dataset attribute for apple flower bud classification | |
CN115272848A (en) | Intelligent change detection method for buildings in multi-cloud and multi-fog farmland protection area | |
CN107578050A (en) | The automatic classifying identification method of rice basal part of stem On Planthopperss and its worm state | |
Islam et al. | Nitrogen fertilizer recommendation for paddies through automating the Leaf Color Chart (LCC) | |
Sun et al. | A recognition method of cattle and sheep based on convolutional neural network | |
Singh et al. | A review on recognition of plant disease using intelligent image retrieval techniques | |
CN116310338A (en) | Single litchi red leaf tip segmentation method based on examples and semantic segmentation | |
CN115511838A (en) | Plant disease high-precision identification method based on group intelligent optimization | |
Ren et al. | Recognition of common pests in agriculture and forestry based on convolutional neural networks |
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 |