CN106874930A - A kind of seed sorting technique and device - Google Patents
A kind of seed sorting technique and device Download PDFInfo
- Publication number
- CN106874930A CN106874930A CN201611250010.6A CN201611250010A CN106874930A CN 106874930 A CN106874930 A CN 106874930A CN 201611250010 A CN201611250010 A CN 201611250010A CN 106874930 A CN106874930 A CN 106874930A
- Authority
- CN
- China
- Prior art keywords
- seed
- characteristic parameter
- parameter vector
- dictionary
- tested
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 31
- 239000013598 vector Substances 0.000 claims abstract description 92
- 241000894007 species Species 0.000 claims description 39
- 230000006870 function Effects 0.000 claims description 21
- 239000011159 matrix material Substances 0.000 claims description 21
- 241000208340 Araliaceae Species 0.000 claims description 2
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 2
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 2
- 238000005284 basis set Methods 0.000 claims description 2
- 235000008434 ginseng Nutrition 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 5
- 230000000877 morphologic effect Effects 0.000 description 4
- 238000000605 extraction Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000003749 cleanliness Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 239000002932 luster Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Agronomy & Crop Science (AREA)
- General Health & Medical Sciences (AREA)
- Marine Sciences & Fisheries (AREA)
- Mining & Mineral Resources (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- Animal Husbandry (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a kind of seed sorting technique and device, including：Original dictionary is set up according to default training data, be stored with seed characteristic parameter vector and the corresponding relation of seed species in the original dictionary；The default training data includes multiple seed characteristic parameter vectors and multiple seed species；Obtain the characteristic parameter vector of seed to be tested；Original dictionary described in characteristic parameter vector query according to the seed to be tested for obtaining, obtains seed species corresponding with the characteristic parameter vector of seed to be tested.The present invention is realized and seed is quickly recognized and is classified；Improve the degree of accuracy of classification and the efficiency of classification.
Description
Technical field
The present invention relates to identification technology field, and in particular to a kind of seed sorting technique and device.
Background technology
The variety ecotype and Quality Identification of seed differentiate and modernize that the fields such as grain depot warehousing management have wide in germplasm
Application prospect.It is main at present that overall merit is carried out in terms of profile, color and luster, cleanliness etc. using artificial means, thus there is master
The shortcoming of the property the seen low and repeatable difference of strong, efficiency, enhances the uncertainty of seed assortment.
Currently, developing rapidly with computer technology, machine vision is able to be widely used in the inspection of crop kernels quality
It is that the efficient production of agricultural implementation provides safeguard in survey and identification classification.Compared with the vision of the mankind, machine vision replaces manually entering
Row Variety identification can not only reduce the subjective factor influence of people, realize automation；Inspection error can also be reduced, improves accurate
Degree and precision.
In existing technology, classification is identified to seed using rarefaction representation grader, in the feelings that training sample is larger
Amount of calculation is larger under condition, and certain influence can be so caused on the identification of seed classification, causes, efficiency not high in the presence of the classification degree of accuracy
Low problem.
The content of the invention
In order to solve the above technical problems, the present invention provides a kind of seed sorting technique and device, realize is carried out to seed
Quick identification and classification；Improve the degree of accuracy of classification and the efficiency of classification.
To achieve the above object, the present invention provides following technical scheme：
On the one hand, the invention provides a kind of seed sorting technique, including：
Original dictionary is set up according to default training data, be stored with seed characteristic parameter vector and seed in the original dictionary
The corresponding relation of grain species；The default training data includes multiple seed characteristic parameter vectors and multiple seed species；
Obtain the characteristic parameter vector of seed to be tested；
Original dictionary described in characteristic parameter vector query according to the seed to be tested for obtaining, obtains and seed to be tested
The corresponding seed species of characteristic parameter vector.
Further, after the step of the characteristic parameter vector for obtaining seed to be tested, also include：
The original dictionary is updated and iterative processing is to obtain characteristics dictionary；It is stored with more in the characteristics dictionary
The corresponding relation of the seed species after new and iterative processing after seed characteristic parameter vector and renewal and iteration；
Characteristics dictionary described in characteristic parameter vector query according to the seed to be tested for obtaining, obtains and seed to be tested
The corresponding seed species of characteristic parameter vector.
Further, it is described the original dictionary to be updated and iterative processing is to obtain characteristics dictionary, specifically include：
Rarefaction representation is carried out to the training sample seed using the original dictionary and obtains rarefaction representation coefficient matrix；
Treatment is iterated to the coefficient matrix and obtains characteristics dictionary.
Further, the default training data of the basis sets up original dictionary, including：
The default training data is mapped in higher dimensional space by the way of kernel function, it is right in the higher dimensional space
Linearly inseparable data become linear separability data in the default training data.
Further, the seed characteristic parameter includes：The chromatic component of seed coat color, the saturation degree of seed coat color point
The area of amount, the luminance component of seed coat color, the rectangular degree of Seed shape, the elongation of Seed shape and Seed shape.
Further, the original dictionary according to the characteristic parameter vector query of the seed to be tested for obtaining obtain with
The corresponding seed species of characteristic parameter vector of seed to be tested, including：
The target signature parameter minimum with the error of the characteristic parameter vector of the test seed is inquired about in original dictionary
Vector；
The corresponding seed kind of characteristic parameter vector of the seed to be tested is determined according to the target signature parameter vector
Class.
On the other hand, the invention provides a kind of seed sorter, including：
Dictionary module, for setting up original dictionary according to default training data, the seed that is stored with the original dictionary is special
Levy the corresponding relation of parameter vector and seed species；The default training data includes that multiple seed characteristic parameters are vectorial and many
Individual seed species；
Acquisition module, the characteristic parameter vector for obtaining seed to be tested；
Enquiry module, for original dictionary described in the characteristic parameter vector query according to the seed to be tested for obtaining, obtains
Seed species corresponding with the characteristic parameter vector of seed to be tested.
Further, described device also includes：
Update module, for being updated to the original dictionary and iterative processing is to obtain characteristics dictionary；The feature
Be stored with dictionary update and iterative processing after seed characteristic parameter vector with renewal and iteration after seed species it is corresponding close
System；
Output module, for characteristics dictionary described in the characteristic parameter vector query according to the seed to be tested for obtaining, obtains
Seed species corresponding with the characteristic parameter vector of seed to be tested.
Further, the dictionary module includes：
Memory cell, for storing the seed characteristic parameter, including：The chromatic component of seed coat color, seed coat color
The face of saturation degree component, the luminance component of seed coat color, the rectangular degree of Seed shape, the elongation of Seed shape and Seed shape
Product.
Further, the enquiry module includes：
Error unit, for being inquired about in original dictionary and the error minimum of the characteristic parameter vector of the test seed
Target signature parameter vector；
Positioning unit, is same as determining according to the target signature parameter vector characteristic parameter vector of the seed to be tested
Corresponding seed species.
As shown from the above technical solution, a kind of seed sorting technique of the present invention and device, by original dictionary
The corresponding seed species of characteristic parameter vector of middle inquiry test seed, realizes seed is quickly known according to original dictionary
Other and classification；Improve the degree of accuracy of classification and the efficiency of classification.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are the present invention
Some embodiments, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis
These accompanying drawings obtain other accompanying drawings.
Fig. 1 is a kind of schematic flow sheet of seed sorting technique that the embodiment of the present invention one is provided；
Fig. 2 is a kind of schematic flow sheet of seed sorting technique that the embodiment of the present invention two is provided；
Fig. 3 is the flow chart of step S203 in a kind of seed sorting technique that the embodiment of the present invention two is provided；
Fig. 4 is a kind of structural representation of seed sorter of the invention.
Specific embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, clear, complete description is carried out to the technical scheme in the embodiment of the present invention, it is clear that described embodiment is
A part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
It is main at present that overall merit is carried out to the kind of seed using artificial means, have that subjectivity is strong, efficiency is low and can
The shortcoming of poor repeatability, enhances the uncertainty of seed assortment.With developing rapidly for computer technology, machine vision
It is able to be widely used in crop kernels Quality Detection and identification classification, is that efficiently production provides safeguard for agricultural implementation.It is existing
Technology in, using rarefaction representation grader to seed be identified classification, in the case where training sample is larger amount of calculation compared with
Greatly, certain influence can be so caused on the identification of seed classification, causes, efficiency low problem not high in the presence of the classification degree of accuracy.For
Above-mentioned technical problem is solved, the embodiment of the present invention provides a kind of seed sorting technique and device.
Embodiment one
The embodiment of the present invention provides a kind of seed sorting technique, and referring to Fig. 1, the method includes：
S101：Original dictionary is set up according to default training data, be stored with the original dictionary seed characteristic parameter to
The corresponding relation of amount and seed species；The default training data includes multiple seed characteristic parameter vectors and multiple seed kinds
Class；
In this step, PCA (Principal Component are used using the image to training sample seed
Analysis dimension-reduction treatment) is carried out and feature is carried out using SIFT (Scale-invariant feature transform) to carry
Take；There is linearly inseparable situation in the data for obtaining.Therefore, entirely trained to enable the dictionary for training preferably to represent
Collection, takes into account these linearly inseparable data：The data after dimension-reduction treatment are mapped in higher dimensional space using kernel function,
That is feature space, in feature space, these data will become linear separability, so as to simplify treatment.
The characteristic parameter of training sample seed is extracted, the species generation data matrix according to characteristic parameter and seed；Using
Original dictionary is obtained after the mode of kernel function is carried out to the data matrix of training sample plus core is processed.
S102：Obtain the characteristic parameter vector of seed to be tested；
In this step, dimension-reduction treatment is carried out using PCA using the image for treating test sample seed and is entered using SIFT
Row feature extraction；And linearly inseparable data are become by linear separability data using kernel function.Generated according to characteristic parameter to be measured
Try the characteristic parameter vector of seed.
S103：According to obtain seed to be tested characteristic parameter vector query described in original dictionary, obtain with it is to be tested
The corresponding seed species of characteristic parameter vector of seed.
In this step, the mesh minimum with the error of the characteristic parameter vector of the test seed is inquired about in original dictionary
Mark characteristic parameter vector；
The corresponding seed kind of characteristic parameter vector of the seed to be tested is determined according to the target signature parameter vector
Class.
Knowable to foregoing description, a kind of seed sorting technique that the present embodiment is provided carries out seed with reference to rarefaction representation
Identification classification, it is to avoid the image of all training seeds is calculated object so as to produce huge amount of calculation as rarefaction representation；Using
PCA carries out dimension-reduction treatment and carries out feature extraction using SIFT, and the amount of calculation in reduction amount processing procedure improves the speed of classification
Degree；Data matrix to training sample by the way of kernel function is processed, and linearly inseparable data become in data matrix
Linear separability data, improve nicety of grading.
Embodiment two
The embodiment of the present invention provides a kind of seed sorting technique, and referring to Fig. 2, the method includes：
S201：Original dictionary is set up according to default training data, be stored with the original dictionary seed characteristic parameter to
The corresponding relation of amount and seed species；The default training data includes multiple seed characteristic parameter vectors and multiple seed kinds
Class；
In this step, PCA (Principal Component are used using the image to training sample seed
Analysis dimension-reduction treatment) is carried out and feature is carried out using SIFT (Scale-invariant feature transform) to carry
Take；There is linearly inseparable situation in the data for obtaining.Therefore, entirely trained to enable the dictionary for training preferably to represent
Collection, takes into account these linearly inseparable data：The data after dimension-reduction treatment are mapped in higher dimensional space using kernel function,
That is feature space, in feature space, these data will become linear separability, so as to simplify treatment.
The characteristic parameter of training sample seed is extracted, the species generation data matrix according to characteristic parameter and seed；Using
Original dictionary is obtained after the mode of kernel function is carried out to the data matrix of training sample plus core is processed.
S202：Obtain the characteristic parameter vector of seed to be tested；
In this step, dimension-reduction treatment is carried out using PCA using the image for treating test sample seed and is entered using SIFT
Row feature extraction；And linearly inseparable data are become by linear separability data using kernel function.Generated according to characteristic parameter to be measured
Try the characteristic parameter vector of seed.
S203：The original dictionary is updated and iterative processing is to obtain characteristics dictionary；Deposited in the characteristics dictionary
Contain update and iterative processing after seed characteristic parameter vector with renewal and iteration after seed species corresponding relation；
In this step, in this step, by extracting the characteristic parameter of seed as the characteristic feature of seed, for group
Build original dictionary；Original dictionary is trained using K-SVD methods, corresponding characteristics dictionary is obtained.
S204：According to obtain seed to be tested characteristic parameter vector query described in characteristics dictionary, obtain with it is to be tested
The corresponding seed species of characteristic parameter vector of seed.
In this step, the mesh minimum with the error of the characteristic parameter vector of the test seed is inquired about in characteristics dictionary
Mark characteristic parameter vector；
The corresponding seed kind of characteristic parameter vector of the seed to be tested is determined according to the target signature parameter vector
Class.
Knowable to foregoing description, a kind of seed sorting technique that the present embodiment is provided, to original on the basis of embodiment one
Beginning dictionary is updated acquisition characteristics dictionary, not only increases the accuracy of dictionary, also improves the precision of classification.
Further, the characteristic parameter of the training sample seed includes：The chromatic component of seed coat color, seed coat color
The face of saturation degree component, the luminance component of seed coat color, the rectangular degree of Seed shape, the elongation of Seed shape and Seed shape
Product；
The number of the cybernetics control number of the data matrix, is classified as the number of training sample seed.
The color characteristic of seed, HIS color model is the color model commonly used in Digital Image Processing.By HIS models
Extract the color characteristic of seed.It is represented with the average value of the chromatic component of seed coat color, saturation degree component and luminance component
Color characteristic, meets measuring multiple parameters requirement of the seed variety ecotype to seed coat color feature, and formula is as follows:
In above formula,WithThe 1st, 2,3 characteristic values of i-th kind, j-th seed are represented respectively, and p is figure
Belong to the pixel count of seed as in,WithThe respectively colourity of k-th point of seed image I, saturation degree and brightness
Component.
The shape facility of seed
The seed of same species different cultivars, appearance difference is one of key character for distinguishing between them.Morphological feature
The no Uniform provisions of selection of parameter, if the morphological differences of object can be effectively distinguished, and the ginseng that energy fast and easy is obtained
Number can be used as morphological feature parameter.Choose area, rectangular degree, 3 parameters of elongation.The morphological feature of seed is expressed as follows：
Rectangular degree：
Elongation：
Area：
In above formula：A is the area of seed；A is major axis, and b is short axle.
In sum, if sample there are n verieties, there is m training sample per veriety, describe the external appearance characteristic vector of seed
Can be expressed as：
Its m column vector constitutes a space, reacts the i-th veriety, then the dictionary square of n verieties training sample composition
Battle array be
A=[A1A2...An] (8)
The line number of A is the characteristic parameter number of training sample, and columns is training sample sum.
When 40 seeds are chosen as training sample, the characteristic vector of these seeds is lined up, the following training of composition
Sample matrix is dictionary:
Matrix A represents that the dictionary line number of 4 veriety features composition represents the Characteristic Number for extracting seed, and columns represents training
Total sample number.
In a kind of optional embodiment, there is provided a kind of specific embodiment of above-mentioned steps S203.Referring to Fig. 3, on
State step S203 and specifically include following steps：
S2031：Rarefaction representation is carried out to the training sample seed using the original dictionary and obtains rarefaction representation coefficient
Matrix；
In this step, rarefaction representation is carried out to training sample y using given original dictionary D` and obtains rarefaction representation system
Number vector α:
Wherein, δ is the number upper limit of nonzero element in rarefaction representation coefficient.Dictionary training process is a continuous renewal and changes
For process, carry out by column.In this regard, can select the sparse table that the classical alternative manner such as OMP obtains training sample y on original dictionary D`
Show coefficient matrix α.
S2032：Acquisition characteristics dictionary is iterated and updated to the coefficient matrix.
In this step, according to the continuous iteration of coefficient matrix α and renewal, characteristics dictionary is determined.Below updating original word
It is any one in allusion quotation D` to be classified as example, if updating the kth row d of original dictionary D`k, make d in coefficient matrix αkCorresponding row k is
Then
It is fixed, remaining k-th for that need to process to have k-1.L is solved for kth2Norm, wherein EkAs
Currently remove dkThe error matrix for being formed afterwards.For this error matrix, accorded with using SVD (singular value decomposition) alignment error
The minimum value of conjunction condition.The characteristics dictionary for finally giving.
Knowable to foregoing description, represented using SVD dictionary training methods selected part " optimal base " in the present embodiment
All training images, to avoid calculating object so as to produce huge amount of calculation using all training images as rarefaction representation.
A kind of seed sorting technique provided in an embodiment of the present invention, in kernel function, simplest method is exactly to consider many
The convex combination of individual basic kernel function, shape is such as:
Here km(x, x') is basic kernel function, βmIt is the corresponding weight coefficient of the basic kernel function, F is the total of basic core
Number, k (x, y) is final kernel function.βmIt is exactly the object of required optimization here.
Broad sense compound kernel function is used herein, and specific manifestation form is as follows：
In the kernel function, optimize kernel function using (n+m) individual coefficient altogether.Wherein, d1,d2,....,dnFor table
Show the weight relationship inside each characteristic block, and dn+1,dn+2,....,dn+mFor representing the relation between different block features.
From above formula it can be seen that, the kernel function substantially combines " gaussian sum " function and " Gauss product " function.Meanwhile, it
The internal organizational structure of feature is have also contemplated that, the image geometry meaning that distinguishing characteristics is included in the form of block.Need afterwards
The gradient of kernel function is calculated in often wheel optimization, and according to gradient descent method undated parameter.
Plus respectively by D in the case of blocking1K is obtained plus unit matrix1；Then l is solved1Minimum norm problem；
S.t. λ ＞ 0
Now, the kind for being affiliated seed minimum with test sample difference in training sample；
Y'=argmin | | Φ (y)-KiW||2。
Embodiment three
The embodiment of the present invention provides a kind of seed sorter, and referring to Fig. 4, the device includes：
Dictionary module 10, for setting up original dictionary according to default training data, be stored with seed in the original dictionary
The corresponding relation of characteristic parameter vector and seed species；The default training data include multiple seed characteristic parameters vectors and
Multiple seed species；
Acquisition module 20, the characteristic parameter vector for obtaining seed to be tested；
Enquiry module 30, for original dictionary described in the characteristic parameter vector query according to the seed to be tested for obtaining, obtains
Take seed species corresponding with the characteristic parameter vector of seed to be tested.
Further, described device also includes：
Update module 40, for being updated to the original dictionary and iterative processing is to obtain characteristics dictionary；The spy
Levy be stored with dictionary update and iterative processing after seed characteristic parameter vector it is corresponding with the seed species after renewal and iteration
Relation；
Output module 50, for characteristics dictionary described in the characteristic parameter vector query according to the seed to be tested for obtaining, obtains
Take seed species corresponding with the characteristic parameter vector of seed to be tested.
Further, the dictionary module 10 includes：
Memory cell 101, for storing the seed characteristic parameter, including：The chromatic component of seed coat color, seed coat color
Saturation degree component, the luminance component of seed coat color, the rectangular degree of Seed shape, the elongation of Seed shape and Seed shape
Area.
Further, the enquiry module 20 includes：
Error unit 201, for inquiring about the error vectorial with the characteristic parameter of the test seed in original dictionary most
Small target signature parameter vector；
Positioning unit 202, is same as determining according to the target signature parameter vector characteristic parameter of the seed to be tested
The corresponding seed species of vector.
By foregoing description, a kind of seed sorter provided in an embodiment of the present invention enters by original dictionary
Row treatment obtains characteristics dictionary, and the calculating during the characteristic parameter of collection seed is reduced while dictionary accuracy is improved
Amount；Seed is quickly recognized and classified according to characteristics dictionary；Improve the degree of accuracy of classification and the efficiency of classification.
Above example is merely to illustrate technical scheme, rather than its limitations；Although with reference to the foregoing embodiments
The present invention has been described in detail, it will be understood by those within the art that：It still can be to foregoing each implementation
Technical scheme described in example is modified, or carries out equivalent to which part technical characteristic；And these are changed or replace
Change, do not make the spirit and scope of the essence disengaging various embodiments of the present invention technical scheme of appropriate technical solution.
Claims (10)
1. a kind of seed sorting technique, it is characterised in that including：
Original dictionary is set up according to default training data, be stored with seed characteristic parameter vector and seed kind in the original dictionary
The corresponding relation of class；The default training data includes multiple seed characteristic parameter vectors and multiple seed species；
Obtain the characteristic parameter vector of seed to be tested；
Original dictionary described in characteristic parameter vector query according to the seed to be tested for obtaining, obtains the feature with seed to be tested
The corresponding seed species of parameter vector.
2. method according to claim 1, it is characterised in that the step of the characteristic parameter vector of the acquisition seed to be tested
After rapid, also include：
The original dictionary is updated and iterative processing is to obtain characteristics dictionary；Be stored with the characteristics dictionary renewal and
After iterative processing seed characteristic parameter vector with update and iteration after seed species corresponding relation；
Characteristics dictionary described in characteristic parameter vector query according to the seed to be tested for obtaining, obtains the feature with seed to be tested
The corresponding seed species of parameter vector.
3. method according to claim 2, it is characterised in that described to be updated and iterative processing to the original dictionary
To obtain characteristics dictionary, specifically include：
Rarefaction representation is carried out to the training sample seed using the original dictionary and obtains rarefaction representation coefficient matrix；
Treatment is iterated to the coefficient matrix and obtains characteristics dictionary.
4. method according to claim 1 and 2, it is characterised in that the default training data of the basis sets up original dictionary,
Including：
The default training data is mapped in higher dimensional space by the way of kernel function, to described in the higher dimensional space
Linearly inseparable data become linear separability data in default training data.
5. method according to claim 1 and 2, it is characterised in that the seed characteristic parameter includes：The color of seed coat color
Degree component, the saturation degree component of seed coat color, the luminance component of seed coat color, the rectangular degree of Seed shape, Seed shape are stretched
The area of length and Seed shape.
6. method according to claim 1, it is characterised in that the characteristic parameter according to the seed to be tested for obtaining to
The amount inquiry original dictionary obtains seed species corresponding with the characteristic parameter vector of seed to be tested, including：
The target signature parameter vector minimum with the error of the characteristic parameter vector of the test seed is inquired about in original dictionary；
The corresponding seed species of characteristic parameter vector of the seed to be tested is determined according to the target signature parameter vector.
7. a kind of seed sorter, it is characterised in that including：
Dictionary module, for setting up original dictionary, the seed feature that is stored with original dictionary ginseng according to default training data
The corresponding relation of number vector and seed species；The default training data includes multiple seed characteristic parameter vectors and multiple seeds
Grain species；
Acquisition module, the characteristic parameter vector for obtaining seed to be tested；
Enquiry module, for original dictionary described in the characteristic parameter vector query according to the seed to be tested for obtaining, obtains and treats
Test the corresponding seed species of characteristic parameter vector of seed.
8. device according to claim 7, it is characterised in that described device also includes：
Update module, for being updated to the original dictionary and iterative processing is to obtain characteristics dictionary；The characteristics dictionary
In be stored with update and iterative processing after seed characteristic parameter vector with renewal and iteration after seed species corresponding relation；
Output module, for characteristics dictionary described in the characteristic parameter vector query according to the seed to be tested for obtaining, obtains and treats
Test the corresponding seed species of characteristic parameter vector of seed.
9. device according to claim 7, it is characterised in that the dictionary module includes：
Memory cell, for storing the seed characteristic parameter, including：The chromatic component of seed coat color, the saturation of seed coat color
The area of degree component, the luminance component of seed coat color, the rectangular degree of Seed shape, the elongation of Seed shape and Seed shape.
10. device according to claim 7, it is characterised in that the enquiry module includes：
Error unit, for inquiring about the target minimum with the error of the characteristic parameter vector of the test seed in original dictionary
Characteristic parameter vector；
Positioning unit, is same as determining according to the target signature parameter vector characteristic parameter vector correspondence of the seed to be tested
Seed species.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611250010.6A CN106874930A (en) | 2016-12-29 | 2016-12-29 | A kind of seed sorting technique and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611250010.6A CN106874930A (en) | 2016-12-29 | 2016-12-29 | A kind of seed sorting technique and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106874930A true CN106874930A (en) | 2017-06-20 |
Family
ID=59164249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611250010.6A Pending CN106874930A (en) | 2016-12-29 | 2016-12-29 | A kind of seed sorting technique and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106874930A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114067105A (en) * | 2022-01-12 | 2022-02-18 | 安徽高哲信息技术有限公司 | Grain density estimation method, storage medium, and grain density estimation apparatus |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101701916A (en) * | 2009-12-01 | 2010-05-05 | 中国农业大学 | Method for quickly identifying and distinguishing variety of corn |
| CN102279976A (en) * | 2011-09-22 | 2011-12-14 | 河南工业大学 | Method for constructing and identifying BP (Back Propagation) neural network for identifying different coarse rice seeds |
| JP2015125662A (en) * | 2013-12-27 | 2015-07-06 | 株式会社リコー | Object identification program and device |
-
2016
- 2016-12-29 CN CN201611250010.6A patent/CN106874930A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101701916A (en) * | 2009-12-01 | 2010-05-05 | 中国农业大学 | Method for quickly identifying and distinguishing variety of corn |
| CN102279976A (en) * | 2011-09-22 | 2011-12-14 | 河南工业大学 | Method for constructing and identifying BP (Back Propagation) neural network for identifying different coarse rice seeds |
| JP2015125662A (en) * | 2013-12-27 | 2015-07-06 | 株式会社リコー | Object identification program and device |
Non-Patent Citations (2)
| Title |
|---|
| 冯丽娟: "基于稀疏表示的小麦品种识别方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 古楠楠 等主编: "《流形学习若干关键问题与算法研究》", 31 July 2015, 首都经济贸易大学出版社 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114067105A (en) * | 2022-01-12 | 2022-02-18 | 安徽高哲信息技术有限公司 | Grain density estimation method, storage medium, and grain density estimation apparatus |
| CN114067105B (en) * | 2022-01-12 | 2022-04-22 | 安徽高哲信息技术有限公司 | Grain density estimation method, storage medium, and grain density estimation apparatus |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107633255B (en) | Rock lithology automatic identification and classification method under deep learning mode | |
| CN106203523B (en) | The hyperspectral image classification method of the semi-supervised algorithm fusion of decision tree is promoted based on gradient | |
| CN106326288B (en) | Image search method and device | |
| CN105046276B (en) | Hyperspectral image band selection method based on low-rank representation | |
| CN112613536B (en) | Near infrared spectrum diesel fuel brand recognition method based on SMOTE and deep learning | |
| CN108460391B (en) | Hyperspectral image unsupervised feature extraction method based on generation countermeasure network | |
| CN108090472B (en) | Pedestrian re-identification method and system based on multi-channel consistency characteristics | |
| CN108846338A (en) | Polarization characteristic selection and classification method based on object-oriented random forest | |
| Nema et al. | Wheat leaf detection and prevention using support vector machine | |
| CN102324038A (en) | A kind of floristics recognition methods based on digital picture | |
| CN110516754B (en) | Hyperspectral image classification method based on multi-scale superpixel segmentation | |
| CN110188225A (en) | A kind of image search method based on sequence study and polynary loss | |
| CN111222545B (en) | Image classification method based on linear programming incremental learning | |
| CN105631474B (en) | Based on Jeffries-Matusita distance and class to the more classification methods of the high-spectral data of decision tree | |
| CN108629373A (en) | A kind of image classification method, system, equipment and computer readable storage medium | |
| CN104573699B (en) | Trypetid recognition methods based on middle equifield intensity magnetic resonance anatomy imaging | |
| CN103064941A (en) | Image retrieval method and device | |
| CN112949738B (en) | Multi-class unbalanced hyperspectral image classification method based on EECNN algorithm | |
| CN109543723A (en) | A kind of image clustering method of robust | |
| Lee et al. | Cross-domain CNN for hyperspectral image classification | |
| CN104699781B (en) | SAR image search method based on double-deck anchor figure hash | |
| Pandey et al. | Automatic seed classification by shape and color features using machine vision technology | |
| CN113435254A (en) | Sentinel second image-based farmland deep learning extraction method | |
| CN104143088B (en) | Face identification method based on image retrieval and feature weight learning | |
| Kang et al. | CST-YOLO: A Novel Method for Blood Cell Detection Based on Improved YOLOv7 and CNN-Swin Transformer |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170620 |