CN108346142A - A kind of plant growth state recognition methods based on plant illumination image - Google Patents
A kind of plant growth state recognition methods based on plant illumination image Download PDFInfo
- Publication number
- CN108346142A CN108346142A CN201810041037.7A CN201810041037A CN108346142A CN 108346142 A CN108346142 A CN 108346142A CN 201810041037 A CN201810041037 A CN 201810041037A CN 108346142 A CN108346142 A CN 108346142A
- Authority
- CN
- China
- Prior art keywords
- plant
- organ
- sample
- image
- laplce
- 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
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000008635 plant growth Effects 0.000 title claims abstract description 30
- 238000005286 illumination Methods 0.000 title claims abstract description 24
- 210000000056 organ Anatomy 0.000 claims abstract description 233
- 239000002689 soil Substances 0.000 claims abstract description 50
- 230000012010 growth Effects 0.000 claims abstract description 28
- 238000012549 training Methods 0.000 claims abstract description 6
- 241000196324 Embryophyta Species 0.000 claims description 390
- 235000013399 edible fruits Nutrition 0.000 claims description 20
- 238000009826 distribution Methods 0.000 claims description 16
- 239000011159 matrix material Substances 0.000 claims description 12
- 230000000007 visual effect Effects 0.000 claims description 12
- 239000000284 extract Substances 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 230000005540 biological transmission Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 238000005259 measurement Methods 0.000 claims description 4
- 239000006185 dispersion Substances 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 2
- 230000007613 environmental effect Effects 0.000 abstract description 11
- 238000001514 detection method Methods 0.000 abstract description 5
- 235000021049 nutrient content Nutrition 0.000 abstract description 3
- 238000003745 diagnosis Methods 0.000 abstract description 2
- 238000005516 engineering process Methods 0.000 description 8
- 238000004458 analytical method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- IJGRMHOSHXDMSA-UHFFFAOYSA-N Atomic nitrogen Chemical compound N#N IJGRMHOSHXDMSA-UHFFFAOYSA-N 0.000 description 4
- 238000012271 agricultural production Methods 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 4
- 235000016709 nutrition Nutrition 0.000 description 3
- 240000008067 Cucumis sativus Species 0.000 description 2
- 235000010799 Cucumis sativus var sativus Nutrition 0.000 description 2
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 description 2
- 240000005860 Portulaca grandiflora Species 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000035558 fertility Effects 0.000 description 2
- 238000003064 k means clustering Methods 0.000 description 2
- 229910052757 nitrogen Inorganic materials 0.000 description 2
- 235000015097 nutrients Nutrition 0.000 description 2
- 230000035764 nutrition Effects 0.000 description 2
- 235000003715 nutritional status Nutrition 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 239000011574 phosphorus Substances 0.000 description 2
- 229910052698 phosphorus Inorganic materials 0.000 description 2
- 230000000243 photosynthetic effect Effects 0.000 description 2
- 230000035882 stress Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 235000013619 trace mineral Nutrition 0.000 description 2
- 239000011573 trace mineral Substances 0.000 description 2
- 235000007173 Abies balsamea Nutrition 0.000 description 1
- 239000002028 Biomass Substances 0.000 description 1
- OYPRJOBELJOOCE-UHFFFAOYSA-N Calcium Chemical compound [Ca] OYPRJOBELJOOCE-UHFFFAOYSA-N 0.000 description 1
- 241000208671 Campanulaceae Species 0.000 description 1
- 241000522254 Cassia Species 0.000 description 1
- 244000050510 Cunninghamia lanceolata Species 0.000 description 1
- 240000000731 Fagus sylvatica Species 0.000 description 1
- 235000010099 Fagus sylvatica Nutrition 0.000 description 1
- 235000014043 Glechoma hederacea var parviflora Nutrition 0.000 description 1
- FYYHWMGAXLPEAU-UHFFFAOYSA-N Magnesium Chemical compound [Mg] FYYHWMGAXLPEAU-UHFFFAOYSA-N 0.000 description 1
- 244000215554 Nepeta hederacea Species 0.000 description 1
- 244000131316 Panax pseudoginseng Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 241000218657 Picea Species 0.000 description 1
- 241000351396 Picea asperata Species 0.000 description 1
- 241000968366 Picea purpurea Species 0.000 description 1
- 235000008331 Pinus X rigitaeda Nutrition 0.000 description 1
- 241000018646 Pinus brutia Species 0.000 description 1
- 235000011613 Pinus brutia Nutrition 0.000 description 1
- 241000219000 Populus Species 0.000 description 1
- ZLMJMSJWJFRBEC-UHFFFAOYSA-N Potassium Chemical compound [K] ZLMJMSJWJFRBEC-UHFFFAOYSA-N 0.000 description 1
- 241000124033 Salix Species 0.000 description 1
- 244000046101 Sophora japonica Species 0.000 description 1
- 235000010586 Sophora japonica Nutrition 0.000 description 1
- NINIDFKCEFEMDL-UHFFFAOYSA-N Sulfur Chemical compound [S] NINIDFKCEFEMDL-UHFFFAOYSA-N 0.000 description 1
- 239000005864 Sulphur Substances 0.000 description 1
- 240000001949 Taraxacum officinale Species 0.000 description 1
- 235000005187 Taraxacum officinale ssp. officinale Nutrition 0.000 description 1
- 241001149649 Taxus wallichiana var. chinensis Species 0.000 description 1
- 241000332824 Tsuga chinensis Species 0.000 description 1
- 235000010183 Tsuga mertensiana Nutrition 0.000 description 1
- 241001148769 Woodwardia Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000010923 batch production Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000011575 calcium Substances 0.000 description 1
- 229910052791 calcium Inorganic materials 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000002405 diagnostic procedure Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000006353 environmental stress Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000009313 farming Methods 0.000 description 1
- 239000003337 fertilizer Substances 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000009605 growth rhythm Effects 0.000 description 1
- 210000004919 hair shaft Anatomy 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 239000011777 magnesium Substances 0.000 description 1
- 229910052749 magnesium Inorganic materials 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000037323 metabolic rate Effects 0.000 description 1
- 244000005700 microbiome Species 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 238000010899 nucleation Methods 0.000 description 1
- 230000021368 organ growth Effects 0.000 description 1
- 230000000149 penetrating effect Effects 0.000 description 1
- 239000011591 potassium Substances 0.000 description 1
- 229910052700 potassium Inorganic materials 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000036387 respiratory rate Effects 0.000 description 1
- 239000004016 soil organic matter Substances 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 230000005477 standard model Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000003971 tillage Methods 0.000 description 1
- 235000013311 vegetables Nutrition 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24133—Distances to prototypes
- G06F18/24137—Distances to cluster centroïds
-
- 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
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/46—Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30181—Earth observation
- G06T2207/30188—Vegetation; Agriculture
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Business, Economics & Management (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Quality & Reliability (AREA)
- Primary Health Care (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Mining & Mineral Resources (AREA)
- Marine Sciences & Fisheries (AREA)
- Animal Husbandry (AREA)
- Agronomy & Crop Science (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
The present invention provides a kind of plant growth state recognition methods based on plant illumination image, including：By the Lighting information for extracting the depth information and plant of collected plant long-time consecutive variations image, in conjunction with soil characteristic information, it obtains the plant coordinate of plant and standardizes the plant coordinate data of the plant, build the conditional random field models based on plant organ's cluster, and it is trained using training sample, obtain the identification model of plant organ's variable condition training, finally according to the sample characteristics of target plant to be identified, utilize the identification model, diagnosis detection plant growth state variation tendency.The present invention can be improved to plant image data and environmental data, and every environmental parameter such as the type to soil, nutrient content and intensity of illumination coloration identification, so as to accurately judge the growth conditions of each plant organ.
Description
Technical field
The present invention relates to field of agricultural production technologies, are given birth to more particularly, to a kind of plant based on plant illumination image
Long status recognition methods.
Background technology
With the development of science and technology, agricultural production is rapidly changing traditional tillage system reform, modern industrial technology
, through gradually penetrating into agricultural production, agricultural technology gradually develops to the direction of more educated informationization for oneself.Especially with industry
Change horizontal raising, modern installations agricultural is come into being, and in global rapid emergence, and gradually forms fund, technology-intensive type
New high-tech industry, become most active one of the industry in the world today.
Modern greenhouse refers generally to carry out the greenhouse of batch production plant production, it is general have automatically controlled using environment, electronics
Technology, biotechnology, robot and new material etc. carry out plant anniversary quantity-produced system.Due to industrialized facility technology
The application of management means can be such that crop is grown under controllable environment.Therefore, greenhouse vegetable production is broken to a certain extent
The region of plant growth and space-time boundary, by the control of the factors such as humiture, illumination, soil, can reach be bordering on whole year can
The condition of farming can greatly increase crop growth period, improve the yield of unit area.
Since crop is more sensitive to environmental factor, environment-stress can directly be reacted on the plant forms of crop.
Such as：In the case of cucumber nitrogen stress, stem lignifying is hardened, yellow leaf；Plant is short and small when cucumber lacks phosphorus, and tender leaf can become smaller,
There is dark gray etc. in hair shaft.In current actual agricultural production management it is mostly be administrative staff rule of thumb, by visual method into
The stress of row crop diagnoses.
It is above-mentioned based on artificial diagnosis, although achieving certain effect on the basis of experience accumulation, automating
It is required that in relatively high modern greenhouse production, not only the efficiency of management is low for this subjective diagnostic method conjestured, but also regulated party
Subjective impact lacks objectivity, can not meet the requirement of automatic management than more serious.Therefore, in modern greenhouse production
Quickly and effectively the various information such as crop nutritional status and environmental factor are diagnosed using modern technological means, are greenhouses
One of automatically control with the important foundation of decision-making management.
As shown in Figure 1, to carry out plant growth shape using plant leaf and growth point image according to a kind of of the prior art
The system diagram of state identification, the system carry out the system diagram of growth conditions detection for plant image information.Within the system, only
By the leaf image information and growing point image information of plant, by extracting the characteristic parameter in the two, based on certain
Judge index and judgment criterion, show whether plant is health status.
The growth conditions for having existed the plants such as some crops based on computer vision technique in the prior art judge
Method, but these methods or the factor of consideration are excessively single, and processing is identified only for certain organs, or ignore outer
The growth effect to plants such as crops such as factor such as soil, illumination is connect, the visual monitoring to the true growing environment of plant is lacked,
And therefore cause the recognition accuracy to vegetative state not high.
Invention content
In order to overcome the above problem or solve the above problems at least partly, the present invention provides a kind of based on plant illumination
The plant growth state recognition methods of image, to effectively improve the identification to plant image data and environmental data,
To more accurately judge the growth conditions of each plant organ.
The present invention provides a kind of plant growth state recognition methods based on plant illumination image, including：S1 is based on sample
The growing environment soil characteristic information of the long-time consecutive variations image of plant and the sample plant, obtains the sample
The plant color texture coordinate of plant, and predict that the blade of target plant absorbs illumination based on the Lighting information of the sample plant
Image；S2 carries out data normalization processing, from the sample plant image after standardization to the plant color texture coordinate
Middle extraction plant organ, and light image is absorbed based on the blade, determine the light type of the target plant；S3 passes through
Analysis counts variation high-lighting of the organ specificity of the plant organ on room and time, and analyzes the target plant and determine
The light type of plan cluster, structure concern organ group and target plant grow light；S4 is based on the concern organ group, meter
Calculate each multidimensional Laplce feature of the sample plant about each plant organ；S5, based on each plant organ's
Multidimensional Laplce's feature, structure Laplce cluster core, are corresponding device by all multidimensional Laplce feature clusterings
Official classifies, and extracts the cluster index center of each organ classes；S6, it is corresponding described for each organ classes
Plant organ builds organ characteristics dictionary based on the cluster index center belonging to each plant organ；S7 is based on
Organ characteristic's dictionary and the corresponding multidimensional Laplce feature of the sample plant build and train plant organ's cluster
Conditional random field models obtain vegetative state identification model；S8 is planted based on target plant growth light and the target
The corresponding plant organ's multidimensional Laplce's feature of light image to be identified of object, using the vegetative state identification model, in advance
Survey the growth conditions variation tendency of the target plant.
Wherein, the step of S1 further comprises：S11 extracts the depth information of the long-time consecutive variations image
With the Lighting information of the sample plant sample is determined using the object identification method based on Naive Bayes Classifier
The plant position of plant；S12, the growing environment soil characteristic of plant position and the sample plant based on the sample plant
Information obtains the plant color texture coordinate of the sample plant；S13 simulates the sample Plant Leaf using higher order filter
The light-absorbing function of piece predicts that the blade of the target plant absorbs light image.
Wherein, the step of carrying out data normalization processing to the plant color texture coordinate described in step S2 is further
Including：The plant size, plant organ's feature and plant SOIL DISTRIBUTION of the sample plant are standardized respectively；Its
In, the step of being standardized to the plant size of the sample plant, further comprises：It chooses in the sample plant
Plant forms model meets the plant of established standards as standard size plant；Keep the corresponding plant image of the sample plant
The direction of characteristic vector is constant, is that the standard size plant pair is answered by the length adjustment of each plant characteristics of image vector
Plant characteristics of image vector length；Organ centered on the designated organ of the sample plant is chosen, plant tree is built, and
According to the length of the corresponding plant characteristics of image vector of the standard size plant, mobile each plant tree adjusts mobile arrow
Amount；The step of being standardized to plant organ's feature of the sample plant further comprises：With the designated organ
As the co-ordinate zero point in new coordinate reference space, mobile all plant trees；To the plant SOIL DISTRIBUTION of the sample plant
The step of being standardized further comprises：With the growing environment soil left margin of the sample plant to right margin vector
As original coordinate system horizontal axis, original coordinate system is determined；It is described with the growing environment Soil Interface center of the sample plant
The co-ordinate zero point in new coordinate reference space, constructs the coordinate perpendicular to the growing environment Soil Interface, as the new coordinate
With reference to the coordinate vertical pivot in space；The growing environment Soil Interface is rotated, the new coordinate reference space is projected into.
Wherein, the step of organ group of structure concern described in step S3 further comprises：S31 is based on the plant organ
Corresponding color texture coordinate calculates the color texture variations distance of contiguous image frame in corresponding sample plant image；S32 is folded
Add all color texture variations distances, obtains the color texture variations figure of the plant organ of corresponding sample plant；S33, profit
With greedy algorithm, number of clusters is preset, uses Hamming distances for the similarity of corresponding plants organ, carried out the color texture and become
Change range measurement；S34 filters out the color texture variations apart from shortest plant organ's video frame figure, with the color texture
Variation changes organ outstanding as organ characteristic apart from longest two plants organ on room and time, builds corresponding life
The concern organ group of long status.
Wherein, the step of organ group of structure concern described in step S3 further comprises：Structure includes the sample plant
Root, stem, leaf, flower, fruit and seed concern organ group.
Wherein, the step of S4 further comprises：The new coordinate reference space is pressed given division rule and drawn by S41
It is divided into multiple subspaces, each plant organ is made to be in different subspaces；S42, in being with stem concern organ group
The heart calculates separately described, the subspace Laplacian density of the concern organ group of leaf, flower, fruit and seed；S43, according to every
The root of a sample plant, stem, leaf, flower, the corresponding subspace Laplce of concern organ group of fruit and seed are close
Degree, constitutes multidimensional Laplce's feature of the sample plant.
Wherein, multidimensional Laplce's feature based on each plant organ described in step S5 builds Laplce
The step of clustering core further comprises：The Laplce is constructed according to the following formula clusters core：
In formula, x, y indicate that multidimensional Laplce's feature of different plants organ, σ indicate the standard deviation of x and y.
Wherein, it is corresponding organ classes by all multidimensional Laplce feature clusterings described in step S5, and carries
The step of cluster index center for taking each organ classes, further comprises：S51 clusters core based on the Laplce,
Calculate the similarity s (x, y) of multidimensional Laplce's feature of each group plant organ；S52 is enabled for dispersion matrixSimilarity be 0, build class between similar matrix；S53, according to the similarity and institute
The total number for stating sample plant determines reference value, and determines cluster numbers by message transmission according to the sample plant；S54,
Using the greedy cluster for supporting similar matrix, the cluster index center of each organ classes is obtained.
Wherein, the step of S6 further comprises：S61 replaces the primitive organ of the sample plant and is planted for the sample
The center organ of strain obtains one group of visual vocabulary string；S62, by the visual word for analyzing plant organ's image pattern
The continuous similarity of remittance string removes the visual vocabulary string that continuous similarity is higher than given threshold, obtains organ characteristics dictionary.
Wherein, the step of S42 further comprises：S421 calculates separately described, the reason of leaf, flower, fruit and seed
By normal distribution subspace Laplacian density；S422, according to theoretical normal distribution subspace Laplacian density, setting
It is more than the Laplacian density in the organ space corresponding to the concern organ group of setpoint distance threshold value apart from subspace centre distance
Value is setting constant.
A kind of plant growth state recognition methods based on plant illumination image provided by the invention, by by the plant of plant
Strain coordinate data standardization, the organ of plant is extracted from standardized image, can effectively be promoted and be grown to plant
The accuracy of state observation detection, testing result is inaccurate caused by the factors such as avoiding image fault, overstepping the bounds.Meanwhile based on system
It counts organ characteristic in each organ and changes organ outstanding, the photometric data of analysis decision cluster, structure on room and time
Concern organ group and plant growth light are built, the omission of plant growth state possible factor can be avoided, objectively and accurately by plant
Growth conditions feed back to user.The present invention contains to plant image data and environmental data, and to the type of soil, nutrition
Every environmental parameter such as amount and intensity of illumination coloration all has higher identification, can effectively improve the knowledge to each organ of plant
It does not spend, to more accurately judge the growth conditions of each plant organ.
Description of the drawings
Fig. 1 is to carry out plant growth state identification using plant leaf and growth point image according to a kind of of the prior art
System diagram；
Fig. 2 is a kind of flow chart of the plant growth state recognition methods based on plant illumination image of the embodiment of the present invention；
Fig. 3 is a kind of leaf obtaining the plant color texture coordinate of sample plant and predict target plant of the embodiment of the present invention
Piece absorbs the flow chart of light image；
Fig. 4 is a kind of flow chart of structure concern organ group of the embodiment of the present invention；
Fig. 5 is a kind of flow chart calculating plant organ's multidimensional Laplce's feature of the embodiment of the present invention；
Fig. 6 is a kind of stream for calculating each plant organ and paying close attention to the subspace Laplacian density of organ group of the embodiment of the present invention
Cheng Tu；
Fig. 7 is a kind of flow chart carrying out multidimensional Laplce's feature clustering of the embodiment of the present invention；
Fig. 8 is a kind of flow chart of structure organ characteristics dictionary of the embodiment of the present invention；
Fig. 9 is a kind of Reducing sugar conditional random field models schematic diagram of the embodiment of the present invention.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached in the embodiment of the present invention
Figure, is clearly and completely described the technical solution in the present invention, it is clear that described embodiment is one of the present invention
Divide embodiment, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making
The every other embodiment obtained under the premise of creative work, shall fall within the protection scope of the present invention.
As one embodiment of the embodiment of the present invention, the plant life based on plant illumination image that the present embodiment provides a kind of
Long status recognition methods is a kind of plant growth state identification side based on plant illumination image of the embodiment of the present invention with reference to figure 2
The flow chart of method, including：
The growing environment soil of S1, long-time consecutive variations image and the sample plant based on sample plant are special
Property information, obtain the plant color texture coordinate of the sample plant, and mesh is predicted based on the Lighting information of the sample plant
The blade for marking plant absorbs light image；
S2 carries out data normalization processing, from the sample plant after standardization to the plant color texture coordinate
Plant organ is extracted in image, and light image is absorbed based on the blade, determines the light type of the target plant；
S3 by analyzing variation high-lighting of the organ specificity for counting the plant organ on room and time, and divides
The light type of the target plant decision cluster is analysed, structure concern organ group and target plant grow light；
S4 is based on the concern organ group, calculates multidimensional La Pula of each sample plant about each plant organ
This feature；
S5, multidimensional Laplce's feature based on each plant organ, structure Laplce clusters core, by all institutes
It is corresponding organ classes to state multidimensional Laplce's feature clustering, and extracts the cluster index center of each organ classes；
S6, for the corresponding plant organ of each organ classes, belonging to each plant organ
The cluster index center builds organ characteristics dictionary；
S7 is based on organ characteristic's dictionary and the corresponding multidimensional Laplce feature of the sample plant, builds and instruct
Practice the conditional random field models of plant organ's cluster, obtains vegetative state identification model；
S8, the corresponding plant of light image to be identified based on target plant growth light and the target plant
Organ multidimensional Laplce's feature predicts the growth conditions variation of the target plant using the vegetative state identification model
Trend.
In wherein step S1, in order to identify the growth conditions of target plant, the plant part of the target plant can be chosen
For sample plant, seek growth rhythm and characteristic, to identify growth state according to the growth characteristics of target plant.Step S1
Based on the long-time consecutive variations image of collected sample plant, the plant position of sample plant is obtained first, then
In conjunction with the soil characteristic information for the soil that the sample plant is grown, the plant color texture of plant is further obtained.
Meanwhile the Lighting information of collecting sample plant, according to the Lighting information, simulated target plant leaf blade absorbs illumination
Function, predict target plant blade absorb light image.
Wherein optional, the S1's is further processed step with reference to figure 3, is that a kind of acquisition sample of the embodiment of the present invention is planted
The plant color texture coordinate of strain simultaneously predicts that the blade of target plant absorbs the flow chart of light image, including：
S11 extracts the Lighting information of the depth information and the sample plant of the long-time consecutive variations image, utilizes
Object identification method based on Naive Bayes Classifier determines the plant position of the sample plant.
It is to be understood that this step carries out depth letter to the long-time consecutive variations image of collected sample plant first
The extraction of breath is known in combination with the Lighting information of collected sample plant using the object based on Naive Bayes Classifier
Other method determines the plant position of the sample plant.That is, extracting the depth letter of collected plant long-time consecutive variations image
The Lighting information of breath and plant confirms the plant position of plant using the object identification method based on Naive Bayes Classifier.
S12, the growing environment soil characteristic information of plant position and the sample plant based on the sample plant, is obtained
Take the plant color texture coordinate of the sample plant.
It is to be understood that according to the plant position of sample plant, it may be determined that the build environment of sample plant judges growth ring
Border soil characteristic information.According to growing environment soil characteristic information, the plant color texture coordinate of sample plant is determined.That is, knot
Close the plant color texture coordinate that soil characteristic information further obtains plant.
S13 simulates the light-absorbing function of sample plant leaf using higher order filter, predicts the target plant
Blade absorb light image.
It is to be understood that for the sample plant Lighting information of acquisition, the sample Plant Leaf is simulated using higher order filter
The light-absorbing function of piece predicts that the blade of target plant absorbs light image according to analog result.
In wherein step S2, the plant color texture coordinate of the sample plant obtained according to above-mentioned steps is subjected to data mark
Standardization extracts the organ of plant from standardized image.
Wherein, in one embodiment, the plant organ of extraction includes such as root, stem, leaf, flower, fruit, seed.
Wherein, in another embodiment, after extracting plant organ, also to wherein redundancy organ and to Plant state
The small organ of detection recognition reaction is filtered out.Obtain the plant organ of relative efficiency.
Meanwhile according to above-mentioned steps obtain target plant blade absorb light image, based on BRDF illumination models into
Row analysis, the light type of Decision Classfication target plant, such as sun life, cloudy life, shade tolerant.
It should be understood that intensity of illumination has built up important role to plant growth and morphosis, such as plant
Aetiolation is since intensity of illumination is inadequate.Light intensity also influences the development of plant simultaneously, subtracts in florescence or young fruit period, such as light intensity
It is weak, it can also cause abnormal seeding or fruit room to develop midway stopping or even shedding.Light also has good action to the quality of fruit.
According to the relationship of plant and intensity of illumination, plant can be divided into sun plant, shade plant and shade plant three
Big ecotype.The compensation point and saturation point of sun plant light are higher, it is desirable that and full exposure, photosynthetic and metabolic rate is all higher,
It is grown in the good place of illumination condition more, includes mainly dandelion, Ji, pine, China fir, poplar, willow, Chinese scholartree etc.；The light of heliophobous plant compensates
Point and saturation point are relatively low, and photosynthetic and respiratory rate is relatively low, are grown in place or the thick forest of moist back of the body sun, frequent species more
There are chain fern, Longtube Ground Ivy Herb, Chinese hemlock spruce, Chinese yew, purplecone spruce (Picea purpurea)；Shade plant grows best under full exposure, but can also restrain oneself suitable
The concealment of degree, or slight shading is needed during fertility, such as Cyclobalanopsis, beech, dragon spruce, campanulaceae, sealwort, Chinese cassia tree, party
Ginseng etc..
The light type for going out plant by above-mentioned plant data normalization means decision, judges the growth of current plant
Whether state is suitable for its existence, and in conjunction with organ growth conditions and soil etc., other environmental factors make further judgement.
Wherein optionally, the step of data normalization processing being carried out to the plant color texture coordinate described in step S2
Further comprise：
Place is standardized to the plant size, plant organ's feature and plant SOIL DISTRIBUTION of the sample plant respectively
Reason；Wherein,
The step of being standardized to the plant size of the sample plant further comprises：
It chooses plant forms model in the sample plant and meets the plant of established standards as standard size plant；
Keep the direction of the corresponding plant characteristics of image vector of the sample plant constant, it will each plant image spy
The length adjustment for levying vector is the length of the corresponding plant characteristics of image vector of the standard size plant；
Organ centered on the designated organ of the sample plant is chosen, builds plant tree, and according to the standard size
The length of the corresponding plant characteristics of image vector of plant, mobile each plant tree, adjusts mobile vector；
The step of being standardized to plant organ's feature of the sample plant further comprises：
Using the designated organ as the co-ordinate zero point in new coordinate reference space, mobile all plant trees；
The step of being standardized to the plant SOIL DISTRIBUTION of the sample plant further comprises：
Using the growing environment soil left margin of the sample plant to right margin vector as original coordinate system horizontal axis, determine
Original coordinate system；
With the co-ordinate zero point that the growing environment Soil Interface center of the sample plant is the new coordinate reference space, structure
The coordinate perpendicular to the growing environment Soil Interface is made, the coordinate vertical pivot as the new coordinate reference space；
The growing environment Soil Interface is rotated, the new coordinate reference space is projected into.
It is to be understood that the present embodiment carries out plant size criteria, plant organ's characteristic standard to sample plant respectively
Change and plant SOIL DISTRIBUTION standardizes.
Wherein, the plant size of sample plant is standardized according to the following steps：
First, select plant forms model as plant size criteria model from available sample；
Secondly, keep the direction of the plant characteristics of image vector of remaining each sample constant, by the length adjustment of each vector
For the plant morphological characteristics vector length in master pattern；
Again, by point centered on base of the plant, plant tree is built, each plant tree is moved according to vector length, adjustment moves
Dynamic vector is：
In formula,Indicate the f of current plant treeiA central point diverse vector, n indicate the variation arrow of Current central point
The number of amount.
Wherein, plant organ's feature of sample plant is standardized according to the following steps：
With according to the designated organ of above-described embodiment, i.e. co-ordinate zero point O ' of the stem center as new coordinate reference space,
The mobile all plant trees built according to above-mentioned steps.
Wherein, the plant SOIL DISTRIBUTION of sample plant is standardized according to the following steps：
First, using initial coordinate system X-axis, make the vector of itself and soil left margin to right marginIt is parallel；
Secondly, it is interface center where soil with new coordinate reference space zero O ', constructs perpendicular to new ground reference
Plane obtains new coordinate reference space Z axis；
Again, interface where rotation soil, which is projected in new reference frame.
In addition, it is to be understood that containing the various nutrients needed for plant growth in the soil organism, it can be directly or simple
Ground connection provides nitrogen, phosphorus, potassium, calcium, magnesium, sulphur and various trace elements for plant growth.The trace element contained in soil is to weigh
One of important indicator of soil fertility, is the important sources of plant nutrient and microorganism lives and the movable energy, is evaluation
The important indicator of cultivated-land.For anthropogenic soil, the key link of fertilizing is exactly to increase various organic fertilizers, carries out stalk also
Field keeps and improves soil organic matter content.
Identification mainly is standardized to soil present image to SOIL DISTRIBUTION standardization, the image measurement based on soil
Level, can the nutrient content current to soil analyze, judge whether the growth for being beneficial to current plant
In wherein step S3, analyzes per class plant organ's classification, be based on K-Means clustering algorithms, count each organ
In, organ characteristic changes organ outstanding on room and time, and the photometric data of analysis decision cluster passes through greedy algorithm structure
Build concern organ group and target plant growth light.
In one embodiment, the step of organ group of structure concern described in step S3 further comprises：Structure includes institute
State the root of sample plant, the concern organ group of stem, leaf, flower, fruit and seed.
It is to be understood that in plant organ's root, stem, leaf, flower, fruit and the kind of extracting sample plant according to above-described embodiment
It on the basis of son, counts in each organ, organ characteristic changes organ outstanding, structure concern organ on room and time
Group.Organ group includes plant：Root, stem, leaf, flower, fruit and seed.
It should be understood that it is exactly to each device of crop itself to obtain the most accurate method of plant growth conditions information
The chemical composition analysis of official, under lab this method be very effective, be study growing way diagnostic experiences method technology
Parameter foundation.Biomass accumulation, plant forms, the size of blade and color of the crops in growth course, flower-shape, fruit shape,
With the nutritional status of crop, environmental factor is closely related for the variation of yield etc..
The embodiment of the present invention from the nutritional need of crop, Appliance computer vision technical research plant growth environment and
Itself organ changes the influence to plant growth, analyzes its quantitative relationship between growing way, is determined to identification crop nutrition
The pictorial information characteristic index of state, informationization and automation to crop management, which have, all to have wide practical use.
Wherein optional, the organ group's of structure concern described in step S3 is further processed step with reference to figure 4, for the present invention
A kind of flow chart of structure concern organ group of embodiment, including：
S31 is based on the corresponding color texture coordinate of the plant organ, calculates contiguous image in corresponding sample plant image
The color texture variations distance of frame.
This step calculates the color texture variations distance of contiguous image frame in plant video image.Assuming that in adjacent image
In frame (i frames, i+1 frames), the color texture coordinate of certain plant organ is respectively (xik,yik,zik) and (xi+1,k,yi+1,k,zi+1,k),
Then color texture variations distance is dik, dikIt acquires as the following formula：
dik 2=(xik-xi+1,k)2+(yik-yi+1,k)2+(zik-zi+1,k)2。
S32 is superimposed all color texture variations distances, obtains the color texture of the plant organ of corresponding sample plant
Variation diagram.
This step is overlapped institute's colored texture variations distance that above-mentioned steps acquire as the following formula, obtains plant device
The color texture variations figure of official：
In formula, DkIndicate the color texture variations figure value of sample k-th of plant organ of plant, dikIndicate sample plant device
Color texture variations distance in k-th of plant organ, i-th of contiguous image frame of official.
S33 presets number of clusters, uses Hamming distances for the similarity of corresponding plants organ using greedy algorithm, carries out
The color texture variations range measurement.
This step is based on greedy algorithm, and prior predetermined clusters quantity is surveyed using the similarity that Hamming distances are organ
Amount.
S34 filters out the color texture variations apart from shortest plant organ's video frame figure, with the color texture variations
Change organ outstanding on room and time as organ characteristic apart from longest two plants organ, builds corresponding growth shape
The concern organ group of state.
This step filters out color texture variations apart from shortest organ video frame figure, color texture variations distance compared with
The organ characteristic of two kinds of long organs changes organ outstanding on room and time, builds the concern organ of the growth conditions
Group.
It should be understood that different types of plant in growth course organ by infringement may coefficient it is different,
It is also not quite similar by infringement degree, and even for same plant, different growth periods is by also can be different the case where infringement.Cause
This, the gap that the present invention can carry out the plant organ of different plants growth conditions identifies, by the variation shape of Different Organs
State carries out cluster differentiation, will change organ outstanding in the plant and regards as easily being encroached on or being handed over by infringement degree high organ,
And then suitable counte-rplan are specified by user.
In wherein step S4, for each organ classes of sample plant according to above-described embodiment, it is based on concern organ group,
Calculate multidimensional Laplce's feature of each organ.
Wherein optional, the S4's is further processed step with reference to figure 5, is a kind of calculating plant device of the embodiment of the present invention
The flow chart of official's multidimensional Laplce's feature, including：
The new coordinate reference space is divided into multiple subspaces by given division rule, makes each plant device by S41
Official is in different subspaces.
Hyperspace is divided into m × n × l (m, n, l ∈ Z) sub-spaces by this step, in this way, each organ is necessarily in
In one sub-spaces.
S42 calculates separately described, the concern device of leaf, flower, fruit and seed centered on the stem pays close attention to organ group
The subspace Laplacian density of official group.
It is to be understood that according to above-described embodiment, the stem of sample plant is chosen as plant center organ, accordingly originally
Step calculates separately the subspace Laplce except excentral remaining 5 organs of stem centered on stem pays close attention to organ group
Density.
Wherein, in one embodiment, the S42 is further processed step with reference to figure 6, is the embodiment of the present invention is a kind of
The flow chart that each plant organ pays close attention to the subspace Laplacian density of organ group is calculated, including：
S421 calculates separately described, the theoretical normal distribution subspace Laplacian density of leaf, flower, fruit and seed.
It is to be understood that each plant organ, its subspace Laplacian density is calculated as follows：
In formula, X indicates that joint coordinates, u indicate that subspace center, ∑ indicate covariance matrix, enable ∑=d/3*n*I, formula
In, d is indicated per the cornerwise length of sub-spaces, and n indicates that subspace dimension, I indicate canonical matrix.
S422, according to theoretical normal distribution subspace Laplacian density, setting is big apart from subspace centre distance
The Laplacian density value in the organ space corresponding to the concern organ group of setpoint distance threshold value is setting constant.
It is to be understood that for normal distribution, 99% information is included in (i.e. d*n*I, n=3 in positive and negative 3 standard deviations
~5).It enables apart from subspace centre distance dJoint, bin＞ ε (ε=d；) organ group correspond to the Laplacian density p in organ space
(X, u, ∑)=0.
S43 is corresponded to respectively according to the concern organ group of the root of each sample plant, stem, leaf, flower, fruit and seed
Subspace Laplacian density, constitute multidimensional Laplce's feature of the sample plant.
It is to be understood that for each sample plant, with 6 its root, stem, leaf, flower, fruit and seed organ space difference
Corresponding subspace Laplacian density constitutes variable condition feature representation.
In wherein step S5, using K-Means clustering algorithms, structure Laplce clusters core, will project to plant space
Multidimensional Laplacian space feature gather for M group organ classes, and extract the cluster index center for representing every group of organ.
Wherein, in one embodiment, the multidimensional Laplce based on each plant organ described in step S5 is special
The step of sign, structure Laplce clusters core, further comprises：
The Laplce is constructed according to the following formula clusters core：
In formula, x, y indicate that multidimensional Laplce's feature of different plants organ, σ indicate the standard deviation of x and y.
Wherein optionally, it is corresponding organ point by all multidimensional Laplce feature clusterings described in step S5
Class, and the step that is further processed for extracting the cluster index center of each organ classes refers to figure 7, for the embodiment of the present invention
A kind of flow chart carrying out multidimensional Laplce's feature clustering, including：
S51 clusters core based on the Laplce, calculates the similarity of multidimensional Laplce's feature of each group plant organ
s(x,y)。
Specifically, the computational methods of core are clustered according to above-mentioned Laplce, the corresponding drawing pula of structure different plants organ
This cluster core, and core is clustered according to structure Laplce, it calculates each group and acts Laplacian density characteristic similarity.
S52 is enabled for dispersion matrixSimilarity be 0, build class between it is similar
Matrix.
S53 determines reference value, and plant according to the sample according to the total number of the similarity and the sample plant
Strain, by message transmission, determines cluster numbers.
Specifically, choosing reference value as the following formula：
In formula, s (k, k) indicates that similarity, n indicate number of samples.
Cluster numbers are automatically determined by message transmission according to above-mentioned sample.
S54 obtains the cluster index center of each organ classes using supporting the greedy of similar matrix to cluster.
Specifically, using the greedy Clustering features for supporting similar matrix, each organ classes that analysis above-mentioned steps obtain obtain
Sample the corresponding cluster index center of each plant organ of this plant.
In wherein step S6, for every group of organ, using the affiliated cluster index center construction organ characteristic's word of each organ
Allusion quotation.In one embodiment, after building organ characteristics dictionary, data cleansing also is carried out to the characteristics dictionary of every group of organ,
Filter out invalidation word.
Wherein optional, the S6's is further processed step with reference to figure 8, is that a kind of structure organ of the embodiment of the present invention is special
The flow chart of dictionary is levied, including：
S61, the primitive organ for replacing the sample plant is the center organ of the sample plant, obtains one group of vision
Vocabulary string.
Specifically, by replacing the center organ that all samples of initial organ characteristic are its affiliated sample, one group is obtained
Visual vocabulary string.
S62, the continuous similarity of the visual vocabulary string by analyzing plant organ's image pattern, removal are continuous similar
Degree obtains organ characteristics dictionary higher than the visual vocabulary string of given threshold.
Specifically, the visual vocabulary string by analyzing each organic image sample, remove in vocabulary string continuous similarity compared with
High vocabulary string obtains organ characteristic's dictionary to reduce space and influence caused by time-shift between isomery sample.
In wherein step S7, the organ characteristic's dictionary built using above-mentioned steps, corresponding different plants organ builds plant
The conditional random field models of organ cluster.It is a kind of Reducing sugar conditional random field models of the embodiment of the present invention with reference to figure 9
Schematic diagram.
Then according to the corresponding multidimensional Laplce latent structure training sample of sample plant, to the item of plant organ's cluster
Part random field models are trained, by Model Parameter Optimization, the final vegetative state identification for obtaining output and meeting preset standard
Model.
Wherein, in one embodiment, when carrying out the training of conditional random field models, model parameter is carried out as the following formula
Optimization：
minθF (θ)=- logpθ(Y|X)+r(θ)；
Wherein：
In formula, m ∈ Y, j ∈ [0, T], gb(yt,yt-1[the y of)=1t=m1∧yt-1=m2], m1,m2∈Y。
By minimizing the value of object function f (θ), continuous correction model parameter θ makes model prediction identification output and reason
Reach established standards by the error of output, completes model training.
Wherein, in one embodiment, Plant state identification model is built in conjunction with Lighting information and soil information.
In wherein step S8, the corresponding plant of light image to be identified of light and target plant is grown according to target plant
Strain organ multidimensional Laplce's feature, builds the state recognition feature of target plant to be identified.By target plant to be identified
Input of the state recognition feature as vegetative state identification model, predict the growth conditions variation tendency of target plant.
A kind of plant growth state recognition methods based on plant illumination image provided in an embodiment of the present invention, by that will plant
The plant coordinate data of object standardizes, and the organ of plant is extracted from standardized image, can effectively be promoted and be planted to plant
The accuracy of strain growth conditions observation detection, testing result is inaccurate caused by the factors such as avoiding image fault, overstepping the bounds.
Meanwhile changing organ outstanding on room and time based on organ characteristic in each organ is counted, analysis is determined
The photometric data of plan cluster, structure concern organ group and plant growth light, can avoid the something lost of plant growth state possible factor
Leakage, objectively and accurately feeds back to user by plant growth state.
The present invention is to plant image data and environmental data, and the type to soil, nutrient content and illumination are strong
Every environmental parameter such as degree coloration all has higher identification, can effectively improve the resolution to each organ of plant, to
More accurately judge the growth conditions of each plant organ.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, and those skilled in the art should understand that：It still can be right
Technical solution recorded in foregoing embodiments is modified or equivalent replacement of some of the technical features；And this
A little modification or replacements, the spirit and model of various embodiments of the present invention technical solution that it does not separate the essence of the corresponding technical solution
It encloses.
Claims (10)
1. a kind of plant growth state recognition methods based on plant illumination image, which is characterized in that including：
S1, the growing environment soil characteristic letter of long-time consecutive variations image and the sample plant based on sample plant
Breath is obtained the plant color texture coordinate of the sample plant, and is planted based on the Lighting information of sample plant prediction target
The blade of object absorbs light image；
S2 carries out data normalization processing, from the sample plant image after standardization to the plant color texture coordinate
Middle extraction plant organ, and light image is absorbed based on the blade, determine the light type of the target plant；
S3 by analyzing variation high-lighting of the organ specificity for counting the plant organ on room and time, and analyzes institute
The light type of target plant decision cluster is stated, structure concern organ group and target plant grow light；
S4 is based on the concern organ group, and the multidimensional Laplce for calculating each sample plant about each plant organ is special
Sign；
S5, multidimensional Laplce's feature based on each plant organ, structure Laplce cluster core, will be all described more
Wella Prast sign cluster is corresponding organ classes, and extracts the cluster index center of each organ classes；
S6, for the corresponding plant organ of each organ classes, described in belonging to each plant organ
Cluster index center builds organ characteristics dictionary；
S7 is based on organ characteristic's dictionary and the corresponding multidimensional Laplce feature of the sample plant, builds and training is planted
The conditional random field models of strain organ cluster, obtain vegetative state identification model；
S8, the corresponding plant organ of light image to be identified based on target plant growth light and the target plant
Multidimensional Laplce's feature predicts the growth conditions variation tendency of the target plant using the vegetative state identification model.
2. according to the method described in claim 1, it is characterized in that, the step of S1 further comprise：
S11 extracts the Lighting information of the depth information and the sample plant of the long-time consecutive variations image, using based on
The object identification method of Naive Bayes Classifier determines the plant position of the sample plant；
S12, the growing environment soil characteristic information of plant position and the sample plant based on the sample plant obtain institute
State the plant color texture coordinate of sample plant；
S13 simulates the light-absorbing function of sample plant leaf using higher order filter, predicts the leaf of the target plant
Piece absorbs light image.
3. method according to claim 1 or 2, which is characterized in that sat to the plant color texture described in step S2
The step of mark progress data normalization processing, further comprises：
The plant size, plant organ's feature and plant SOIL DISTRIBUTION of the sample plant are standardized respectively；Its
In,
The step of being standardized to the plant size of the sample plant further comprises：
It chooses plant forms model in the sample plant and meets the plant of established standards as standard size plant；
Keep the direction of the corresponding plant characteristics of image vector of the sample plant constant, it will each plant characteristics of image arrow
The length adjustment of amount is the length of the corresponding plant characteristics of image vector of the standard size plant；
Organ centered on the designated organ of the sample plant is chosen, builds plant tree, and according to the standard size plant
The length of corresponding plant characteristics of image vector, mobile each plant tree, adjusts mobile vector；
The step of being standardized to plant organ's feature of the sample plant further comprises：
Using the designated organ as the co-ordinate zero point in new coordinate reference space, mobile all plant trees；
The step of being standardized to the plant SOIL DISTRIBUTION of the sample plant further comprises：
Using the growing environment soil left margin of the sample plant to right margin vector as original coordinate system horizontal axis, determine original
Coordinate system；
With the co-ordinate zero point that the growing environment Soil Interface center of the sample plant is the new coordinate reference space, construction hangs down
Directly in the coordinate of the growing environment Soil Interface, the coordinate vertical pivot as the new coordinate reference space；
The growing environment Soil Interface is rotated, the new coordinate reference space is projected into.
4. according to the method described in claim 3, it is characterized in that, building the step of paying close attention to organ group described in step S3 into one
Step includes：
S31 is based on the corresponding color texture coordinate of the plant organ, calculates contiguous image frame in corresponding sample plant image
Color texture variations distance；
S32 is superimposed all color texture variations distances, obtains the color texture variations of the plant organ of corresponding sample plant
Figure；
S33 presets number of clusters, uses Hamming distances for the similarity of corresponding plants organ, described in progress using greedy algorithm
Color texture variations range measurement；
S34 filters out the color texture variations apart from shortest plant organ's video frame figure, with the color texture variations distance
Longest two plants organ changes organ outstanding as organ characteristic on room and time, builds corresponding growth conditions
Pay close attention to organ group.
5. according to the method described in claim 4, it is characterized in that, building the step of paying close attention to organ group described in step S3 into one
Step includes：
Structure includes the concern organ group of the root of the sample plant, stem, leaf, flower, fruit and seed.
6. according to the method described in claim 5, it is characterized in that, the step of S4 further comprise：
The new coordinate reference space is divided into multiple subspaces by given division rule, made at each plant organ by S41
In in different subspaces；
S42 calculates separately described, the concern organ group of leaf, flower, fruit and seed centered on the stem pays close attention to organ group
Subspace Laplacian density；
S43, according to the root of each sample plant, the corresponding son of concern organ group of stem, leaf, flower, fruit and seed
Space Laplacian density constitutes multidimensional Laplce's feature of the sample plant.
7. according to the method described in claim 6, it is characterized in that, based on the more of each plant organ described in step S5
The step of Wella Prast is levied, and structure Laplce clusters core further comprises：
The Laplce is constructed according to the following formula clusters core：
In formula, x, y indicate that multidimensional Laplce's feature of different plants organ, σ indicate the standard deviation of x and y.
8. the method according to the description of claim 7 is characterized in that described in step S5 that all multidimensional Laplces are special
Sign cluster is corresponding organ classes, and the step of extracting the cluster index center of each organ classes further comprises：
S51, based on the Laplce cluster core, calculate each group plant organ multidimensional Laplce's feature similarity s (x,
y)；
S52 is enabled for dispersion matrixSimilarity be 0, build class between similar matrix；
S53 determines reference value, and according to the sample plant, lead to according to the total number of the similarity and the sample plant
Message transmission is crossed, determines cluster numbers；
S54 obtains the cluster index center of each organ classes using supporting the greedy of similar matrix to cluster.
9. according to the method described in claim 3, it is characterized in that, the step of S6 further comprise：
S61, the primitive organ for replacing the sample plant is the center organ of the sample plant, obtains one group of visual vocabulary
String；
S62, the continuous similarity of the visual vocabulary string by analyzing plant organ's image pattern, it is high to remove continuous similarity
In the visual vocabulary string of given threshold, organ characteristics dictionary is obtained.
10. according to the method described in claim 6, it is characterized in that, the step of S42 further comprise：
S421 calculates separately described, the theoretical normal distribution subspace Laplacian density of leaf, flower, fruit and seed；
S422, according to theoretical normal distribution subspace Laplacian density, setting is more than apart from subspace centre distance to be set
The Laplacian density value in the organ space corresponding to the concern organ group of set a distance threshold value is setting constant.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810041037.7A CN108346142B (en) | 2018-01-16 | 2018-01-16 | Plant growth state identification method based on plant illumination image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810041037.7A CN108346142B (en) | 2018-01-16 | 2018-01-16 | Plant growth state identification method based on plant illumination image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108346142A true CN108346142A (en) | 2018-07-31 |
| CN108346142B CN108346142B (en) | 2020-08-14 |
Family
ID=62960365
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810041037.7A Expired - Fee Related CN108346142B (en) | 2018-01-16 | 2018-01-16 | Plant growth state identification method based on plant illumination image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108346142B (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109815846A (en) * | 2018-12-29 | 2019-05-28 | 腾讯科技（深圳）有限公司 | Image processing method, device, storage medium and electronic device |
| CN111027375A (en) * | 2019-10-29 | 2020-04-17 | 厦门迈信物联科技股份有限公司 | Automatic identification method for plant growth quality |
| CN111462301A (en) * | 2020-03-05 | 2020-07-28 | 清华大学 | Method for constructing generation model for multi-view plant three-dimensional reconstruction |
| CN114266734A (en) * | 2021-12-03 | 2022-04-01 | 深圳市桑瑞生物科技有限公司 | Plant growth stage identification method, device, equipment and storage medium |
| CN114565846A (en) * | 2022-02-24 | 2022-05-31 | 杭州睿胜软件有限公司 | Plant growing environment identification method and equipment and computer readable storage medium |
| CN114973001A (en) * | 2022-06-17 | 2022-08-30 | 东北林业大学 | Wetland plant growth information monitoring method |
| CN115067097A (en) * | 2022-07-26 | 2022-09-20 | 江苏桁壹生物科技有限公司 | Plant cultivation artificial light supplementing method, system and equipment |
| CN115601413A (en) * | 2022-10-25 | 2023-01-13 | 广州市嘉卉园林绿化建筑工程有限公司（Cn） | Landscaping maintenance monitoring and early warning method and system |
| CN116076253A (en) * | 2022-12-19 | 2023-05-09 | 河源职业技术学院 | Novel grafting technique 'scissors' type branch grafting method |
| CN116311086A (en) * | 2023-05-23 | 2023-06-23 | 苏州浪潮智能科技有限公司 | Plant monitoring method, training method, device and equipment for plant monitoring model |
| CN116824581A (en) * | 2023-06-08 | 2023-09-29 | 中国科学院空间应用工程与技术中心 | Method, system and storage medium for generating spatial rice leaf tip detection and growth track |
| CN117150785A (en) * | 2023-09-05 | 2023-12-01 | 之江实验室 | Soybean growth full-growth period simulation method and system |
| CN117236070A (en) * | 2023-11-08 | 2023-12-15 | 吉林大学 | Regional corn plant phosphorus demand quantitative acquisition method based on optimal yield |
| CN117910662A (en) * | 2024-03-19 | 2024-04-19 | 中国建筑西南设计研究院有限公司 | Urban building group green land arrangement optimization method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103955938A (en) * | 2014-05-15 | 2014-07-30 | 安徽农业大学 | Wheat growing status diagnosing method based on mobile internet mode and leaf color analysis |
| US9652840B1 (en) * | 2014-10-30 | 2017-05-16 | AgriSight, Inc. | System and method for remote nitrogen monitoring and prescription |
| CN106682570A (en) * | 2016-11-04 | 2017-05-17 | 东莞市隆声智能科技有限公司 | Method and device for monitoring growing situations of plants |
| CN106971409A (en) * | 2017-02-23 | 2017-07-21 | 北京农业信息技术研究中心 | Maize canopy leaf color modeling and method |
| CN107295310A (en) * | 2017-07-31 | 2017-10-24 | 深圳前海弘稼科技有限公司 | Planting monitoring method and planting monitoring device |
-
2018
- 2018-01-16 CN CN201810041037.7A patent/CN108346142B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103955938A (en) * | 2014-05-15 | 2014-07-30 | 安徽农业大学 | Wheat growing status diagnosing method based on mobile internet mode and leaf color analysis |
| US9652840B1 (en) * | 2014-10-30 | 2017-05-16 | AgriSight, Inc. | System and method for remote nitrogen monitoring and prescription |
| CN106682570A (en) * | 2016-11-04 | 2017-05-17 | 东莞市隆声智能科技有限公司 | Method and device for monitoring growing situations of plants |
| CN106971409A (en) * | 2017-02-23 | 2017-07-21 | 北京农业信息技术研究中心 | Maize canopy leaf color modeling and method |
| CN107295310A (en) * | 2017-07-31 | 2017-10-24 | 深圳前海弘稼科技有限公司 | Planting monitoring method and planting monitoring device |
Non-Patent Citations (3)
| Title |
|---|
| HAN LI等: "Identifying blueberry fruit of different growth stages using natural outdoor color images", 《COMPUTERS AND ELECTRONICS IN AGRICULTURE》 * |
| 贾彪: "基于计算机视觉技术的棉花长势监测***构建", 《中国博士学位论文全文数据库农业科技辑》 * |
| 赵晓兰: "作物长势监控图像中绿色植物的识别方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109815846A (en) * | 2018-12-29 | 2019-05-28 | 腾讯科技（深圳）有限公司 | Image processing method, device, storage medium and electronic device |
| CN111027375A (en) * | 2019-10-29 | 2020-04-17 | 厦门迈信物联科技股份有限公司 | Automatic identification method for plant growth quality |
| CN111462301A (en) * | 2020-03-05 | 2020-07-28 | 清华大学 | Method for constructing generation model for multi-view plant three-dimensional reconstruction |
| CN111462301B (en) * | 2020-03-05 | 2022-10-28 | 清华大学 | Method for constructing generation model for multi-view plant three-dimensional reconstruction |
| CN114266734A (en) * | 2021-12-03 | 2022-04-01 | 深圳市桑瑞生物科技有限公司 | Plant growth stage identification method, device, equipment and storage medium |
| CN114266734B (en) * | 2021-12-03 | 2023-04-07 | 深圳市桑瑞生物科技有限公司 | Plant growth stage identification method, device, equipment and storage medium |
| CN114565846A (en) * | 2022-02-24 | 2022-05-31 | 杭州睿胜软件有限公司 | Plant growing environment identification method and equipment and computer readable storage medium |
| CN114973001A (en) * | 2022-06-17 | 2022-08-30 | 东北林业大学 | Wetland plant growth information monitoring method |
| CN115067097A (en) * | 2022-07-26 | 2022-09-20 | 江苏桁壹生物科技有限公司 | Plant cultivation artificial light supplementing method, system and equipment |
| CN115601413B (en) * | 2022-10-25 | 2023-06-16 | 广州市嘉卉园林绿化建筑工程有限公司 | Afforestation maintenance monitoring and early warning method and system |
| CN115601413A (en) * | 2022-10-25 | 2023-01-13 | 广州市嘉卉园林绿化建筑工程有限公司（Cn） | Landscaping maintenance monitoring and early warning method and system |
| CN116076253A (en) * | 2022-12-19 | 2023-05-09 | 河源职业技术学院 | Novel grafting technique 'scissors' type branch grafting method |
| CN116311086A (en) * | 2023-05-23 | 2023-06-23 | 苏州浪潮智能科技有限公司 | Plant monitoring method, training method, device and equipment for plant monitoring model |
| CN116311086B (en) * | 2023-05-23 | 2023-08-22 | 苏州浪潮智能科技有限公司 | Plant monitoring method, training method, device and equipment for plant monitoring model |
| CN116824581A (en) * | 2023-06-08 | 2023-09-29 | 中国科学院空间应用工程与技术中心 | Method, system and storage medium for generating spatial rice leaf tip detection and growth track |
| CN117150785A (en) * | 2023-09-05 | 2023-12-01 | 之江实验室 | Soybean growth full-growth period simulation method and system |
| CN117150785B (en) * | 2023-09-05 | 2024-05-24 | 之江实验室 | Soybean growth full-growth period simulation method and system |
| CN117236070A (en) * | 2023-11-08 | 2023-12-15 | 吉林大学 | Regional corn plant phosphorus demand quantitative acquisition method based on optimal yield |
| CN117236070B (en) * | 2023-11-08 | 2024-01-30 | 吉林大学 | Regional corn plant phosphorus demand quantitative acquisition method based on optimal yield |
| CN117910662A (en) * | 2024-03-19 | 2024-04-19 | 中国建筑西南设计研究院有限公司 | Urban building group green land arrangement optimization method |
| CN117910662B (en) * | 2024-03-19 | 2024-05-28 | 中国建筑西南设计研究院有限公司 | Urban building group green land arrangement optimization method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108346142B (en) | 2020-08-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108346142A (en) | A kind of plant growth state recognition methods based on plant illumination image | |
| Zhou et al. | Strawberry maturity classification from UAV and near-ground imaging using deep learning | |
| Sonohat et al. | Three-dimensional reconstruction of partially 3D-digitized peach tree canopies | |
| Alejandrino et al. | Visual classification of lettuce growth stage based on morphological attributes using unsupervised machine learning models | |
| Chen et al. | Diagnosing of rice nitrogen stress based on static scanning technology and image information extraction | |
| CN105653895B (en) | A kind of strain of sargassum fusifome differentiates and sorting technique | |
| CN113469112B (en) | Crop growth condition image identification method and system | |
| CN106651149A (en) | Plant growth behavior analyzing method | |
| Wang et al. | Digital image processing technology under backpropagation neural network and K-Means Clustering algorithm on nitrogen utilization rate of Chinese cabbages | |
| CN111967441A (en) | Crop disease analysis method based on deep learning | |
| CN112215522A (en) | Crop growth monitoring system, method, computer equipment and storage medium | |
| CN115728249A (en) | Prediction method for chlorophyll content of tomato seedlings and processing terminal | |
| CN112364710B (en) | Plant electric signal classification and identification method based on deep learning algorithm | |
| CN105184234B (en) | A kind of Wheat Straw burns the measuring method and device of pollutant discharge amount | |
| CN116778477B (en) | Corn ear character index calculation method based on image processing | |
| Tobias et al. | Hybrid tree-fuzzy logic for aquaponic lettuce growth stage classification based on canopy texture descriptors | |
| Cui et al. | Phenotypic characteristics of ramie (Boehmeria nivea L) germplasm resources based on UAV remote sensing | |
| Yang et al. | Feature extraction of cotton plant height based on DSM difference method | |
| Feng et al. | A real-time monitoring and control system for crop | |
| CN103125371B (en) | Method for carrying out field rapid assistant screening on crop drought-resisting mutants | |
| ZHANG et al. | Nutrient deficiency diagnosis in whole hydroponic lettuce based on random forest. | |
| Guo et al. | High-throughput estimation of plant height and above-ground biomass of cotton using digital image analysis and Canopeo | |
| Butler | Making the replant decision: utilization of an aerial platform to guide replant decisions in Tennessee cotton | |
| Burton | Phenotypic evaluation and genetic basis of anatomical and architectural root traits in the genus Zea | |
| J Hilli | An overview on phenomics applications in different agriculture disciplines |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200814 Termination date: 20210116 |