CN108399400A - A kind of early stage crop recognition methods and system based on high-definition remote sensing data - Google Patents
A kind of early stage crop recognition methods and system based on high-definition remote sensing data Download PDFInfo
- Publication number
- CN108399400A CN108399400A CN201810246796.7A CN201810246796A CN108399400A CN 108399400 A CN108399400 A CN 108399400A CN 201810246796 A CN201810246796 A CN 201810246796A CN 108399400 A CN108399400 A CN 108399400A
- Authority
- CN
- China
- Prior art keywords
- crop
- data
- sentinel
- preferred feature
- remote sensing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/188—Vegetation
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The present invention proposes a kind of early stage crop recognition methods based on high-definition remote sensing data and system, the method includes：S1 collects the ground data of 1 data of Sentinel and Sentinel 2 data and crop location, and the ground data includes crop geographical location and type；S2 obtains the generated time sequence of T time length by Data Synthesis method；S3 extracts the corresponding characteristic of division of each crop；S4, for the corresponding feature of each crop obtained in S3, evaluate the separability of contribution and crop that each characteristic of division identifies crop, obtain the preferred feature of identification crop；S5, using the preferred feature wave band of the preferred feature time series and image obtained in S4, be based on immune system network algorithm, the shorter preferred feature time series of usage time sequence length is identified into row crop.The present invention is identified using multidate high-resolution data into row crop, and under the physical condition of China's farmland massif opposed breaker, this method has stronger applicability.
Description
Technical field
The present invention relates to remote sensing technologies, more specifically, are related to carrying out early stage crop based on high-definition remote sensing data
The technology of identification.
Background technology
In past many decades, world population speedup is very fast, is predicted according to the United Nations, and world population will be broken through in the year two thousand thirty
8700000000, this will carry out huge challenge to world food safety belt.Since crop-planting remote sensing monitoring business can support world food
Cultivate administration, the sustainable development for pushing resource, the important leverage of world food safety；And crops planting area is crop-planting
Remote sensing monitoring important component is the input data of crop growth monitoring；So being badly in need of timely, accurate high spatial at present
The plantation of several main cereal crops of resolution ratio crop-planting distributed data, especially wheat, corn, rice and soybean etc. point
Cloth.
In order to which the growth course to crop is more effectively managed with harvest, so better Guarantee Grain Production, mesh
It is preceding to be badly in need of acquisition crop pattern map as early as possible.But most of crop recognition methods needs the remote sensing using the whole year at present
Data are identified into row crop, thus can only could obtain crop pattern map in the annual end of the year or second year, these crops identification knot
Fruit is difficult to support crop growth monitoring and crop management.
In order to realize crop ahead of time identify, it usually needs using multi-temporal remote sensing data (under normal circumstances data when
Between resolution ratio should be higher than that 15 days/phase)；And the remotely-sensed data spatial resolution of most of high time resolutions is relatively low, for example,
Resolution data is daily observation between NOAAAVHRR data and the time of MODIS data, but its spatial resolution is respectively
1km and 250m.Since the universal plot in China farmland is broken, if extracted into row crop using these data, often will appear big
It measures " mixed pixel ", influences crop extraction accuracy.On the other hand, it is relatively broken to be suitable for China for the remotely-sensed data of high spatial resolution
The extraction in farmland, the usual temporal resolution of these data is relatively low, for example, the U.S. Landsat data, although its spatial discrimination
Rate is 30m, but its temporal resolution is 16 days.And these satellites are observed using wave bands such as visible light, near-infrared and short-wave infrareds
Earth's surface cannot usually obtain effective observation data due to the influence of cloud when observation.So being obtained using Landsat data are practical
The image obtained cannot be guaranteed to obtain for 1 phase in every 16 days.This " invalid observation " further reduced the temporal resolution of remotely-sensed data,
Difficulty is brought to identification crop ahead of time, still, most of Classification in Remote Sensing Image algorithms (such as maximum likelihood method, decision tree method, branch at present
Hold vector machine method, random forest method etc.) the case where " invalid observation " in image cannot be handled.So being needed at present in China
It is extracted into row crop using multidate high-resolution (spatial resolution is better than 30m) data；And for the " nothing of wherein generally existing
The problem of effect observation ", proposes corresponding solution.
When carrying out Crops Classification, there are a large amount of characteristic of division to can be used for the identification of crop, such as visible light-near infrared band
(feux rouges, green light, blue light etc.), geometric properties (height, porosity, coverage etc.), phenology feature (time started in growth period, life
Long-term end time etc.).Some researches show that less phase and less wave band can realize accurate classification, but have one
A little phases and wave band have larger contribution to Accurate classification.Maximum spy is contributed to classification so being filtered out from a large amount of feature
Sign, and the feasibility that Crops Classification is carried out using these features is proved, it, can be notable to extract crop using only optimal characteristics
Improve the efficiency of Crops Classification.But larger Classification in Remote Sensing Image feature is contributed in identifying the work of specific crop ahead of time at present
It is still not clear.
Invention content
The purpose of the present invention is using remotely-sensed data to identify agrotype (such as cotton, spring maize) ahead of time, the present invention carries
Go out in one and know method for distinguishing ahead of time into row crop using high-definition remote sensing data, realizes high-resolution crop identification, institute
The method of stating includes：
A kind of early stage crop recognition methods based on high-definition remote sensing data, which is characterized in that including：
S1 collects the ground data of remotely-sensed data and crop location, and the remotely-sensed data is that remotely-sensed data is
Sentinel-1 data and Sentinel-2 data, the ground data include crop geographical location and type；
S2 obtains synthesis Sentinel-1 and the Sentinel-2 time series of T time length by Data Synthesis method, makees
For generated data；
It is special to extract the corresponding classification of each crop according to the generated data obtained in the training sample and S2 obtained in S1 by S3
Sign；
S4, for the corresponding feature of each crop obtained in S3, evaluate the contribution that each characteristic of division identifies crop
With the separability of crop, the preferred feature of identification crop is obtained；
S5, using the preferred feature wave band of the preferred feature time series and image obtained in S4, be based on immune system net
Network algorithm, according to the needs of Different Crop, the shorter preferred feature time series of usage time sequence length is identified into row crop.
The present invention also proposes a kind of early stage crop identifying system based on high-definition remote sensing data, which is characterized in that packet
Processor and memory are included, the program that the processor is able to carry out wherein is stored on memory, described program is performed
Realize following steps：Computer program
S1 collects the ground data of remotely-sensed data and crop location, and the remotely-sensed data is that remotely-sensed data is
Sentinel-1 data and Sentinel-2 data, the ground data include crop geographical location and type；
S2 obtains synthesis Sentinel-1 and the Sentinel-2 time series of T time length by Data Synthesis method, makees
For generated data；
It is special to extract the corresponding classification of each crop according to the generated data obtained in the training sample and S2 obtained in S1 by S3
Sign；
S4, for the corresponding feature of each crop obtained in S3, evaluate the contribution that each characteristic of division identifies crop
With the separability of crop, the preferred feature of identification crop is obtained；
S5, using the preferred feature wave band of the preferred feature time series and image obtained in S4, be based on immune system net
Network algorithm, according to the needs of Different Crop, the shorter preferred feature time series of usage time sequence length is identified into row crop.
Beneficial effects of the present invention include：
1, the present invention is identified using multidate high-resolution data into row crop, in the reality of China's farmland massif opposed breaker
Under the conditions of border, this method has stronger applicability.
2, the present invention solves the crop identification based on multi-temporal data under the conditions of " shortage of data ".
3, the present invention is when extract crop, has carried out the assessment of feature importance, and it is preferred to identify winter wheat, cotton and
Corn contributes larger feature memory crop to identify, and then improves the efficiency of crop identification.
Description of the drawings
Fig. 1 is the flow chart of the method for the present invention.
Fig. 2 is the sample of the corresponding feature of each crop obtained according to the method for the present invention.
Fig. 3 is the sample of the preferred feature time series of the training sample obtained according to the method for the present invention.
Fig. 4 is the flow chart of an embodiment of the method for the present invention.
Fig. 5 is " classification " process in improved immune system network (IAIN) algorithm.
Fig. 6 is using the Hengshui City winter wheat distribution map that 4~May, data obtained.
Fig. 7 is Hengshui City cotton, spring maize and the summer corn distribution map obtained using 4~August data.
Fig. 8 is part " shortage of data " territorial classification result figure.
Specific implementation mode
Embodiments of the present invention are described with reference to the accompanying drawings, wherein identical component is presented with like reference characters.
In the absence of conflict, the technical characteristic in following embodiment and embodiment can be combined with each other.
Crop do sth. in advance identify main flow be：
S1, remotely-sensed data and ground data are collected.
When identifying crop using remotely-sensed data, need to carry out field investigation, specific method be in farmland where crop,
Using GPS positioning, and the agrotype that record location point is planted.These GPS location datas and agrotype data are ground
Face data, ground data can be used as training sample and verification sample.The composition of ground data, such as [record 1, longitude and latitude：
116.45E, 38.48N, the crop of the position：Cotton].In addition, ground data is equally divided into two parts, a part is as instruction
Practice sample, a part as verification sample.
The remotely-sensed data that the present invention collects is Sentinel-1 data and Sentinel-2 data, such as on April 1st, 2017
To August all Sentinel-1 data and Sentinel-2 data on the 31st.The wave band that Sentinel-1 data include is " the poles VV
Change " wave band and " VH polarization " wave band；The wave band that Sentinel-2 data include be " blue light ", " green light ", " feux rouges " and " closely it is red
Four wave bands outside ".Later, " NDVI " is calculated according to " feux rouges " and " near-infrared " two wave bands.
NDVI=(NIR-Red)/(NIR+Red) (1)
Formula (1) is the calculation formula of NDVI, wherein NIR represents near infrared band, Red represents red spectral band.
S2, for the remotely-sensed data in S1, pass through the synthesis that Data Synthesis method obtains T time length (such as 15 days)
Sentinel-1 and Sentinel-2 time serieses, as generated data.
Specifically, by Data Synthesis method in step S1 Sentinel-1 data and Sentinel-2 data close
At the time span of simultaneous selection is, for example, T=15 days, obtains Sentinel-1 the and Sentinel-2 time sequences of synthesis in T days
Row.
Such as obtain n issue evidences altogether in T time, then shown in building-up process such as formula (2)：
Xcomposition=Median (X1,X2,...Xn) (2)
Median () function is to take median function.
For example, going out August generated data on the 1st~15 with August 1 day to August Data Synthesis on the 15th, that is, obtain T time length
Generated data.If using 4-8 month data, T=15, then the data of the 4-8 months are the data of 10 15 days time spans of phase, are passed through
Synthesis Sentinel-1 and the Sentinel-2 time series of formula (2) movable 15 day time as above.
The time series data of different time sequence length can carry out synthesizing for T days.With reference to Fig. 4,4~May in this patent,
4~August and complete time sequence data are all made of 15 days synthetic schemes.
Since image is influenced by cloud, the case where there are shortage of data, by the synthesis of more phase images, in the image of synthesis
In, the picture dot of " shortage of data " will be reduced.In building-up process, need the setting for considering the length of T emphatically, if the length of T compared with
It is long, such as 30 days or 60 days, the interior data for Data Synthesis obtained are more during this period, remove " shortage of data " picture dot
Effect it is higher；But the disadvantage is that the temporal resolution of composite result is relatively low, such as T is 60 days, is only capable of within 1 year obtaining 6 synthesis knots
Fruit is unfavorable for crop identification work.It is preferred, therefore, that being synthesized using 15 days, while ensureing generated data temporal resolution, have
Ensure the effect of removal " shortage of data " phenomenon as far as possible.
S3, the corresponding characteristic of division of each crop is extracted.
According to the generated data obtained in the training sample and S2 obtained in S1, it is special to obtain the corresponding classification of each crop
Sign, the results are shown in Figure 2 for step acquisition.
In the present invention, characteristic of division refer to T time length (such as 15 days) synthesis Sentinel-1 and
The data of 7 wave bands in Sentinel-2 time serieses, 7 wave bands refer to：" VV polarization ", " VH polarization ", " feux rouges ",
" green light ", " blue light ", " near-infrared " and " NDVI " wave band.
For example, if using 1 day~August April in 2017 all Sentinel-1 data and Sentinel-2 numbers on the 31st
According to.By Data Synthesis, 1 day~August April in 2017 every 15 days generated datas (totally 10 phase) on the 31st can be obtained, per issue
According to including " VV polarization ", " VH polarization ", " feux rouges ", " green light ", " blue light ", " near-infrared " and " NDVI " wave band.All these waves
Section is exactly that crop is corresponding " characteristic of division ", such as first-phase NDVI in April, is exactly a feature.So sharing 70 here
Characteristic of division.
S4, for the corresponding feature of each crop obtained in S3, evaluated using random forest method and JM Furthest Neighbors each
The separability of contribution (prominence score) and crop that characteristic of division identifies crop obtains the preferred feature of identification crop.
Specifically, when identifying crop, prominence score is carried out to each characteristic of division and calculates JM distances, selects work accordingly
The best features of object identification.
From " VV polarization " of the present invention, " VH polarization ", " feux rouges ", " green light ", " blue light ", " near-infrared " and " NDVI "
In 7 features of wave band, by calculating Gini scorings and the JM distances of random forest, 2 higher and JM of Gini scorings are selected
Apart from larger feature as preferred feature.
For example, the Gini of the 10th phase " NDVI " scores, for summation JM apart from summation highest, " near-infrared " second is high, then selects
" NDVI " and " near-infrared " is used as preferred feature.The preferred feature time series (as shown in Figure 3) of the training sample finally obtained
With in image by the wave band of these preferred features.
S5, using the preferred feature time series of the training sample obtained in S4 and the preferred feature wave band of image, be based on
Improved immune system network algorithm (such as IAIN), according to the needs of Different Crop, shorter excellent of usage time sequence length
Characteristic time sequence is selected to be identified into row crop.Herein, length of time series refers to the time series data of crop for identification
Length, such as identify crop with data in April~June.
For example, winter wheat harvests in June, winter wheat is identified using 4~5 monthly series data (4 phase)；Cotton 9~
October harvest, spring maize was harvested in late August, summer corn is in late September~October harvest, using 4~August data (10 phase),
Identify spring maize, cotton and summer corn.Finally obtain：It is several that this is obtained before winter wheat, spring maize, cotton and summer corn harvest
The 10m resolution space distribution maps of kind crop, specific experiment result are described below with regard to Fig. 6-8.
Grader is obtained by operation.Specifically, when identifying winter wheat, first by the preferred feature sequence (4 of training sample
~May part) input grader (IAIN)；Using " classifier training " process of immune system network method, obtaining classification needs
" antibody " to be used is used herein improved immune system network (IAIN) method and is identified into row crop.The innovation of the present invention
It is to improve algorithm " classification " process, improved though is as shown in Figure 4.Concrete principle is as follows：
Each picture dot in the classification image is " antigen " (see Fig. 5), ideally, the intrinsic dimensionality of antigen and antibody
It is identical, thus identify antigen, the i.e. distance of the characteristic sequence of the characteristic sequence of calculating antigen to each antibody using antibody, and will
The agrotype of the minimum antibody of distance, the classification results as picture dot (antigen).But when appearance " shortage of data " in image
When phenomenon, antigen is different with the dimension of antibody, thus antibody cannot identify antigen.In this case, positioning " has in " antigen "
The position of effect data ", and for all " antibody ", effective sequence corresponding with " antigen " is filtered out according to " valid data " position
Row are used as " new antibodies ", calculate the characteristic sequence of " antigen " to the distance of all " new antibodies ", and will be apart from minimum antibody
Agrotype, the classification results as picture dot (antigen).This method can be adaptive the selection from " antibody " and remote sensing shadow
" valid data " corresponding feature is classified as in, to exclude influence of " shortage of data " phenomenon to classification.
Specifically, step S5 includes：
S51 inputs the excellent of the training sample obtained in S4 using " classifier training " process of immune system network method
Characteristic time sequence is selected, obtaining classification needs " antibody " to be used.
Specific method please refers to " An Adaptive Artificial Immune Network for Supervised
Classification of Multi-/Hyperspectral Remote Sensing Imagery”(Yanfei Zhong,
Member,IEEE,and Liangpei Zhang,IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE
SENSING,VOL.50,NO.3,MARCH 2012)
S52, by each picture dot in image of classifying as " antigen ", the position of " valid data " in positioning " antigen ".
Such as the sequence in " antigen " is A (a1, a2, a3 ... an), wherein a2 is " shortage of data value ", can be used
The where functions in find functions or idl softwares in matlab tools, list entries A, these functions are effective by auto-returned
The sequence (1,3 ... n) of Data Position.
S53 filters out ordered sequence conduct corresponding with " antigen " to owning " antibody " according to " valid data " position
" new antibodies ".
Such as the sequence of antibody is B (b1, b2, b3 ... bn), and the sequence of valid data position is (1,3 ... n), that
" new antibodies " are B (b1, b3 ... bn), and automatically generating for " new antibodies " has been achieved.
S54, the characteristic sequence for calculating " antigen " arrive the distance of all " new antibodies ", and by the crop of the antibody of distance minimum
Type, the classification results as each picture dot (" antigen ").
In Hebei, Hengshui City is tested with regard to method proposed by the present invention, the experimental results showed that, NIR (near-infrared) wave band
With NDVI (normalized differential vegetation index) be extract winter wheat, cotton, spring maize and summer corn best features.It is adjusted on the spot through ground
Sample verification is looked into, when extracting winter wheat using the data in 4~May, the cartographic accuracy and user's precision of winter wheat are respectively
99.28% and 95.42%, the spatial distribution of Hengshui City winter wheat is as shown in Figure 6.Use the data extraction cotton of 4~August, spring
When corn and summer corn, the user's precision and cartographic accuracy of cotton are respectively 95.42% and 99.30%；The drawing essence of spring maize
Degree and user's precision are respectively 97.92% and 98.95,；The cartographic accuracy and user's precision of summer corn are respectively 95.81% He
99.88%.The spatial distribution of these three crops is as shown in Figure 7.In addition, the IAIN methods in the present invention can effectively solve " number
According to missing " on being influenced caused by crop identification.In Fig. 8, the region within red line is shortage of data, in the classification results enjoyed
In, crop identifies the influence for not receiving shortage of data.
Embodiment described above, the only present invention more preferably specific implementation mode, those skilled in the art is at this
The usual variations and alternatives carried out within the scope of inventive technique scheme should be all included within the scope of the present invention.
Claims (10)
1. a kind of early stage crop recognition methods based on high-definition remote sensing data, which is characterized in that including：
S1 collects the ground data of remotely-sensed data and crop location, and the remotely-sensed data is that remotely-sensed data is Sentinel-1
Data and Sentinel-2 data, the ground data include crop geographical location and type；
S2 obtains synthesis Sentinel-1 and the Sentinel-2 time series of T time length by Data Synthesis method, as conjunction
At data；
S3 extracts the corresponding characteristic of division of each crop according to the generated data obtained in the training sample and S2 obtained in S1；
S4, for the corresponding feature of each crop obtained in S3, evaluate the contribution and work that each characteristic of division identifies crop
The separability of object obtains the preferred feature of identification crop；
S5, using the preferred feature wave band of the preferred feature time series and image obtained in S4, calculated based on immune system network
Method, according to the needs of Different Crop, the shorter preferred feature time series of usage time sequence length is identified into row crop.
2. the early stage crop recognition methods according to claim 1 based on high-definition remote sensing data, which is characterized in that
In S1, the wave band that the Sentinel-1 data in the remotely-sensed data of collection include is VV polarization wave bands and VH polarization
Wave band；The wave band that Sentinel-2 data include is blue wave band, green light band, red spectral band and near infrared band, Yi Jigen
The NDVI wave bands calculated according to following formula：NDVI=(NIR-Red)/(NIR+Red), wherein NIR represent near infrared band, Red
Represent red spectral band.
3. the early stage crop recognition methods according to claim 2 based on high-definition remote sensing data, which is characterized in that
In S2, the T time length is 15 days；
In S3, the characteristic of division refers to 7 in synthesis Sentinel-1 and the Sentinel-2 time series of T time length
The data of a wave band.
4. the early stage crop recognition methods according to claim 3 based on high-definition remote sensing data, which is characterized in that
In S4, when identifying crop, using random forest method and JM Furthest Neighbors to each characteristic of division carry out prominence score and
JM distances are calculated, select the best features of crop identification accordingly.
5. the early stage crop recognition methods according to claim 4 based on high-definition remote sensing data, which is characterized in that
By calculating Gini scorings and the JM distances of random forest, two Gini is selected to score higher and JM apart from larger spy
Sign is used as preferred feature.
6. the early stage crop recognition methods according to claim 3 based on high-definition remote sensing data, which is characterized in that step
Suddenly S5 includes：
S51 inputs the preferred feature of the training sample obtained in S4 using the classifier training process of immune system network method
Time series, obtaining classification needs antibody to be used；
S52 positions the position of valid data in antigen using each picture dot in image of classifying as antigen；
S53 filters out ordered sequence corresponding with antigen as new antibodies to all antibody according to valid data position；
S54 calculates the characteristic sequence of antigen to the distance of all new antibodies, and by the agrotype of the minimum antibody of distance, makees
For the classification results of each antigen.
7. a kind of early stage crop identifying system based on high-definition remote sensing data, which is characterized in that including processor and storage
Device, is wherein stored with the program that the processor is able to carry out on memory, described program is performed realization following steps：
S1 collects the ground data of remotely-sensed data and crop location, and the remotely-sensed data is that remotely-sensed data is Sentinel-1
Data and Sentinel-2 data, the ground data include crop geographical location and type；
S2 obtains synthesis Sentinel-1 and the Sentinel-2 time series of T time length by Data Synthesis method, as conjunction
At data；
S3 extracts the corresponding characteristic of division of each crop according to the generated data obtained in the training sample and S2 obtained in S1；
S4, for the corresponding feature of each crop obtained in S3, evaluate the contribution and work that each characteristic of division identifies crop
The separability of object obtains the preferred feature of identification crop；
S5, using the preferred feature wave band of the preferred feature time series and image obtained in S4, calculated based on immune system network
Method, according to the needs of Different Crop, the shorter preferred feature time series of usage time sequence length is identified into row crop.
8. the early stage crop identifying system according to claim 7 based on high-definition remote sensing data, which is characterized in that
In S1, the wave band that the Sentinel-1 data in the remotely-sensed data of collection include is VV polarization wave bands and VH polarization
Wave band；The wave band that Sentinel-2 data include is blue wave band, green light band, red spectral band and near infrared band, Yi Jigen
The NDVI wave bands calculated according to following formula：NDVI=(NIR-Red)/(NIR+Red), wherein NIR represent near infrared band, Red
Represent red spectral band.
9. the early stage crop identifying system according to claim 8 based on high-definition remote sensing data, which is characterized in that
In S2, the T time length is 15 days；
In S3, the characteristic of division refers to 7 in synthesis Sentinel-1 and the Sentinel-2 time series of T time length
The data of a wave band；
In S4, when identifying crop, using random forest method and JM Furthest Neighbors to each characteristic of division carry out prominence score and
Crop separability is assessed, and selects the best features of crop identification accordingly.
10. the early stage crop identifying system according to claim 9 based on high-definition remote sensing data, which is characterized in that
By calculating Gini scorings and the JM distances of random forest, two Gini is selected to score higher and JM apart from larger spy
Sign is used as preferred feature；Step S5 includes：
S51 inputs the preferred feature of the training sample obtained in S4 using the classifier training process of immune system network method
Time series, obtaining classification needs antibody to be used；
S52 positions the position of valid data in antigen using each picture dot in image of classifying as antigen；
S53 filters out ordered sequence corresponding with antigen as new antibodies to all antibody according to valid data position；
S54 calculates the characteristic sequence of antigen to the distance of all new antibodies, and by the agrotype of the minimum antibody of distance, makees
For the classification results of each antigen.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810246796.7A CN108399400A (en) | 2018-03-23 | 2018-03-23 | A kind of early stage crop recognition methods and system based on high-definition remote sensing data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810246796.7A CN108399400A (en) | 2018-03-23 | 2018-03-23 | A kind of early stage crop recognition methods and system based on high-definition remote sensing data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108399400A true CN108399400A (en) | 2018-08-14 |
Family
ID=63091651
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810246796.7A Pending CN108399400A (en) | 2018-03-23 | 2018-03-23 | A kind of early stage crop recognition methods and system based on high-definition remote sensing data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108399400A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110909652A (en) * | 2019-11-16 | 2020-03-24 | 中国水利水电科学研究院 | Method for dynamically extracting monthly scale of crop planting structure with optimized textural features |
| CN112084991A (en) * | 2020-09-18 | 2020-12-15 | 中国农业科学院农业资源与农业区划研究所 | Crop early identification method based on multi-source remote sensing time sequence image and convolutional neural network |
| CN112101256A (en) * | 2020-09-21 | 2020-12-18 | 河南大学 | Garlic crop identification method based on coupling active and passive remote sensing images of cloud platform |
| CN112446397A (en) * | 2019-09-02 | 2021-03-05 | 中国林业科学研究院资源信息研究所 | Grass yield estimation method and device based on remote sensing and random forest and storage medium |
| CN113673628A (en) * | 2021-09-07 | 2021-11-19 | 中国气象科学研究院 | Corn planting distribution extraction method based on high-resolution satellite data |
| CN114494882A (en) * | 2022-02-07 | 2022-05-13 | 中国科学院空天信息创新研究院 | Winter wheat remote sensing identification analysis method and system based on random forest |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102592181A (en) * | 2011-12-30 | 2012-07-18 | 中国农业科学院农业资源与农业区划研究所 | Method for optimizing spatial distribution of statistical data about crop planting area |
| CN105005782A (en) * | 2014-04-24 | 2015-10-28 | 中国科学院遥感与数字地球研究所 | Fine method for global vegetation classification based on multi-temporal remote sensing data and spectroscopic data |
| WO2017129940A1 (en) * | 2016-01-29 | 2017-08-03 | Global Surface Intelligence Limited | System and method for earth observation and analysis |
| CN107688777A (en) * | 2017-07-21 | 2018-02-13 | 同济大学 | A kind of urban green space extracting method for cooperateing with multi-source Remote Sensing Images |
-
2018
- 2018-03-23 CN CN201810246796.7A patent/CN108399400A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102592181A (en) * | 2011-12-30 | 2012-07-18 | 中国农业科学院农业资源与农业区划研究所 | Method for optimizing spatial distribution of statistical data about crop planting area |
| CN105005782A (en) * | 2014-04-24 | 2015-10-28 | 中国科学院遥感与数字地球研究所 | Fine method for global vegetation classification based on multi-temporal remote sensing data and spectroscopic data |
| WO2017129940A1 (en) * | 2016-01-29 | 2017-08-03 | Global Surface Intelligence Limited | System and method for earth observation and analysis |
| CN107688777A (en) * | 2017-07-21 | 2018-02-13 | 同济大学 | A kind of urban green space extracting method for cooperateing with multi-source Remote Sensing Images |
Non-Patent Citations (3)
| Title |
|---|
| JAN HAAS 等: "Sentinel-1A SAR and sentinel-2A MSI data fusion for urban ecosystem", 《REMOTE SENSING APPLICATIONS: SOCIETY AND ENVIRONMENT》 * |
| 李舰 等: "《数据科学中的R语言》", 31 July 2015 * |
| 郝鹏宇: "基于多时相遥感数据的农作物分类研究", 《中国博士学位论文全文数据库农业科技辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112446397A (en) * | 2019-09-02 | 2021-03-05 | 中国林业科学研究院资源信息研究所 | Grass yield estimation method and device based on remote sensing and random forest and storage medium |
| CN110909652A (en) * | 2019-11-16 | 2020-03-24 | 中国水利水电科学研究院 | Method for dynamically extracting monthly scale of crop planting structure with optimized textural features |
| CN110909652B (en) * | 2019-11-16 | 2022-10-21 | 中国水利水电科学研究院 | Method for dynamically extracting monthly scale of crop planting structure with optimized textural features |
| CN112084991A (en) * | 2020-09-18 | 2020-12-15 | 中国农业科学院农业资源与农业区划研究所 | Crop early identification method based on multi-source remote sensing time sequence image and convolutional neural network |
| CN112101256A (en) * | 2020-09-21 | 2020-12-18 | 河南大学 | Garlic crop identification method based on coupling active and passive remote sensing images of cloud platform |
| CN112101256B (en) * | 2020-09-21 | 2021-07-09 | 河南大学 | Garlic crop identification method based on coupling active and passive remote sensing images of cloud platform |
| CN113673628A (en) * | 2021-09-07 | 2021-11-19 | 中国气象科学研究院 | Corn planting distribution extraction method based on high-resolution satellite data |
| CN114494882A (en) * | 2022-02-07 | 2022-05-13 | 中国科学院空天信息创新研究院 | Winter wheat remote sensing identification analysis method and system based on random forest |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108399400A (en) | A kind of early stage crop recognition methods and system based on high-definition remote sensing data | |
| Waldhoff et al. | Multi-Data Approach for remote sensing-based regional crop rotation mapping: A case study for the Rur catchment, Germany | |
| Iqbal et al. | Gray level co-occurrence matrix (GLCM) texture based crop classification using low altitude remote sensing platforms | |
| Karlson et al. | Assessing the potential of multi-seasonal WorldView-2 imagery for mapping West African agroforestry tree species | |
| Zheng et al. | Early season detection of rice plants using RGB, NIR-GB and multispectral images from unmanned aerial vehicle (UAV) | |
| George et al. | Forest tree species discrimination in western Himalaya using EO-1 Hyperion | |
| Ok et al. | Evaluation of random forest method for agricultural crop classification | |
| Arvor et al. | Classification of MODIS EVI time series for crop mapping in the state of Mato Grosso, Brazil | |
| Reis et al. | Identification of hazelnut fields using spectral and Gabor textural features | |
| González-Yebra et al. | Methodological proposal to assess plastic greenhouses land cover change from the combination of archival aerial orthoimages and Landsat data | |
| Gerts et al. | Multi-temporal monitoring of cotton growth through the vegetation profile classification for Tashkent province, Uzbekistan | |
| CN109635731A (en) | It is a kind of to identify method and device, storage medium and the processor effectively ploughed | |
| Qian et al. | Mapping regional cropping patterns by using GF-1 WFV sensor data | |
| Zhu et al. | Identification of soybean based on Sentinel-1/2 SAR and MSI imagery under a complex planting structure | |
| Kang et al. | Construction of multidimensional features to identify tea plantations using multisource remote sensing data: A case study of Hangzhou city, China | |
| Kupková et al. | Towards reliable monitoring of grass species in nature conservation: Evaluation of the potential of UAV and PlanetScope multi-temporal data in the Central European tundra | |
| Lass et al. | Detecting the locations of Brazilian pepper trees in the Everglades with a hyperspectral sensor | |
| Venkatesan et al. | Multi-temporal feature extraction for precise maize area mapping using time-series Sentinel 1A SAR data | |
| Okubo et al. | Land use/cover classification of a complex agricultural landscape using single-dated very high spatial resolution satellite-sensed imagery | |
| Pinto-Ledezma et al. | Mapping oak wilt disease from space using land surface phenology | |
| Dibs et al. | Estimation and Mapping the Rubber Trees Growth Distribution using Multi Sensor Imagery With Remote Sensing and GIS Analysis | |
| Ormeci et al. | Identification of crop areas Using SPOT–5 data | |
| Udgata et al. | Remote Sensing and Machine Learning techniques for acreage estimation of mango (Mangifera indica) | |
| Yang et al. | Monitoring winter wheat in ShanDong province using Sentinel data and Google Earth Engine platform | |
| Dehkordi et al. | Performance Evaluation of Temporal and Spatial-Temporal Convolutional Neural Networks for Land-Cover Classification (A Case Study in Shahrekord, Iran) |
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 |
Application publication date: 20180814 |
|
| RJ01 | Rejection of invention patent application after publication |