CN109919207A - Consider the crop growing mode classification method and system of phenology feature - Google Patents
Consider the crop growing mode classification method and system of phenology feature Download PDFInfo
- Publication number
- CN109919207A CN109919207A CN201910137051.1A CN201910137051A CN109919207A CN 109919207 A CN109919207 A CN 109919207A CN 201910137051 A CN201910137051 A CN 201910137051A CN 109919207 A CN109919207 A CN 109919207A
- Authority
- CN
- China
- Prior art keywords
- serial position
- evi
- time
- pixel
- sorted
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Pretreatment Of Seeds And Plants (AREA)
- Image Processing (AREA)
Abstract
The present invention provides a kind of crop growing mode classification method and system for considering phenology feature, is related to remote sensing of vegetation field.By making full use of effect of the difference of crucial phenological period EVI time-serial position in identification cropping pattern type, enhanced by the weight to the crucial phenological period, the information in crucial phenological period can be efficiently used, inhibit the information in non-key phenological period, improves the precision of crop growing mode classification.
Description
Technical field
The present invention relates to remote sensing of vegetation technical fields, and in particular to a kind of crop growing mode for considering phenology feature point
Class method and system.
Background technique
The agricultural natural resources such as crop growing mode and the light in region, heat, water, soil fertility are closely bound up.With complete
The variation of ball weather, the Distribution Pattern of the agricultural natural resources such as light, heat, water also gradually change, and necessarily affect cropping pattern
Spatial and temporal distributions.Cropping pattern is largely limited by local weather conditions, and as forefathers contradiction causes to cultivated land utilization
Intensity is continuously increased, and on the one hand to be improved grain yield, on the one hand also to be guaranteed the sustainable development of agricultural.Therefore, timely, quasi-
Really monitoring agricultural planting patterns and its change in time and space, to the plantation situation, prediction grain yield, formulation agricultural hair for grasping crops
Exhibition policy and ensure that grain security is of great significance.
Usually agricultural planting patterns are monitored and are classified using remote sensing technology at present, the planting die based on remote sensing technology
Formula research uses various filtering and noise reduction reconstructions mostly based on VI (Vegetation Index) time series data first
Then smooth crop growth curve carries out the extraction of cropping pattern.Cropping pattern monitoring is carried out using time series data
The advantages of be the rule that can quickly judge crop planting, that is, remote sensing time series data just can satisfy relevant art
In only need plantation strength information of ploughing, the problem of being indifferent to specific crop type, but the disadvantage is that be not easy to judge
Crop type.The method of discrimination of remote sensing cropping pattern mainly includes classification, Peak Intensity Method and phenology parametric method.
However, classification and Peak Intensity Method require researcher very familiar to the crops shortening in research area, excessively rely on
The experience of researcher causes its versatility weak.Phenology parametric method is also needed when dividing crop and non-crop by threshold value,
Therefore threshold value determination method influences final result very big.How to obtain optimal threshold improve accuracy of identification be influence object
Parametric method is waited in the key of widespread adoption, but how most to have threshold value without specific perfect method.It follows that existing agriculture
The generalization of crop planting model Classification in Remote Sensing Image method is poor, not applied to crop growing mode nicety of grading when monitoring on a large scale
It is high.
Summary of the invention
(1) the technical issues of solving
In view of the deficiencies of the prior art, the present invention provides a kind of crop growing mode classification sides for considering phenology feature
Method and system solve crop growing mode Classification in Remote Sensing Image method in the prior art and are being applied to a wide range of prison time difference method
Low technical problem.
(2) technical solution
In order to achieve the above object, the present invention is achieved by the following technical programs:
The present invention provides a kind of crop growing mode classification method for considering phenology feature, and this method includes following step
It is rapid:
A, the standard of every class cropping pattern in the EVI time-serial position and sample of each pixel to be sorted on image is obtained
EVI time-serial position;
B, intersecting for the EVI time-serial position and the standard EVI time-serial position is calculated based on CCSM algorithm
Related coefficient obtains maximum cross-correlation and the corresponding matching position of the maximum cross-correlation;
C, significance test is carried out to the maximum cross-correlation, obtains significance test result;
D, the significance test result is analyzed, if in the EVI time-serial position and the sample extremely
The maximum correlation coefficient of the standard EVI time-serial position of certain few a kind of cropping pattern is greater than 0.5 and passes through significance test, that
Step F is gone to, step E is otherwise gone to;
E, it is calculated in the original position of the EVI time-serial position every in the EVI time-serial position and the sample
The Euclidean distance of the crucial phenological period weight enhancing of the standard EVI time-serial position of class cropping pattern, is based on Euclidean distance pair
The pixel to be sorted is classified, and classification results are obtained;
F, judge EVI time-serial position whether the standard EVI time with two classes in sample or the above cropping pattern of two classes
The maximum correlation coefficient of sequence curve is greater than 0.5 and by significance test, according to judging result to EVI time-serial position into
Row classification.
Preferably, maximum cross-correlation is obtained in the step B method particularly includes:
Keep the standard EVI time-serial position of every class cropping pattern motionless, it is bent by the movement EVI time series
Line, and the two is calculated in the coefficient R of different location according to formulam, remembered by the maximum correlation coefficient that cross-matched obtains
For Rmax;Specific formula is as follows:
Wherein:
RmFor the related coefficient of EVI time-serial position and standard EVI time-serial position in matching position m;
N indicates the data point number of two curves overlapping;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample;
VIi,tFor the vegetation index value of i-th of data point of EVI time-serial position of pixel to be sorted.
Preferably, Euclidean distance is calculated in the step E method particularly includes:
The EVI time-serial position of pixel to be sorted is calculated in the original position of the EVI time-serial position of pixel to be sorted
With the Euclidean distance D of the standard EVI time-serial position of every class cropping patternw0, calculation formula is as follows:
Wherein:
wiFor the original weighted value of i-th of data point;
w′iFor the weighted value of i-th of data point after standardization;
K is the position in crucial phenological period;
N is the number of time series data point;
nkFor the number in crucial phenological period;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample;
VIi0,tFor the vegetation index of i-th of data point of EVI time-serial position of pixel to be sorted under non-situation of movement
Value;
Dw0For in the standard EVI of the EVI time-serial position of original position pixel to be sorted and certain class cropping pattern
Between sequence curve weight enhance Euclidean distance.
Preferably, include according to the detailed process that judging result classifies to EVI time-serial position in the step F
Step F1 and step F2;
The standard EVI time-serial position related coefficient of F1, EVI time-serial position and cropping pattern a kind of in sample
Greater than 0.5 and by significance test, gone inside this kind then directly assigning to pixel to be sorted;
The standard EVI time-serial position of two classes or the above cropping pattern of two classes in F2, EVI time-serial position and sample
Related coefficient is greater than 0.5 and by significance test, then calculates in the matching position for obtaining maximum cross-correlation to be sorted
The Europe of the weight enhancing of the standard EVI time-serial position of every class cropping pattern in the EVI time-serial position and sample of pixel
Family name's distance, and the comprehensive distance for combining maximum correlation coefficient and weight enhancing Euclidean distance is constructed, then carry out most narrow spacing
From classification, the standard EVI time-serial position of any class cropping pattern in the EVI time-serial position and sample of pixel to be sorted
Comprehensive distance is minimum, then pixel to be sorted is assigned to which kind of the inside is gone.
Preferably, the calculating comprehensive distance of the step F2 method particularly includes:
The EVI time-serial position and sample of pixel to be sorted are calculated at the matching position for obtaining maximum cross-correlation
The Euclidean distance D of the weight enhancing of the standard EVI time-serial position of every class cropping pattern in thiswm, construct one and combine maximum
Coefficient RmaxEnhance Euclidean distance D with weightwmComprehensive distance D, then carry out minimum distance classification, pixel to be sorted
The comprehensive distance of the standard EVI time-serial position of which class cropping pattern is minimum in EVI time-serial position and sample, then will be to
Which kind of the inside classification pixel, which assigns to, is gone;The calculation formula of comprehensive distance is as follows:
D=Dwm*(1.0-Rmax)
Wherein:
wiFor the original weighted value of i-th of data point;
w′iFor the weighted value of i-th of data point after standardization;
K is the position in crucial phenological period;
N is the number of time series data point;
nkFor the number in crucial phenological period;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample;
VIim,tThe vegetation index value of i-th of data point behind the position m is moved to for pixel EVI time-serial position to be sorted;
DwmEVI time-serial position and certain class planting die behind the position m are moved to for pixel EVI time-serial position to be sorted
The weight of the standard EVI time-serial position of formula enhances Euclidean distance;
D is comprehensive distance;
RmaxFor the maximum correlation coefficient obtained when CCSM algorithmic match.
The present invention also provides a kind of crop growing mode categorizing systems for considering phenology feature, comprising:
Acquiring unit, the acquiring unit are used to obtain every class in the EVI time-serial position and sample of pixel to be sorted
The standard EVI time-serial position of cropping pattern;
First computing unit, first computing unit are used to calculate each pixel to be sorted on image based on CCSM algorithm
EVI time-serial position and every class cropping pattern standard EVI time-serial position cross-correlation, obtain maximum friendship
Pitch related coefficient and the corresponding matching position of the maximum cross-correlation;
Significance test unit, the significance test unit are used for the maximal correlation that the first computing unit is calculated
Coefficient carries out significance test, obtains significance test result;
First judging unit, first judging unit is for analyzing significance test result, if to be sorted
The EVI time-serial position of pixel is at least big with the standard EVI time-serial position maximum correlation coefficient of certain a kind of cropping pattern
In 0.5 and then going to second judgment unit otherwise going to the second computing unit by significance test;
Second computing unit, second computing unit are used to calculate the EVI time series of pixel to be sorted in the original location
The Euclidean of weight enhancing at the crucial phenological period key position of curve and every class cropping pattern standard EVI time-serial position away from
From calculating Euclidean distance is classified with minimum distance classification again, obtains the classification results of pixel to be sorted;
Second judgment unit, the second judgment unit be used for judge pixel to be sorted EVI time-serial position whether
It is greater than 0.5 with the maximum correlation coefficient of two classes or the standard EVI time-serial position of the above cropping pattern of two classes and by significant
Property examine, if it is not, then explanation from the EVI time-serial position of pixel to be sorted in shape only with certain a kind of cropping pattern
Standard EVI time-serial position is similar, goes inside this kind of cropping patterns then directly assigning to pixel to be sorted, if so,
Then go to third computing unit;
Third computing unit, the third computing unit is based at the matching position for obtaining maximum cross-correlation
The Europe that the weight of the EVI time-serial position and every class cropping pattern standard EVI time-serial position of calculating pixel to be sorted enhances
Family name's distance;And by weight enhance Euclidean distance combination maximum correlation coefficient construct comprehensive distance, then utilize comprehensive distance into
Row minimum distance classification, the standard EVI time-serial position of the EVI time-serial position of pixel to be sorted and any class cropping pattern
Comprehensive distance it is minimum, then pixel to be sorted is assigned to go inside which kind of cropping pattern;
Classification results output unit, the classification results output unit are used to export the EVI time series of pixel to be sorted
The classification results of curve.
(3) beneficial effect
The present invention provides a kind of crop growing mode classification methods and system for considering phenology feature.With the prior art
Compare, have it is following the utility model has the advantages that
The present invention is based on CCSM algorithm calculate described image on each pixel to be sorted EVI time-serial position with it is described
The cross-correlation of the standard EVI time-serial position of every class cropping pattern in sample, CCSM algorithm can find to be sorted
The standard EVI time-serial position of the EVI time-serial position of pixel and class cropping pattern every in sample obtains maximum and intersects phase
Matching position when relationship number can eliminate the advances of plant phenophase of crops or the influence to classification of postponement, while phase relation
The size of number logarithm is insensitive, can eliminate influence of the growing way difference to classification.
Enhanced simultaneously by the weight to the crucial phenological period, the information in crucial phenological period can be efficiently used, inhibited
The information in non-key phenological period improves the precision of crop growing mode classification.
Above-mentioned maximum cross-correlation can describe the similitude of spectrum, and the Euclidean distance of above-mentioned weight enhancing can
To describe the amplitude difference of spectrum, the present invention has comprehensively considered spectral similarity and amplitude difference, and crop planting can be improved
The nicety of grading of mode.
Since the main cropping pattern in region and crucial phenological period can be from passing through the vegetation index time sequence to region
Column tracing analysis obtains, therefore the present invention is readily applied to different research areas, and the classification side of crop growing mode can be improved
The universality of method.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is a kind of flow chart for the crop growing mode classification method for considering phenology feature of the embodiment of the present invention;
A kind of block diagram for the crop growing mode categorizing system for considering phenology feature of Fig. 2 another embodiment of the present invention;
Fig. 3 is the EVI time-serial position and its schematic diagram in 6 crucial phenological periods of different type cropping pattern;
Fig. 4 is CCSM algorithm schematic diagram;
Wherein, in Fig. 3:
The period of seedling establishment of S1 expression winterness plant;
The heading stage of S2 expression winterness plant;
S3 indicates the harvest time of winterness plant and the sowing time of summer sown crop;
The heading stage of S4 expression summer sown crop;
S5 indicates the harvest time of summer sown crop and the sowing time of winterness plant;
The previous peaks phase in winter of S6 expression winterness plant.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, to the technology in the embodiment of the present invention
Scheme is clearly and completely described, it is clear that and described embodiments are some of the embodiments of the present invention, rather than whole
Embodiment.Based on the embodiments of the present invention, those of ordinary skill in the art are obtained without creative efforts
The every other embodiment obtained, shall fall within the protection scope of the present invention.
The embodiment of the present application is by providing a kind of crop growing mode classification method and system for considering phenology feature, solution
The method of discrimination of remote sensing crop growing mode in the prior art of having determined is being applied to the low technology of a wide range of prison time difference method
Problem effectively improves the precision of crop growing mode classification.
Technical solution in the embodiment of the present application is in order to solve the above technical problems, general thought is as follows:
Based on described in background, the method for discrimination of existing remote sensing cropping pattern mainly includes classification, Peak Intensity Method and phenology
Parametric method.But classification, Peak Intensity Method and phenology parametric method do not fully take into account the variation and growing way in crops phenological period
Difference does not make full use of the information in crucial phenological period yet, causes the generalization of crop growing mode Classification in Remote Sensing Image method poor,
Crop growing mode nicety of grading not high technical problem when applied to a wide range of monitoring.
To solve the above problems, present applicant proposes it is a kind of consider phenology feature crop growing mode classification method and
System, for being classified based on remote sensing technology to crop growing mode.By making full use of crucial phenological period EVI time series bent
Effect of the difference of line in identification cropping pattern type, is enhanced by the weight to the crucial phenological period, can be effectively sharp
With the information in crucial phenological period, inhibit the information in non-key phenological period, improves the precision of crop growing mode classification.
In order to better understand the above technical scheme, in conjunction with appended figures and specific embodiments to upper
Technical solution is stated to be described in detail.
The embodiment of the invention provides a kind of crop growing mode classification methods for considering phenology feature, as shown in Figure 1,
Detailed process is as follows:
A, the standard of every class cropping pattern in the EVI time-serial position and sample of each pixel to be sorted on image is obtained
EVI time-serial position, EVI refer to enhancement mode meta file (Enhanced Vegetation Index);
B, the mark of every class cropping pattern in the EVI time-serial position and sample of pixel to be sorted is calculated based on CCSM algorithm
It is corresponding with maximum cross-correlation to obtain maximum cross-correlation for the cross-correlation of quasi- EVI time-serial position
Matching position, CCSM here refer to that cross-correlation spectroscopy matches (Cross Correlogram Spectral Matching);
C, significance test is carried out to above-mentioned maximum cross-correlation, obtains significance test result;
D, significance test result is analyzed, if in the EVI time-serial position of pixel to be sorted and sample extremely
The maximum correlation coefficient of the standard EVI time-serial position of certain few a kind of cropping pattern is greater than 0.5 and passes through significance test, that
Step F is gone to, step E is otherwise gone to;
E, the EVI time series song of pixel to be sorted is calculated in the original position of the EVI time-serial position of pixel to be sorted
The Euclidean distance of the crucial phenological period weight enhancing of the standard EVI time-serial position of every class cropping pattern, base in line and sample
In Euclidean distance, is classified using minimum distance classification to the pixel to be sorted, obtain classification results;
F, judge pixel to be sorted EVI time-serial position whether with two classes in sample or the above cropping pattern of two classes
The maximum correlation coefficient of standard EVI time-serial position is greater than 0.5 and by significance test, according to judging result to be sorted
The EVI time-serial position of pixel is classified.
The embodiment of the present invention calculates the EVI time-serial position of each pixel to be sorted in described image based on CCSM algorithm
With the cross-correlation of the standard EVI time-serial position of class cropping pattern every in the sample, CCSM algorithm can be found
The standard EVI time-serial position of every class cropping pattern obtains maximum in the EVI time-serial position and sample of pixel to be sorted
The matching position of cross-correlation can eliminate the advances of plant phenophase of crops or the influence to classification of postponement, same to phase
The size of relationship number logarithm is insensitive, can eliminate influence of the growing way difference to classification.
Enhanced simultaneously by the weight to the crucial phenological period, the information in crucial phenological period can be efficiently used, inhibited
The information in non-key phenological period improves the precision of crop growing mode classification.
Above-mentioned maximum cross-correlation can describe the similitude of spectrum, and the Euclidean distance of above-mentioned weight enhancing
The amplitude difference of spectrum can be described, the embodiment of the present invention has comprehensively considered spectral similarity and amplitude difference, and agriculture can be improved
The nicety of grading of crop planting model.
Since the main cropping pattern in region and crucial phenological period can be from passing through the vegetation index time sequence to region
Column tracing analysis obtains, therefore the present invention is readily applied to different research areas, and the classification side of crop growing mode can be improved
The universality of method.
A step is described in detail below with reference to the crop growing mode in certain research area:
The crop growing mode in certain research area is as shown in figure 3, include winter sowing, summer sowing, two crops a year and industrial crops kind
It plants.It is the maximum cropping pattern of region area due to yielding two crops a year, one shares 6 keys in two crops a year cropping pattern
Phenological period, be respectively the period of seedling establishment (S1) of winterness plant, the heading stage (S2) of winterness plant, winterness plant harvest time and summer
Broadcast the sowing time (S3) of crop, heading stage (S4), the harvest time of summer sown crop and the sowing time of winterness plant of summer sown crop
(S5), the previous peaks phase in winter (S6) of winterness plant.As can be seen that EVI value of the different cropping patterns this 6 crucial phenological periods
Variation is different, can increase the weight in this 6 key phenological periods when calculating Euclidean distance, thus more efficiently estimate to
The EVI time-serial position of classification pixel and the standard EVI time-serial position similitude of a certain cropping pattern.As shown in Figure 1,
Specific assorting process is as follows:
A, the EVI time-serial position and sample set of each pixel to be sorted on image are obtained, and from the sample set
Selecting sample calculates the standard EVI time-serial position of every class cropping pattern in the sample, to be sorted in order to avoid interference
The EVI time-serial position of pixel is by filtering processing.
B, the EVI time-serial position and the sample of each pixel to be sorted in described image are calculated based on CCSM algorithm
In every class cropping pattern standard EVI time-serial position cross-correlation, record acquired maximum crosscorrelation system
Matching position when number and acquirement maximum cross-correlation.As shown in figure 4, giving CCSM algorithm calculates cross-correlation
Schematic diagram: Fig. 4 (a) shows two EVI time-serial positions, and one is sample Plays EVI time-serial position, another
Item indicates the EVI time-serial position of pixel to be sorted.Keep standard curve motionless, when by the EVI of movement pixel to be sorted
Between sequence curve and according to formula calculate both different location related coefficient.In Fig. 4 (b), position 0 indicates not move,
Position, which is negative, indicates that curve to be matched is moved to the left, and position, which is positive, indicates that curve to be matched moves right, it is shown that in different positions
Set the related coefficient of curve and standard curve to be matched, it can be seen that when matching position is -1, the related coefficient of the two is maximum,
When curve namely to be matched is moved to the left a position, the related coefficient of the two is maximum.Specific formula is as follows:
Wherein:
RmFor the phase of the EVI time-serial position and standard EVI time-serial position of pixel to be sorted in matching position m
Relationship number;
N indicates the data point number of two curves overlapping;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample;
VIi,tFor the vegetation index value of i-th of data point of EVI time-serial position of pixel to be sorted.
C, significance test is carried out to the maximum cross-correlation that the step B is calculated, obtains conspicuousness
Inspection result.
D, significance test result is analyzed, if the EVI time-serial position of the pixel to be sorted at least with
The maximum correlation coefficient of the EVI time-serial position of a kind of cropping pattern of certain in the sample is greater than 0.5 and is examined by conspicuousness
It tests, then going to step F, otherwise goes to step E;
E, due in the EVI time-serial position of pixel to be sorted and sample when the standard EVI of any kind cropping pattern
Between sequence curve significant correlativity is all not present, illustrate any in the shape and sample of the EVI timing curve of pixel to be sorted
The shape of the EVI timing curve of a kind of cropping pattern is all dissimilar, from being difficult to judge its classification in shape, it is therefore desirable to according to width
Degree difference is classified, the Euclidean distance D that amplitude difference can be reinforced by calculating weightw0It obtains, in the EVI of pixel to be sorted
When the original position of time-serial position calculates the standard EVI of EVI time-serial position and every class cropping pattern of pixel to be sorted
Between sequence curve Euclidean distance Dw0, it is based on Euclidean distance Dw0, classified using minimum distance classification, it is more to be sorted
The EVI timing curve of which kind of cropping pattern is nearest in the shape and sample of the EVI timing curve of pixel, obtains classification results.
Calculation formula is as follows:
Wherein:
wiFor the original weighted value of i-th of data point;
w′iFor the weighted value of i-th of data point after standardization;
K is the position in crucial phenological period;
N is the number of time series data point;
nkFor the number in crucial phenological period;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample;
VIi0,tFor the vegetation index of i-th of data point of EVI time-serial position of pixel to be sorted under non-situation of movement
Value;
Dw0For in the standard EVI of the EVI time-serial position of original position pixel to be sorted and certain class cropping pattern
Between sequence curve weight enhance Euclidean distance.
F, judge pixel to be sorted EVI time-serial position whether the standard with two classes or the above cropping pattern of two classes
The maximum correlation coefficient of EVI time-serial position is greater than 0.5 and by significance test, if it is not, then step F1 is gone to, if
It is to go to step F2;
F1, due to pixel to be sorted EVI time-serial position a kind of cropping pattern only and in sample the standard EVI time
Sequence curve related coefficient is greater than 0.5 and by significance test, then illustrates the EVI time series from pixel to be sorted in shape
Curve is only similar to the standard EVI time-serial position of a kind of cropping pattern of certain in sample, then directly by picture to be sorted
Member is assigned to goes inside this kind;
F2, the EVI time-serial position of pixel to be sorted are related to the standard EVI time-serial position of multiclass cropping pattern
Coefficients RmaxGreater than 0.5 and by significance test, illustrate the shape and multiclass planting die of the EVI timing curve of pixel to be sorted
The shape of the EVI timing curve of formula all has similitude, cannot directly judge its type merely with related coefficient.Therefore further
Calculating the weight that pixel to be sorted meets the classification of condition with these enhances Euclidean distance Dwm.Current DwmBe with every class kind
It is calculated in the maximum matching position of implant model.For example, certain pixel obtains maximum with winter sowing cropping pattern in m=1 intersects phase
Relationship number then calculates the Euclidean distance reinforced with the weight of winter sowing cropping pattern in the position of m=1;If the same pixel with
Summer sowing cropping pattern obtains maximum cross-correlation in m=2, then calculates itself and summer sowing cropping pattern in the position of m=2
The Euclidean distance D of weight enhancingwm.Obtain Euclidean distance DwmAfterwards, it constructs one and combines maximum correlation coefficient and weight enhancing Euclidean
Then the comprehensive distance D of distance carries out minimum distance classification, which class in the EVI time-serial position and sample of pixel to be sorted
The comprehensive distance of the standard EVI time-serial position of cropping pattern is minimum, then pixel to be sorted is assigned to which kind of the inside is gone.It is comprehensive
The calculation formula for closing distance is as follows:
D=Dwm*(1.0-Rmax)
Wherein:
wiFor the original weighted value of i-th of data point;
w′iFor the weighted value of i-th of data point after standardization;
K is the position in crucial phenological period;
N is the number of time series data point;
nkFor the number in crucial phenological period;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample,
VIim,tFor the vegetation index value for being moved to i-th of data point of pixel EVI time-serial position to be sorted behind the position m;
DwmThe weight increasing of two EVI time-serial positions behind the position m is moved to for pixel EVI time-serial position to be sorted
Strong Euclidean distance;
D is comprehensive distance;
RmaxFor the maximum correlation coefficient obtained when CCSM algorithmic match.
Another embodiment of the present invention additionally provides a kind of crop growing mode categorizing system for considering phenology feature, such as
Shown in Fig. 2, comprising: acquiring unit, the first computing unit, significance test unit, the first judging unit, the second computing unit,
Second judgment unit, third computing unit, classification results output unit.
Acquiring unit is used to obtain the standard of every class cropping pattern in the EVI time-serial position and sample of pixel to be sorted
EVI time-serial position;
First computing unit is used to calculate the EVI time-serial position of each pixel to be sorted on image based on CCSM algorithm
With the cross-correlation of the standard EVI time-serial position of every class cropping pattern, maximum cross-correlation and described is obtained
The corresponding matching position of maximum cross-correlation;
The maximum correlation coefficient that significance test unit is used to that the first computing unit to be calculated carries out significance test,
Obtain significance test result;
First judging unit is for analyzing significance test result, if the EVI time series of pixel to be sorted
Curve is at least greater than 0.5 with the standard EVI time-serial position maximum correlation coefficient of certain a kind of cropping pattern and passes through conspicuousness
It examines, then going to second judgment unit, otherwise goes to the second computing unit;
Second computing unit is used to calculate the EVI time-serial position and every class cropping pattern of pixel to be sorted in the original location
The Euclidean distance of weight enhancing at the crucial phenological period key position of standard EVI time-serial position, calculates Euclidean distance again
Classified with minimum distance classification, obtains the classification results of pixel to be sorted;
Second judgment unit be used for judge pixel to be sorted EVI time-serial position whether with more than two classes or two classes plant
The maximum correlation coefficient of the standard EVI time-serial position of implant model is greater than 0.5 and by significance test, if it is not, then saying
The bright EVI time-serial position from pixel to be sorted in shape is only bent with the standard EVI time series of certain a kind of cropping pattern
Line is similar, goes inside this kind of cropping patterns then directly assigning to pixel to be sorted, if it is, going to third calculates list
Member;
Third computing unit is used at the matching position for obtaining maximum cross-correlation calculate the EVI of pixel to be sorted
The Euclidean distance that the weight of time-serial position and every class cropping pattern standard EVI time-serial position enhances;And pass through weight
Enhance Euclidean distance combination maximum correlation coefficient and construct comprehensive distance, then carries out minimum distance classification using comprehensive distance, to
The comprehensive distance of the standard EVI time-serial position of the EVI time-serial position and which class cropping pattern of classification pixel is minimum, then
Pixel to be sorted is assigned to and is gone inside which kind of cropping pattern;
Classification results output unit is used to export the classification results of the EVI time-serial position of pixel to be sorted.
In conclusion compared with prior art, have it is following the utility model has the advantages that
The embodiment of the present invention calculates the EVI time-serial position of each pixel to be sorted in described image based on CCSM algorithm
With the cross-correlation of the standard EVI time-serial position of class cropping pattern every in the sample, CCSM algorithm can be found
The standard EVI time-serial position of every class cropping pattern obtains maximum in the EVI time-serial position and sample of pixel to be sorted
Matching position when cross-correlation can eliminate the advances of plant phenophase of crops or the influence to classification of postponement, simultaneously
The size of related coefficient logarithm is insensitive, can eliminate influence of the growing way difference to classification.
Enhanced simultaneously by the weight to the crucial phenological period, the information in crucial phenological period can be efficiently used, inhibited
The information in non-key phenological period improves the precision of crop growing mode classification.
Above-mentioned maximum cross-correlation can describe the similitude of spectrum, and the Euclidean distance of above-mentioned weight enhancing
The amplitude difference of spectrum can be described, the embodiment of the present invention has comprehensively considered spectral similarity and amplitude difference, and agriculture can be improved
The nicety of grading of crop planting model.
Since the main cropping pattern in region and crucial phenological period can be from passing through the vegetation index time sequence to region
Column tracing analysis obtains, therefore the present invention is readily applied to different research areas, and the classification side of crop growing mode can be improved
The universality of method.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including above-mentioned element.
The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations;Although with reference to the foregoing embodiments
Invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each implementation
Technical solution documented by example is modified or equivalent replacement of some of the technical features;And these modification or
Replacement, the spirit and scope for technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution.
Claims (6)
1. a kind of crop growing mode classification method for considering phenology feature, which is characterized in that method includes the following steps:
A, the standard EVI of every class cropping pattern in the EVI time-serial position and sample of each pixel to be sorted on image is obtained
Time-serial position;
B, the crosscorrelation of the EVI time-serial position Yu the standard EVI time-serial position is calculated based on CCSM algorithm
Coefficient obtains maximum cross-correlation and the corresponding matching position of the maximum cross-correlation;
C, significance test is carried out to the maximum cross-correlation, obtains significance test result;
D, the significance test result is analyzed, if in the EVI time-serial position and the sample at least certain
The maximum correlation coefficient of the standard EVI time-serial position of a kind of cropping pattern is greater than 0.5 and by significance test, then turning
To step F, step E is otherwise gone to;
E, every class kind in the EVI time-serial position and the sample is calculated in the original position of the EVI time-serial position
The Euclidean distance of the crucial phenological period weight enhancing of the standard EVI time-serial position of implant model, based on Euclidean distance to described
Pixel to be sorted is classified, and classification results are obtained;
F, judge EVI time-serial position whether the standard EVI time series with two classes in sample or the above cropping pattern of two classes
The maximum correlation coefficient of curve is greater than 0.5 and by significance test, is divided according to judging result EVI time-serial position
Class.
2. the crop growing mode classification method according to claim 1 for considering phenology feature, which is characterized in that described
Maximum cross-correlation is obtained in step B method particularly includes:
Keep the standard EVI time-serial position of every class cropping pattern motionless, by the movement EVI time-serial position, and
The two is calculated in the coefficient R of different location according to formulam, R is denoted as by the maximum correlation coefficient that cross-matched obtainsmax;
Specific formula is as follows:
Wherein:
RmFor the related coefficient of EVI time-serial position and standard EVI time-serial position in matching position m;
N indicates the data point number of two curves overlapping;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample;
VIi,tFor the vegetation index value of i-th of data point of EVI time-serial position of pixel to be sorted.
3. the crop growing mode classification method according to claim 2 for considering phenology feature, which is characterized in that described
Euclidean distance is calculated in step E method particularly includes:
The EVI time-serial position of pixel to be sorted and every is calculated in the original position of the EVI time-serial position of pixel to be sorted
The Euclidean distance D of the standard EVI time-serial position of class cropping patternw0, calculation formula is as follows:
Wherein:
wiFor the original weighted value of i-th of data point;
w′iFor the weighted value of i-th of data point after standardization;
K is the position in crucial phenological period;
N is the number of time series data point;
nkFor the number in crucial phenological period;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample;
VIi0,tFor the vegetation index value of i-th of data point of EVI time-serial position of pixel to be sorted under non-situation of movement;
Dw0For in the standard EVI time sequence of the EVI time-serial position of original position pixel to be sorted and certain class cropping pattern
The weight of column curve enhances Euclidean distance.
4. the crop growing mode classification method according to claim 1 for considering phenology feature, which is characterized in that described
It according to judging result include step F1 and step F2 to the detailed process that EVI time-serial position is classified in step F;
Standard EVI time-serial position related coefficient of F1, EVI time-serial position only with cropping pattern a kind of in sample is greater than
0.5 and by significance test, it is gone inside this kind then directly assigning to pixel to be sorted;
F2, EVI time-serial position are related to the standard EVI time-serial position of two classes in sample or the above cropping pattern of two classes
Coefficient is greater than 0.5 and by significance test, then calculates pixel to be sorted in the matching position for obtaining maximum cross-correlation
EVI time-serial position and sample in every class cropping pattern standard EVI time-serial position weight enhancing Euclidean away from
From, and the comprehensive distance for combining maximum correlation coefficient and weight enhancing Euclidean distance is constructed, then carry out minimum range point
Class, the synthesis of the EVI time-serial position and the standard EVI time-serial position of any class cropping pattern in sample of pixel to be sorted
Distance is minimum, then pixel to be sorted is assigned to which kind of the inside is gone.
5. the crop growing mode classification method according to claim 4 for considering phenology feature, which is characterized in that described
The calculating comprehensive distance of step F2 method particularly includes:
In the EVI time-serial position and sample for calculating pixel to be sorted at the matching position for obtaining maximum cross-correlation
The Euclidean distance D of the weight enhancing of the standard EVI time-serial position of every class cropping patternwm, building one combines maximal correlation
Coefficients RmaxEnhance Euclidean distance D with weightwmComprehensive distance D, minimum distance classification is then carried out, when the EVI of pixel to be sorted
Between in sequence curve and sample the standard EVI time-serial position of any class cropping pattern comprehensive distance it is minimum, then will be to be sorted
Which kind of the inside pixel, which assigns to, is gone;The calculation formula of comprehensive distance is as follows:
D=Dwm*(1.0-Rmax)
Wherein:
wiFor the original weighted value of i-th of data point;
w′iFor the weighted value of i-th of data point after standardization;
K is the position in crucial phenological period;
N is the number of time series data point;
nkFor the number in crucial phenological period;
VIi,rFor the vegetation index value of standard EVI i-th of data point of time-serial position of certain class cropping pattern in sample,
VIim,tThe vegetation index value of i-th of data point behind the position m is moved to for pixel EVI time-serial position to be sorted;
DwmEVI time-serial position and certain class cropping pattern behind the position m are moved to for pixel EVI time-serial position to be sorted
The weight of standard EVI time-serial position enhances Euclidean distance;
D is comprehensive distance;
RmaxFor the maximum correlation coefficient obtained when CCSM algorithmic match.
6. a kind of crop growing mode categorizing system for considering phenology feature characterized by comprising
Acquiring unit, the acquiring unit are used to obtain every class plantation in the EVI time-serial position and sample of pixel to be sorted
The standard EVI time-serial position of mode;
First computing unit, first computing unit are used to calculate each pixel to be sorted on image based on CCSM algorithm
The cross-correlation of the standard EVI time-serial position of EVI time-serial position and every class cropping pattern, obtains maximum intersection
Related coefficient and the corresponding matching position of the maximum cross-correlation;
Significance test unit, the significance test unit are used for the maximum correlation coefficient that the first computing unit is calculated
Significance test is carried out, significance test result is obtained;
First judging unit, first judging unit is for analyzing significance test result, if pixel to be sorted
EVI time-serial position be at least greater than 0.5 with the standard EVI time-serial position maximum correlation coefficient of certain a kind of cropping pattern
And otherwise the second computing unit is gone to then going to second judgment unit by significance test;
Second computing unit, second computing unit are used to calculate the EVI time-serial position of pixel to be sorted in the original location
The Euclidean distance enhanced with the weight at the crucial phenological period key position of every class cropping pattern standard EVI time-serial position,
It calculates Euclidean distance to be classified with minimum distance classification again, obtains the classification results of pixel to be sorted;
Whether second judgment unit, the second judgment unit are used to judge the EVI time-serial position of pixel to be sorted with two
The maximum correlation coefficient of the standard EVI time-serial position of class or the above cropping pattern of two classes is greater than 0.5 and is examined by conspicuousness
It tests, if it is not, then explanation is from the EVI time-serial position of the pixel to be sorted in shape only standard with certain a kind of cropping pattern
EVI time-serial position is similar, goes inside this kind of cropping patterns then directly assigning to pixel to be sorted, if it is, turning
To third computing unit;
Third computing unit, the third computing unit are used to calculate in the matching position for obtaining maximum cross-correlation wait divide
The Euclidean distance that the weight of the EVI time-serial position of class pixel and every class cropping pattern standard EVI time-serial position enhances;
And Euclidean distance combination maximum correlation coefficient is enhanced by weight and constructs comprehensive distance, then most narrow spacing is carried out using comprehensive distance
From classification, the synthesis of the standard EVI time-serial position of the EVI time-serial position of pixel to be sorted and any class cropping pattern away from
From minimum, then pixel to be sorted is assigned to and gone inside which kind of cropping pattern;
Classification results output unit, the classification results output unit are used to export the EVI time-serial position of pixel to be sorted
Classification results.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910137051.1A CN109919207B (en) | 2019-02-25 | 2019-02-25 | Crop planting mode classification method and system considering phenological characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910137051.1A CN109919207B (en) | 2019-02-25 | 2019-02-25 | Crop planting mode classification method and system considering phenological characteristics |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109919207A true CN109919207A (en) | 2019-06-21 |
CN109919207B CN109919207B (en) | 2020-11-17 |
Family
ID=66962193
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910137051.1A Active CN109919207B (en) | 2019-02-25 | 2019-02-25 | Crop planting mode classification method and system considering phenological characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109919207B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102592134A (en) * | 2011-11-28 | 2012-07-18 | 北京航空航天大学 | Multistage decision fusing and classifying method for hyperspectrum and infrared data |
CN102668899A (en) * | 2012-03-28 | 2012-09-19 | 北京师范大学 | Crop planting mode recognition method |
-
2019
- 2019-02-25 CN CN201910137051.1A patent/CN109919207B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102592134A (en) * | 2011-11-28 | 2012-07-18 | 北京航空航天大学 | Multistage decision fusing and classifying method for hyperspectrum and infrared data |
CN102668899A (en) * | 2012-03-28 | 2012-09-19 | 北京师范大学 | Crop planting mode recognition method |
Non-Patent Citations (4)
Title |
---|
JIANHONG LIU等: "A Phenology-Based Method to Map Cropping Patterns under a Wheat-Maize Rotation Using Remotely Sensed Time-Series Data", 《REMOTE SENSING》 * |
JIN CHEN等: "A Simple Method for Detecting Phenological Change From Time Series of Vegetation Index", 《IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING》 * |
K.R. MANJUNATH等: "Mapping of rice-cropping pattern and cultural type using remote-sensing and ancillary data: a case study for South and Southeast Asian countries", 《INTERNATIONAL JOURNAL OF REMOTE SENSING》 * |
张霞等: "利用 MODIS_EVI 图像时间序列提取作物种植模式初探", 《农业工程学报》 * |
Also Published As
Publication number | Publication date |
---|---|
CN109919207B (en) | 2020-11-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Medar et al. | Crop yield prediction using machine learning techniques | |
Kontgis et al. | Mapping rice paddy extent and intensification in the Vietnamese Mekong River Delta with dense time stacks of Landsat data | |
Latham et al. | A method for quantifying vertical forest structure | |
Brown et al. | Classifying multiyear agricultural land use data from Mato Grosso using time-series MODIS vegetation index data | |
Taylor et al. | Establishing management classes for broadacre agricultural production | |
Diao et al. | Use of vegetation change tracker, spatial analysis, and random forest regression to assess the evolution of plantation stand age in Southeast China | |
Li et al. | Definition of management zones for enhancing cultivated land conservation using combined spatial data | |
CN105893977A (en) | Rice mapping method based on self-adaptive feature selection | |
CN102156886A (en) | Statistical data and remote sensing image data based regional fertilizer application spatialization method | |
Padmapriya et al. | Deep learning based multi-labelled soil classification and empirical estimation toward sustainable agriculture | |
LI et al. | Spatio-temporal changes in rice area at the northern limits of the rice cropping system in China from 1984 to 2013 | |
Wu et al. | Investigation of remote sensing imageries for identifying soil texture classes using classification methods | |
Ramesh et al. | Comparison and analysis of Rice Blast disease identification in Greenhouse Controlled Environment and Field Environment using ML Algorithms | |
Chen et al. | A novel Greenness and Water Content Composite Index (GWCCI) for soybean mapping from single remotely sensed multispectral images | |
Chen et al. | Delineating rice cropping activities from MODIS data using wavelet transform and artificial neural networks in the Lower Mekong countries | |
Hampannavar et al. | Prediction of crop fertilizer consumption | |
Pruthviraj et al. | Crop and fertilizer recommendation system based on soil classification | |
Krishna et al. | Analysis of crop yield prediction using machine learning algorithms | |
Navidi et al. | Ecological potential assessment and land use area estimation of agricultural lands based on multi-time images of Sentinel-2 using ANP-WLC and GIS in Bastam, Iran | |
Lin et al. | Optimal soil-sampling design for rubber tree management based on fuzzy clustering | |
Mahendran et al. | Application of Machine Learning and Internet of Things for Identification of Nutrient Deficiencies in Oil Palm | |
Gao et al. | How can agricultural water production be promoted? A review on machine learning for irrigation | |
Taylor et al. | Temporal and spatial relationships of vine pruning mass in C oncord grapes | |
CN109919207A (en) | Consider the crop growing mode classification method and system of phenology feature | |
Mistry et al. | Brief survey of data mining techniques applied to applications of agriculture |
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 |