CN1677085A - Agricultural application integrating system for earth observation technique and its method - Google Patents
Agricultural application integrating system for earth observation technique and its method Download PDFInfo
- Publication number
- CN1677085A CN1677085A CN 200410029826 CN200410029826A CN1677085A CN 1677085 A CN1677085 A CN 1677085A CN 200410029826 CN200410029826 CN 200410029826 CN 200410029826 A CN200410029826 A CN 200410029826A CN 1677085 A CN1677085 A CN 1677085A
- Authority
- CN
- China
- Prior art keywords
- data
- information
- wave band
- band
- spectrum
- 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
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses an agricultural application system and technique for surveying the soil. Above system comprises: data storing unit including spectrum database, remote database, basal database; module unit including parameter selecting module base, agriculture conversion module base; control and calculating processing unit which is for processing figure data with the module in base; and information processing unit for displaying or output in other ways the agriculture information according to the gathered parameter. The invention is able to perform a accurate modern management to the agriculture with the aid of advanced soil survey technique. The technique includes : gather high spectrum data, process the gathered data; select the data by wave band; feature abstract the selected data and obtain the parameter according to which the system output corresponding information.
Description
Technical field
The present invention relates to the application technology of information resources, relate in particular to and utilize the obtained information material of imaging spectral technology for using the application technology that supplementary is provided such as the agricultural of aspects such as agricultural experiment, crop type identification and agricultural feelings diagnosis.
Background technology
Imaging spectral technology is the cutting edge technology in present earth observation field.Because it can obtain the continuous spectrum of atural objects such as surface vegetation, soil, water body, is used to analyze their physical and chemical process, therefore huge application potential is arranged in agriculture application facet.
The civil aerospace technology remote sensor that has moved as: have only 5-6 wave band Landsat TM, SPOT and the NOAA, more than the spectral resolution 50nm.Be difficult to identify the various crop type.And the about 20nm of principal element peak width of vegetation, vegetation is injured and coerces red-shifted component is 5-17nm, these phenomenons are that low spectrum resolution remote sensor is difficult to detect.The superiority of space flight or aerial imagery spectral remote sensing device is to segment out tens or a hundreds of wave band in 0.4-14 μ m spectral range, spectral resolution is 5-10nm.So just can improve the crops recognition capability, monitor the growth change information of crop simultaneously, help the crops precision management.
China is large agricultural country in the world, also is the country that minority is grasped imaging spectral technology.Current, how selecting in the measured high-spectral data of aviation, satellite imagery spectrometer useful Agricultural Information is handled and how to be extracted to the useful spectrum parameter of agricultural, imaging spectrometer data how is key technical problem anxious to be solved.
Summary of the invention
Therefore, the object of the present invention is to provide a kind of agriculture application integrating system and method thereof of earth observation technology, can imaging data be handled, to extract useful Agricultural Information by from the data that imaging spectral and other airborne remote sensing data are obtained, selecting suitable spectrum parameter.
For achieving the above object, the invention provides a kind of agriculture application integration method of earth observation technology, comprising: obtain the earth observation data; The earth observation data that obtained are carried out the data band selection; Earth observation data after described band selection are carried out feature extraction, obtain agricultural feelings parameter; Utilize the agricultural feelings parameter of being obtained to carry out signature analysis, obtain the information needed in the described earth observation data.
In a preferred embodiment of the invention, the earth observation data are high-spectral data, and also comprise in the said method high-spectral data is carried out pretreated step.
On the other hand, the present invention also provides a kind of agriculture application integrating system of earth observation technology, comprising: data storage cell comprises:
Spectra database, it is with the data mode storage different crops of vector graphics or the curve of spectrum of ground object target;
The Multi-Band Remote Sensing Images that airborne imaging spectrometer obtains is stored with the data mode of grating image in the remotely-sensed data storehouse;
Basic database is used for storing other the complementary geographical spatial datas (figure of grid or vector form, view data) and the attribute data (statistics of text or form) that are complementary with the survey region remote sensing images.As the regional Administrative boundaries of vector quantization, meteorological element, soil cover type map etc.;
Model unit, comprise selection model storehouse and agricultural feelings inverse model storehouse, described selection model storehouse is used for image is carried out rebuilding spectrum, the analysis of data compound characteristics and band selection, and described agricultural feelings inverse model storehouse is used to provide polytype agricultural feelings information model;
Control and operation processing unit are used to utilize the model of described model unit that view data is handled accordingly; With
Information output unit is used for according to the agricultural feelings parameter of being extracted, the Agricultural Information that demonstration or output otherwise are correlated with.
The present invention can utilize advanced earth observation technology, provides abundant and information accurately to agricultural production, test and management.
Description of drawings
Fig. 1 is the schematic flow diagram of agricultural application method of the earth observation technology of one embodiment of the present of invention;
Fig. 2 is grouping conversion, the feature selecting schematic flow diagram of an embodiment of the inventive method;
Fig. 3 is the schematic flow diagram of the application of method of the present invention in wheat growing way grading evaluation;
Fig. 4 is the schematic flow diagram of the application of method of the present invention in the wheat leaf area coefficient extracts;
Fig. 5 is the structural representation block scheme of the agriculture application integrating system of earth observation technology of the present invention;
Fig. 6 is a synoptic diagram of setting up the process of model bank in the system of the present invention
Fig. 7 has shown the process of land use classes in system of the present invention.
。
Embodiment
As shown in Figure 1, the invention provides a kind of method of agricultural application of earth observation technology, comprising: obtain high-spectral data; The high-spectral data that is obtained is handled; Described high-spectral data is carried out the data band selection; Extract described high-spectral data middle peasant feelings parameter, show relevant Agricultural Information.
In a preferred embodiment of the invention, need to use spectroscopic data, basic data and remotely-sensed data.
Spectroscopic data refers to the curve of spectrum of different crops or ground object target, and promptly crops or atural object is to the curve that the electromagnetic reflectivity of different-waveband is constituted, and is that the remote sensing images by spectroscopic assay and airborne imaging spectrometer obtain.
Need explanation, real object spectrum is that the reflectivity that utilizes spectrometer to measure by ground obtains; Remotely-sensed data directly obtains is the radiance value (be atural object to external reflection and emittance what) of atural object, rather than reflectivity.Rebuilding spectrum is exactly that radiance value with remote sensing image is converted to reflectivity, and then obtains the process of object spectrum curve.
High-spectral data (image) is a kind of of remotely-sensed data (image), and principal feature be for can distinguish the very thin spectrum in wavelength interval (as 10nm), because of the spectrum segment of its covering a lot, so be called high spectrum.Imaging spectral technology is a kind of high spectral technique.The used remotely-sensed data of the present invention is an imaging spectrometer data.
Remotely-sensed data: refer to the Multi-Band Remote Sensing Images that airborne imaging spectrometer obtains.Data mode storage with grating image.
Basic data: refer to other the complementary geographical spatial datas (figure of grid or vector form, view data) and the attribute data (statistics of text or form) that are complementary with the survey region remote sensing images.As the regional Administrative boundaries of vector quantization, meteorological element, soil cover type map etc.
Basic data provides background information and the information relevant with study area; The remotely-sensed data storehouse provides the multi-band image data of ground object target; Spectra database has then comprised the selected specific atural object clarification of objective curve of spectrum on the above two bases; Usually, different ground object targets have self specific curve of spectrum.
High-spectral data can utilize airborne aerial imagery spectrometer, airborne (boat-carrying) space borne imagery spectrometer etc. to obtain.In certain embodiments, for realizing the present invention better, also should utilize devices such as visible-near infrared intelligence spectrometer, visible-near infrared spectrometer, infrared radiometer, the portable camera of GPS to obtain terrestrial information.Certainly these information also can provide by other source, for example obtain by the informant of internet from other.
After obtaining high-spectral data, need carry out pre-service to image.As required, this pre-service can comprise radiant correction, geometry correction etc.In addition, can carry out also that spectrum strengthens, spectrum identification, grouping KL conversion processing such as image classification.
Further, also can carry out rebuilding spectrum to image, multi-source data is compound and band selection is handled.Wherein, the effect of rebuilding spectrum is the curve of spectrum that obtains atural object by remote sensing images, and these curves of spectrum can be used for image classification, atural object identification.The compound extracting parameter feature that then is used for of multi-source data.These processing will illustrate in greater detail below.
According to one embodiment of present invention, can also after rebuilding spectrum, carry out image transformation.
Image transformation refers to the process of image being changed to transform domain from transform of spatial domain.The purpose of carrying out image transformation is exactly for the treatment of picture process being simplified, by conversion, being made vector have some better character in new space, thereby more help the analysis and the solution of problem.
Feature selecting is generally finished by linear transformation.The expression formula of linear transformation is:
Y=AX (formula 1)
X is the n dimension random vector before the conversion in the formula, and Y is the n dimension random vector after the conversion, and A is the transformation matrix of a n * n, and the difference of A has determined the different in kind of conversion.
By image transformation be with the basic ideas of carrying out feature selecting: select in the transform domain subclass of forming by m the component of Y (1≤m<n), when only using m the representation in components X that is kept when leaving out a remaining n-m component, the error minimum that causes.So just realized tieing up the compression of m dimension by n.The size of error is generally weighed with mean-square error criteria.Mean-square error criteria is: keep m and have the component subset of maximum variance, leave out all the other n-m component.
The present invention adopts the method for grouping KL conversion.For explanation grouping KL conversion, the KL conversion is described at first.
The KL conversion is a kind of orthogonal linear transformation, be the most frequently used in the process in remote sensing digital image processing also be one of the most useful mapping algorithm, be decorrelation, carry out the effective ways of feature extraction, data compression.
In order to derive conveniently, the expression formula of linear transformation is written as:
Wish through after the KL conversion, in new space, only just can under the condition of error minimum, reflect original n dimension information with its preceding m dimensional vector.Now Y is divided into two parts, front m item is a first, and back n-m item is a second portion, then has:
With the x in the second portion in the formula
iBe designated as b
iAnd note
If the error between Y and the Y (m) is ε, then:
Its mean square deviation is:
For making the mean square deviation minimum, to b
iDifferentiate gets:
b
i=E{x
i}=m
x(formula 7)
For new stochastic variable Y, get its preceding m item and omit back n-m item, should be expressed as
y
i=A
i{ x
i-m
x) be Y=A (X-m
x) (formula 8)
The covariance ∑ of new stochastic variable Y
y=E{ (Y-m
y) (Y-m
y)
t, m wherein
yBe the average of Y, the vector after the conversion is the random vector with zero-mean, so m
y=0, therefore can get:
∑
y=E{YY
t}=E{[A(X-m
x)][A(X-m
x)]
t}=A∑
xA
t
This explanation is the covariance matrix diagonalization of stochastic variable X, and the gained diagonal matrix is exactly the covariance square ∑ of new stochastic variable Y
yEach element is exactly a ∑ in the diagonal matrix
xAn eigenwert.With eigenwert series arrangement by size, λ
1>λ
2>λ
3λ
nLike this,, m+1 is omitted to the n item, can make when we only get front m item
Hence one can see that, if transformation matrix A is an orthogonal matrix, and it is the covariance matrix ∑ by raw image data matrix X
xProper vector form, then be transformed to the KL conversion.
Carry out the KL conversion, at first, obtain its covariance matrix ∑ according to original image matrix X
xSecondly, by secular equation
(λ I-∑
x) u=0 (formula 10)
Wherein λ is an eigenwert, and I is a unit matrix, and u is a proper vector, obtains the covariance matrix ∑
xEach eigenvalue
i=(i=1,2 ..., n), it is pressed λ
1〉=λ
2〉=... 〉=λ
nArrange, obtain each eigenwert characteristic of correspondence vector u
1
u
i=[u
1i, u
2i, u
Ni]
t(formula 11)
Then, get transformation matrix A=U
TPromptly obtained the formula that embodies of KL conversion:
Through after the KL conversion, obtain one group of new variable (being each row vector of Y), they be called as successively first principal component, Second principal component, ..., the n major component.Because the row of transformation matrix is a ∑
xProper vector, so it is the weighted sum of power with each component of i proper vector that i the component of Y is actually each component of X, each component of whole Y all is the linear combination of information of each component of X, it combines the information of original each feature rather than accepts or rejects simply, and this makes new n dimension random vector can reflect the feature of original things well.
It is best orthogonal transformation on the square error least meaning as can be known from the principle of KL conversion, and it has following characteristics:
Because the KL conversion is orthogonal linear transformation,, just original variance is redistributed in the new major component image unequally so the variance summation before and after the conversion remains unchanged.
The KL conversion is equivalent to carry out the rotation of volume coordinate on geometric meaning, first principal component is got the most concentrated direction of data scatter in the spectral space, and Second principal component, is got with the first principal component quadrature and fetched data and scatter time concentrated direction, and the rest may be inferred by analogy.Therefore, first principal component has comprised the overwhelming majority (generally more than 80%) of population variance, and variance is consistent with quantity of information, so the result of KL conversion makes first principal component almost comprise the overwhelming majority of original each band image information, the information that all the other major components comprised reduces rapidly successively.
The KL conversion is decorrelation, the effective ways of eliminating data redundancy.Each component is mutual oblique in former space, has bigger correlativity, through the KL conversion, each component is an orthogonal in new space, and is separate, related coefficient is zero, and owing to information concentrates on preceding several components, so under the prerequisite of information loss minimum, available less component replaces original high dimensional data, reach the effect of dimensionality reduction, thereby made the time of deal with data and expense reduce greatly.On the other hand,,, reduced difference in the class, can improve nicety of grading so increased the class spacing because each major component is orthogonal.
Though it should be noted that preceding several major component has often comprised the information more than 98%, but can not think simply that the principal component of back is otiose, sometimes the information that comprises in the very little major component of information is required information just, therefore should do concrete analysis according to concrete application target when major component is accepted or rejected.
Introduce grouping KL of the present invention conversion below.
For conventional remotely-sensed data, because its wave band is few, feature selecting is carried out than being easier to.But for imaging spectrometer data, because the wave band number increases severely, feature extraction just is difficult to realize with conventional method.
The KL conversion is directly used in imaging spectrometer data still faces difficulty.Carry out the KL conversion and mainly comprise two steps, the first step is to generate transformation matrix, and second step was to utilize transformation matrix that image is carried out conversion.The first step does not need too big operand, but second step was with transformation matrix each pixel to be carried out computing, and this is a unusual time-consuming procedure, because be the individual addition of N * N multiplication and N * (N-1) for the calculated amount of each pixel.Therefore, if conversion is used for all wave bands of imaging spectrometer data, low the well imagining of efficient.Because the spectral resolution of imaging spectrometer data is very high, it is subjected to the influence of sun absorption spectrum more obvious, and this influence is different to different wavelength band, reduces with the increase of wavelength.Therefore, if imaging spectrometer data is not done radiation and is corrected, perhaps radiation is corrected not ideal enoughly (this is ever-present), radiation distortion will have influence on the variance size of wave band so, the variance of the wave band that the raying distortion effect is little often is higher than the variance of influenced big wave band, the size of variance can influence the result of KL conversion again, so should not directly the KL conversion be used for imaging spectrometer data.
By the correlation analysis of imaging spectrometer data having been understood the characteristics of its band grouping, the correlativity of adjacent band is very strong in other words, is boundary with certain spectral range, constitutes in several groups the high relevant and relatively independent band group between group of wave band.Grouping KL conversion is on the basis that is based upon this feature understanding of imaging spectrometer data.Change the method that traditional KL transfer pair total data is carried out conversion, but the starting point is placed on each relatively independent band group.Its process as shown in Figure 2.At first raw data is carried out correlation analysis, set a threshold value, they are divided into K band group according to the big young pathbreaker of related coefficient between the wave band, and the wave band number of each group is respectively n
1, n
2... n
kSecondly, each band group is carried out the KL conversion respectively, produce the major component image of each group; Then, carry out feature selecting.If the characteristic parameter of selecting is many, also can repeats top step the characteristic parameter of selecting is carried out principal component transform and feature selecting again, till reaching concrete requirement.
The conventional KL conversion of grouping KL ratio of transformation has tangible advantage, and as previously mentioned, for conventional KL conversion, the multiplying amount of each pixel point is N * N, adopts packet type KL conversion, and the multiplying of each point is n in each group
k* n
kIt is individual,
Whole multiplying amounts of K group are
The ratio of two kinds of distinct methods multiplying amounts is
Promptly with group in wave band number and original image wave band count square being directly proportional of ratio, wave band is counted n in the group
kMore little, top multiplication is than more little, and the time of saving is many more.
Suppose to be divided into three groups, every group of size (be K=3, n
1-n
2=n
3), only consider that then the multiplying amount just can save for 2/3 time.Simultaneously, the result of grouping KL conversion process is more rational.Why wave band can form the group of height correlation, and its inner link is arranged, and they are positioned at specific spectrum segment, has certain common similar character, i.e. consistance.Solar spectrum is different to the size that influences of different spectral range wave bands, can cause variance unusual, thereby influences the result of KL conversion, but in the narrow group of spectral range, this influence is consistent relatively, so the KL conversion of dividing into groups can be avoided this mistake, obtains more rational result.
Preferably, method of the present invention also comprises the step that high spectrum image is handled, and comprising: figure image intensifying, spectrum strengthen, spectrum is discerned and image classification.These methods can adopt technology known in the field, for example can (spatial information obtains and handle serial monograph, calendar year 2001, Beijing: Science Press) referring to chief editors' such as Wang Changyao " earth observation technology and precision agriculture ".Repeat no more in this article.
According to one embodiment of present invention, can handle image in the following order: radiant correction, geometry correction, rebuilding spectrum, band selection, multi-source data are compound.
Band selection is to choose a small amount of wave band that comprises the atural object target information of enough being correlated with in numerous wave bands, finds the wave band that can reflect terrestrial object information on a small quantity.
The band selection problem can be summed up as establishes raw data n wave band is arranged, and concentrate the data subset of the selecting the m dimension (information that m<n) and don't weight losses are wanted in this n dimension data.People wish to select the least possible wave band reflects former wave band information as much as possible.When conventional remotely-sensed data Optimal Bands Selection, method commonly used is according to certain algorithm, criterion all possible combination to be estimated, thereby obtains the best band combination.For conventional remotely-sensed data, the wave band number is fewer, is 7 wave bands as the TM data, is feasible with such method.But for high-spectral data, because the wave band number increases severely, it obviously is inappropriate indiscriminately imitating such method again.The n dimension data concentrate select m dimension subclass all combinations may for
This is a googol.Only the imaging spectrometer data with 30 wave bands is an example, if select 10 wave bands in 30 wave bands, possible combination just has 30045015 kinds.And actual imaging spectrometer data, tens wave bands at least, a up to a hundred at most even hundreds of wave band.Therefore, for the very many high-spectral datas of wave band, might situation all consider the institute of selecting m wave band from n wave band one by one, be unpractical, also there is no need in fact.Because spectrum segmentation, different with conventional remotely-sensed data, high-spectral data no longer is one, two at the different wave band numbers that spectral coverage had, but tens even tens, they constitute several spectrum groups, the spectrum range that belongs to different in kind in the electromagnetic wave, each wave band has bigger similarity in the group, then has bigger otherness between the different groups.This shows to such an extent that be perfectly clear on correlation matrix.High-spectral data all can be divided into several groups significantly, and the correlativity of each wave band is very strong in the group, and the correlativity between on the same group is very not weak.The starting point that spectral band is selected should be placed on the evaluation of each wave band in the group, rather than between group.Therefore, when carrying out the high-spectral data band selection, should earlier high-spectral data be divided into some groups by character, be the superior and inferior evaluating that the unit carries out spectral band then with the group, so both avoid a large amount of unnecessary operands, can improve the accuracy and the efficient of band selection again.
Based on above consideration, provided the method for calculating the band index that is used for the evaluation of imaging spectrometer data wave band below.If ρ
IjBe the related coefficient between wave band i and the j, imaging spectrometer data is divided into the k group, and every group wave band number is respectively n
1, n
2N
k, the definition band index is:
Wherein
R
i=R
w+R
a
σ in the formula
iBe the mean square deviation of i wave band, R
wBe the mean value of the absolute value sum of other wave band related coefficient in i wave band and the place group, Ra is the absolute value sum of the related coefficient between i wave band and place group other wave band in addition.In other words, band index be wave band mean square deviation with this wave band in group average correlation coefficient and this wave band with organize outside the ratio of wave band related coefficient absolute value sum.
Owing to the correlativity of wave band between relevant very strong group of each wave band in the group is very weak, the overall relevancy of a wave band is strong and weak mainly by its correlativity size decision with interior each wave band of group, and the varying in size of each group, promptly the wave band of formation group is counted difference, therefore the mean value of a wave band and other wave band related coefficient absolute value sum more can reasonably reflect the whole good and bad level of this wave band as one of this band index denominator in the use group.
The meaning of band index is very clear and definite, it be show this wave band and other wave band between dispersion degree.Band index is big more, and the duplicate message of the quantity of information that it and other wave band are contained is few more.In other words, because mean square deviation is big more, show that the dispersion degree of wave band is big more, contained quantity of information is abundant more, and the absolute value of the population correlation coefficient of wave band is more little, shows that the independence of wave band data is strong more, and information redundance is more little.So band index Pi can synthetically reflect band class information content and two factors of correlativity, one of important parameter of wave band alternatively.Obviously, should select the big wave band of Pi.
After band selection, can carry out various application to data and handle, for example carry out that multi-source data is compound, feature extraction, signature analysis etc.
Multi-source data is compound to be meant the figure or the image data layer stack of separate sources or to merge into a figure layer, to obtain abundanter, complete information.Be compound from a plurality of figure layers be the process of a figure layer, promptly a plurality of data are merged into data.Purpose is in order to increase quantity of information.
Feature extraction be in image identification and extract that some is concerned about have the special characteristic information of (comprising geometric properties, spectral signature etc.), as river, the road of wire, the water body that reflectivity is very low etc.This is a process of extracting certain information needed from numerous information of image selectively.Can utilize known genetic algorithm that remotely-sensed datas such as imaging spectrometer and radar are carried out feature extraction and neuroid classification.In feature extraction, in case of necessity, it is compound to carry out multi-source data.
Signature analysis is meant the space distribution of the various features of analysis and mutual relationship to each other, utilize methods such as statistics, mathematics or literal that they are measured and describe, and object is done data analysis, understanding, identification by the background knowledge (basic data) of object.The signature analysis technology has been used for image recognition, and these technology comprise use always at present known to the edge of reflection geographic entity and detection, prospect and background separation, texture analysis and the spectrum etc. at turning in the cartographic feature.
As a rule, processing sequence be that multi-source data is compound, feature extraction, signature analysis.But, there is no strict sequencing between the three.Such as, be used for the data of feature extraction, may pass through the data Combined Processing, may not need yet; The data Layer that feature extraction obtains also may carry out compound with other data.
For agricultural feelings monitoring, the parameter that only has the spectrum after rebuilding to provide is not enough.That is to say that agricultural feelings monitoring model not only needs spectrum parameter, the supplementary that the ground that also needs some to match with spectral information obtains.This will adopt above-mentioned multi-source data compound.This process by known figure stackedly add, step such as AIT merging realizes.
In the present invention, can be compound by the figure of remotely-sensed data and basic data, obtain abundanter integrated information.
By the stack of different figure layers between basic data and remotely-sensed data, can obtain different types of terrestrial information, the data of these separate sources are formed composite information, for ground object target spectrum is judged and analysis provides auxiliary reference and support.The characteristic spectrum curve is applied to remote sensing images, can in big regional extent, differentiates different ground object targets, and further excavate the deep layer ground object target information that image comprised.
Because the different pieces of information source may comprise the attribute information of repetition, therefore can carry out AIT and merge, so that, form unified complete attribute information these information merger arrangements.
This AIT merges can utilize known GIS technology, the graphic attribute table of separate sources is edited arrangement get final product.
The information characteristics of data can be analyzed according to relevant statistical parameter.Entropy and variance can be used for the evaluation foundation of measurement information size, and their value is big more, and descriptive information is abundant more; The average brightness level of average reflection image, the contrast size of the dynamic range of image (being the poor of the maximal value of brightness and minimum value) reflection image, the contrast greatly then separability of atural object are strong; The correlativity power of the related coefficient reflection wave band between the wave band, the redundancy of relevant strong explanation data is big, and the independence of the weak explanation of correlativity wave band is strong.Time different mutually, the underlying surface situation difference of remote sensing images, the entropy of image, variance, average and dynamic range can change, but the variation of the related coefficient between the wave band is very little.
The wave band that a little information amount that band selection obtains is abundant is the data source that is used for feature extraction.Classification and feature selecting all are to carry out on the basis of band selection, and the information that feature extraction obtains can be used for classification.
In a preferred embodiment of the invention, the step that provides Agricultural Information to show.Introduce the method that the data of utilizing the earth observation technology to obtain of the present invention (spectrum, remotely-sensed data etc.) provide agricultural mutual affection to analyse below in conjunction with specific embodiment.
As shown in Figure 3, in one embodiment of the invention, agricultural feelings are the wheat growing way.The steps include:
Utilize the characteristics of high spectrum, promptly the characteristics of high spectrum are made up of the very little abundant spectral band in wavelength interval, and promptly the spectrum branch is very thin; As 10nm at interval, promptly the wavelength coverage that each wave band covered has only 10nm.These subdivided spectrals can be band selection and combination provides abundant selection, pass through subdivided spectral, according to the ground observation data, carry out principal component analysis (PCA) with spectroscopic data, obtain reflecting the band class information of three main factors (chlorophyll, protein, moisture) of wheat growing way, and respectively it is carried out normalized.According to the contribution of these three main factors (chlorofucine, protein P, moisture W) to the wheat growing way, adopt the method for linear combination, ask and calculate wheat growing way function G,
G=aC+bP+cW
A=2 wherein, b=1, c=2; Be respectively chlorophyll, protein, moisture factor contribution rate to the wheat growing way.
Chlorophyll information: the 2nd wave band (0.46-0.48um) can fine reflection chlorophyll information.
Protein information: the 15th wave band (1.00-1.02um) can fine reflection protein information.
Moisture information: the 12nd wave band (0.94-0.96um) can fine reflection moisture information.
And the 8th wave band (0.64-0.66um) all is reflection low value district for chlorophyll, protein, moisture, utilize the 2nd wave band and the 8th wave band respectively, the 15th wave band and the 8th wave band, the 12nd wave band and the 8th wave band carry out normalized, the influence of elimination band, outstanding simultaneously chlorophyll, protein, moisture information, reflection wheat growing way.
Then wheat growing way function G is carried out grading evaluation, obtain wheat growing way result.
Utilize basic data, superimposed crop distribution thematic information is removed non-wheat zone.
Input database.
As shown in Figure 4, in one embodiment of the invention, can calculate leaf-area coefficient.
At first high-spectral data is carried out radiant correction and geometry correction, classifies then,, try to achieve wheat greenness index NI by normalized,
Further obtain wheat coverage rate fv, wherein, CHANNEL8 and CHANNEL12 represent the value in wave band 8 and 12 respectively.Utilize LAI=K
-1In (1-fv)
-1
Wherein K is the extinction coefficient of wheat,
Obtaining the wheat leaf area coefficient at last distributes.
Fig. 5 has shown the schematic block diagram that is used to realize system for carrying out said process according to one embodiment of the present invention, and this system comprises:
One, data storage cell comprises:
1) spectra database, it is with the data layout storage different crops of vector graphics or the curve of spectrum of ground object target;
2) Multi-Band Remote Sensing Images that airborne imaging spectrometer obtains is stored with the data layout of grating image in remotely-sensed data storehouse.
3) basic database is used for storing other the complementary geographical spatial datas (figure of grid or vector form, view data) and the attribute data (statistics of text or form) that are complementary with the survey region remote sensing images.As the regional Administrative boundaries of vector quantization, meteorological element, soil cover type map etc.
Specifically, for vector data, can use x, y coordinate mode is as the sign of provider location, and also the available topologies structure reflects the mutual relationship between each entity.Mainly comprise:
The point key element is as well, the observation station etc.
Line feature is as public (iron) road, rivers etc.
The polygon key element, as zoning, field piece etc.
The annotation key element.
For vector data file
Employing is carried out according to special topic one by one towards the mode of map sheet, forms corresponding thematic data file, comprising:
Coordinate file, the memory space geometric data
Property file, the memory space attribute data
The multiple thematic data file of same application can constitute a file group.
In system of the present invention, the vector data form can adopt ARC/INFO (E00) data layout or/and Auto CAD (DXF) data layout
For raster data, be that map or image are divided into the graticule mesh that some row and columns are formed, as a some full figure scanning sample is obtained every attribute data by each graticule mesh.Full figure after grid is finished is the array of rule, so the coordinate position of entity is hidden in the storage address of grid.
Native system need store multiple topic space data, choose onesize zone, same engineer's scale.At this moment each grid comprises two or more attribute, and following two kinds of recording modes can be arranged:
The file group mode.The independent file that forms of each special topic, all associated documents constitute a file group.
The multiband mode.Each different special topic is placed in the different wave bands, forms a file.
The raster data form can adopt ranks matrix data form, pci data form, TIFF data layout, BMP data layout.
For the multiband raster data, can take the following manner storage.
BIP form, file are arranged each wave band data of the 1st point earlier in proper order, arrange each wave band data of the 2nd point again, drain into last point according to this.
BIL form, file are arranged the 1st wave band data of the 1st row earlier, arrange the 1st row the 2nd wave band data again, and after having arranged the 1st all wave band datas of row, order is arranged the 2nd each wave band data of row according to this again, until last column.
BSQ form, file are arranged the data that the 1st wave band is had a few earlier, arrange the data that the 2nd wave band is had a few again, according to this until last wave band.
In one embodiment of the invention, above-mentioned database can adopt ARC/INFO Geographic Information System, FOXPRO database processing software to design.
Two, control and operation processing unit comprise the image pretreatment unit, include conventional Flame Image Process instrument, for example known GIS function.Be used for to the remotely-sensed data image carry out that radiant correction, geometry correction, noise remove, mosaic splicing, projective transformation, figure image intensifying, spectrum strengthen, processing such as spectrum identification and image classification; Also can only carry out above-mentioned some of them handles.The method of these processing can adopt prior art to realize, therefore repeats no more.
In one embodiment of the invention, can adopt each functional module of pci system Version6.2 of Canadian PCI Company exploitation is core, and in conjunction with above-mentioned ARC/INFO Geographic Information System, FOXPRO database processing software, and the C/C++ software of Microsoft etc. forms the tool storage room of system, is used to data acquisition typing, data management maintenance, data conventional processing, utilizes model in the model unit described later that the data in the database are handled computing, the output of information map etc. the support of system software instrument is provided.
Three, model unit is used for the memory modify model algorithm in system of the present invention, several data is had purpose, specialized computing, obtains the central processing module of information needed.Model in the model unit is that institute of general Geographic Information System and remote sensing image processing system is uncollected, specially the algoritic module of developing at this problem the objectives and finishing of task.
Model unit in the system of the present invention comprises selection model storehouse and agricultural feelings inverse model storehouse.
Selection model is used for processed images is carried out rebuilding spectrum, the analysis of data compound characteristics and band selection etc.These processing procedures describe in detail at preamble, are omitted herein.
In model unit, to set up agricultural feelings inverse model storehouse.At first need to collect other each sub-subject study, employing for this reason, and be proved to be model in practice with good result, on the basis of analyzing its algorithm structure, from tool storage room, select the necessary function module, the algorithm of taking according to model links in proper order, be aided with program development in case of necessity, form from data input, model calculation to the electronic flow process of a whole set of model of information output.When the user uses a model, only need pending data file information of input and destination file name, and needn't know the mid-module calculating process.
Referring to Fig. 6, be example with simple NDVI vegetation index computation model below, the performance history of each model in the agricultural feelings inverse model storehouse is described.
As mentioned above, when setting up model, at first to be in this example according to the Model Calculation formula that other sub-problem provided
Carry out the algorithm structure analysis, the calculation step that comprises in promptly analyzing in formula (wave band subtracts each other, wave band addition and carry out division arithmetic etc.).
Then, according to resulting specific algorithm structure, the electronization of carrying out model makes up.When the user imported pending image, system called the wave band computing module in control and the operation processing unit automatically, and two wave bands of image are carried out additive operation, the result was outputed to first wave band of new images.
System call tool storage room medium wave band computing module carries out sum operation to two wave bands of image, and the result is outputed to new images second wave band.
System call tool storage room medium wave band computing module carries out the phase division operation to two wave bands of new images, obtains net result figure.
Then, system deletes the figure in the intermediate treatment process automatically, exports result images to the user.
Agricultural feelings information model in the native system farming feelings inverse model storehouse mainly comprises:
High spectrum crop physicochemical property extraction model
Crop growing state dynamic monitoring model
High spectrum flying quality and existing spaceborne data composite model
The imaging spectrometer selection model
The concrete development approach of these models is similar substantially, and this paper repeats no more.
Four, information output unit is used for according to the agricultural feelings parameter of being extracted, the Agricultural Information that demonstration or output otherwise are correlated with.
System of the present invention can use for agricultural multiple information is provided, and comprises land-use map, crop distribution plan, crop growing state data, damage caused by a drought Monitoring Data, multiple information combined result, agricultural parameter selection scheme etc.Below to utilize information with the soil be the method for example explanation system of the present invention.
The soil utilizes information to be used to study the spectral characteristic of atural object on high spectrum image, inquires into and uses high-spectral data to carry out the method and the potentiality of land use classes.
Land-use map is the basis of maps such as crop distribution plan, wheat growing way figure, and the quality of land-use map directly influences the quality of other thematic map.Therefore, when carrying out the remote sensing mapping of land-use map, system of the present invention can adopt multiple known method such as visual interpretation, Computer Automatic Recognition classification and comprehensive classification, thereby has guaranteed the accuracy of this basic map.Fig. 7 has shown that the present landuse map with Beijing Shunyi County is the detailed process of example.
High-spectral data: imaging time on April 24th, 1998, wheat jointing, head sprouting season.Spectral range 0.44~2.40 μ m, the wavelength band of totally 32 each wave bands of wave band sees Table 5, because the instrument problem wherein has seven wave bands (1,9,17,22,23,25,32 wave band) poor quality, therefore actual spendable wave band is 25.
Non-remote sensing data: 1: 5 ten thousand ground coloured picture is the thematic maps such as land-use map that forefathers finish.
Geometric correction and image mosaic: test site, Shunyi is made up of seven air strips of East and West direction, the side of certain width is all arranged to overlapping between each air strips.On the same air strips, geometric distortion increases to the west gradually from middle, in order to make full use of air strips intermediate mass better image, to be in the other image that geometry deformation is bigger in the overlapping scope earlier cuts away, contrast high spectrum image and 1: 5 ten thousand topomap then, on image, select the reference mark equably, set up the polynomial expression transfer equation, image is carried out geometric correction through taking a sample again.At last will be through geometric correction, have common map projection (Gauss-Ke Lvge projection), seven air strips under the same coordinate system are set into piece image.
Set up interpret tag: carry out open-air on-the-spot investigation, with the Hyperspectral imaging contrast, the image feature of each atural object of analysis and research test block is set up index of image interpretation.Formulate drawing norm and categorizing system on this basis.
Present status of land utilization classification: comprise methods such as visual interpretation, computer automatic sorting.
(1) visual interpretation method
The 12nd wave band, the 10th wave band and the 6th wave band of choosing high spectrum flying quality carry out the RGB colour and synthesize;
With CorelDraw software is operating platform, according to the interpret tag of being set up, with reference to the interrelated datas such as land-use map that topomap, forefathers do, learns the professional knowledge analysis-by-synthesis in combination, carries out the man-machine interaction interpretation.
The decipher result is derived in CorelDraw, save as the acceptable DXF form of Arc/Info.
In Arc/Info to the decipher result make amendment, editor, ascription code, the row-coordinate of going forward side by side conversion.
At the knotty problem that runs in the decipher, go on-the-spot investigation to solve, and to decipher result verification, modification.
Input database.
(2) computer automatic sorting
The A supervised classification
Spectral band is selected
At the classification of certain purpose, the correct selection of best band and combination thereof is very important.Compare with other remotely-sensed data, the characteristics of high-spectral data are, wavelength band is very narrow, and spectral information is extremely abundant, and selectable wave band is more, and the ability of recognition object is stronger.But this does not also mean that the wave band number that uses at minute time-like is The more the better.This be because, the first, the spectral band number does not equal the intelligence wave hop count simply.Correlativity very strong (table 4.1) between some wave band datas is arranged.If use all wave bands to classify,, also can influence classification results on the contrary because the phase mutual interference of related data not only can not improve nicety of grading; The second, select wave band too much, can influence the speed of calculating, computer hardware is proposed higher requirement; The 3rd, the conventional images process software can't satisfy this requirement.Various softwares all have certain qualification to the wave band number, and at the branch time-like that exercises supervision, the image wave hop count can not be greater than 16 as PCI software.
When selecting spectral band, we have mainly considered two factors, the correlativity of each wave band of the first, and it two is spectral response characteristics of each atural object of test block.
Table 1 is the related coefficient between each wave band of high-spectral data, following as can be seen from Table 1 several characteristics:
A. wave band 2~11 visible-ranges, the correlativity between each wave band is very strong, and its medium wave band 2 is relevant with other wave band the most weak, secondly is wave band 11.
B. the infrared band of wave band 12~31 is all relevant with visible light wave range not too strong, wave band 12~16 relevant with visible light very little (related coefficient is less than 0.2) especially, and the related coefficient between the wave band 18~31 is less than 0.2.
C. the facies relationship number average between the wave band 12~16 is more than 0.94, and the related coefficient between they and the wave band 18~31 is less than 0.2.
D. the related coefficient between the wave band 18~31 is 0.7~0.85, relevant the strongest be wave band 28 and 31, the most weak is wave band 19 and 29, wave band 19 relevant with each wave band a little less than.
Become response from the light of atural object, the wheat field of test block and orchard, orchard and vegetable plot and forest land, reservoir are very similar on some wave band to the spectral characteristic between atural objects such as river and pond, if but removed to observe their curve of spectrum from all wave bands, their variant on other wave bands (seeing Table 2) could be found.
Each wave band correlation matrix of table 1 high-spectral data
Wave band | ????2 | ????3 | ????4 | ????5 | ????6 | ????7 | ????8 | ????10 | ????11 | ????12 | ????13 | ????14 |
????2 | ????1.00 | ????0.84 | ????0.86 | ????0.85 | ????0.85 | ????0.84 | ????0.85 | ????0.82 | ????0.75 | ????-0.04 | ????-0.06 | ????-0.11 |
????3 | ????0.84 | ????1.00 | ????0.92 | ????0.95 | ????0.95 | ????0.95 | ????0.95 | ????0.93 | ????0.86 | ????-0.08 | ????-0.08 | ????-0.11 |
????4 | ????0.86 | ????0.92 | ????1.00 | ????0.92 | ????0.94 | ????0.93 | ????0.94 | ????0.91 | ????0.86 | ????-0.02 | ????-0.03 | ????-0.06 |
????5 | ????0.85 | ????0.95 | ????0.92 | ????1.00 | ????0.95 | ????0.96 | ????0.96 | ????0.94 | ????0.86 | ????-0.11 | ????-0.12 | ????-0.14 |
????6 | ????0.85 | ????0.95 | ????0.94 | ????0.95 | ????1.00 | ????0.96 | ????0.98 | ????0.96 | ????0.86 | ????-0.16 | ????-0.16 | ????-0.19 |
????7 | ????0.84 | ????0.95 | ????0.93 | ????0.96 | ????0.96 | ????1.00 | ????0.99 | ????0.97 | ????0.87 | ????-0.17 | ????-0.17 | ????-0.20 |
????8 | ????0.85 | ????0.95 | ????0.94 | ????0.96 | ????0.98 | ????0.99 | ????1.00 | ????0.98 | ????0.87 | ????-0.18 | ????-0.18 | ????-0.20 |
????10 | ????0.82 | ????0.93 | ????0.91 | ????0.94 | ????0.96 | ????0.97 | ????0.98 | ????1.00 | ????0.86 | ????-0.16 | ????-0.15 | ????-0.17 |
????11 | ????0.75 | ????0.86 | ????0.86 | ????0.86 | ????0.86 | ????0.87 | ????0.87 | ????0.86 | ????1.00 | ????0.15 | ????0.18 | ????0.17 |
????12 | ????-0.04 | ????-0.08 | ????-0.02 | ????-0.11 | ????-0.16 | ????-0.17 | ????-0.18 | ????-0.16 | ????0.15 | ????1.00 | ????0.94 | ????0.95 |
????13 | ????-0.06 | ????-0.08 | ????-0.03 | ????-0.12 | ????-0.16 | ????-0.17 | ????-0.18 | ????-0.15 | ????0.18 | ????0.94 | ????1.00 | ????0.97 |
????14 | ????-0.11 | ????-0.11 | ????-0.06 | ????-0.14 | ????-0.19 | ????-0.20 | ????-0.20 | ????-0.17 | ????0.17 | ????0.95 | ????0.97 | ????1.00 |
????15 | ????-0.12 | ????-0.12 | ????-0.07 | ????-0.16 | ????-0.20 | ????-0.21 | ????-0.22 | ????-0.19 | ????0.15 | ????0.95 | ????0.97 | ????0.97 |
????16 | ????-0.09 | ????-0.11 | ????-0.05 | ????-0.14 | ????-0.19 | ????-0.20 | ????-0.21 | ????-0.18 | ????0.16 | ????0.97 | ????0.98 | ????0.98 |
????18 | ????0.51 | ????0.61 | ????0.59 | ????0.61 | ????0.63 | ????0.65 | ????0.66 | ????0.68 | ????0.63 | ????-0.02 | ????-0.01 | ????0.00 |
????19 | ????0.49 | ????0.58 | ????0.56 | ????0.58 | ????0.59 | ????0.60 | ????0.61 | ????0.60 | ????0.53 | ????-0.07 | ????-0.06 | ????-0.07 |
????20 | ????0.55 | ????0.61 | ????0.60 | ????0.60 | ????0.61 | ????0.63 | ????0.63 | ????0.64 | ????0.65 | ????0.15 | ????0.16 | ????0.15 |
????21 | ????0.53 | ????0.60 | ????0.59 | ????0.59 | ????0.60 | ????0.62 | ????0.63 | ????0.64 | ????0.67 | ????0.19 | ????0.21 | ????0.21 |
????24 | ????0.67 | ????0.76 | ????0.74 | ????0.76 | ????0.79 | ????0.81 | ????0.82 | ????0.81 | ????0.71 | ????-0.16 | ????-0.15 | ????-0.16 |
????26 | ????0.65 | ????0.72 | ????0.70 | ????0.71 | ????0.73 | ????0.75 | ????0.76 | ????0.76 | ????0.68 | ????-0.07 | ????-0.08 | ????-0.09 |
????27 | ????0.66 | ????0.72 | ????0.71 | ????0.72 | ????0.74 | ????0.76 | ????0.77 | ????0.76 | ????0.69 | ????-0.07 | ????-0.08 | ????-0.09 |
????28 | ????0.65 | ????0.70 | ????0.69 | ????0.70 | ????0.71 | ????0.73 | ????0.73 | ????0.73 | ????0.65 | ????-0.07 | ????-0.08 | ????-0.10 |
????29 | ????0.65 | ????0.69 | ????0.69 | ????0.68 | ????0.70 | ????0.71 | ????0.72 | ????0.71 | ????0.65 | ????-0.03 | ????-0.04 | ????-0.07 |
????30 | ????0.58 | ????0.64 | ????0.64 | ????0.64 | ????0.65 | ????0.67 | ????0.67 | ????0.67 | ????0.63 | ????0.03 | ????0.03 | ????0.01 |
????31 | ????0.65 | ????0.70 | ????0.69 | ????0.69 | ????0.71 | ????0.72 | ????0.73 | ????0.72 | ????0.65 | ????-0.06 | ????-0.07 | ????-0.09 |
Continuous table 1
????15 | ????16 | ????18 | ????19 | ????20 | ????21 | ????24 | ????26 | ????27 | ????28 | ????29 | ????30 | ????31 | |
????2 | ????-0.12 | ????-0.09 | ????0.51 | ????0.49 | ????0.55 | ????0.53 | ????0.67 | ????0.65 | ????0.66 | ????0.65 | ????0.65 | ????0.58 | ????0.65 |
????3 | ????-0.12 | ????-0.11 | ????0.61 | ????0.58 | ????0.61 | ????0.60 | ????0.76 | ????0.72 | ????0.72 | ????0.70 | ????0.69 | ????0.64 | ????0.70 |
????4 | ????-0.07 | ????-0.05 | ????0.59 | ????0.56 | ????0.60 | ????0.59 | ????0.74 | ????0.70 | ????0.71 | ????0.69 | ????0.69 | ????0.64 | ????0.69 |
????5 | ????-0.16 | ????-0.14 | ????0.61 | ????0.58 | ????0.60 | ????0.59 | ????0.76 | ????0.71 | ????0.72 | ????0.70 | ????0.68 | ????0.64 | ????0.69 |
????6 | ????-0.20 | ????-0.19 | ????0.63 | ????0.59 | ????0.61 | ????0.60 | ????0.79 | ????0.73 | ????0.74 | ????0.71 | ????0.70 | ????0.65 | ????0.71 |
????7 | ????-0.21 | ????-0.20 | ????0.65 | ????0.60 | ????0.63 | ????0.62 | ????0.81 | ????0.75 | ????0.76 | ????0.73 | ????0.71 | ????0.67 | ????0.72 |
????8 | ????-0.22 | ????-0.21 | ????0.66 | ????0.61 | ????0.63 | ????0.63 | ????0.82 | ????0.76 | ????0.77 | ????0.73 | ????0.72 | ????0.67 | ????0.73 |
????10 | ????-0.19 | ????-0.18 | ????0.68 | ????0.60 | ????0.64 | ????0.64 | ????0.81 | ????0.76 | ????0.76 | ????0.73 | ????0.71 | ????0.67 | ????0.72 |
????11 | ????0.15 | ????0.16 | ????0.63 | ????0.53 | ????0.65 | ????0.67 | ????0.71 | ????0.68 | ????0.69 | ????0.65 | ????0.65 | ????0.63 | ????0.65 |
????12 | ????0.95 | ????0.97 | ????-0.02 | ????-0.07 | ????0.15 | ????0.19 | ????-0.16 | ????-0.07 | ????-0.07 | ????-0.07 | ????-0.03 | ????0.03 | ????-0.06 |
????13 | ????0.97 | ????0.98 | ????-0.01 | ????-0.06 | ????0.16 | ????0.21 | ????-0.15 | ????-0.08 | ????-0.08 | ????-0.08 | ????-0.04 | ????0.03 | ????-0.07 |
????14 | ????0.97 | ????0.98 | ????0.00 | ????-0.07 | ????0.15 | ????0.21 | ????-0.16 | ????-0.09 | ????-0.09 | ????-0.10 | ????-0.07 | ????0.01 | ????-0.09 |
????15 | ????1.00 | ????0.98 | ????-0.02 | ????-0.08 | ????0.14 | ????0.19 | ????-0.18 | ????-0.11 | ????-0.11 | ????-0.11 | ????-0.08 | ????0.00 | ????-0.10 |
????16 | ????0.98 | ????1.00 | ????-0.03 | ????-0.08 | ????0.14 | ????0.19 | ????-0.17 | ????-0.10 | ????-0.10 | ????-0.10 | ????-0.06 | ????0.01 | ????-0.09 |
????18 | ????-0.02 | ????-0.03 | ????1.00 | ????0.74 | ????0.79 | ????0.82 | ????0.80 | ????0.77 | ????0.79 | ????0.76 | ????0.75 | ????0.74 | ????0.76 |
????19 | ????-0.08 | ????-0.08 | ????0.74 | ????1.00 | ????0.78 | ????0.75 | ????0.68 | ????0.67 | ????0.68 | ????0.69 | ????0.65 | ????0.69 | ????0.69 |
????20 | ????0.14 | ????0.14 | ????0.79 | ????0.78 | ????1.00 | ????0.91 | ????0.70 | ????0.72 | ????0.74 | ????0.73 | ????0.72 | ????0.76 | ????0.74 |
????21 | ????0.19 | ????0.19 | ????0.82 | ????0.75 | ????0.91 | ????1.00 | ????0.71 | ????0.72 | ????0.74 | ????0.73 | ????0.71 | ????0.76 | ????0.73 |
????24 | ????-0.18 | ????-0.17 | ????0.80 | ????0.68 | ????0.70 | ????0.71 | ????1.00 | ????0.86 | ????0.87 | ????0.84 | ????0.84 | ????0.76 | ????0.84 |
????26 | ????-0.11 | ????-0.10 | ????0.77 | ????0.67 | ????0.72 | ????0.72 | ????0.86 | ????1.00 | ????0.88 | ????0.84 | ????0.85 | ????0.76 | ????0.84 |
????27 | ????-0.11 | ????-0.10 | ????0.79 | ????0.68 | ????0.74 | ????0.74 | ????0.87 | ????0.88 | ????1.00 | ????0.86 | ????0.88 | ????0.78 | ????0.88 |
????28 | ????-0.11 | ????-0.10 | ????0.76 | ????0.69 | ????0.73 | ????0.73 | ????0.84 | ????0.84 | ????0.86 | ????1.00 | ????0.86 | ????0.79 | ????0.97 |
????29 | ????-0.08 | ????-0.06 | ????0.75 | ????0.65 | ????0.72 | ????0.71 | ????0.84 | ????0.85 | ????0.88 | ????0.86 | ????1.00 | ????0.80 | ????0.87 |
????30 | ????0.00 | ????0.01 | ????0.74 | ????0.69 | ????0.76 | ????0.76 | ????0.76 | ????0.76 | ????0.78 | ????0.79 | ????0.80 | ????1.00 | ????0.78 |
????31 | ????-0.10 | ????-0.09 | ????0.76 | ????0.69 | ????0.74 | ????0.73 | ????0.84 | ????0.84 | ????0.88 | ????0.97 | ????0.87 | ????0.78 | ????1.00 |
The spectrum separability of table 2 test block part atural object
Annotate: √ represents that atural object can divide at this wave band
Take all factors into consideration the band characteristic of atural object and the correlativity of each wave band data, select wave band 2,6,11,12,19,24 and 30 to participate in classification.
Select training area
Selecting to represent the typical sample district of its characteristic according to the spectrum characteristic of every class atural object is training area.Consider the existence of the different spectrum of jljl, more than one of the training area of selected every class atural object.
Classification
Adopt maximum likelihood method to classify
B, based on standard spectrum database classification
Set up test site main atural object standard spectrum database
The database file form that the standard spectrum database is transferred to PCI software
Transfer high spectrum image to reflectance value by gray-scale value according to calibration coefficient
With the reflectance curve of pixel on the high spectrum image and standard spectrum curve ratio, the ownership class of each pixel is judged in pointwise, obtains classification chart.
C, comprehensive classification
More than various sorting techniques advantage is respectively arranged, but exist not enough simultaneously.Visual interpretation method nicety of grading height, but the decipher result is subjected to decipher personnel specialty quality influence bigger, and spended time is more, though supervised classification speed is fast, but, can't solve the problem of the different spectrum of jljl, same spectrum foreign matter, so nicety of grading is restricted owing to be a kind of pure spectrum sorting technique; Based on the classification of standard spectrum database, must with the spectra database of abundance prerequisite, simultaneously the pre-service of image, the quality of image there are higher requirement.Based on above consideration, we propose comprehensive classification.So-called comprehensive classification organically combines various classification exactly, is satisfying under the prerequisite of precision, saves time for asking, laborsaving, practical, reaches whole structure the best.
Concrete steps are:
For emphasis atural object, classification under the standard spectrum database is supported
The atural object that only depends on spectrum just can separate is classified with supervised classification
On the basis of two subseries,, other method is difficult for atural object decipher separately comes out in the above by man-machine interaction.
Do to guarantee cartographic accuracy like this, the classification time is significantly reduced.
More than the preferred embodiments of the present invention are described in detail for illustrative purposes; but those of ordinary skill in the art is to be appreciated that; in scope and spirit of the present invention; various improvement, interpolation and replacement all are possible, and all in the protection domain that claim of the present invention limited.
Claims (10)
1. the agriculture application integration method of an earth observation technology comprises:
Obtain the earth observation data;
The earth observation data that obtained are carried out the data band selection;
Earth observation data after described band selection are carried out feature extraction, obtain agricultural feelings parameter;
Utilize the agricultural feelings parameter of being obtained to carry out signature analysis, obtain the information needed in the described earth observation data.
2. method according to claim 1, it is characterized in that, described earth observation data are high-spectral data, described method further comprises carries out pretreated step to high-spectral data, and described pre-treatment step comprises one or more following processing: radiant correction, geometry correction, rebuilding spectrum, image transformation.
3. method according to claim 2, it is characterized in that, the step of described feature extraction comprises the compound step of obtaining the variety classes terrestrial information by multi-source data, and described multi-source data is compound to carry out figure stacked Calais realization by described high-spectral data and terrestrial information.
4. method according to claim 2 is characterized in that, described image transformation is grouping KL conversion.
5. method according to claim 2 is characterized in that, described band selection comprises:
Wave band to described high-spectral data carries out band grouping; With
Utilize institute to divide each batch total calculation band index;
Wherein, establish ρ
IjBe the related coefficient between wave band i and the j, imaging spectrometer data is divided into the k group, and every group wave band number is respectively n
1, n
2N
k, the definition band index is:
R
i=R
w+R
a
σ in the formula
iBe the mean square deviation of i wave band, R
wBe the mean value of the absolute value sum of other wave band related coefficient in i wave band and the place group, Ra is the absolute value sum of the related coefficient between i wave band and place group other wave band in addition.
6. method according to claim 1 is characterized in that, described agricultural feelings parameter comprises chlorophyll information, protein information and wheat moisture content information, and described relevant Agricultural Information comprises the wheat growth information; Perhaps
Described agricultural feelings parameter information is a vegetation index, and described relevant Agricultural Information is a crop yield trend.
7. the agriculture application integrating system of an earth observation technology comprises:
Data storage cell comprises:
Spectra database, it is with the data mode storage different crops of vector graphics or the curve of spectrum of ground object target;
The Multi-Band Remote Sensing Images that airborne imaging spectrometer obtains is stored with the data mode of grating image in the remotely-sensed data storehouse;
Model unit, comprise selection model storehouse and agricultural feelings inverse model storehouse, described selection model storehouse is used for image is carried out rebuilding spectrum, the analysis of data compound characteristics and band selection, and described agricultural feelings inverse model storehouse is used to provide polytype agricultural feelings information model;
Control and operation processing unit are used to utilize the model of described model unit that view data is handled accordingly; With
Information output unit is used for according to the agricultural feelings parameter of being extracted, the Agricultural Information that demonstration or output otherwise are correlated with.
8. the agriculture application integrating system of earth observation technology according to claim 7 is characterized in that, further comprises basic database, is used for storing complementary geographical spatial data and the attribute data that is complementary with the survey region remote sensing images.
9. the agriculture application integrating system of earth observation technology according to claim 7, it is characterized in that, figure, view data that described complementary geographical spatial data is grid or vector form, described attribute data is the statistics of text or form, the regional Administrative boundaries that comprise vector quantization, meteorological element, soil cover type map.
10. the agriculture application integrating system of earth observation technology according to claim 7, it is characterized in that described agricultural feelings information model comprises: high spectrum crop physicochemical property extraction model, crop growing state dynamic monitoring model, high spectrum flying quality and existing spaceborne data composite model and imaging spectrometer selection model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200410029826 CN1677085A (en) | 2004-03-29 | 2004-03-29 | Agricultural application integrating system for earth observation technique and its method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200410029826 CN1677085A (en) | 2004-03-29 | 2004-03-29 | Agricultural application integrating system for earth observation technique and its method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN1677085A true CN1677085A (en) | 2005-10-05 |
Family
ID=35049726
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200410029826 Pending CN1677085A (en) | 2004-03-29 | 2004-03-29 | Agricultural application integrating system for earth observation technique and its method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1677085A (en) |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1794286A (en) * | 2006-01-09 | 2006-06-28 | 江苏省农业科学院 | Remote sensing estimation method allowance of crop multimetadata crop rotation cycle |
CN102385694A (en) * | 2010-09-06 | 2012-03-21 | 邬明权 | Hyperspectral identification method for land parcel-based crop variety |
CN102426153A (en) * | 2011-11-21 | 2012-04-25 | 南京农业大学 | Wheat plant moisture monitoring method based on canopy high spectral index |
CN101477036B (en) * | 2009-01-13 | 2012-06-06 | 中国科学院遥感应用研究所 | Inland water chlorophyll a concentration remote-sensing monitoring method based on segmenting cooperation model |
CN101960291B (en) * | 2008-02-27 | 2012-08-22 | 亚洲大学校产学协力团 | Method for realtime target detection based on reduced complexity hyperspectral processing |
CN102706876A (en) * | 2012-04-28 | 2012-10-03 | 中国神华能源股份有限公司 | Method and device for monitoring solid pollution source region as well as data processing equipment |
CN102855495A (en) * | 2012-08-22 | 2013-01-02 | 苏州多捷电子科技有限公司 | Method for implementing electronic edition standard answer, and application system thereof |
CN103761674A (en) * | 2014-01-27 | 2014-04-30 | 林兴志 | Crop growing period alarming and intervening method based on remote sensing and mass climate information |
CN104266982A (en) * | 2014-09-04 | 2015-01-07 | 浙江托普仪器有限公司 | Large-area insect pest quantization monitoring system |
CN104703464A (en) * | 2013-07-17 | 2015-06-10 | 吴炎东 | Method for stimulating plant growth, apparatus and methods for computing cumulative light quantity |
CN105067116A (en) * | 2015-07-15 | 2015-11-18 | 北京农业信息技术研究中心 | Method and system for splicing frame imaging spectral data |
CN105277491A (en) * | 2015-09-24 | 2016-01-27 | 中国农业科学院农业资源与农业区划研究所 | Chlorophyll content measurement method and apparatus thereof |
CN105654524A (en) * | 2014-11-11 | 2016-06-08 | 孙义 | Method for establishing land data spectrum library |
CN105678281A (en) * | 2016-02-04 | 2016-06-15 | 中国农业科学院农业资源与农业区划研究所 | Plastic film mulching farmland remote sensing monitoring method based on spectrum and texture features |
CN105678280A (en) * | 2016-02-04 | 2016-06-15 | 中国农业科学院农业资源与农业区划研究所 | Plastic film mulching farmland remote sensing monitoring method based on texture features |
CN105758806A (en) * | 2016-02-04 | 2016-07-13 | 中国农业科学院农业资源与农业区划研究所 | Spectral characteristic based remote sensing monitoring method of plastic film mulched farmland |
CN108399355A (en) * | 2017-02-08 | 2018-08-14 | 广东交通职业技术学院 | A kind of hyperspectral image classification method that spatial information adaptively merges |
CN108759903A (en) * | 2018-04-02 | 2018-11-06 | 深圳万智联合科技有限公司 | A kind of quick electrical equipment malfunction detecting system of detection |
CN111145235A (en) * | 2019-12-26 | 2020-05-12 | 长光禹辰信息技术与装备(青岛)有限公司 | Crop medicine spraying method, device, equipment and computer readable storage medium |
-
2004
- 2004-03-29 CN CN 200410029826 patent/CN1677085A/en active Pending
Cited By (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1794286B (en) * | 2006-01-09 | 2014-09-24 | 江苏省农业科学院 | Remote sensing estimation method allowance of crop multimetadata crop rotation cycle |
CN1794286A (en) * | 2006-01-09 | 2006-06-28 | 江苏省农业科学院 | Remote sensing estimation method allowance of crop multimetadata crop rotation cycle |
CN101960291B (en) * | 2008-02-27 | 2012-08-22 | 亚洲大学校产学协力团 | Method for realtime target detection based on reduced complexity hyperspectral processing |
CN101477036B (en) * | 2009-01-13 | 2012-06-06 | 中国科学院遥感应用研究所 | Inland water chlorophyll a concentration remote-sensing monitoring method based on segmenting cooperation model |
CN102385694A (en) * | 2010-09-06 | 2012-03-21 | 邬明权 | Hyperspectral identification method for land parcel-based crop variety |
CN102426153A (en) * | 2011-11-21 | 2012-04-25 | 南京农业大学 | Wheat plant moisture monitoring method based on canopy high spectral index |
CN102426153B (en) * | 2011-11-21 | 2015-09-16 | 南京农业大学 | A kind of Wheat plant moisture monitoring method based on canopy high spectral index |
CN102706876B (en) * | 2012-04-28 | 2014-12-31 | 中国神华能源股份有限公司 | Method and device for monitoring solid pollution source region as well as data processing equipment |
CN102706876A (en) * | 2012-04-28 | 2012-10-03 | 中国神华能源股份有限公司 | Method and device for monitoring solid pollution source region as well as data processing equipment |
CN102855495A (en) * | 2012-08-22 | 2013-01-02 | 苏州多捷电子科技有限公司 | Method for implementing electronic edition standard answer, and application system thereof |
CN102855495B (en) * | 2012-08-22 | 2015-06-24 | 南京蒙渡电子科技有限公司 | Method for implementing electronic edition standard answer, and application system thereof |
CN104703464A (en) * | 2013-07-17 | 2015-06-10 | 吴炎东 | Method for stimulating plant growth, apparatus and methods for computing cumulative light quantity |
CN103761674A (en) * | 2014-01-27 | 2014-04-30 | 林兴志 | Crop growing period alarming and intervening method based on remote sensing and mass climate information |
CN103761674B (en) * | 2014-01-27 | 2016-11-23 | 林兴志 | The crop growth phase based on remote sensing with magnanimity climatic information alerts and interference method |
CN104266982B (en) * | 2014-09-04 | 2017-03-15 | 浙江托普仪器有限公司 | A kind of large area insect pest quantifies monitoring system |
CN104266982A (en) * | 2014-09-04 | 2015-01-07 | 浙江托普仪器有限公司 | Large-area insect pest quantization monitoring system |
CN105654524A (en) * | 2014-11-11 | 2016-06-08 | 孙义 | Method for establishing land data spectrum library |
CN105067116A (en) * | 2015-07-15 | 2015-11-18 | 北京农业信息技术研究中心 | Method and system for splicing frame imaging spectral data |
CN105277491A (en) * | 2015-09-24 | 2016-01-27 | 中国农业科学院农业资源与农业区划研究所 | Chlorophyll content measurement method and apparatus thereof |
CN105678280A (en) * | 2016-02-04 | 2016-06-15 | 中国农业科学院农业资源与农业区划研究所 | Plastic film mulching farmland remote sensing monitoring method based on texture features |
CN105758806A (en) * | 2016-02-04 | 2016-07-13 | 中国农业科学院农业资源与农业区划研究所 | Spectral characteristic based remote sensing monitoring method of plastic film mulched farmland |
CN105678281A (en) * | 2016-02-04 | 2016-06-15 | 中国农业科学院农业资源与农业区划研究所 | Plastic film mulching farmland remote sensing monitoring method based on spectrum and texture features |
CN105678280B (en) * | 2016-02-04 | 2020-06-16 | 中国农业科学院农业资源与农业区划研究所 | Mulching film mulching farmland remote sensing monitoring method based on textural features |
CN105678281B (en) * | 2016-02-04 | 2020-06-16 | 中国农业科学院农业资源与农业区划研究所 | Remote sensing monitoring method for mulching film farmland based on spectrum and texture characteristics |
CN108399355A (en) * | 2017-02-08 | 2018-08-14 | 广东交通职业技术学院 | A kind of hyperspectral image classification method that spatial information adaptively merges |
CN108399355B (en) * | 2017-02-08 | 2022-01-14 | 广东交通职业技术学院 | Hyperspectral image classification method based on spatial information adaptive fusion |
CN108759903A (en) * | 2018-04-02 | 2018-11-06 | 深圳万智联合科技有限公司 | A kind of quick electrical equipment malfunction detecting system of detection |
CN111145235A (en) * | 2019-12-26 | 2020-05-12 | 长光禹辰信息技术与装备(青岛)有限公司 | Crop medicine spraying method, device, equipment and computer readable storage medium |
CN111145235B (en) * | 2019-12-26 | 2023-10-31 | 长光禹辰信息技术与装备(青岛)有限公司 | Crop drug spraying method, device, equipment and computer readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1677085A (en) | Agricultural application integrating system for earth observation technique and its method | |
Li et al. | Multi-LUTs method for canopy nitrogen density estimation in winter wheat by field and UAV hyperspectral | |
Wulder et al. | High spatial resolution remotely sensed data for ecosystem characterization | |
Fan et al. | Hyperspectral-based estimation of leaf nitrogen content in corn using optimal selection of multiple spectral variables | |
Dash et al. | Land cover classification using multi‐temporal MERIS vegetation indices | |
Apan et al. | Formulation and assessment of narrow-band vegetation indices from EO-1 Hyperion imagery for discriminating sugarcane disease | |
CN111144250B (en) | Land coverage classification method integrating radar and optical remote sensing data | |
CN102609944B (en) | Hyper-spectral remote sensing image mixed pixel decomposition method based on distance geometry theory | |
Henkel et al. | Cantharellaceae of Guyana II: New species of Craterellus, new South American distribution records for Cantharellus guyanensis and Craterellus excelsus, and a key to the Neotropical taxa | |
Buyck et al. | Cantharellus texensis sp. nov. from Texas, a southern lookalike of C. cinnabarinus revealed by tef-1 sequence data | |
Guo et al. | Partitioning beta diversity in a tropical karst seasonal rainforest in Southern China | |
CN108830312A (en) | A kind of integrated learning approach adaptively expanded based on sample | |
CN111626224B (en) | Near infrared spectrum and SSA optimization-based ELM (enzyme-linked immunosorbent assay) quick coal gangue identification method | |
Wagner et al. | Validation scheme for solar coronal models: Constraints from multi-perspective observations in EUV and white light | |
CN113887493A (en) | Black and odorous water body remote sensing image identification method based on ID3 algorithm | |
Rosales-Ortega et al. | Integrated spectra extraction based on signal-to-noise optimization using integral field spectroscopy | |
Bossung et al. | Estimation of canopy nitrogen content in winter wheat from Sentinel-2 images for operational agricultural monitoring | |
Li et al. | Improving estimation of forest aboveground biomass using Landsat 8 imagery by incorporating forest crown density as a dummy variable | |
Pereira Martins-Neto et al. | Tree Species Classification in a Complex Brazilian Tropical Forest Using Hyperspectral and LiDAR Data | |
Zhang et al. | UAV mission height effects on wheat lodging ratio detection | |
Kangas et al. | Re-calibrating stem volume models–is there change in the tree trunk form from the 1970s to the 2010s in Finland? | |
Mirandilla et al. | Leaf Spectral Analysis for Detection and Differentiation of Three Major Rice Diseases in the Philippines | |
CN115063610B (en) | Soybean planting area identification method based on Sentinel-1 and 2 images | |
Li et al. | A technique system for the measurement, reconstruction and character extraction of rice plant architecture | |
Simic et al. | Testing the top-down model inversion method of estimating leaf reflectance used to retrieve vegetation biochemical content within empirical approaches |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
WD01 | Invention patent application deemed withdrawn after publication |