CN106650811A - Hyperspectral mixed pixel classification method based on neighbor cooperation enhancement - Google Patents
Hyperspectral mixed pixel classification method based on neighbor cooperation enhancement Download PDFInfo
- Publication number
- CN106650811A CN106650811A CN201611218275.8A CN201611218275A CN106650811A CN 106650811 A CN106650811 A CN 106650811A CN 201611218275 A CN201611218275 A CN 201611218275A CN 106650811 A CN106650811 A CN 106650811A
- Authority
- CN
- China
- Prior art keywords
- class
- classification
- atural object
- pixel
- neighbour
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
Abstract
The invention provides a hyperspectral mixed pixel classification method based on neighbor cooperation enhancement, and the method comprises the steps: calculating a spectrum signature matrix of a plurality of target ground features through marked sample ground features; designing a multi-class classifier based on spectrum characteristics, and carrying out the classification of the ground features; carrying out the fusion of spatial structure features in a classification result, and extracting neighbor pixels; carrying out the class marking of unmarked hyperspectral ground features through the neighbor pixels; carrying out the classification and marking of the unmarked hyperspectral ground features through an interaction method; carrying out the further fusion of the spatial features of the target ground features in a mode of neighbor expansion, and completing the final classification and marking. According to the invention, the multi-class classifier is used for the simultaneous classification of ground features, and a problem that a conventional classification method cannot carry out the classification of background features is solved. Moreover, a mode of neighbor cooperation enhancement is employed for marking the unmarked ground objects step by step, thereby achieving the effective fusion of the spectrum features and spatial features of the ground features. The classification effect is good.
Description
Technical field
Classification hyperspectral imagery technical field of the present invention, more particularly to it is a kind of based on the enhanced EO-1 hyperion mixing of neighbour's collaboration
Pixel sorting technique.
Background technology
Classification hyperspectral imagery as Hyperspectral imagery processing in an important application, its final target is to image
In each pixel carry out the ownership of classification.High spectrum resolution remote sensing technique makes it in atural object category classification using more spectral bands
Aspect has big advantage, but the accuracy of object spectrum information also causes interference, background parts to have in hyperspectral classification
Certain impact；On the other hand, the characteristics of there is big high dimensional data amount and little training sample due to high-spectral data, when making classification
It is also easy to produce Hughes phenomenons.
The sorting technique that in recent years the empty spectrum signature of high spectrum image is combined is taken seriously, based on the m- space characteristics of spectrum
Hyperspectral image classification method has become current study hotspot, and this kind of method is by combining space information feature and Spectral Properties
Levy, improve the precision of classification hyperspectral imagery.At present major applications be carried out using methods such as SVMs it is one-to-one, or
The classification of person's one-to-many；The problem of one maximum of this kind of method is that the background parts of high spectrum image cannot be classified,
Therefore generally all it is using removing by the way of background, cause such method to lack accuracy in classification of assessment method；In addition,
This kind of method is classified using pure unit, causes the limitation of sorting technique, lacks versatility.
The content of the invention
The present invention provides a kind of based on the enhanced EO-1 hyperion mixed pixel sorting technique of neighbour's collaboration, solves above-mentioned technology and asks
Topic.
One kind of the invention cooperates with enhanced EO-1 hyperion mixed pixel sorting technique based on neighbour, including：
The spectrum signature matrix of multiple target atural object is calculated according to marked sample atural object；
The Target scalar is entered using the multi-class grader for being based on the spectrum signature character matrix and constraint matrix
Row classification；
Neighbour's pixel is extracted again after the Abundances fusion spatial structure characteristic that the grader is obtained；
The Target scalar to unmarked EO-1 hyperion is cooperateed with to carry out category label according to neighbour's pixel, using alternative manner
Progressively unlabelled atural object is classified respectively.
Further, the spectrum signature matrix that multiple target atural object is calculated according to marked sample atural object, including：
According to formula
The spectral signature vector of Target scalar is calculated, wherein, dkFor the spectral signature vector of kth class Target scalar, dk=
{dk1,dk2,...dkL, L be wave band number, { Cset(k) } it is marked kth class sample atural object set, NkFor { Cset(k) } in
Pixel number, HkJ () is { Cset(k) } in j-th pixel spectrum signature；
The spectrum signature matrix D of Target scalar end member, D=[d are calculated according to spectral signature vector1,d2,...dp],
Wherein, p is atural object species number to be sorted, d1Spectrum for first kind atural object is signed.
Further, the multi-class grader using based on the spectrum signature character matrix and constraint matrix is by institute
State Target scalar to be classified, including：
According to the class number p of the Target scalar, the constraint matrix of grader is set as C=[c1…cp], wherein, cj
For the constrained vector of j-th Target scalar, 1≤j≤p, for entering row constraint to jth class Target scalar；
Matrix is signed using the spectral signature and constraint matrix definition is carried out point simultaneously to p Target scalars
The multi-class grader T of class, the grader is：
Wherein, R is the sample autocorrelation matrix of high spectrum image, and the R is：
Wherein, r=[r1r2...rn]。
Further, spatial structure characteristic is merged in classification results, and extracts neighbour's pixel, including：
Using formula
Convolutional calculation is carried out to classification results and obtains TG(k), wherein, the σ is the standard deviation of gaussian filtering, and r is high
The filtration radius of this filtering；
To TGK the abundance Value Data in () is ranked up, the individual pixel composition atural object classifications of the maximum 2*n (k) of extraction of values
Neighbour gathers { MCset(k) }, wherein, n (k) is the pixel number that kth class atural object increases every time mark newly.
Further, the employing neighbour collaboration strengthens, and unlabelled atural object is progressively carried out respectively classification annotation, including：
Defining similitude between the class of various Target scalars is
dis(Hk(j))=| | M (k)-Hk(j)||2 (5)
Wherein, M (k) is the sample clustering center of the kth class Target scalar for having marked classification, HkJ () is { MCset(k) } in
J-th neighbour's pixel；
Similitude between the class of atural object is calculated in neighbour's set in sample and marked set according to similarity criterion between class；
The individual pixels of the n (k) of similitude maximum between the class are labeled.
Further, after the employing alternative manner is progressively respectively classified unlabelled atural object, also include：
Local expansion is carried out to classification results using neighbor operator, morphology expansion, for strengthening the space of data atural object
It is regional.
It is of the invention a kind of based on the enhanced hyperspectral image classification method of neighbour's collaboration, by progressively right with the pixel for marking
Unlabelled pixel is marked, sharp by using partly the sample atural object of label has been calculated Target scalar spectral signature
Preliminary classification result is carried out with the multi-class grader of design, then classification results is merged into spatial structure characteristic, and using near
Unlabelled atural object is progressively carried out respectively the enhanced mode of neighbour's collaboration classification annotation, so as to reach the classification of EO-1 hyperion mixed pixel
Purpose.Cooperate with enhanced EO-1 hyperion mixed pixel sorting technique multi-class based on spectral signature by defining one based on neighbour
Grader can classify to all of target classification, and having effectively eliminated conventional sorting methods cannot be carried out to background atural object
The problem of classification, enhanced mode is cooperateed with method using neighbour, and is done step-by-step to unlabelled by merging space characteristics
Ground object target is marked, and classifying quality is preferable.
Description of the drawings
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are these
Some bright embodiments, for those of ordinary skill in the art, without having to pay creative labor, can be with
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 cooperates with enhanced EO-1 hyperion mixed pixel sorting technique flow chart for the present invention based on neighbour；
Fig. 2 cooperates with enhanced EO-1 hyperion mixed pixel sorting technique overall schematic for the present invention based on neighbour；
Fig. 3 a and Fig. 3 b are that the spatial structure characteristic of atural object to be sorted in the present invention extracts schematic diagram；
Fig. 4 a, Fig. 4 b and Fig. 4 c are that target to be sorted based on the enhanced pixel of neighbour marks schematic diagram in the present invention；
Fig. 5 is the neighborhood extending schematic diagram of target to be sorted in the present invention；
Fig. 6 a and Fig. 6 b is Purdue data classification results figures in the present invention；
Fig. 7 a and Fig. 7 b is Salinas data classification results figures in the present invention；
Fig. 8 a and Fig. 8 b is Pavia data classification results figures in the present invention.
Specific embodiment
To make purpose, technical scheme and the advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
The a part of embodiment of the present invention, rather than the embodiment of whole.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
Fig. 1 cooperates with enhanced EO-1 hyperion mixed pixel sorting technique flow chart, the present embodiment side for the present invention based on neighbour
Method, including：
Step 101, the spectrum signature matrix that multiple target atural object is calculated according to marked sample atural object；
Specifically, the present embodiment hyperspectral image data r=(r1,r2,...rn)T, wherein, n is high spectrum image
Pixel number, ri(1 ＜=i ＜=n) represents i-th pixel of high spectrum image, ri=(ri1,ri2,...riL), L represents bloom
The wave band number of spectrogram picture.
Known k is atural object 1 ＜ of class-mark=k ＜=p, NumkFor whole numbers of samples of k class atural objects, selected using random fashion
Taking the individual sample data of EO-1 hyperion n (k) carries out pixel mark, and constitutes marked set { Cset(k)}.Iterations is set as U=
Numk/ n (k), makes primary iteration number of times u=1.
According to marked set { Cset(k) }, NkFor { Cset(k) } in kth class atural object marked pixel number, according to public affairs
Formula (1) calculates the spectral vector d of ground object target to be sortedk：
Wherein, dkFor the spectral vector of Target scalar, { Cset(k) } be marked sample atural object set, NkFor { Cset(k)}
The marked pixel number of middle kth class atural object, HkJ () is { Cset(k) } in mark pixel spectrum signature.
Step 102, using the multi-class grader based on the spectrum signature character matrix and constraint matrix by the mesh
Mark atural object is classified；
Specifically, the spectral signature of EO-1 hyperion is its most important information, and the present embodiment grader is effective by design
The multi-class spectral signature of utilization grader identification earth's surface object.Specifically, according to the class number p of classification, generation is treated
The spectrum signature matrix D of class object end member1=[d1,d2,..dk...dp]；. based on spectrum statistical nature D, by arranging
Constraint matrix constrains all Target scalars, and various atural objects are classified simultaneously using FIR filter.The p atural object same time-division of design
The multi-class grader T of classkIt is as follows：
Wherein,Wherein Cp×p=[c1,c2,..cp] it is with 1 as cornerwise diagonal
Battle array, its column vector ciFor constraining i-th ground object target；R is the spectrum autocorrelation matrix of high spectrum image, and it is defined as：Wherein r=[r1r2...rn], on the one hand multi-class grader utilizes inverse matrix R-1High spectrum image
Sample spectrum constrained, reach the effect for weakening background, on the other hand can be realized simultaneously by constraint matrix D multiple
The other classification of species；
Neighbour's pixel is extracted again after step 103, the Abundances for obtaining grader fusion spatial structure characteristic；
Specifically, by high-spectral data after multi-class grader T, the classification results of every kind of atural object classification are obtained
T(k).There is substantial amounts of classification noise excessively based on the classification results of the grader T of spectral information, need to merge empty in classification results
Between feature strengthen the spatial information of classification atural object, so as to eliminate merely with the spectral signature noise problem brought of classification.Specifically
Way is to carry out convolutional calculation to classification results using gaussian filtering, and the classification " particle " of crossing to classification results is eliminated, its
Formula is：
Wherein σ is the standard deviation of gaussian filtering, and r is the filtration radius of gaussian filtering.
Step 104, cooperateed with according to neighbour's pixel category label is carried out to the Target scalar of unmarked EO-1 hyperion, adopted
Alternative manner is progressively respectively classified unlabelled atural object.
Specifically, the present embodiment is to improve labeling effciency and precision, using in classification results per the neighbour of class atural object
Pixel collaboration carries out ground substance markers.Extract neighbour's pixel specific practice be：To TGK the abundance Value Data of () is ranked up, carry
Take TGNeighbour set { the MC of the individual pixel composition atural object classifications of the maximum 2*n (k) of (k) intermediate valueset(k)}。
The mode cooperateed with using neighbour carries out the individual pixel marks of n (k) to unlabelled ground object target, and specific practice is：It is first
First define similitude between the class of every kind of atural object as follows：
dis(Hk(j))=| | D (k)-Hk(j)||2 (4)
{ MC is calculated using formula (4)set(k) } in sample and { Cset(k) } in ground object target class between similitude, by dis
Minimum n (k) samples of value carry out correspondence classification mark, and put it into marked set { Cset(k) } in.
When being marked to unlabelled pixel using marked sample, according to spectral signature and spatial structure characteristic phase
The mode of fusion selects neighbour's sample in classification results, then by select in neighbour's sample classification similitude it is maximum i.e. away from
It is marked from minimum pixel.Enhanced mode is cooperateed with by this neighbour, the mark precision of unmarked sample is improve, is had
Help the classification of high spectrum image.
U=u+1 is made, if u<U, goes to step B, and continuation carries out EO-1 hyperion mixed pixel mark using iterative manner, from
And progressively strengthen the number of marked sample.
The present invention is exemplified below cooperates with enhanced EO-1 hyperion mixed pixel sorting technique, sample data based on neighbour
Come from real high spectrum image：The high-spectral data of Indian Pine test blocks, hereinafter referred to as Purdue data.The data
It is the farmland image obtained in the state of Indiana northwestward by AVIRIS sensors, image size is 145 × 145, spatial discrimination
20 meters of rate, original wave band number is 220, including 16 class atural object classifications；Its pseudocolour picture and true terrestrial object information such as Fig. 3 a and
Shown in Fig. 3 b.
Iterations 10 times is set first, and the initial value of u is 1.Randomly select from the truly survey data of experimental data
10% marked sample, and constitute marked set k { Cset(k)},1<=k<=16, table 1 to table 3 is every kind of ground species
Not marked number.
Table 1
The spectrum signature calculation of sample atural object is carried out according to formula (1), and by the classification light of calculated three width image
Spectrum signature matrix D=[d1,d2,...d16]；
Constraint matrix C is set:C16×16, its column vector ciFor constraining i-th atural object classification of purdue；
Simultaneously D is constrained using constrained vector Matrix C, simultaneously can simultaneously be entered 16 class ground object targets in purdue data
Row classification；
According to DkWith constraint matrix C, according to formula (2), the multi-class grader T of definitionk(i)(1<=i<=16) calculate
The classification results of ground object target.
σ=1.5 are set, and r=11 extracts the spatial structure characteristic T of Purdue data classification results figures according to formula (3)
(iG)(1<=i<=16).As shown in Figure 3 a and Figure 3 b shows, wherein Fig. 3 a are the spatial extraction schematic diagram of the 2nd class of Purdue data
The Equations of The Second Kind first time classification results of corn-notil classifications, Fig. 3 b are the result figure after gaussian filtering, color depth therein
Superficial shows the height of pixel Abundances.It can be seen that after space characteristics filtering, in having effectively eliminated classification results
" particle " noise, the space characteristics of corn-notil atural objects are incorporated to suffer with spectral classification result.
To T (iG)(1<=i<=16) be ranked up, the individual neighbour's pixel of each class 2*n (k) is selected, set up neighbour's set
MCset。
Below with the classification 1 of Purdue data, alfalfa data instances are illustrated.The enhanced pixel of neighbour is marked such as
Shown in Fig. 4 a, Fig. 4 b and Fig. 4 c.The size of example image is (the 90 of artwork:111,60:81) part.Wherein Fig. 4 a are marked
Alfalfa class atural objects pixel, the specific random mark pixel collection for generating is combined into Cset, specific coordinate for (96,73)
(98,68)(100,73)(100,74)(101,73)；Neighbour's pixel of the Abundances sequence after spatial structure characteristic is merged
As shown in Fig. 4 b figures.Neighbour's pixel collection is combined into Mcset, concrete coordinate value for (101,73), (100,73), (101,74), (100,
74), (101,72), (100,72), (102,73), (99,73), (102,74), (99,72)；
Using formula (4) calculate neighbour set Mcset and Cset in marked alfalfa atural objects class between similitude, its
Specific value is as shown in table 2.
Table 2
Then the data of table 2 are ranked up, select 5 pixels (pixel number is 1,2,3,5,6) is marked, and by its
Marked set is put into update Cset.New mark alfalfa data are as illustrated in fig. 4 c.
To three width images using more than operate, can with by EO-1 hyperion atural object classification by neighbour cooperate with enhanced mode by
Step expands marked pixel number.
U=u+1 is made, if u<10, update all class ground object target matrix Ds according to formula (1)；Turn next to step E after
It is continuous to perform；When u=11 is reached, iteration stopping, neighbour's collaboration strengthens mark and completes, and the pixel in Cset set is every
The classification results of class atural object.
Finally the classification results of Purdue images are carried out with local expansion for the operator of 2*2 using Size of Neighborhood, strengthen number
According to the area of space of atural object, the schematic diagram of extension is as shown in Figure 5.Wherein Fig. 5 left figures are the atural object after neighbour's collaboration
Classification results, the result figure after the neighbor operator that size is 2*2 is shown in Fig. 5 right figures, it can be seen that expanded through neighborhood
After exhibition, the space characteristics of atural object have been characterized out, represent the form and size of atural object.One group of Purdue data is final
As shown in figures 6 a and 6b, Fig. 6 a are ground truth images to classification results, and Fig. 6 b are one group of classification results of the present invention.
Other two groups of real high-spectral datas, hereinafter referred to as Salinas data and Pavia numbers are also used in experiment
According to.Wherein：One group is by the high-spectrum in the Salinas mountain valleys obtained in California, USA southern areas using AVIRIS sensors
Picture, hereinafter referred to as Salinas data.The size of the image is 512 × 217, and spatial resolution is 3.7 meters, containing 224 ripples
Section, altogether including 16 class atural objects, its pseudocolour picture and true terrestrial object information as shown in Figure 3 a and Figure 3 b shows, the number such as table of initial Cset
Shown in 3.
Another group is the city district image Pavia obtained in Pavia universities overhead using ROSIS-03 sensors
University, hereinafter referred to as pavia data, the data image size is 610 × 340, and spatial resolution is 1.3 meters, is contained
103 wave bands, altogether including 9 class atural objects, its pseudocolour picture and true terrestrial object information as shown in Figure 3 a and Figure 3 b shows, the number of initial Cset
Mesh is as shown in table 4.
Table 3
Table 4
As shown in Fig. 7 a and Fig. 8 a, one group of final classification result is as schemed for the grountruth images of this two groups of group True Datas
Shown in 7b and Fig. 8 b.By more than based on neighbour cooperate with three groups of true classification results obtained by enhanced sorting technique with
The result of Groundtruth image labelings can significantly find out that the sorting technique that this invention is proposed is devised with Spectral Properties
All atural object classifications once can be classified by the multi-class grader based on levying, and effectively solve tradition with SVM
Based on the problem that image background cannot be classified of grader.
This invents proposed classification side to carry out quantitative and objective appraisal following with nicety of grading and kappa coefficients
Method, classification rate OA is defined as follows：
Wherein p be atural object classification number, SiFor the i-th class in classification results, ground truth are also the pixel of the i-th class atural object
Number, NiFor the number of samples of the i-th class atural object in Ground truth results.
The nicety of grading of every kind of ground object target classification, computing formula is as follows：
The computing formula of Kappa coefficients is：
Wherein：
N1iPresentation class result be divided into for the i-th class atural object be other atural objects pixel number；N2iRepresent other atural objects
Mistake is divided into the pixel number of the i-th class atural object.
Table 5,6,7 is the nicety of grading of three panel height spectrum pictures and the concrete numerical value of kappa coefficients.It can be seen that of the invention
Sorting technique pretty good nicety of grading is all achieved for every kind of atural object classification, while kappa coefficients are also higher, show this
The result of sorting technique is higher with real classification results uniformity.
Table 5
Table 6
Table 7
Present invention utilizes the multi-categorizer testing result of spectrum statistical nature design, by unlabelled sample neighbour is carried out
Strengthen mark from collaboration, be gradually completing mixed pixel classification.In order to improve nicety of grading in method, employing will compose between feature with it is empty
Between feature integration mode carry out EO-1 hyperion mixed pixel category feature judgement.Multi-class hyperspectral classification device is designed first,
After part sample is marked, by multi-class hyperspectral classification device, preliminary classification result is detected, then by testing result
It is compared with marked atural object, the minimum sample collaboration of chosen distance is labeled, and progressively strengthens marked sample
Number, then by updating the spectrum signature of Target scalar, using the mode of iteration the classification of high spectrum image is completed.
Finally it should be noted that：Various embodiments above only to illustrate technical scheme, rather than a limitation；To the greatest extent
Pipe has been described in detail with reference to foregoing embodiments to the present invention, it will be understood by those within the art that：Its according to
So the technical scheme described in foregoing embodiments can be modified, either which part or all technical characteristic are entered
Row equivalent；And these modifications or replacement, do not make the essence disengaging various embodiments of the present invention technology of appropriate technical solution
The scope of scheme.
Claims (6)
1. it is a kind of that enhanced EO-1 hyperion mixed pixel sorting technique is cooperateed with based on neighbour, it is characterised in that to include：
The spectrum signature matrix of multiple target atural object is calculated according to marked sample atural object；
The Target scalar is carried out point using the multi-class grader based on the spectrum signature character matrix and constraint matrix
Class；
Neighbour's pixel is extracted again after the Abundances fusion spatial structure characteristic that the grader is obtained；
Cooperate with the Target scalar to unmarked EO-1 hyperion to carry out category label according to neighbour's pixel, using alternative manner progressively
Unlabelled atural object is classified respectively.
2. method according to claim 1, it is characterised in that described that multiple target ground is calculated according to marked sample atural object
The spectrum signature matrix of thing, including：
According to formula
The spectral signature vector of Target scalar is calculated, wherein, dkFor the spectral signature vector of kth class Target scalar, dk={ dk1,
dk2,...dkL, L be wave band number, { Cset(k) } it is marked kth class sample atural object set, NkFor { Cset(k) } in picture
First number, HkJ () is { Cset(k) } in j-th pixel spectrum signature；
The spectrum signature matrix D of Target scalar end member, D=[d are calculated according to spectral signature vector1,d2,...dp], wherein,
P is atural object species number to be sorted, d1Spectrum for first kind atural object is signed.
3. method according to claim 1, it is characterised in that described using being based on the spectrum signature character matrix peace treaty
The multi-class grader of beam matrix is classified the Target scalar, including：
According to the class number p of the Target scalar, set the constraint matrix of grader asWherein, cjFor j-th
The constrained vector of Target scalar, 1≤j≤p, for entering row constraint to jth class Target scalar；
Define what the p Target scalar was classified simultaneously using spectral signature signature matrix and the constraint matrix
Multi-class grader T, the grader is：
Wherein, R is the sample autocorrelation matrix of high spectrum image, and the R is：
Wherein, r=[r1r2...rn]。
4. method according to claim 1, it is characterised in that merge spatial structure characteristic in classification results, and extract
Neighbour's pixel, including：
Using formula
Convolutional calculation is carried out to classification results and obtains TG(k), wherein, the σ is the standard deviation of gaussian filtering, and r is gaussian filtering
Filtration radius；
To TGK the abundance Value Data in () is ranked up, neighbour's collection of the individual pixel composition atural object classifications of the maximum 2*n (k) of extraction of values
Close { MCset(k) }, wherein, n (k) is the pixel number that kth class atural object increases every time mark newly.
5. method according to claim 1, it is characterised in that the employing neighbour collaboration strengthens progressively will be unlabelled
Thing carries out respectively classification annotation, including：
Defining similitude between the class of various Target scalars is
dis(Hk(j))=| | M (k)-Hk(j)||2 (5)
Wherein, M (k) is the sample clustering center of the kth class Target scalar for having marked classification, HkJ () is { MCset(k) } in jth
Individual neighbour's pixel；
Similitude between the class of atural object is calculated in neighbour's set in sample and marked set according to similarity criterion between class；
The individual pixels of the n (k) of similitude maximum between the class are labeled.
6. method according to claim 1, it is characterised in that the employing alternative manner is progressively by unlabelled atural object point
After not classified, also include：
Local expansion is carried out to classification results using neighbor operator, morphology expansion, for strengthening the area of space of data atural object
Property.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611218275.8A CN106650811B (en) | 2016-12-26 | 2016-12-26 | A kind of EO-1 hyperion mixed pixel classification method cooperateing with enhancing based on neighbour |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611218275.8A CN106650811B (en) | 2016-12-26 | 2016-12-26 | A kind of EO-1 hyperion mixed pixel classification method cooperateing with enhancing based on neighbour |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106650811A true CN106650811A (en) | 2017-05-10 |
| CN106650811B CN106650811B (en) | 2019-08-13 |
Family
ID=58827438
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611218275.8A Active CN106650811B (en) | 2016-12-26 | 2016-12-26 | A kind of EO-1 hyperion mixed pixel classification method cooperateing with enhancing based on neighbour |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106650811B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967694A (en) * | 2017-12-22 | 2018-04-27 | 大连海事大学 | A kind of EO-1 hyperion object detection method, system, storage medium and processor based on feedback abundance constraint |
| CN108171270A (en) * | 2018-01-05 | 2018-06-15 | 大连海事大学 | A kind of hyperspectral image classification method based on Hash study |
| CN108470192A (en) * | 2018-03-13 | 2018-08-31 | 广东工业大学 | A kind of hyperspectral classification method and device |
| CN110033039A (en) * | 2019-04-10 | 2019-07-19 | 大连海事大学 | A kind of EO-1 hyperion real-time grading method based on goal constraint detection |
| CN111739036A (en) * | 2020-07-22 | 2020-10-02 | 吉林大学 | Hyperspectrum-based file handwriting counterfeiting detection method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060251324A1 (en) * | 2004-09-20 | 2006-11-09 | Bachmann Charles M | Method for image data processing |
| CN101692125A (en) * | 2009-09-10 | 2010-04-07 | 复旦大学 | Fisher judged null space based method for decomposing mixed pixels of high-spectrum remote sensing image |
| CN103679210A (en) * | 2013-12-03 | 2014-03-26 | 西安电子科技大学 | Ground object recognition method based on hyperspectral image unmixing |
| CN104751191A (en) * | 2015-04-23 | 2015-07-01 | 重庆大学 | Sparse self-adaptive semi-supervised manifold learning hyperspectral image classification method |
-
2016
- 2016-12-26 CN CN201611218275.8A patent/CN106650811B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060251324A1 (en) * | 2004-09-20 | 2006-11-09 | Bachmann Charles M | Method for image data processing |
| CN101692125A (en) * | 2009-09-10 | 2010-04-07 | 复旦大学 | Fisher judged null space based method for decomposing mixed pixels of high-spectrum remote sensing image |
| CN103679210A (en) * | 2013-12-03 | 2014-03-26 | 西安电子科技大学 | Ground object recognition method based on hyperspectral image unmixing |
| CN104751191A (en) * | 2015-04-23 | 2015-07-01 | 重庆大学 | Sparse self-adaptive semi-supervised manifold learning hyperspectral image classification method |
Non-Patent Citations (2)
| Title |
|---|
| 余淞淞: "基于丰度划分的高光谱遥感图像解混", 《图形图像 现代计算机》 * |
| 倪鼎，马洪兵: "基于近邻协同的高光谱图像谱−空联合分类", 《自动化学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107967694A (en) * | 2017-12-22 | 2018-04-27 | 大连海事大学 | A kind of EO-1 hyperion object detection method, system, storage medium and processor based on feedback abundance constraint |
| CN108171270A (en) * | 2018-01-05 | 2018-06-15 | 大连海事大学 | A kind of hyperspectral image classification method based on Hash study |
| CN108171270B (en) * | 2018-01-05 | 2021-08-27 | 大连海事大学 | Hyperspectral image classification method based on Hash learning |
| CN108470192A (en) * | 2018-03-13 | 2018-08-31 | 广东工业大学 | A kind of hyperspectral classification method and device |
| CN108470192B (en) * | 2018-03-13 | 2022-04-19 | 广东工业大学 | Hyperspectral classification method and device |
| CN110033039A (en) * | 2019-04-10 | 2019-07-19 | 大连海事大学 | A kind of EO-1 hyperion real-time grading method based on goal constraint detection |
| CN110033039B (en) * | 2019-04-10 | 2022-10-11 | 大连海事大学 | Hyperspectral real-time classification method based on target constraint detection |
| CN111739036A (en) * | 2020-07-22 | 2020-10-02 | 吉林大学 | Hyperspectrum-based file handwriting counterfeiting detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106650811B (en) | 2019-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109753885B (en) | Target detection method and device and pedestrian detection method and system | |
| CN106650811A (en) | Hyperspectral mixed pixel classification method based on neighbor cooperation enhancement | |
| Huang et al. | A new building extraction postprocessing framework for high-spatial-resolution remote-sensing imagery | |
| CN105389550B (en) | It is a kind of based on sparse guide and the remote sensing target detection method that significantly drives | |
| CN103049763B (en) | Context-constraint-based target identification method | |
| CN105809198B (en) | SAR image target recognition method based on depth confidence network | |
| CN103577475B (en) | A kind of picture mechanized classification method, image processing method and its device | |
| CN106446933B (en) | Multi-target detection method based on contextual information | |
| CN104239902B (en) | Hyperspectral image classification method based on non local similitude and sparse coding | |
| CN106845418A (en) | A kind of hyperspectral image classification method based on deep learning | |
| CN106529508A (en) | Local and non-local multi-feature semantics-based hyperspectral image classification method | |
| CN104299009B (en) | License plate character recognition method based on multi-feature fusion | |
| CN104778476B (en) | A kind of image classification method | |
| CN109409384A (en) | Image-recognizing method, device, medium and equipment based on fine granularity image | |
| CN105139028A (en) | SAR image classification method based on hierarchical sparse filtering convolutional neural network | |
| CN105574527A (en) | Quick object detection method based on local feature learning | |
| CN106339674A (en) | Hyperspectral image classification method based on edge preservation and graph cut model | |
| CN105574063A (en) | Image retrieval method based on visual saliency | |
| CN105513066B (en) | It is a kind of that the generic object detection method merged with super-pixel is chosen based on seed point | |
| CN103778443B (en) | Method for achieving scene analysis description based on theme model method and field rule library | |
| CN109948593A (en) | Based on the MCNN people counting method for combining global density feature | |
| CN106846322A (en) | Based on the SAR image segmentation method that curve wave filter and convolutional coding structure learn | |
| CN107527054A (en) | Prospect extraction method based on various visual angles fusion | |
| CN107767416A (en) | The recognition methods of pedestrian's direction in a kind of low-resolution image | |
| CN108090485A (en) | Display foreground extraction method based on various visual angles fusion |
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 |