CN102110227A - Compound method for classifying multiresolution remote sensing images based on context - Google Patents
Compound method for classifying multiresolution remote sensing images based on context Download PDFInfo
- Publication number
- CN102110227A CN102110227A CN 201010560334 CN201010560334A CN102110227A CN 102110227 A CN102110227 A CN 102110227A CN 201010560334 CN201010560334 CN 201010560334 CN 201010560334 A CN201010560334 A CN 201010560334A CN 102110227 A CN102110227 A CN 102110227A
- Authority
- CN
- China
- Prior art keywords
- resolution
- low
- remote sensing
- classification
- random field
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to a compound method for classifying multiresolution remote sensing images based on context. The method comprises the following steps of: firstly, registering in a local training area; secondly extracting classification characteristics from a low-resolution image; then establishing a context based on a conditional random field model by utilizing the classification characteristics in a training area based on the previous two steps; and finally, carrying out global classification on the conditional random field model obtained by the previous three steps, popularizing the trained model to the entire coverage area of low-resolution images, and classifying the wide area low-resolution images. In the invention, the multi-resolution remote sensing images are comprehensively utilized, the context between pixels is constructed, the space continuity of ground object distribution is considered, and the conditional random field model provides support for multi-classification characteristics, thus the high-accuracy classification problem of wide-area low-resolution remote sensing images is solved.
Description
Technical field
The invention belongs to pattern-recognition and computer vision field, also relate to remote sensing and agriculture field, be specifically related to multiresolution remote sensing images combined entry method based on context relation.
Background technology
Face of land cover classification is to obtain the basic technology of soil covering and present status of land utilization, in fields such as environmental assessment, map renewal, agricultural output assessments significant application value is arranged.The remotely-sensed data source is increasing in recent years, and the remote sensing images of different spatial resolutions have provided more face of land information on different scale.How to make full use of the multiple spatial resolution remotely-sensed data of areal, different spaces coverage rate, further improve the challenge that wide area face of land cover classification precision has become remote Sensing Image Analysis.
For wide area face of land classification problem, no doubt use high-resolution remote sensing image can obtain more accurate face of land classification results in general, but the high-definition remote sensing data intrinsic long, many restrictions such as coverage is little, data price height of heavily visit cycle, restricted it or the practical application in the monitoring of the long-term face of land on a large scale.Therefore, adopt remote sensing images combined entry method can fully utilize the complementarity of multiple remotely-sensed data on coverage and spatial resolution, when keeping, improve comprehensive nicety of grading than large coverage.Existing multiresolution remote sensing images combined entry method instructs overall low spatial resolution classification of Data process by the high spatial resolution data of selecting some little coverages in the large coverage of low spatial resolution data for use, but because existing method supposes that in assorting process pixel is independent, promptly carry out in pixel level or inferior pixel level, ignored the influence of the locus of pixel and neighborhood of pixels atural object classification, so classification results is subject to the remote sensing images noise effect to classification results.
Summary of the invention
In order to overcome above-mentioned the deficiencies in the prior art, the object of the present invention is to provide a kind of multiresolution remote sensing images combined entry method based on context relation, this method comprehensive utilization multiresolution remote sensing images, make up the context relation between pixel, considered the space continuity that atural object distributes, and provide support, thereby solve the classification problem of high-precision wide area low resolution remote sensing images to many characteristic of divisions by conditional random field models.
To achieve these goals, the technical solution used in the present invention is:
Multiresolution remote sensing images combined entry method based on context relation may further comprise the steps,
Step 1, carry out the registration in the local training zone: at first, the regional area of selecting to have high low-resolution image more than one or one simultaneously and comprising all kinds of atural object classifications is as the training zone, secondly, resolution proportionate relationship according to high low-resolution image is carried out the adjustment of resolution integer multiplying power to high-definition picture, make that high low resolution ratio is an integral multiple, last, high low-resolution image is carried out the local space registration;
Step 2 is carried out characteristic of division to low-resolution image and is extracted;
Step 3, based on preceding two steps, in the training zone, utilize the further context relation of setting up based on conditional random field models of characteristic of division: at first, based on context relation makes up the characteristic of division sequence of low-resolution image, as the input observation random series of conditional random field models, secondly, generate the mark sequence corresponding behind the high-definition picture classification results resolution decreasing with the characteristic of division sequence, last, obtain the conditional random field models parameter by training set;
Step 4, obtain carrying out overall situation classification behind the conditional random field models according to first three step, the model that trains is extended to whole low-resolution image overlay area, the wide area low-resolution image is classified: at first, context relation during according to training generates corresponding low resolution image characteristic of division sequence, again according to formula
With
The conditional random field models parameter that utilization trains obtains the conditional probability of all categories of pixel, adopt the maximal condition canon of probability to carry out the final atural object classification that overall situation classification obtains each pixel correspondence at last, wherein, E (Y, X) be the potential function of condition random field, by monobasic potential function g
i(y
i, X) with binary potential function f
Ij(y
j, y
i, X) form, wherein monobasic potential function g
i(y
i, the X) relation of presentation class feature and class label, binary potential function f
Ij(y
j, y
i, X) the spatial context relation between the adjacent picture elements in the expression remote sensing images, λ
IjAnd μ
iBe respectively potential function f
Ij(y
j, y
i, X) and g
i(y
i, weight X), Z (X) be to might sequence normalized factor,
Local space registration in the step 1 specifically is meant,
At first, the coupling pre-service uses the Sobel operator that high-resolution aerophotograph is carried out edge extracting, after edge extracting is finished, carries out once or once above diffusion, obtains image boundary as further process object;
Secondly, carry out thick coupling based on manual coupling, select suggestion according to the reference mark earlier, adopt the method for visualization on high low-resolution image subject to registration, to determine 6～7 groups of reference mark one to one, extract its position coordinates, carry out again based on the manually geometric transformation of thick coupling, the interpolation processing after then slightly mating, described interpolation processing is handled for the bidirectional linear interpolation carries out interpolation;
At last, adopt full-automatic match pattern to carry out the essence coupling, in automatic smart matching process, by having choosing automatically once more of optimal spatial distribution reference mark, carry out optimal registration according to the gray scale matching criterior, the volume coordinate corresponding relation of accurate Calculation high-definition picture and low-resolution image realizes that the pixel level be equal to high-definition picture accurately mates, and described automatic smart coupling is finished by the tolerance of calculating certain similarity between two images or dissimilarity.At first, in high-definition picture, select " window " zone, in low-resolution image, select " search " zone, in the region of search, all possible overlapping relation is carried out correlation computations by displacement.Again with the region of search high-definition picture by being averaged the method resolution decreasing to low resolution, moving window in the region of search calculates the gray variance between two width of cloth image blocks afterwards, the point of variance minimum is optimal match point.Such region of search can be the basis according to the reference mark that manual coupling is chosen the time, and the result of coupling finds optimal match point in the near zone of original taken point automatically.
Select normalized differential vegetation index feature or enhancement mode vegetation index feature or textural characteristics or spectral signature as characteristic of division according to the remotely-sensed data form in the step 2.
The pixel classification information that employing pixel level sorter obtains in the step 2 is as the sequence characteristic of division.
Conditional random field models is meant the modeling to the relation of spectral signature and classification and the context relation function between pixel in the step 3, make G=(S, E) be a non-directed graph, wherein S is a node set among the figure, E is the nonoriented edge set between S, list entries X be one can be observed the sequence of random variables set, output node value Y is one and can be gathered by the stochastic variable of model prediction, nonoriented edge by the indication dependence between output node connects Y={y
i| i ∈ S|}, if given X is and each stochastic variable y
iSatisfy
Wherein, S-{i} represents to remove among the S all node set of node i, N
iBe the neighborhood node set of node i, then (X Y) constitutes a condition random field.
The mode of obtaining the conditional random field models parameter in the step 3 is, at first the classification results of the pairing N of low resolution pixel * N high resolving power pixel piece voted, and the highest classification of ratio generates class label as the atural object classification of this low resolution pixel; Generate the characteristic of division sequence then, form with class label and demarcate sample set; Utilize the gradient descent method that the fundamental function weight is carried out the maximum-likelihood method parameter estimation again.
The present invention compared with prior art has the following advantages:
One) nicety of grading height.
Two) transplantability is strong, applicability is high, can use on existing remote sensing images combined entry system-based, and nicety of grading is further promoted.
Three) can carry out cascade with other remote sensing images combined entry devices, set up the disaggregated model of integrating based on context relation.
Four) avoided the high-definition remote sensing data intrinsic long, many restrictions such as coverage is little, data price height of heavily visit cycle, can realize high precision table sort in large area.
Five) in overall low resolution remote sensing image classification process, do not need the user to carry out manual intervention, have the automaticity advantages of higher.
Description of drawings
Fig. 1 is a system flow block diagram of the present invention.
Fig. 2 is a condition random field sequence modeling synoptic diagram, and wherein X shows and decides observation sequence, and Y refers to the terrain classification sequence label of each pixel correspondence, x
I-2～x
I+2The observation characteristic of division value that refers to position i-2～i+2 pixel respectively, y
I-2～y
I+2The terrain classification label value that refers to position i-2～i+2 correspondence respectively.
Embodiment
The present invention is described in detail below in conjunction with accompanying drawing.
The present invention is divided into local training and overall classification two large divisions, is made up of the registration in the training zone, part, characteristic of division extraction, condition random field modeling and overall situation classification four big basic modules, and Fig. 1 has provided general frame of the present invention.Below, will be set forth respectively module functions separately and the specific algorithm that is adopted thereof.
Step 1 is carried out the registration in the local training zone
The main effect of this module is the inferior pixel level spatial relationship coupling that realizes between the height image in different resolution of training zone, comprises following process:
The first step: the training zone is selected
The regional area of selecting to have high low-resolution image more than one or one simultaneously and comprising all kinds of atural object classifications guarantees the rationality of training sample as the training zone;
Second step: high low-resolution image pixel integer multiplying power is adjusted
Resolution proportionate relationship according to high low-resolution image is carried out the adjustment of resolution integer multiplying power to high-definition picture, makes that high low resolution ratio is an integral multiple, is convenient to registration and sets up high low-resolution image many-one spatial relationship;
The 3rd step: high precision spatial registration
The present invention adopts multistage method for registering, solves the high registration accuracy problem between different sensor multiresolution remote sensing images, sets up accurate high low-resolution image many-one spatial relationship, and high low-resolution image is carried out the local space registration.Concrete thinking is, at first adopt semi-automatic match pattern slightly to mate with certain interactivity, by manually choosing of minority reference mark, guestimate high-resolution remote sensing image and low resolution remote sensing images in the observation area, the basic corresponding parameter of relative scale, relative rotation angle etc.; Adopt full-automatic match pattern to carry out the essence coupling then, in automatic smart matching process, by having choosing automatically once more of optimal spatial distribution reference mark, carry out optimal registration according to the gray scale matching criterior, the volume coordinate corresponding relation of accurate Calculation high-definition picture and low-resolution image realizes that the pixel level be equal to high-definition picture accurately mates.
Detailed process is as follows:
1. coupling pre-service
In the multi-source remote sensing coupling, the mode of choosing at reference mark and rationality thereof are one of key factors of decision matching precision.If the reference mark is chosen in variation of image grayscale slowly or do not have the zone of limbus feature, then can cause pseudo-coupling phenomenon, that the mode by automatic coupling of being difficult to finds is best, reasonable match point.Therefore, in the choosing of reference mark, introduce the peculiar local edge of image-region, by treating the mode of matching image edge analysis, advise comparatively significantly zone of first-selected those marginal textures to the user, at last by the user manually selected be concerned about that the unique point of target area is as optimum controlling point.
Here, at first use conventional Sobel operator that high-resolution boat sheet is carried out edge extracting.It is simple that the Sobel operator has method, and processing speed is fast, and the continuous edge of gained is smooth.The computation process of conventional Sobel operator is as follows, establishes,
A＝|(f(i-1，j-1)+2f(i-1，j)+f(i-1，j+1))-(f(i+1，j-1)+2f(i+1，j)+f(i+1，j+1))| (1)
B＝|(f(i-1，j-1)+2f(i，j-1)+f(i+1，j-1))-(f(i-1，j+1)+2f(i，j+1)+f(i+1，j+1))| (2)
Then, and S (i, j)=max (A, B).
Suitably choose thresholding TH, judge if S (i, j)＞TH, then (i j) is marginal point to pixel.After the edge extracting process is finished, carry out the diffusion of one or many, make little border disappear, only stay the further process object of border conduct than large scale.The selection principle of optimum controlling point is to select to change precipitous point on these borderline regions, guarantees that simultaneously the reference mark is evenly distributed on the entire image.
2. based on the thick coupling of manual coupling
At first carry out selecting: select suggestion according to the reference mark, adopt the method for visualization on high low-resolution image subject to registration, to determine 6～7 groups of reference mark one to one, extract its position coordinates based on the manual reference mark of optimizing suggestion.
Carry out based on the manually geometric transformation of thick coupling: suppose to exist between two width of cloth images certain optimum linear coordinate transformation relation to come the nonlinear transformation of close approximation complexity, coordinate transform can be obtained by the least square fitting reference mark with one, low order linear polynomial is simulated again.This low order affined transformation can be expressed displacement, rotation, deflection and the engineer's scale difference between this two width of cloth image, simultaneously can by least square handle avoid indivedual reference mark the excessive influence that may bring.
Interpolation processing after slightly mating at last: to each pixel of image after the thick matching treatment (X ', Y '), need to calculate pairing coordinate in its original image (X, Y).Since coordinate transformation equation given therefore (X, Y) value not necessarily accurately appears in the coordinate place of original pixel, need handle the calculating original image in (X, the pixel value of Y) locating by interpolation.The present invention has adopted the bidirectional linear interpolation to carry out interpolation and has handled, and this is because the interpolating image of this method has continuous gray scale and precision height preferably.
3. full-automatic smart coupling
Automatic digital registration can be finished by the tolerance of certain similarity or dissimilarity between two images of calculating, and this tolerance is the function of relative displacement between the image, and a kind of tolerance of similarity is being correlated with between two superimposed images zones.If these two zones are to have better correspondence, then can produce bigger correlation properties.Because the correlation computations between big area image can be brought bigger operand,, employing carries out dependent evaluation than the zonule so being distributed in interior some of two whole overlapping scopes of image.Concrete grammar is as follows: at first, select " window " zone in high-definition picture, select " search " zone in low-resolution image, in the region of search all possible overlapping relation is carried out correlation computations by displacement.Again with the region of search high-definition picture by being averaged the method resolution decreasing to low resolution, moving window in the region of search calculates the variance between two width of cloth image blocks afterwards, the point of variance minimum is optimal match point.
Such region of search can be the basis according to the reference mark that manual coupling is chosen the time, and the result of coupling finds optimal match point in the near zone of original taken point automatically.Then, carry out geometric transformation and interpolation again, method is with manually the coupling corresponding steps is identical.
Step 2, characteristic of division extracts
The main effect of characteristic of division extraction step is that the low-resolution image spectral signature is handled, and obtains good classification effect, and the characteristic of division of discrimination is arranged.This method also adopts other characteristic of divisions to characteristic of division without limits.Can select normalized differential vegetation index feature, textural characteristics, spectral signature etc. as characteristic of division according to the remotely-sensed data form, also can adopt pixel classification information that other pixel level sorters obtain as the sequence characteristic of division.
The present invention is an example with normalized differential vegetation index NDVI, enhancement mode vegetation index EVI.
The computing formula of NDVI and EVI is as follows:
Wherein, ρ
NIR, ρ
RedAnd ρ
BlueIt is respectively the slin emissivity value of near infrared, redness and blue sensor wave band.
NDVI is the influence of elimination and the relevant radiation variation with atmospheric conditions such as sun altitude, moonscope angle, landform, Yun Ying partly, and NDVI is the best indicator of vegetation growth state and vegetation coverage.The weak point of NDVI is that its numerical value is saturated easily, high vegetation density area susceptibility is reduced, and be subjected to planting the influence of hat background easily, does not eliminate atmospheric scattering, absorbs the influence that spectrum is produced.The EVI index is more stable than NDVI, because it has the ability of gasoloid resistivity, normalization Soil Background and undersaturated ability under the high-biomass condition.NDVI and EVI complement one another in global vegetation study, and NDVI is responsive more to the vegetation in barren zone, and EVI can play effective function more in the zone that is coated with dense vegetation.
Step 3, the condition random field MBM
In combined entry, what play is the effect of supervising and guiding a kind of subregion in because high-definition picture is actual, so the method that can adopt supervisory sequence to classify.Condition random field (CRF, Conditional Random Fields) model is a kind of non-directed graph model based on conditional probability, when given observation sequence X, can be used for calculating corresponding mark sequence Y, it is a kind of discriminant model, directly conditional probability P (y|x) is carried out modeling.Compare with Markov random field model, conditional random field models can reduce the probability distribution hypothesis, and has the advantage that to select the context dependent feature, better comprise neighborhood information, and, all features carry out global normalization by being carried out sequence form, obtain global optimum, solve many characteristic sequences classification problem based on context relation.
In view of above-mentioned analysis, the present invention has constructed a conditional random field models that is used for the remote sensing images combined entry, and two class potential functions of definite condition random field models are used for describing respectively the relation of spectral signature and classification and the context relation between pixel.Utilize conditional random field models that these potential functions are carried out modeling, thereby infer globally optimal solution realization remote sensing images combined entry by model.The main effect of condition random field MBM is the context relation at the built-in conditional random field models that is based in training zone.
For the ease of condition random field is described, make G=that (S E) is a non-directed graph, and wherein S is a node set among the figure, and E is the nonoriented edge set between S.List entries X be one can be observed sequence of random variables set, output node value Y be one can be by the set of the stochastic variable of model prediction, the nonoriented edge by the indication dependence between output node connect Y={y
i| i ∈ S|}.If given X is and each stochastic variable y
iSatisfy
Wherein, S-{i} represents to remove among the S all node set of node i, N
iBe the neighborhood node set of node i, then (X Y) constitutes a condition random field.
In combined entry, the atural object context relation between pixel is represented by the nonoriented edge in the condition random field.The characteristic of division set that observation sequence X is formed corresponding to n neighborhood pixel.As shown in Figure 2, the present invention forms observation sequence with 5 pixel features on row, column and diagonal angle 4 directions.The tag along sort Y of condition random field node is the atural object class label of each pixel correspondence, and the present invention is with y
i=0,1,2,3,4} is an example, wherein, 0 expression forest land, 1 expression cities and towns, 2 expressions are ploughed, 3 expression wastelands, 4 expression water bodys.Conditional random field models provides the model framework that calculates tag along sort Y conditional probability behind the given observation sequence X, and this conditional probability is given by following formula:
Wherein, (Y X) is the potential function of condition random field to E, by monobasic potential function g
i(y
i, X) with binary potential function f
Ij(y
j, y
i, X) form, wherein monobasic potential function g
i(y
i, the X) relation of presentation class feature and class label, binary potential function f
Ij(y
j, y
i, X) the spatial context relation between the adjacent picture elements in the expression remote sensing images is not only relevant with the characteristic of division vector of position i, also relevant with the label of neighborhood location point; λ
IjAnd μ
iBe respectively potential function f
Ij(y
j, y
i, X) and g
i(y
i, weight X) has reflected the significance level of this function; Z (X) be to might sequence normalized factor,
After determining conditional random field models, need train the zone to train, estimate model parameter θ=(λ by following process in the part
Ij, μ
i):
The first step: the classification results to the pairing N of low resolution pixel * N high resolving power pixel piece is voted, and the highest classification of ratio generates class label Y as the atural object classification of this low resolution pixel;
Second step: generate the characteristic of division sequence X, form with class label and demarcate sample set T;
The 3rd step: utilize the gradient descent method to the fundamental function weight λ in the formula (6)
kAnd μ
lCarry out the maximum-likelihood method parameter estimation:
Train resulting conditional probability model to reflect that to train the zone be this observation area atural object distribution character of representative, this conditional probability model extends to overall low-resolution image overlay area, obtain each low resolution pixel with respect to conditional probability of all categories, adopt the maximal condition canon of probability can judge the atural object classification that each low resolution pixel is affiliated.
Step 4, overall sort module
The main effect of overall situation sort module is that the model that will train extends to whole low-resolution image overlay area, realizes by following process:
The first step: according in the condition random field MBM to the definition of context relation, generate the low image classification characteristic sequence of differentiating in the remote sensing images coverage of wide area;
Second step:, utilize the conditional random field models parameter that trains to obtain the corresponding conditional probability of all categories of pixel according to formula (6);
The 3rd step: adopt the maximal condition canon of probability to carry out the final atural object classification that overall situation classification obtains each pixel correspondence.
The present invention by setting up the space continuity that context relation between pixel has considered that atural object distributes, and utilizes the condition random field modeling that support to many characteristic of divisions is provided in assorting process, thereby further improves the precision of classification.With respect to other combined entry method, the present invention fully utilizes pixel spatial neighborhood relation in the remote sensing image classification process, simultaneously, because model can be supported self-defined many features, and on realization flow, has independence, the present invention can be used as the useful of existing remote sensing images combined entry method and replenishes, and by carrying out cascade with other combined entry models, realizes the high precision wide area combined entry of the interval level of comprehensive inferior pixel level, pixel level and neighborhood.
The present invention provides a kind of brand-new thinking for remote sensing images combined entry method.
Claims (7)
1. based on the multiresolution remote sensing images combined entry method of context relation, may further comprise the steps,
Step 1, carry out the registration in the local training zone: at first, the regional area of selecting to have high low-resolution image more than one or one simultaneously and comprising all kinds of atural object classifications is as the training zone, secondly, resolution proportionate relationship according to high low-resolution image is carried out the adjustment of resolution integer multiplying power to high-definition picture, make that high low resolution ratio is an integral multiple, last, high low-resolution image is carried out the local space registration;
Step 2 is carried out characteristic of division to low-resolution image and is extracted;
Step 3, based on preceding two steps, in the training zone, utilize the further context relation of setting up based on conditional random field models of characteristic of division: at first, based on context relation makes up the characteristic of division sequence of low-resolution image, as the input observation random series of conditional random field models, secondly, generate the mark sequence corresponding behind the high-definition picture classification results resolution decreasing with the characteristic of division sequence, last, obtain the conditional random field models parameter by training set;
Step 4, obtain carrying out overall situation classification behind the conditional random field models according to first three step, the model that trains is extended to whole low-resolution image overlay area, the wide area low-resolution image is classified: at first, context relation during according to training generates corresponding low resolution image characteristic of division sequence, again according to formula
With
The conditional random field models parameter that utilization trains obtains the conditional probability of all categories of pixel, adopt the maximal condition canon of probability to carry out the final atural object classification that overall situation classification obtains each pixel correspondence at last, wherein, E (Y, X) be the potential function of condition random field, by monobasic potential function g
i(y
i, X) with binary potential function f
Ij(y
j, y
i, X) form, wherein monobasic potential function g
i(y
i, the X) relation of presentation class feature and class label, binary potential function f
Ij(y
j, y
i, X) the spatial context relation between the adjacent picture elements in the expression remote sensing images, λ
IjAnd μ
iBe respectively potential function f
Ij(y
j, y
i, X) and g
i(y
i, weight X), Z (X) be to might sequence normalized factor,
2. the multiresolution remote sensing images combined entry method based on context relation according to claim 1 is characterized in that, the local space registration in the step 1 specifically is meant,
At first, the coupling pre-service uses the Sobel operator that high-resolution aerophotograph is carried out edge extracting, after edge extracting is finished, carries out once or once above diffusion, obtains image boundary as further process object;
Secondly, carry out thick coupling based on manual coupling, select suggestion according to the reference mark earlier, adopt the method for visualization on high low-resolution image subject to registration, to determine 6～7 groups of reference mark one to one, extract its position coordinates, carry out again based on the manually geometric transformation of thick coupling, the interpolation processing after then slightly mating;
At last, adopt full-automatic match pattern to carry out the essence coupling, in high-definition picture, select " window " zone, in low-resolution image, select " search " zone, the region of search can be the basis according to the reference mark that manual coupling is chosen the time, automatically the result of coupling finds optimal match point in the near zone of original taken point, in the region of search, all possible overlapping relation is carried out correlation computations by displacement, again with the region of search high-definition picture by being averaged the method resolution decreasing to low resolution, moving window in the region of search afterwards, calculate the variance between two width of cloth image blocks, the point of variance minimum is optimal match point.
3. the multiresolution remote sensing images combined entry method based on context relation according to claim 2 is characterized in that, described interpolation processing is handled for the bidirectional linear interpolation carries out interpolation.
4. the multiresolution remote sensing images combined entry method based on context relation according to claim 1, it is characterized in that, select normalized differential vegetation index feature or enhancement mode vegetation index feature or textural characteristics or spectral signature as characteristic of division according to the remotely-sensed data form in the step 2.
5. the multiresolution remote sensing images combined entry method based on context relation according to claim 1 is characterized in that, the pixel classification information that employing pixel level sorter obtains in the step 2 is as the sequence characteristic of division.
6. the multiresolution remote sensing images combined entry method based on context relation according to claim 1, it is characterized in that, conditional random field models is meant the relation to spectral signature and classification in the step 3, and the modeling of the context relation function between pixel, make G=(S, E) be a non-directed graph, wherein S is a node set among the figure, E is the nonoriented edge set between S, list entries X be one can be observed sequence of random variables set, output node value Y is one and can be gathered by the stochastic variable of model prediction, nonoriented edge by the indication dependence between output node connects Y={y
i| i ∈ S|}, if given X is and each stochastic variable y
iSatisfy
Wherein, S-{i} represents to remove among the S all node set of node i, N
iBe the neighborhood node set of node i, then (X Y) constitutes a condition random field.
7. the multiresolution remote sensing images combined entry method based on context relation according to claim 1, it is characterized in that, the mode of obtaining the conditional random field models parameter in the step 3 is, at first the classification results of the pairing N of low resolution pixel * N high resolving power pixel piece is voted, the highest classification of ratio generates class label as the atural object classification of this low resolution pixel; Generate the characteristic of division sequence then, form with class label and demarcate sample set; Utilize the gradient descent method that the fundamental function weight is carried out the maximum-likelihood method parameter estimation again.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105603346A CN102110227B (en) | 2010-11-24 | 2010-11-24 | Compound method for classifying multiresolution remote sensing images based on context |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010105603346A CN102110227B (en) | 2010-11-24 | 2010-11-24 | Compound method for classifying multiresolution remote sensing images based on context |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102110227A true CN102110227A (en) | 2011-06-29 |
| CN102110227B CN102110227B (en) | 2012-06-06 |
Family
ID=44174381
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010105603346A Expired - Fee Related CN102110227B (en) | 2010-11-24 | 2010-11-24 | Compound method for classifying multiresolution remote sensing images based on context |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102110227B (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103179424A (en) * | 2013-03-26 | 2013-06-26 | 深圳市掌网立体时代视讯技术有限公司 | Method and device for synthesising three-dimensional image |
| CN104331686A (en) * | 2014-10-28 | 2015-02-04 | 中国农业大学 | Manual assisting type recognition system of coverage rate of straw on soil surface |
| CN104331698A (en) * | 2014-11-19 | 2015-02-04 | 中国农业科学院农业资源与农业区划研究所 | Remote sensing type urban image extracting method |
| CN104794424A (en) * | 2014-01-20 | 2015-07-22 | 北京天合数维科技有限公司 | Novel middle-low resolution remote sensing data combined cultivated land identification method |
| CN105740891A (en) * | 2016-01-27 | 2016-07-06 | 北京工业大学 | Target detection method based on multilevel characteristic extraction and context model |
| CN106056128A (en) * | 2016-04-20 | 2016-10-26 | 北京航空航天大学 | Remote sensing image classification marking method based on composite graph conditional random field |
| CN106373150A (en) * | 2016-08-15 | 2017-02-01 | 中国农业大学 | Identification method of corns for seed on the basis of high resolution remote sensing data texture analysis |
| CN108629788A (en) * | 2018-05-08 | 2018-10-09 | 苏州大学 | A kind of method for detecting image edge, device, equipment and readable storage medium storing program for executing |
| CN110035687A (en) * | 2016-11-30 | 2019-07-19 | 卡普索影像公司 | The method and device of image mosaic is carried out to the image for using capsule cameras to be captured |
| CN110084203A (en) * | 2019-04-29 | 2019-08-02 | 北京航空航天大学 | Full convolutional network aircraft level detection method based on context relation |
| CN110580469A (en) * | 2019-09-10 | 2019-12-17 | 广州麦仑信息科技有限公司 | palm vein recognition system and method based on embedded equipment |
| CN112183209A (en) * | 2020-08-27 | 2021-01-05 | 中国农业大学 | Regional crop classification method and system based on multi-dimensional feature fusion |
| CN113358154A (en) * | 2021-06-03 | 2021-09-07 | 广东工业大学 | Agricultural machinery equipment fault monitoring and tracing system and method |
| CN113935928A (en) * | 2020-07-13 | 2022-01-14 | 四川大学 | Rock core image super-resolution reconstruction based on Raw format |
| CN117115679A (en) * | 2023-10-25 | 2023-11-24 | 北京佳格天地科技有限公司 | Screening method for space-time fusion remote sensing image pairs |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1484039A (en) * | 2003-07-24 | 2004-03-24 | 上海交通大学 | Image merging method based on inseparable wavelet frame |
| CN101777125A (en) * | 2010-02-03 | 2010-07-14 | 武汉大学 | Method for supervising and classifying complex class of high-resolution remote sensing image |
-
2010
- 2010-11-24 CN CN2010105603346A patent/CN102110227B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1484039A (en) * | 2003-07-24 | 2004-03-24 | 上海交通大学 | Image merging method based on inseparable wavelet frame |
| CN101777125A (en) * | 2010-02-03 | 2010-07-14 | 武汉大学 | Method for supervising and classifying complex class of high-resolution remote sensing image |
Non-Patent Citations (2)
| Title |
|---|
| 《电子与信息学报》 20030131 黄宁等 一种采用高斯隐马尔可夫随机场模型的遥感图像分类算法 第50-53页 1-7 第25卷, 第1期 * |
| 《遥感学报》 20080331 郑玮等 基于模糊马尔可夫随机场的无监督遥感图像分割算法 第246-252页 1-7 第12卷, 第2期 * |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103179424A (en) * | 2013-03-26 | 2013-06-26 | 深圳市掌网立体时代视讯技术有限公司 | Method and device for synthesising three-dimensional image |
| CN103179424B (en) * | 2013-03-26 | 2015-03-11 | 深圳市掌网立体时代视讯技术有限公司 | Method and device for synthesising three-dimensional image |
| CN104794424A (en) * | 2014-01-20 | 2015-07-22 | 北京天合数维科技有限公司 | Novel middle-low resolution remote sensing data combined cultivated land identification method |
| CN104331686A (en) * | 2014-10-28 | 2015-02-04 | 中国农业大学 | Manual assisting type recognition system of coverage rate of straw on soil surface |
| CN104331686B (en) * | 2014-10-28 | 2017-08-08 | 中国农业大学 | A kind of soil surface improving straw mulching rate human assistance identifying system |
| CN104331698A (en) * | 2014-11-19 | 2015-02-04 | 中国农业科学院农业资源与农业区划研究所 | Remote sensing type urban image extracting method |
| CN104331698B (en) * | 2014-11-19 | 2017-05-03 | 中国农业科学院农业资源与农业区划研究所 | Remote sensing type urban image extracting method |
| CN105740891A (en) * | 2016-01-27 | 2016-07-06 | 北京工业大学 | Target detection method based on multilevel characteristic extraction and context model |
| CN105740891B (en) * | 2016-01-27 | 2019-10-08 | 北京工业大学 | Target detection based on multi level feature selection and context model |
| CN106056128A (en) * | 2016-04-20 | 2016-10-26 | 北京航空航天大学 | Remote sensing image classification marking method based on composite graph conditional random field |
| CN106056128B (en) * | 2016-04-20 | 2019-03-22 | 北京航空航天大学 | Remote sensing image classification mask method based on composite diagram condition random field |
| CN106373150A (en) * | 2016-08-15 | 2017-02-01 | 中国农业大学 | Identification method of corns for seed on the basis of high resolution remote sensing data texture analysis |
| CN110035687A (en) * | 2016-11-30 | 2019-07-19 | 卡普索影像公司 | The method and device of image mosaic is carried out to the image for using capsule cameras to be captured |
| CN110035687B (en) * | 2016-11-30 | 2022-02-11 | 卡普索影像公司 | Method and device for image stitching of images captured by capsule camera |
| CN108629788A (en) * | 2018-05-08 | 2018-10-09 | 苏州大学 | A kind of method for detecting image edge, device, equipment and readable storage medium storing program for executing |
| CN108629788B (en) * | 2018-05-08 | 2022-03-15 | 苏州大学 | Image edge detection method, device and equipment and readable storage medium |
| CN110084203A (en) * | 2019-04-29 | 2019-08-02 | 北京航空航天大学 | Full convolutional network aircraft level detection method based on context relation |
| CN110580469A (en) * | 2019-09-10 | 2019-12-17 | 广州麦仑信息科技有限公司 | palm vein recognition system and method based on embedded equipment |
| CN110580469B (en) * | 2019-09-10 | 2022-08-09 | 广州麦仑信息科技有限公司 | Palm vein recognition system and method based on embedded equipment |
| CN113935928A (en) * | 2020-07-13 | 2022-01-14 | 四川大学 | Rock core image super-resolution reconstruction based on Raw format |
| CN113935928B (en) * | 2020-07-13 | 2023-04-11 | 四川大学 | Rock core image super-resolution reconstruction based on Raw format |
| CN112183209A (en) * | 2020-08-27 | 2021-01-05 | 中国农业大学 | Regional crop classification method and system based on multi-dimensional feature fusion |
| CN112183209B (en) * | 2020-08-27 | 2023-12-22 | 中国农业大学 | Regional crop classification method and system based on multidimensional feature fusion |
| CN113358154A (en) * | 2021-06-03 | 2021-09-07 | 广东工业大学 | Agricultural machinery equipment fault monitoring and tracing system and method |
| CN117115679A (en) * | 2023-10-25 | 2023-11-24 | 北京佳格天地科技有限公司 | Screening method for space-time fusion remote sensing image pairs |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102110227B (en) | 2012-06-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102110227B (en) | Compound method for classifying multiresolution remote sensing images based on context | |
| CN102254319B (en) | Method for carrying out change detection on multi-level segmented remote sensing image | |
| Huang et al. | A new building extraction postprocessing framework for high-spatial-resolution remote-sensing imagery | |
| CN102609682B (en) | Feedback pedestrian detection method for region of interest | |
| CN103632363B (en) | Object level high-resolution remote sensing image change detecting method based on Multiscale Fusion | |
| CN101551856B (en) | SAR target recognition method based on sparse least squares support vector machine | |
| CN103971115A (en) | Automatic extraction method for newly-increased construction land image spots in high-resolution remote sensing images based on NDVI and PanTex index | |
| Liu et al. | Deep multi-level fusion network for multi-source image pixel-wise classification | |
| CN103353988B (en) | Allos SAR scene Feature Correspondence Algorithm performance estimating method | |
| Ok et al. | 2-D delineation of individual citrus trees from UAV-based dense photogrammetric surface models | |
| CN104867150A (en) | Wave band correction change detection method of remote sensing image fuzzy clustering and system thereof | |
| CN102096826B (en) | Compound classification method for multi-resolution remote sensing image on basis of real likelihood characteristic | |
| Sun et al. | Wheat head counting in the wild by an augmented feature pyramid networks-based convolutional neural network | |
| CN104156943A (en) | Multi-target fuzzy cluster image variance detecting method based on non-control-neighborhood immune algorithm | |
| Zhou et al. | Autonomous detection of crop rows based on adaptive multi-ROI in maize fields | |
| CN105069451A (en) | License plate identifying and positioning method based on binocular camera | |
| CN105488541A (en) | Natural feature point identification method based on machine learning in augmented reality system | |
| CN102855485A (en) | Automatic wheat earing detection method | |
| Bai et al. | A fast and robust method for plant count in sunflower and maize at different seedling stages using high-resolution UAV RGB imagery | |
| Li et al. | High-throughput phenotyping analysis of maize at the seedling stage using end-to-end segmentation network | |
| Wang et al. | Recognition of rice seedling rows based on row vector grid classification | |
| Lou et al. | Recognition of thaw slumps based on machine learning and UAVs: A case study in the Qilian Mountains, northeastern Qinghai-Tibet Plateau | |
| CN116188943A (en) | Solar radio spectrum burst information detection method and device | |
| Tahraoui et al. | Land change detection in sentinel-2 images using ir-mad and deep neural network | |
| Wang et al. | Underwater Terrain Image Stitching Based on Spatial Gradient Feature Block. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20120606 Termination date: 20141124 |
|
| EXPY | Termination of patent right or utility model |