CN104851113B - The urban vegetation extraction method of multiple-spatial resolution remote sensing image - Google Patents
The urban vegetation extraction method of multiple-spatial resolution remote sensing image Download PDFInfo
- Publication number
- CN104851113B CN104851113B CN201510186167.6A CN201510186167A CN104851113B CN 104851113 B CN104851113 B CN 104851113B CN 201510186167 A CN201510186167 A CN 201510186167A CN 104851113 B CN104851113 B CN 104851113B
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- mtd
- msup
- vegetation
- 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.)
- Expired - Fee Related
Links
Landscapes
- Other Investigation Or Analysis Of Materials By Electrical Means (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of urban vegetation extraction method of multiple-spatial resolution remote sensing image, including the initial patch extraction of multispectral image vegetation based on high spatial resolution, the multispectral image computation vision perceptual parameters to high spatial resolution, the panchromatic wave-band image to high spatial resolution calculate textural characteristics parameters, combination features vegetation area automatic increase, obtain the steps such as vegetation area distribution map.Image block dividing processing is employed, accelerates image processing speed.According to NDVI features, using automatically selecting for the adaptive dynamic threshold of maximal mathematical expectation algorithm.The panchromatic wave-band of remote sensing image, multi light spectrum hands are made full use of, the NDVI features extracted from panchromatic wave-band and multi light spectrum hands, visual signature, textural characteristics are integrated, vegetation area judgement is carried out to vegetation, accuracy is improved.
Description
Technical field
The present invention relates to a kind of urban vegetation extraction method of multiple-spatial resolution remote sensing image, belong to remote sensing image neck
Domain.
Background technology
Urban vegetation has carbon sequestration, releases foster function, plays an important roll in terms of urban ecological environment is kept.Based on high score
The urban vegetation covering that resolution remote sensing image is extracted, is to assess ecocity, the important indicator of garden-like cities.Height is utilized at this stage
The method that resolution remote sense image extracts urban vegetation mainly has following several:
(1) artificial visual interpretation interpretation.This method is main with artificial interpretation vegetation area, vegetation border of then sketching.
The feature that Main Basiss vegetation shows in the picture in visual interpretation includes shape, size, color and tint, shade, position, line
Reason relation etc..The method that visual interpretation uses manual work screen digitization, automaticity is relatively low, and subjectivity is strong, it is desirable to have
Abundant image interpretation practical experience, and the thematic information extracted is better understood from and recognized.But being due to can be fully
Ground utilizes various information in remote sensing image, is still the important method of remote sensing image processing.
(2) supervised classification method of object-oriented.This method mainly utilizes color, space line of the vegetation area in image
Characteristic is managed, on the basis of the image segmentation of object-oriented, the various features of each object are calculated, including (it includes spectrum:Spectrum
Average, brightness, mean square deviation, ratio, the spectral signature related to field, the spectral signature related with parent object etc.) shape (bag
Include area, girth, the length of side, tight ness rating, length-width ratio and the shape facility related with parent object to subobject), texture is (according to ash
Second moment (energy), contrast, correlation, entropy, unfavourable balance square that degree co-occurrence matrix is defined) and topological characteristic etc., finally select phase
The feature answered carries out object-based classification and extracted.
These above-mentioned extracting methods all existing defects and problem, including:
(1) when extracting vegetation area, it is necessary to select sample in advance, priori is generated, and it is special to be not based on image itself
Property.There is man-machine interaction during this, artificial influence factor is more.
(2) automaticity is not high.The yardstick split in assorting process, threshold value, the feature of vegetation classification are required for people
Work interactive selection, these pilot process have had a strong impact on the automaticity of vegetation extraction.In addition, in cutting procedure, view picture
Image uses same threshold value, does not take into full account the difference (such as meadow, coniferous forest, broad-leaf forest) between different vegetation types
Feature difference on image.
(3) multispectral data in high spatial resolution image is only make use of, not using with more high spatial resolution
The characteristics of panchromatic wave-band, and in practice, as the remotely-sensed data of the high spatial resolutions such as WorldView, QuickBird it is panchromatic with it is many
Spectrum image is bundled.The multispectral data in high spatial resolution image is only make use of, not using with more
The characteristics of high spatial resolution panchromatic wave-band, and in practice, as the remote sensing of the high spatial resolutions such as WorldView, QuickBird
Data PAN and multi-spectral image is bundled.
The content of the invention
In order to solve the above problems, the present invention has panchromatic and light more for IKONOS, QuickBird, WorldView etc.
The high-resolution remote sensing image of wave band is composed, it is full-automatic to extract urban vegetation region, it is proposed that a kind of multiple-spatial resolution remote sensing image
Urban vegetation extraction method.
The present invention provides a kind of urban vegetation extraction method of multiple-spatial resolution remote sensing image, mainly including following step
Suddenly:
Step 1: the initial patch of multispectral image vegetation based on high spatial resolution is extracted;
Step 2: to the multispectral image computation vision perceptual parameters of high spatial resolution;
Step 3: calculating textural characteristics parameter to the panchromatic wave-band image of high spatial resolution;
Step 4: the vegetation area automatic increase of combination features, obtains vegetation area distribution map.
It is preferred that, above-mentioned steps one specifically include following steps:
Step 1.1 calculates NDVI vegetation spectral indexes, and calculation formula is
Wherein NIR represents the pixel value of near infrared band, and R represents the pixel value of red band;
Step 1.2 carries out piecemeal to NDVI images, per block size 300*300 pixels, in image edge part less than 300*
During 300 pixel, it is defined by actual size;
Step 1.3 the NDVI block datas of each 300*300 pixel are obtained automatically segmentation threshold distinguish vegetation with it is non-
Vegetation, obtains initial vegetation area;
Step 1.4 repeat step 1.3, obtains the vegetation prime area of all piecemeals, merges the vegetation region in all piecemeals
Domain, obtains initial vegetation area, and connective mark is carried out to whole image vegetation area, the pixel of each vegetative patch is obtained
Coordinate.
It is preferred that, the acquisition of segmentation threshold is obtained by the following method in above-mentioned steps 1.3:
(1) the several ground mulching type of correspondence in each piecemeal NDVI, with ωiRepresent, it is assumed that each atural object classification
NDVI distribution obey Gaussian density function distribution, then probability-distribution function p (x) total NDVI in each piecemeal can be represented
For:
Wherein, n represents the total quantity of classification, ωiI-th of atural object classification is represented, average and variance use m respectivelyi、Represent,
Its probability density function is expressed as
Wherein, p (ωi) be each atural object classification probability density, its meet condition be:
Under these conditions, the quantity n of estimation atural object classification, each class can be converted into by solving the threshold value of vegetation and non-vegetation
Other ωiAverage and variance point mi、Comprise the following steps that:
(2) maximum of NDVI in the frequency histogram f (x) of NDVI in piecemeal image, search piecemeal image is counted
NDVImaxWith minimum value NDVImin, by NDVIminTo NDVImax256 parts are equally divided into, statistics NDVI values appear in every a interval
Pixel quantity, obtain NDVI histogram f (x);
(3) single order for calculating NDVI histogram f (x) is reciprocal, straight for 0 position searching NDVI frequencies according to single order inverse
Maximum point, minimum point in square figure;
(4) quantity of maximum point is counted, ground mulching categorical measure n is determined, using the NDVI values at maximum as often
The initial mean value m of one ground mulching classificationi, initial σ is calculated with two neighboring minimum pointiFormula it is as follows:
σi=(NDVI2-NDVI1)/4
Wherein, NDVI2, NDVI1Corresponding NDVI values at two neighboring minimum are represented respectively;
Wherein, wherein ncount is the number of all pixels of the NDVI values between two neighboring minimum, and n_total is
Total number of pixels of piecemeal;
(5) maximal mathematical expectation (EM) algorithm is based on, p (ω are estimated using loop iterationi)、mi、EM algorithms are by following
Ring iterative, each iteration is by asking two steps of desired value and expectation maximization to constitute;The former according to the currency of parameter to be estimated,
The desired value of direct estimation probability density from observation data, the latter is by maximizing this expectation come the estimation of undated parameter
Amount, this two step in whole iterative process successively alternately, until iterative process convergence;Specific formula for calculation is as follows:
The estimation of above-mentioned three formula is that k represents k-th of atural object classification, t and t in prior probability, average and standard deviation, formula respectively
+ 1 represents the estimate used in current and next iteration respectively, and i, j represents the line number and columns of NDVI images, X respectively
The NDVI values that the i-th row j is arranged in (i, j) expression NDVI images, and conditional probability p (X (i, j) | ωi) calculating and full probability p (X (i,
J) value) is drawn by the respective formula in step (1);As the adjacent p (ω iterated to calculate twicei)、miAnd σiValue be less than it is given
Threshold epsilon (ε=10-8) when iteration ends;
(6) the initial vegetation area of piecemeal is determined, it is assumed that the NDVI averages value of each classification obtained in upper step is by from big
To small sequence, m1, m2, mn, corresponding variance is σi, then initial vegetation distribution map is represented by VI (i, j), wherein 1 represents vegetation
Region, 0 represents nonvegetated area domain, then
Wherein threshold value T is tried to achieve by equation below
It is preferred that, the connective mark in above-mentioned steps 1.4 uses 4 neighborhoods or 8 neighborhoods.
It is preferred that, above-mentioned steps two specifically include following steps:
Step 2.1 has RGB color space to transform to HSI color spaces to multispectral image, is calculated and planted by below equation
By brightness, colourity, the result of variations of saturation degree
Step 2.2 computation vision characteristic perception parameter, according to initial vegetative patch as sample, calculates each patch
The brightness of vegetation, colourity, the mean μ of saturation degreex(H, S, I), variances sigmax(H, S, I), calculates normalization tone parameter of saturation
Wherein, H (i, j) is tone, and (i, j) represents pixel coordinate, μkThe average of k-th of vegetative patch, its corresponding side
Difference is σk, HL, HUFor the lower and upper limit of vegetation tone
It is preferred that, above-mentioned steps three are specially:
For panchromatic wave-band video generation gray level co-occurrence matrixes, 6 parametric textures of gray level co-occurrence matrixes are calculated:Contrast,
Diversity, angle energy, entropy, homogeneity, auto-correlation coefficient, specific formula for calculation are as follows:
The selection of textural characteristics computation window mouthful and the resolution ratio of panchromatic image and multispectral image have relation, it is assumed that panchromatic
The spatial resolution of wave band is n times of multispectral image, then the window that texture is calculated is 2n-l, is calculated once every n pixel.
It is preferred that, above-mentioned steps four are specially:Using each initial vegetative patch as " seed ", the method increased using region
The pixel that will be adjacent to " seed " patch is attached on seed region.
It is preferred that, the criterion that above-mentioned zone increases is decided by vote using textural characteristics, visually-perceptible characteristic synthetic using multiple features
To determine.
It is preferred that, the criterion that above-mentioned zone increases is as follows:
(1) to each initial vegetative patch, brightness (I), normalization tone (H), saturation degree (S), contrast are counted respectively
(Contrast), diversity (Dissimilarity), angle energy (Energy), entropy (Entropy), homogeneity
(Homogeneity), the average of auto-correlation coefficient and features described above parameter, variance, it is assumed that its value is μ, σ;To initial vegetation
8 neighborhood territory pixels of patch, according to each feature, whether be vegetation, discriminant function is defined as follows if judging it
Wherein, M (i, j) is the characteristic parameter calculated, in T2Judgment threshold is represented, i, j represents the ranks number of pixel, 1
Vegetation area is represented, 0 represents nonvegetated area domain, and then obtains the corresponding Potential vegetation administrative division map of above-mentioned 9 characteristic parameters difference;
(2) multiple features voting determines final vegetation area, and this 9 width Potential vegetation administrative division map is voted, that is, existed
Each pixel is when when having 6 and its above, feature is collectively labeled as 1 on vegetation area figure, it is believed that be final vegetation area.
Relative to prior art, the urban vegetation extraction method of the multiple-spatial resolution remote sensing image that the present invention is provided have with
Lower advantage:
(1) present invention employs the vegetation threshold value automatic selecting method based on maximal mathematical expectation, data processing is improved
Middle automaticity.Piecemeal treatment technology is employed in addition, it is to avoid view picture image uses same threshold decision, it is to avoid difference
The feature difference that vegetation pattern (such as meadow, coniferous forest, broad-leaf forest) is showed in remotely-sensed data, enhance vegetation threshold value from
Adaptability.
(2) present invention proposes initial vegetation area using NDVI, and high-resolution multi-spectral data are combined on this basis
Visually-perceptible feature, the textural characteristics of panchromatic wave-band, the method voted using multiple features determines final vegetation area.
This method considers the vision of high spatial resolution, textural characteristics, improves the accuracy of vegetation extraction.
Brief description of the drawings
Fig. 1 is schematic flow sheet of the present invention.
Embodiment
Understand for the ease of those of ordinary skill in the art and implement the present invention, below in conjunction with the accompanying drawings and embodiment
The present invention is described in further detail.
This programme has panchromatic and multi light spectrum hands high-resolution for IKONOS, QuickBird, WorldView etc.
Remote sensing image, it is full-automatic to extract urban vegetation region.Specific step includes:
Step 1: the initial patch of multispectral image vegetation based on high spatial resolution is extracted.
Calculate NDVI vegetation spectral indexes.Specific formula for calculation is shown in (1)
Wherein NIR represents the pixel value of near infrared band, and R represents the pixel value of red band.
Piecemeal is carried out to NDVI images, per block size 300*300 pixels, in image edge part less than 300*300 pixels
When, it is defined by actual size.
Segmentation threshold differentiation vegetation and non-vegetation are obtained automatically to the NDVI block datas of each 300*300 pixel, obtained
Obtain initial vegetation area.The method for obtaining segmentation threshold is as follows:
Because vegetation has strong absorption in red band, and there is strong reflection in near-infrared, by red, near infrared band meter
The NDVI indexes of calculation enhance this characteristic of vegetation, can preferably react the difference of vegetation and other types of ground objects.NDVI's
Codomain scope [- 1,1], for vegetation, NDVI is typically larger.
The several ground mulching type of correspondence is (with ω in each piecemeal NDVIiRepresent), it is assumed that each atural object classification
Gaussian density function distribution is obeyed in NDVI distributions, then probability-distribution function p (x) total NDVI in each piecemeal is represented by
Wherein, n represents the total quantity of classification, ωiI-th of atural object classification is represented, average and variance use m respectivelyi、Table
Show, its probability density function is expressed as formula (3), p (ωi) be each atural object classification probability density, its meet
Condition see formula (4).
Under these conditions, the quantity n of estimation atural object classification, each class can be converted into by solving the threshold value of vegetation and non-vegetation
Other ωiAverage and variance point mi、Detailed process is as follows:
Count the frequency histogram f (x) of NDVI in piecemeal image.Search for the maximum NDVI of NDVI in piecemeal imagemaxWith
Minimum value NDVImin, by NDVIminTo NDVImax256 parts are equally divided into, statistics NDVI values appear in every a interval pixel count
Amount, obtains NDVI histogram f (x).
The single order for calculating NDVI histogram f (x) is reciprocal, and NDVI frequency histograms are found for 0 position according to single order inverse
Maximum point, minimum point in figure.
The quantity of maximum point is counted, ground mulching categorical measure n is determined.Each is used as using the NDVI values at maximum
The initial mean value m of ground mulching classificationi, initial σ is calculated by formula (5) with two neighboring minimum pointi.
σi=(NDVI2-NDVI1)/4 (5)
NDVI2, NDVI1Corresponding NDVI values at two neighboring minimum are represented respectively.
Wherein ncount is the number of all pixels of the NDVI values between two neighboring minimum, and n_total is piecemeal
Total number of pixels.
Based on maximal mathematical expectation (EM) algorithm, p (ω are estimated using loop iterationi)、mi、EM algorithms pass through circulation
Iteration, each iteration is by asking two steps of desired value and expectation maximization to constitute;The former according to the currency of parameter to be estimated, from
Observe data in direct estimation probability density desired value, the latter by maximizing this expectation come the estimator of undated parameter,
This two step in whole iterative process successively alternately, until iterative process convergence;Specific formula for calculation is as follows:
Three formulas estimation above is that k represents k-th of atural object classification, t and t in prior probability, average and standard deviation, formula respectively
+ 1 represents the estimate used in current and next iteration respectively, and i, j represents the line number and columns of NDVI images, X respectively
The NDVI values that the i-th row j is arranged in (i, j) expression NDVI images, and conditional probability p (X (i, j) | ωi) calculating see (3), full probability p
The value of (X (i, j)) is provided by formula (2);As the adjacent p (ω iterated to calculate twicei)、miAnd σiValue be less than given threshold epsilon (ε
=10-8) when iteration ends.
Determine the initial vegetation area of piecemeal.Assuming that the NDVI average values of each classification obtained in upper step are by from big to small
Sequence, m1, m2, mn, corresponding variance is σi, then initial vegetation distribution map is represented by VI (i, j), wherein 1 represents vegetation region
Domain, 0 represents nonvegetated area domain.
Then
Wherein threshold value T is shown in formula
1.3 steps are repeated, the vegetation prime area of all piecemeals is obtained.Merge the vegetation area in all piecemeals, obtain just
Beginning vegetation area.Then connective mark is carried out to whole image vegetation area, the pixel coordinate of each vegetative patch is obtained.
Connectedness mark can use 4 neighborhoods or 8 neighborhoods.
Step 2: to the multispectral image computation vision perceptual parameters of high spatial resolution.
2.1 pairs of multispectral images have RGB color space to transform to HSI color spaces.
The image of corresponding red, green, blue wave band in high-resolution multi-spectral image, calculates vegetation bright according to formula (10)
Degree, colourity, the result of variations of saturation degree.
2.2 computation vision characteristic perception parameters.
According to initial vegetative patch as sample, the brightness of the vegetation of each patch, colourity, the average of saturation degree are calculated
μx(H, S, I), variances sigmaxStatistical parameter values such as (H, S, I).Calculate normalization tone parameter of saturation
In formula, (i, j) represents pixel coordinate, μkThe average of k-th of vegetative patch, its corresponding variance is σ k, HL, HUFor
The lower and upper limit of vegetation tone, are shown in formula (11)
Step 3: calculating textural characteristics parameter to the panchromatic wave-band image of high spatial resolution
For panchromatic wave-band video generation gray level co-occurrence matrixes, 6 parametric textures of gray level co-occurrence matrixes are calculated:Contrast
(Con), diversity (Dis), angle energy (ASM), entropy (Ent), homogeneity (Hom), auto-correlation coefficient (Cor), specific formula are shown in
(11-17).The selection of textural characteristics computation window mouthful and the resolution ratio of panchromatic image and multispectral image have relation, and false color is panchromatic
The spatial resolution of wave band is n times of multispectral image, then the window that texture is calculated is 2n-1, is calculated once every n pixel.
Such as resolution ratio of quickbird images panchromatic wave-band image is 0.6 1m, and the resolution ratio of multispectral image is 2.4 4m, meter
The size of window is 7 × 7 when calculating panchromatic wave-band texture, and a textural characteristics are calculated every 4 pixels.So just can will be panchromatic
The textural characteristics of image are corresponding with the visually-perceptible feature and NDVI of multispectral image.
Step 4: the vegetation area automatic increase of combination features, obtains vegetation area distribution map
Using each initial vegetative patch as " seed ", the method increased using region will be adjacent to the pixel of " seed " patch
It is attached on seed region.Region increase criterion using textural characteristics, visually-perceptible characteristic synthetic using multiple features voting come
It is determined that.
The criterion that region increases is as follows
To each initial vegetative patch, brightness (I), normalization tone (H), saturation degree (S), contrast are counted respectively
(Contrast), diversity (Dissimilarity), angle energy (Energy), entropy (Entropy), homogeneity
(Homogeneity), auto-correlation coefficient, average, the variance of this 9 characteristic parameters, it is assumed that its value is μ, σ.To initial vegetation spot
8 neighborhood territory pixels of block, according to each feature, whether be vegetation, discriminant function is defined as follows if judging it
Wherein, M (i, j) refers to the characteristic parameter calculated, is exactly 3 visual signature parameters described in (1), 6 lines
Characteristic parameter is managed, 9, i.e. brightness, normalization tone, saturation degree, contrast, diversity, angle energy, an entropy is had, homogeneity, from phase
Relation number.Herein with M come instead of any one variable therein.Judgment threshold, i are represented in T2, j represents the ranks number of pixel, 1
Vegetation area is represented, 0 represents nonvegetated area domain.
So, can with brightness, normalization tone, saturation degree, contrast (Contrast), diversity (Dissimilarity),
Angle energy (Energy), entropy (Entropy), homogeneity (Homogeneity), auto-correlation coefficient, this 9 characteristic parameters can divide
A width Potential vegetation administrative division map is not obtained.
Multiple features voting determines final vegetation area.This 9 width Potential vegetation administrative division map is voted, i.e., in vegetation
Each pixel is when having 6 and its features above is collectively labeled as 1 on administrative division map, it is believed that be final vegetation area.
The urban vegetation extraction method for the multiple-spatial resolution remote sensing image that the present invention is provided, employs image block segmentation
Processing, accelerates image processing speed.According to NDVI features, using the automatic of the adaptive dynamic threshold of maximal mathematical expectation algorithm
Selection.The panchromatic wave-band of remote sensing image, multi light spectrum hands are made full use of, to what is extracted from panchromatic wave-band and multi light spectrum hands
NDVI features, visual signature, textural characteristics are integrated, and are carried out vegetation area judgement to vegetation, are improved accuracy.The present invention is provided
Method also there is certain applicability, the NDVI vegetation indexs in such as this method can change EVI indexes, SR indexes into.
It is described above, be only illustrate the present invention specific implementation case, be not limited to the present invention can
Practical range, such as those skilled in the art without departing from indicated by the present invention spirit with principle under completed all
Equivalent change or modification, should be covered by the scope of the claims in the present invention.
Claims (7)
1. a kind of urban vegetation extraction method of multiple-spatial resolution remote sensing image, it is characterised in that methods described mainly include with
Lower step:
Step 1: the initial patch of multispectral image vegetation based on high spatial resolution is extracted;Specifically include following steps:
Step 1.1 calculates NDVI vegetation spectral indexes, and calculation formula is
<mrow>
<mi>N</mi>
<mi>D</mi>
<mi>V</mi>
<mi>I</mi>
<mo>=</mo>
<mfrac>
<mrow>
<mi>N</mi>
<mi>I</mi>
<mi>R</mi>
<mo>-</mo>
<mi>R</mi>
</mrow>
<mrow>
<mi>N</mi>
<mi>I</mi>
<mi>R</mi>
<mo>+</mo>
<mi>R</mi>
</mrow>
</mfrac>
</mrow>
Wherein NIR represents the pixel value of near infrared band, and R represents the pixel value of red band;
Step 1.2 carries out piecemeal to NDVI images, per block size 300*300 pixels, in image edge part less than 300*300 pictures
When plain, it is defined by actual size;
Step 1.3 obtains segmentation threshold to the NDVI block datas of each 300*300 pixel and distinguishes vegetation and non-vegetation automatically,
Obtain initial vegetation area;
Step 1.4 repeat step 1.3, obtains the vegetation prime area of all piecemeals, merges the vegetation area in all piecemeals, obtains
To initial vegetation area, connective mark is carried out to whole image vegetation area, the pixel coordinate of each vegetative patch is obtained;
The acquisition of segmentation threshold is obtained by the following method in the step 1.3:
(1) the several ground mulching type of correspondence in each piecemeal NDVI, with ωiRepresent, it is assumed that the NDVI of each atural object classification
Gaussian density function distribution is obeyed in distribution, then probability-distribution function p (x) total NDVI in each piecemeal is represented by:
<mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<mi>p</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>|</mo>
<msub>
<mi>&omega;</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
Wherein, n represents the total quantity of classification, ωiI-th of atural object classification is represented, average and variance use m respectivelyi、Represent, its is general
Rate density fonction is expressed as
<mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>|</mo>
<msub>
<mi>&omega;</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<msqrt>
<mrow>
<mn>2</mn>
<msubsup>
<mi>&pi;&sigma;</mi>
<mi>i</mi>
<mn>2</mn>
</msubsup>
</mrow>
</msqrt>
</mfrac>
<mi>exp</mi>
<mo>{</mo>
<mo>-</mo>
<mfrac>
<msup>
<mrow>
<mo>(</mo>
<mi>x</mi>
<mo>-</mo>
<msub>
<mi>m</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mrow>
<mn>2</mn>
<msubsup>
<mi>&sigma;</mi>
<mi>i</mi>
<mn>2</mn>
</msubsup>
</mrow>
</mfrac>
<mo>}</mo>
</mrow>
Wherein, p (ωi) be each atural object classification probability density, its meet condition be:
<mrow>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<mi>p</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mn>1</mn>
</mrow>
Under these conditions, the quantity n of estimation atural object classification can be converted into by solving the threshold value of vegetation and non-vegetation, each classification
ωiAverage and variance point mi、Comprise the following steps that:
(2) the maximum NDVI of NDVI in the frequency histogram f (x) of NDVI in piecemeal image, search piecemeal image is countedmaxWith
Minimum value NDVImin, by NDVIminTo NDVImax256 parts are equally divided into, statistics NDVI values appear in every a interval pixel count
Amount, obtains NDVI histogram f (x);
(3) single order for calculating NDVI histogram f (x) is reciprocal, and NDVI frequency histograms are found for 0 position according to single order inverse
In maximum point, minimum point;
(4) quantity of maximum point is counted, ground mulching categorical measure n is determined, each is used as using the NDVI values at maximum
The initial mean value m of ground mulching classificationi, initial σ is calculated with two neighboring minimum pointiFormula it is as follows:
σi=(NDVI2-NDVI1)/4
Wherein, NDVI2, NDVIiCorresponding NDVI values at two neighboring minimum are represented respectively;
<mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mi>n</mi>
<mi>c</mi>
<mi>o</mi>
<mi>u</mi>
<mi>n</mi>
<mi>t</mi>
</mrow>
<mrow>
<mi>n</mi>
<mo>_</mo>
<mi>t</mi>
<mi>o</mi>
<mi>t</mi>
<mi>a</mi>
<mi>l</mi>
</mrow>
</mfrac>
</mrow>
1
Wherein, wherein ncount is the number of all pixels of the NDVI values between two neighboring minimum, and n_total is piecemeal
Total number of pixels;
(5) maximal mathematical expectation (EM) algorithm is based on, p (ω are estimated using loop iterationi)、mi、By loop iteration, every time
Iteration is by asking two steps of desired value and expectation maximization to constitute;The step of seeking desired value according to the currency of parameter to be estimated,
Updated from observation data the step of the desired value of direct estimation probability density, expectation maximization by maximizing this expectation
The estimator of parameter, this two step in whole iterative process successively alternately, until iterative process convergence;Specific formula for calculation
It is as follows:
<mrow>
<msup>
<mi>p</mi>
<mrow>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&Element;</mo>
<msub>
<mi>M</mi>
<mi>D</mi>
</msub>
</mrow>
</munder>
<mfrac>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>/</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>)</mo>
</mrow>
</mfrac>
</mrow>
<mrow>
<mi>I</mi>
<mo>*</mo>
<mi>I</mi>
</mrow>
</mfrac>
</mrow>
<mrow>
<msubsup>
<mi>m</mi>
<mi>k</mi>
<mrow>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&Element;</mo>
<msub>
<mi>M</mi>
<mi>D</mi>
</msub>
</mrow>
</munder>
<mfrac>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>/</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>)</mo>
</mrow>
</mfrac>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&Element;</mo>
<msub>
<mi>M</mi>
<mi>D</mi>
</msub>
</mrow>
</munder>
<mfrac>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>/</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
</mfrac>
</mrow>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>k</mi>
<mn>2</mn>
</msubsup>
<mo>)</mo>
</mrow>
<mrow>
<mi>t</mi>
<mo>+</mo>
<mn>1</mn>
</mrow>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&Element;</mo>
<msub>
<mi>M</mi>
<mi>D</mi>
</msub>
</mrow>
</munder>
<mfrac>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>/</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>)</mo>
</mrow>
</mfrac>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msubsup>
<mi>m</mi>
<mi>k</mi>
<mi>t</mi>
</msubsup>
<mo>&rsqb;</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>X</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&Element;</mo>
<msub>
<mi>M</mi>
<mi>D</mi>
</msub>
</mrow>
</munder>
<mfrac>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>/</mo>
<msub>
<mi>&omega;</mi>
<mi>k</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msup>
<mi>p</mi>
<mi>t</mi>
</msup>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
</mfrac>
</mrow>
</mfrac>
</mrow>
The estimation of above-mentioned three formula is that k represents k-th of atural object classification in prior probability, average and standard deviation, formula respectively, t and t+1 points
The estimate used in current and next iteration is not represented, and i, j represents the line number and columns of NDVI images, X (i, j) respectively
The NDVI values that the i-th row j is arranged in expression NDVI images, and conditional probability p (X (i, j) | ωi) calculating and full probability p (X (i, j))
Value is drawn by the respective formula in step (1);As the adjacent p (ω iterated to calculate twicei)、miAnd σiValue be less than given threshold
Iteration ends during value ε, threshold epsilon=10-8;
(6) the initial vegetation area of piecemeal is determined, it is assumed that the NDVI average values of each classification obtained in upper step are by from big to small
Sequence, m1, m2, mn, corresponding variance is σi, then initial vegetation distribution map is represented by VI (i, j), wherein 1 represents vegetation region
Domain, 0 represents nonvegetated area domain, then
<mrow>
<mi>V</mi>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mrow>
<mi>N</mi>
<mi>D</mi>
<mi>V</mi>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>></mo>
<mi>T</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mrow>
<mi>N</mi>
<mi>D</mi>
<mi>V</mi>
<mi>I</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>&le;</mo>
<mi>T</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein threshold value T is tried to achieve by equation below
<mrow>
<mi>T</mi>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>m</mi>
<mn>1</mn>
</msub>
<mo>-</mo>
<mn>1.5</mn>
<msub>
<mi>&sigma;</mi>
<mn>1</mn>
</msub>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>m</mi>
<mn>1</mn>
</msub>
<mo>-</mo>
<msub>
<mi>m</mi>
<mn>2</mn>
</msub>
<mo>></mo>
<mn>1.5</mn>
<mrow>
<mo>(</mo>
<msub>
<mi>&sigma;</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<msub>
<mi>&sigma;</mi>
<mn>2</mn>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>m</mi>
<mn>2</mn>
</msub>
<mo>+</mo>
<mn>1.5</mn>
<msub>
<mi>&sigma;</mi>
<mn>2</mn>
</msub>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>m</mi>
<mn>1</mn>
</msub>
<mo>-</mo>
<msub>
<mi>m</mi>
<mn>2</mn>
</msub>
<mo>&le;</mo>
<mn>1.5</mn>
<mrow>
<mo>(</mo>
<msub>
<mi>&sigma;</mi>
<mn>1</mn>
</msub>
<mo>+</mo>
<msub>
<mi>&sigma;</mi>
<mn>2</mn>
</msub>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>;</mo>
</mrow>
Step 2: to the multispectral image computation vision perceptual parameters of high spatial resolution;
Step 3: calculating textural characteristics parameter to the panchromatic wave-band image of high spatial resolution;
Step 4: the vegetation area automatic increase of combination features, obtains vegetation area distribution map.
2. the urban vegetation extraction method of multiple-spatial resolution remote sensing image according to claim 1, it is characterised in that:Institute
The connective mark stated in step 1.4 uses 4 neighborhoods or 8 neighborhoods.
3. the urban vegetation extraction method of multiple-spatial resolution remote sensing image according to claim 1, it is characterised in that:Institute
State step 2 and specifically include following steps:
Step 2.1 has RGB color space to transform to HSI color spaces to multispectral image, and it is bright to calculate vegetation by below equation
Degree, colourity, the result of variations of saturation degree
<mrow>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mtable>
<mtr>
<mtd>
<mrow>
<mi>H</mi>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>&theta;</mi>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<mi>G</mi>
<mo>&GreaterEqual;</mo>
<mi>B</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>2</mn>
<mi>&pi;</mi>
<mo>-</mo>
<mi>&theta;</mi>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>o</mi>
<mi>t</mi>
<mi>h</mi>
<mi>e</mi>
<mi>r</mi>
<mi>w</mi>
<mi>i</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>&theta;</mi>
<mo>=</mo>
<mi>arccos</mi>
<mfrac>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>-</mo>
<mi>G</mi>
<mo>)</mo>
<mo>+</mo>
<mo>(</mo>
<mi>R</mi>
<mo>-</mo>
<mi>B</mi>
<mo>)</mo>
</mrow>
<mrow>
<mn>2</mn>
<msqrt>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>-</mo>
<mi>G</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>-</mo>
<mi>B</mi>
<mo>)</mo>
</mrow>
<mrow>
<mo>(</mo>
<mi>G</mi>
<mo>-</mo>
<mi>B</mi>
<mo>)</mo>
</mrow>
</mrow>
</msqrt>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
</mtable>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>S</mi>
<mo>=</mo>
<mn>1</mn>
<mo>-</mo>
<mfrac>
<mn>3</mn>
<mrow>
<mi>R</mi>
<mo>+</mo>
<mi>G</mi>
<mo>+</mo>
<mi>B</mi>
</mrow>
</mfrac>
<mi>min</mi>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>,</mo>
<mi>G</mi>
<mo>,</mo>
<mi>B</mi>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>I</mi>
<mo>=</mo>
<mrow>
<mo>(</mo>
<mi>R</mi>
<mo>+</mo>
<mi>G</mi>
<mo>+</mo>
<mi>B</mi>
<mo>)</mo>
</mrow>
<mo>/</mo>
<mn>3</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>;</mo>
</mrow>
Step 2.2 computation vision characteristic perception parameter, according to initial vegetative patch as sample, calculates the vegetation of each patch
Brightness, colourity, the mean μ of saturation degreex(H, S, I), variances sigmax(H, S, I), calculates normalization tone parameter of saturation
<mrow>
<msup>
<mi>H</mi>
<mo>&prime;</mo>
</msup>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>-</mo>
<mfrac>
<mrow>
<mo>|</mo>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>-</mo>
<msub>
<mi>&mu;</mi>
<mi>k</mi>
</msub>
<mo>|</mo>
</mrow>
<msub>
<mi>&mu;</mi>
<mi>k</mi>
</msub>
</mfrac>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<msub>
<mi>H</mi>
<mi>L</mi>
</msub>
<mo><</mo>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo><</mo>
<msub>
<mi>H</mi>
<mi>U</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>0</mn>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>o</mi>
<mi>t</mi>
<mi>h</mi>
<mi>e</mi>
<mi>r</mi>
<mi>w</mi>
<mi>i</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, H (i, j) is tone, and (i, j) represents pixel coordinate, μkThe average of k-th of vegetative patch, its corresponding variance is
σk, HL, HUFor the lower and upper limit of vegetation tone
<mrow>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>H</mi>
<mi>L</mi>
</msub>
<mo>=</mo>
<msub>
<mi>&mu;</mi>
<mi>k</mi>
</msub>
<mo>-</mo>
<mn>2</mn>
<msub>
<mi>&sigma;</mi>
<mi>k</mi>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>H</mi>
<mi>U</mi>
</msub>
<mo>=</mo>
<msub>
<mi>&mu;</mi>
<mi>k</mi>
</msub>
<mo>+</mo>
<mn>2</mn>
<msub>
<mi>&sigma;</mi>
<mi>k</mi>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>.</mo>
</mrow>
4. the urban vegetation extraction method of multiple-spatial resolution remote sensing image according to claim 1, it is characterised in that institute
Stating step 3 is specially:
For panchromatic wave-band video generation gray level co-occurrence matrixes, 6 parametric textures of gray level co-occurrence matrixes are calculated:It is contrast, different
Property, angle energy, entropy, homogeneity, auto-correlation coefficient, specific formula for calculation are as follows:
<mrow>
<mi>C</mi>
<mi>O</mi>
<mi>N</mi>
<mo>=</mo>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</munder>
<msup>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>-</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>D</mi>
<mi>I</mi>
<mi>S</mi>
<mo>=</mo>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</munder>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>|</mo>
<mi>i</mi>
<mo>-</mo>
<mi>j</mi>
<mo>|</mo>
</mrow>
<mrow>
<mi>A</mi>
<mi>S</mi>
<mi>M</mi>
<mo>=</mo>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</munder>
<mi>p</mi>
<msup>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<mi>E</mi>
<mi>N</mi>
<mi>T</mi>
<mo>=</mo>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</munder>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mi>log</mi>
<mrow>
<mo>(</mo>
<mi>p</mi>
<mo>(</mo>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
<mo>)</mo>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mi>H</mi>
<mi>O</mi>
<mi>M</mi>
<mo>=</mo>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</munder>
<mfrac>
<mrow>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mn>1</mn>
<mo>+</mo>
<msup>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>-</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</mfrac>
</mrow>
<mrow>
<mi>C</mi>
<mi>O</mi>
<mi>R</mi>
<mo>=</mo>
<munder>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
</mrow>
</munder>
<mi>p</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mfrac>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>-</mo>
<msub>
<mi>&mu;</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
<mo>(</mo>
<mi>j</mi>
<mo>-</mo>
<msub>
<mi>&mu;</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mrow>
<msub>
<mi>&sigma;</mi>
<mi>x</mi>
</msub>
<msub>
<mi>&sigma;</mi>
<mi>y</mi>
</msub>
</mrow>
</mfrac>
</mrow>
The selection of textural characteristics computation window mouthful and the resolution ratio of panchromatic image and multispectral image have relation, it is assumed that panchromatic wave-band
Spatial resolution be n times of multispectral image, then the window that texture is calculated is 2n-1, every n pixel calculating once.
5. the urban vegetation extraction method of multiple-spatial resolution remote sensing image according to claim 1, it is characterised in that institute
Stating step 4 is specially:Using each initial vegetative patch as " seed ", the method increased using region will be adjacent to " seed " spot
The pixel of block is attached on seed region.
6. the urban vegetation extraction method of multiple-spatial resolution remote sensing image according to claim 5, it is characterised in that institute
The criterion for stating region growth is decided by vote to determine using textural characteristics, visually-perceptible characteristic synthetic using multiple features.
7. the urban vegetation extraction method of multiple-spatial resolution remote sensing image according to claim 6, it is characterised in that institute
The criterion for stating region growth is as follows:
(1) to each initial vegetative patch, brightness, normalization tone, saturation degree, contrast, diversity, angle energy are counted respectively
Amount, entropy, homogeneity, the average of auto-correlation coefficient and features described above parameter, variance, it is assumed that its value is μ, σ;To initial vegetation spot
8 neighborhood territory pixels of block, according to each feature, whether be vegetation, discriminant function is defined as follows if judging it
<mrow>
<msub>
<mi>T</mi>
<mn>2</mn>
</msub>
<mo>=</mo>
<mi>&mu;</mi>
<mo>-</mo>
<msqrt>
<mn>2</mn>
</msqrt>
<mi>&sigma;</mi>
</mrow>
<mrow>
<mi>V</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mn>1</mn>
<mo>,</mo>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<mi>M</mi>
<mrow>
<mo>(</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>)</mo>
</mrow>
<mo>></mo>
<msub>
<mi>T</mi>
<mn>2</mn>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>0</mn>
<mo>,</mo>
<mi>o</mi>
<mi>t</mi>
<mi>h</mi>
<mi>e</mi>
<mi>r</mi>
<mi>w</mi>
<mi>i</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, M (i, j) is the characteristic parameter calculated, in T2Judgment threshold, i are represented, j represents the ranks number of pixel, and 1 represents to plant
Nonvegetated area domain is represented by region, 0, and then obtains the corresponding Potential vegetation administrative division map of above-mentioned 9 characteristic parameters difference;
(2) multiple features voting determines final vegetation area, this 9 width Potential vegetation administrative division map is voted, i.e., in vegetation
Each pixel is when the feature for having 6 and its above is collectively labeled as 1 on administrative division map, it is believed that be final vegetation area.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510186167.6A CN104851113B (en) | 2015-04-17 | 2015-04-17 | The urban vegetation extraction method of multiple-spatial resolution remote sensing image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510186167.6A CN104851113B (en) | 2015-04-17 | 2015-04-17 | The urban vegetation extraction method of multiple-spatial resolution remote sensing image |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104851113A CN104851113A (en) | 2015-08-19 |
CN104851113B true CN104851113B (en) | 2017-11-03 |
Family
ID=53850740
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510186167.6A Expired - Fee Related CN104851113B (en) | 2015-04-17 | 2015-04-17 | The urban vegetation extraction method of multiple-spatial resolution remote sensing image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104851113B (en) |
Families Citing this family (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR102623403B1 (en) * | 2015-11-08 | 2024-01-12 | 어그로잉 리미티드 | Methods for aerial image acquisition and analysis |
CN105405148B (en) * | 2015-11-17 | 2017-11-14 | 中国科学院遥感与数字地球研究所 | A kind of remote sensing image Chinese white poplar recognition methods of combination trees shadow character |
CN105651384B (en) * | 2016-01-13 | 2018-01-16 | 南京大学 | A kind of all-optical information acquisition system |
CN107657631B (en) * | 2016-12-23 | 2018-09-25 | 航天星图科技(北京)有限公司 | A kind of generation method of the microwave spectrum mixed image based on vegetation distribution |
CN107194187B (en) * | 2017-06-06 | 2019-08-06 | 国家基础地理信息中心 | Take the ground mulching sample size calculation method of space composition and configuration into account |
CN108009542B (en) * | 2017-11-01 | 2021-06-15 | 华中农业大学 | Weed image segmentation method in rape field environment |
CN108089850B (en) * | 2018-01-02 | 2020-11-24 | 北京建筑大学 | Method for updating earth surface coverage product increment based on image collaborative segmentation and ecological geographical partitioning rule base |
CN109033543B (en) * | 2018-06-29 | 2020-06-30 | 北京师范大学 | Method, device and equipment for estimating vegetation coverage of heterogeneous land areas |
CN108986113A (en) * | 2018-07-06 | 2018-12-11 | 航天星图科技(北京)有限公司 | A kind of block parallel multi-scale division algorithm based on LLTS frame |
CN109583378A (en) * | 2018-11-30 | 2019-04-05 | 东北大学 | A kind of vegetation coverage extracting method and system |
CN110826480A (en) * | 2019-11-04 | 2020-02-21 | 湖南城市学院 | Water body automatic extraction method based on ant colony algorithm |
CN111402182B (en) * | 2020-03-18 | 2023-04-28 | 中国资源卫星应用中心 | Middle-divided image synthesis method based on land coverage information |
CN111814809B (en) * | 2020-07-17 | 2024-07-02 | 国家基础地理信息中心 | Remote sensing image vegetation information extraction method and device |
CN114273252A (en) * | 2021-11-26 | 2022-04-05 | 山东安信种苗股份有限公司 | Intelligent vegetable seedling grading method |
CN116188809B (en) * | 2023-05-04 | 2023-08-04 | 中国海洋大学 | Texture similarity judging method based on visual perception and sequencing driving |
CN116740580A (en) * | 2023-08-16 | 2023-09-12 | 山东绿博园市政工程有限公司 | Garden engineering data processing method based on remote sensing technology |
CN117132423B (en) * | 2023-08-22 | 2024-04-12 | 深圳云创友翼科技有限公司 | Park management system |
CN117152634B (en) * | 2023-09-12 | 2024-06-04 | 广东省水利水电科学研究院 | Multi-source satellite image floating plant identification method and system based on chromaticity index |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101832769A (en) * | 2010-03-30 | 2010-09-15 | 中国农业大学 | Method and system for estimating vegetation coverage degree in diggings based on close-up photography |
CN103093233A (en) * | 2012-12-03 | 2013-05-08 | 中国环境科学研究院 | Forest classification method based on object-oriented high-resolution remote sensing image |
CN103593858A (en) * | 2013-11-13 | 2014-02-19 | 北京林业大学 | Method for filtering out green vegetation data in ground laser radar scanning data |
EP2801951A1 (en) * | 2013-05-08 | 2014-11-12 | Honeywell International Inc. | Aerial image segmentation for refineries |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7038694B1 (en) * | 2002-03-11 | 2006-05-02 | Microsoft Corporation | Automatic scenery object generation |
JP5481140B2 (en) * | 2009-09-28 | 2014-04-23 | 株式会社日立ソリューションズ | Spectrum analyzer |
-
2015
- 2015-04-17 CN CN201510186167.6A patent/CN104851113B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101832769A (en) * | 2010-03-30 | 2010-09-15 | 中国农业大学 | Method and system for estimating vegetation coverage degree in diggings based on close-up photography |
CN103093233A (en) * | 2012-12-03 | 2013-05-08 | 中国环境科学研究院 | Forest classification method based on object-oriented high-resolution remote sensing image |
EP2801951A1 (en) * | 2013-05-08 | 2014-11-12 | Honeywell International Inc. | Aerial image segmentation for refineries |
CN103593858A (en) * | 2013-11-13 | 2014-02-19 | 北京林业大学 | Method for filtering out green vegetation data in ground laser radar scanning data |
Non-Patent Citations (2)
Title |
---|
《一种面向对象的遥感影像城市绿地提取方法》;李成范等;《测绘科学》;20110930;第36卷(第5期);全文 * |
《面向对象的城市绿地信息提取方法研究》;熊轶群等;《华东师范大学学报(自然科学版)》;20060731(第4期);第85页第1-2段、第5-6段, 第86页第8段以及图3,第85页最后一段 * |
Also Published As
Publication number | Publication date |
---|---|
CN104851113A (en) | 2015-08-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104851113B (en) | The urban vegetation extraction method of multiple-spatial resolution remote sensing image | |
CN108573276B (en) | Change detection method based on high-resolution remote sensing image | |
CN103971115B (en) | Automatic extraction method for newly-increased construction land image spots based on NDVI and PanTex index | |
CN101840581B (en) | Method for extracting profile of building from satellite remote sensing image | |
CN102902956B (en) | A kind of ground visible cloud image identifying processing method | |
CN107862667A (en) | A kind of city shadow Detection and minimizing technology based on high-resolution remote sensing image | |
CN108009542A (en) | Weed images dividing method under rape field environment | |
CN105550685B (en) | The large format remote sensing image area-of-interest exacting method of view-based access control model attention mechanism | |
CN104951799B (en) | A kind of SAR remote sensing image oil spilling detection recognition method | |
CN103077515B (en) | Multi-spectral image building change detection method | |
CN106611421B (en) | The SAR image segmentation method constrained based on feature learning and sketch line segment | |
Zheng et al. | Large-scale oil palm tree detection from high-resolution remote sensing images using faster-rcnn | |
CN107832797B (en) | Multispectral image classification method based on depth fusion residual error network | |
CN104217440B (en) | A kind of method extracting built-up areas from remote sensing images | |
CN112488050A (en) | Color and texture combined aerial image scene classification method and system | |
CN104102928B (en) | A kind of Classifying Method in Remote Sensing Image based on texture primitive | |
CN104835196B (en) | A kind of vehicle mounted infrared image colorization three-dimensional rebuilding method | |
CN107818303A (en) | Unmanned plane oil-gas pipeline image automatic comparative analysis method, system and software memory | |
CN107292328A (en) | The remote sensing image shadow Detection extracting method and system of multiple dimensioned multiple features fusion | |
CN110879992A (en) | Grassland surface covering object classification method and system based on transfer learning | |
CN107992856A (en) | High score remote sensing building effects detection method under City scenarios | |
JP6358862B2 (en) | Forest phase analysis apparatus and program | |
CN110458208A (en) | Hyperspectral image classification method based on information measure | |
CN107564016A (en) | A kind of Multi-Band Remote Sensing Images segmentation of integrally object light spectrum information and labeling method | |
CN107169946A (en) | Image interfusion method based on non-negative sparse matrix Yu hypersphere color transformation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
DD01 | Delivery of document by public notice |
Addressee: Dian Yuanyong Document name: Notification of Acceptance of Patent Application |
|
C06 | Publication | ||
PB01 | Publication | ||
EXSB | Decision made by sipo to initiate substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171103 Termination date: 20180417 |