CN105405133A - Remote sensing image alteration detection method - Google Patents

Abstract

The invention discloses a remote sensing image alteration detection method comprising the steps that two time phase high-resolution optical remote sensing images X1 and X2 are obtained; image registration is performed on X1 and X2; radiation normalization correction is performed on X1 and X2 by utilizing a multivariate alteration detection method; alteration vector amplitude XM and spectral angle information XSA are respectively calculated according to X1 and X2 after radiation normalization correction; an optimal segmentation threshold T is calculated by utilizing the Bayes theory and an expectation maximization algorithm according to the XM; a pseudo training sample area is selected according to T and XM; XM and XSA are combined to act as input of a core FCM, and optimal model parameter value selection is performed on a core FCM combining spatial neighborhood information model according to the pseudo training sample area; and the alteration area and the non-alteration area of the optical remote sensing images are determined by adopting the method of core FCM combining spatial neighborhood information according to the selected optimal model parameter values. The remote sensing image alteration detection method is more robust and higher in precision.

Description

A kind of remote sensing image variation detection method

Technical field

The present invention relates to remote sensing image change detection techniques field, particularly relate to a kind of remote sensing image variation detection method.

Background technology

Along with the continuous accumulation of multidate high-definition remote sensing data and the foundation in succession of spatial database, how to extract from these remotely-sensed datas and change detected information has become the important subject of remote sensing science and Geographical Information Sciences.According to the remote sensing image of the different phase of the same area, the information of the dynamic change such as city, environment can be extracted, for resource management and the department such as planning, environmental protection provide the foundation of science decision.China " 12 " implements strengthening to expand the high resolving power earth observation engineering that Eleventh Five-Year Plan has started enforcement, concern comprises high-definition remote sensing target and basic theory and the key technology research such as space environment signature analysis and high reliability automatic interpretation, is becoming the research focus solving national security and the great demand of socio-economic development.

It is exactly in remotely-sensed data never of the same period that the change of remote sensing image detects, and analyzes quantitatively and determines the feature that earth's surface changes and process.Scholars proposes many effective detection algorithms from different angles and applied research, as Change vector Analysis method (ChangeVectorAnalysis, CVA), clustering method etc. based on FuzzyC-means (FCM).Wherein, traditional change of the multidate remote optical sensing based on FCM detects, and how first to carry out CVA conversion, then carries out FCM cluster to the amplitude of diverse vector, and then obtains changing testing result.In such technology, the deficiency using FCM is only applicable to spherical or ellipsoid shape cluster, and to noise and outlier (Outlier) very sensitivity thereof.In addition, only use the amplitude of diverse vector, original multispectral information is not excavated fully, not steadily and surely, precision is not high.

For the problems referred to above, many scholars attempt by adding that in FCM objective function the constraint of different spatial neighborhoods solves, but the complicated and target prior imformation of high resolution image testing environment is deficient, and cause these algorithms all to there is certain limitation, precision is not high.For this reason, be necessary that the new High Resolution Visible Light Remote Sensing Imagery Change Detection technology of research effectively overcomes above-mentioned difficult point.

Summary of the invention

Technical matters to be solved by this invention is, a kind of remote sensing image variation detection method is provided, the method is a kind of multi-temporal remote sensing image change detecting method of combining the self-adaptive kernel FCM of CVA and SAM, and the present invention changes that testing result is more sane, precision is higher.

In order to solve the problems of the technologies described above, the invention provides a kind of remote sensing image variation detection method, comprising:

Obtain two phase high-resolution optical remote sensing image X ₁and X ₂;

To optical remote sensing image X ₁and X ₂carry out Image registration;

Utilize Multivariate alteration detection method to optical remote sensing image X ₁and X ₂carry out radiation normalization correction;

Optical remote sensing image X after correcting according to radiation normalization ₁and X ₂calculate diverse vector amplitude X respectively _mwith spectral modeling information X _sA;

According to diverse vector amplitude X _mbayes principle and EM algorithm is utilized to calculate optimum segmentation threshold value T;

According to optimum segmentation threshold value T and diverse vector amplitude X _mselect pseudo-training sample region;

By X _mand X _sAcombine the input as core FCM, according to described pseudo-training sample region, in conjunction with space neighborhood information model, the selection of optimization model parameter value is carried out to core FCM;

According to the optimization model parameter value selected, adopt core FCM in conjunction with the method for space neighborhood information, determine region of variation and the non-changing region of optical remote sensing image.

Implement the present invention, there is following beneficial effect: the present invention combines the diverse vector amplitude of multi-temporal remote sensing image and the spectral modeling mapping graph (SpectralAngleMapper of multidate, SAM) as the input of core FCM, again based on the method for core FCM in conjunction with space neighborhood information, obtain final change testing result.Wherein, the nuclear parameter etc. in core FCM objective function, is selected by the pseudo-training sample based on CVA technical limit spacing, and change testing result is more sane, precision is higher.

Accompanying drawing explanation

In order to be illustrated more clearly in the embodiment of the present invention or technical scheme of the prior art, be briefly described to the accompanying drawing used required in embodiment or description of the prior art below, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.

Fig. 1 is the schematic flow sheet of an embodiment of remote sensing image variation detection method provided by the invention;

Fig. 2 is original high resolution optical remote sensing image figure;

Fig. 3 is the experimental result comparison diagram of the inventive method and additive method

Embodiment

Below in conjunction with the accompanying drawing in the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.

Fig. 1 is the schematic flow sheet of an embodiment of remote sensing image variation detection method provided by the invention, and the present invention is a kind of multi-temporal remote sensing image change detecting method, is mainly applicable to high-resolution optical remote sensing image, as shown in Figure 1, the present invention includes step:

S101, obtain two phase high-resolution optical remote sensing image X ₁and X ₂.

Wherein, X ₁, X ₂two panel height resolution Optical remote sensing images of the different phase of the same area.

S102, to optical remote sensing image X ₁and X ₂carry out Image registration.

Concrete, step S102 specifically comprises step:

S1021, employing ENVI14.8 remote sensing software are to optical remote sensing image X ₁and X ₂carry out geometric approximate correction.

Geometric approximate correction concrete operation step is: (1) display reference images and image to be corrected; (2) ground control point GCPs is gathered; GCPs should be evenly distributed in entire image, and the number of GCPs is at least more than or equal to 9; (3) error of calculation; (4) multinomial model is selected; (5) bilinear interpolation is adopted to carry out resampling output.Bilinearity differential technique is wherein: if ask unknown function f in a value of P=(x, y), supposes that our known function f is at Q ₁₁=(x ₁, y ₁), Q ₁₂=(x ₁, y ₂), Q ₂₁=(x ₂, y ₁), and Q ₂₂=(x ₂, y ₂) value of four points.If select a coordinate system to make the coordinate of these four points be respectively (0,0), (0,1), (1,0) and (1,1), so bilinear interpolation formula just can be expressed as:

f(x,y)≈f(0,0)(1-x)(1-y)+f(1,0)x(1-y)+f(0,1)(1-x)y+f(1,1)xy。

S1022, utilize Auto-matching and Triangulation Method to the X after geometric approximate correction ₁and X ₂carry out geometric accurate correction.

Wherein, Triangulation Method is, incremental algorithm is adopted to build Delaunay triangulation network, to each triangle, utilize the geographic coordinate of the ranks on its three summits number reference images same place corresponding with it to determine the affine Transform Model parameter of this triangle interior, treat correcting image to correct, obtain the remote sensing image after correcting.

S103, utilize Multivariate alteration detection method (MultivariateAlterationDetection, MAD) to optical remote sensing image X ₁and X ₂carry out radiation normalization correction.

Concrete, step S103 specifically comprises step:

S1031, acquisition optical remote sensing image X ₁and X ₂the linear combination of each wave band brightness value, obtains the difference image that change information strengthens;

S1032, according to described difference image by threshold value determination region of variation and non-region of variation;

S1033, by the right mapping equation of two phase pixels corresponding to non-region of variation, complete relative detector calibration.

S104, according to radiation normalization correct after optical remote sensing image X ₁and X ₂calculate diverse vector amplitude X respectively _mwith spectral modeling information X _sA.

Concrete, step S104 comprises step:

S1041, according to radiation normalization correct after optical remote sensing image X ₁and X ₂calculate diverse vector amplitude X _m.

Wherein, $X_M(i,j)=\sqrt{\underset{b=1}{\overset{B}{\Σ}}{\left(X_{1b}\right(i,j)-X_{2b}(i,j\left)\right)}^2},$ In formula, B represents the wave band number of each phase remote sensing image, and (i, j) is the coordinate of image, X _1brepresent X ₁b wave band image, X _2brepresent X ₂b wave band image;

S1042, according to radiation normalization correct after optical remote sensing image X ₁and X ₂calculate diverse vector amplitude X _m,

Wherein, $X_{SA}(i,j)=arccos\left(\underset{b=1}{\overset{B}{\Σ}}\right(X_{1b}\left(i,j\right)X_{2b}\left(i,j\right)\left)\sqrt{\underset{b=1}{\overset{B}{\Σ}}{X}_{1b}^2(i,j)}\sqrt{\underset{b=1}{\overset{B}{\Σ}}{X}_{2b}^2(i,j)}\right).$

S105, according to diverse vector amplitude X _mbayes principle and EM algorithm (Expectation-Maximization, EM) is utilized to calculate optimum segmentation threshold value T.

Concrete, step S105 specifically comprises step:

S1051, employing EM algorithm estimate X _mimage does not change class ω _naverage m _nand variances sigma _n, change class ω _caverage m _cbe σ with variance _c, wherein,

$m_n^{t+1}=\left\{\underset{X(i,j)\∈X_M}{\mathrm{\Σ}}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}X(i,j)\right\}/\left\{\underset{X(i,j)\∈X_M}{\mathrm{\Σ}}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}\right\}$

${\left({\mathrm{\σ}}_n^2\right)}^{t+1}=\{\underset{X(i,j)\∈X_M}{\mathrm{\Σ}}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}\[X(i,j)-m_n^t\]\}/\left\{\underset{X(i,j)\∈X_M}{\mathrm{\Σ}}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}\right\}$

$m_c^{t+1}=\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j\left|{\mathrm{\ω}}_c\right))}{p^t\left(X\right(i,j))}X(i,j)\right\}/\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j\left|{\mathrm{\ω}}_c\right))}{p^t\left(X\right(i,j))}\right\}$

${\left({\mathrm{\σ}}_c^2\right)}^{t+1}=\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j\left|{\mathrm{\ω}}_c\right))}{p^t\left(X\right(i,j))}\[X(i,j)-m_c^t\]\}/\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j\left|{\mathrm{\ω}}_c\right))}{p^t\left(X\right(i,j))}\right\}$

In formula, t represents iterations, and t subscript represents value during the t time iteration of Current Content, such as, represent m _nvalue during the t+1 time iteration, other representation classes seemingly, represent value during the t+1 time iteration, $p^{t+1}\left({\mathrm{\ω}}_n\right)=\frac{\underset{X(i,j)\∈X_M}{\mathrm{\Σ}}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}}{IJ},p^{t+1}\left({\mathrm{\ω}}_c\right)=\frac{\underset{X(i,j)\∈X_M}{\mathrm{\Σ}}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_c\left)\right)}{p^t\left(X\right(i,j\left)\right)}}{IJ},$ I and J represents line number and the columns of image respectively, represent X _mimage does not change class ω _nthe Gaussian distribution of obeying, represent X _mimage changes class ω _cthe Gaussian distribution of obeying;

S1052, according to Bayes minimum error principle, solution formula $({\mathrm{\σ}}_n^2-{\mathrm{\σ}}_c^2)T^2+2(m_n{\mathrm{\σ}}_c^2-m_c{\mathrm{\σ}}_n^2)T+m_c^2{\mathrm{\σ}}_n^2-m_n^2{\mathrm{\σ}}_c^2-2{\mathrm{\σ}}_n^2{\mathrm{\σ}}_c^{2}ln\[\frac{{\mathrm{\σ}}_{c}p\left({\mathrm{\ω}}_n\right)}{{\mathrm{\σ}}_{n}p\left({\mathrm{\ω}}_c\right)}\]=0,$ Obtain optimum segmentation threshold value T.

S106, according to optimum segmentation threshold value T and diverse vector amplitude X _mselect pseudo-training sample region.

Concrete, step S106 comprises step:

S1061, according to optimum segmentation threshold value T and diverse vector amplitude X _mselect not change the pseudo-training set sample of class and be

S1062, according to optimum segmentation threshold value T and diverse vector amplitude X _mselect to change the pseudo-training set sample of class and be wherein, δ is X _m15% of dynamic range.

S107, by X _mand X _sAcombine the input as core FCM, according to described pseudo-training sample region, in conjunction with space neighborhood information model, the selection of optimization model parameter value is carried out to core FCM.

Concrete, step S107 specifically comprises step:

S1071, by X _mand X _sAcombine the input as core FCM, building core FCM in conjunction with space neighborhood information model is: $J_m=\underset{j=1}{\overset{C}{\mathrm{\Σ}}}\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{u}_{jk}^m(1-K(X_{MS}\left(k\right),v_j\left)\right)+\mathrm{\α}\underset{j=1}{\overset{C}{\mathrm{\Σ}}}\underset{k=1}{\overset{N}{\mathrm{\Σ}}}{u}_{jk}^m(1-K({\stackrel{\‾}{X}}_{MS}\left(k\right),v_j\left)\right),$

In formula, C is clusters number, and N is the sum of sample, represent the fuzzy membership of kth sample for jth class cluster centre, m is the weighted index of degree of membership, parameter alpha controls punishment effect, for X _mlocal mean value image and X _sAthe combination of local mean value image, $K(x,y)=\exp \{-\frac{{(x-y)}^2}{g^2}\}.$

S1072, setup parameter α and nuclear parameter g span, utilize pseudo-training sample set, search variability index C _indexfor the value of α and g time minimum is as optimization model parameter value.

Wherein, variability index C _index=D _index/ k _t, k _trepresent the Kappa coefficient of model parameter on pseudo-training sample set, N _n(α, g) represents the non-changing number of pixels of the whole image utilizing the minimization of object function to obtain when given α and g; N _c(α, g) represents when given α and g, the change number of pixels of whole image; TN _n(α, g) represents when given α and g, the non-changing number of pixels that pseudo-training sample is concentrated; TN _c(α, g) represents when given α and g, the change number of pixels that pseudo-training sample is concentrated.

S108, according to the optimization model parameter value selected, adopt core FCM in conjunction with the method for space neighborhood information, determine region of variation and the non-changing region of optical remote sensing image.

Concrete, step S108 specifically comprises:

S1081, setting core FCM, in conjunction with the clusters number C=2 in space neighborhood information model, as the initial center not changing class and change class, select and diverse vector amplitude X _mthe vector that minimum value is corresponding with maximal value; If Weighting exponent m=2 of degree of membership, ε be greater than 0 constant, the value of parameter alpha and nuclear parameter g is selected described optimization model parameter value;

S1082, calculating X _m, X _sAlocal window average, window size is set to 3 × 3;

S1083, employing formula $u_{jk}=\frac{{\left(\right(1-K\left(X_{MS}\right(k),v_j))+\mathrm{\α}(1-K\left({\stackrel{\‾}{X}}_{MS}\right(k),v_j)\left)\right)}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\Σ}}{\left(\right(1-K\left(X_{MS}\right(k),v_j))+\mathrm{\α}(1-K\left({\stackrel{\‾}{X}}_{MS}\right(k),v_j)\left)\right)}^{-1/(m-1)}}$ Upgrade fuzzy partition matrix;

S1084, employing formula $v_j=\frac{\underset{k=1}{\overset{N}{\Σ}}{u}_{jk}^m\left(K\right(X_{MS}\left(k\right),v_j)X_{MS}\left(k\right)+\mathrm{\α}K\left({\stackrel{\‾}{X}}_{MS}\right(k),v_j){\stackrel{\‾}{X}}_{MS}\left(k\right))}{\underset{k=1}{\overset{N}{\Σ}}{u}_{jk}^m\left(K\right(X_{MS}\left(k\right),v_j)+\mathrm{\α}K\left({\stackrel{\‾}{X}}_{MS}\right(k),v_j))}$ Upgrade cluster centre;

S1085, repeat to upgrade fuzzy partition matrix and cluster centre until the cluster centre cluster of adjacent twice iteration is less than ε;

S1086, according to fuzzy partition matrix u _jkdetermine final change detection figure, obtain region of variation and the non-changing region of optical remote sensing image.

Effect of the present invention further illustrates by following experimental result and analysis:

Experimental data of the present invention is the multidate SPOT high-resolution image data in French Littoral area, and image size is 400 × 400, uses B1, B2 and B3 tri-wave bands.In order to verify validity of the present invention, change detecting method of the present invention and following change detecting method are compared:

(1) based on the EM method (CVA-EM) of CVA, [gondola BruzzoneL. etc. are at article " Automaticanalysisofdifferenceimageforunsupervisedchanged etection " (IEEETransactionsonGeoscienceandRemoteSensing, 2000,38 (3): 1171-1182.) detection method carried in].

(2) [Chensongchan etc. are at article " RobustImageSegmentationUsingFCMWithSpatialConstraintsBas edonNewKernel-InducedDistanceMeasure " (IEEETransactionsonSystems in conjunction with the sorting technique (FCM-S) of space neighborhood information for FCM, Man, andCybernetics-PartB:Cybernetics, 2004,34 (4): 1907-1916.) method carried in]

(3) the inventive method.

Detection perform false retrieval number FP, undetected several FN, total error number OE and Kappa coefficient four indexs are weighed.FP, FN and OE more close to 0, Kappa coefficient more close to 1, show that the performance of change detecting method is better.Testing result is as shown in table 1.From Fig. 2, Fig. 3 and table 1, the detection method performance that the present invention carries is better than other two kinds of detection methods, and this shows that the change detecting method that the present invention carries is effective.

The multidate SPOT5 remote sensing imagery change detection results contrast in table 1Littoral area

Method	FP	FN	OE	k
					CVA-EM	7919	3882	11801	0.705
FCM-S	1822	6928	8750	0.737
					The inventive method	2511	4689	7200	0.797
Desirable	0	0	0	1

It should be noted that, in this article, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or device and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or device.When not more restrictions, the key element limited by statement " comprising ... ", and be not precluded within process, method, article or the device comprising this key element and also there is other identical element.

In several embodiments that the application provides, should be understood that, disclosed method can realize by another way.Such as, system embodiment described above is only schematic, professional can also recognize further, in conjunction with unit and the algorithm steps of each example of embodiment disclosed herein description, can realize with electronic hardware, computer software or the combination of the two, in order to the interchangeability of hardware and software is clearly described, generally describe composition and the step of each example in the above description according to function.These functions perform with hardware or software mode actually, depend on application-specific and the design constraint of technical scheme.Professional and technical personnel can use distinct methods to realize described function to each specifically should being used for, but this realization should not thought and exceeds scope of the present invention.Software module can be placed in the storage medium of other form any known in random access memory (RAM), internal memory, ROM (read-only memory) (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technical field.

To the above-mentioned explanation of the disclosed embodiments, professional and technical personnel in the field are realized or uses the present invention.To be apparent for those skilled in the art to the multiple amendment of these embodiments, General Principle as defined herein can without departing from the spirit or scope of the present invention, realize in other embodiments.Therefore, the present invention can not be restricted to these embodiments shown in this article, but will meet the widest scope consistent with principle disclosed herein and features of novelty.

Claims (8)

1. a remote sensing image variation detection method, is characterized in that, comprising:

Obtain two phase high-resolution optical remote sensing image X ₁and X ₂;

To optical remote sensing image X ₁and X ₂carry out Image registration;

Utilize Multivariate alteration detection method to optical remote sensing image X ₁and X ₂carry out radiation normalization correction;

Optical remote sensing image X after correcting according to radiation normalization ₁and X ₂calculate diverse vector amplitude X respectively _mwith spectral modeling information X _sA;

According to diverse vector amplitude X _mbayes principle and EM algorithm is utilized to calculate optimum segmentation threshold value T;

According to optimum segmentation threshold value T and diverse vector amplitude X _mselect pseudo-training sample region;

By X _mand X _sAcombine the input as core FCM, according to described pseudo-training sample region, in conjunction with space neighborhood information model, the selection of optimization model parameter value is carried out to core FCM;

According to the optimization model parameter value selected, adopt core FCM in conjunction with the method for space neighborhood information, determine region of variation and the non-changing region of optical remote sensing image.

2. remote sensing image variation detection method as claimed in claim 1, is characterized in that, described to optical remote sensing image X ₁and X ₂carry out Image registration, specifically comprise:

Adopt ENVI14.8 remote sensing software to optical remote sensing image X ₁and X ₂carry out geometric approximate correction;

Utilize Auto-matching and Triangulation Method to the X after geometric approximate correction ₁and X ₂carry out geometric accurate correction.

3. remote sensing image variation detection method as claimed in claim 1, it is characterized in that, the described Multivariate alteration detection method that utilizes is to optical remote sensing image X ₁and X ₂carry out radiation normalization correction, specifically comprise:

Obtain optical remote sensing image X ₁and X ₂the linear combination of each wave band brightness value, obtains the difference image that change information strengthens;

According to described difference image by threshold value determination region of variation and non-region of variation;

By the mapping equation that the two phase pixels that non-region of variation is corresponding are right, complete relative detector calibration.

4. remote sensing image variation detection method as claimed in claim 1, is characterized in that, described according to the optical remote sensing image X after radiation normalization correction ₁and X ₂calculate diverse vector amplitude X respectively _mwith spectral modeling information X _sA, specifically comprise:

Optical remote sensing image X after correcting according to radiation normalization ₁and X ₂calculate diverse vector amplitude X _m, wherein, $X_M(i,j)=\sqrt{\underset{b=1}{\overset{B}{\Σ}}{\left(X_{1b}\right(i,j)-X_{2b}(i,j\left)\right)}^2},$ In formula, B represents the wave band number of each phase remote sensing image, and (i, j) is the coordinate of image, X _1brepresent X ₁b wave band image, X _2brepresent X ₂b wave band image;

Optical remote sensing image X after correcting according to radiation normalization ₁and X ₂calculate diverse vector amplitude X _m, wherein, $X_{SA}(i,j)=arccos\left(\underset{b=1}{\overset{B}{\Σ}}\right(X_{1b}\left(i,j\right)X_{2b}\left(i,j\right))/\sqrt{\underset{b=1}{\overset{B}{\Σ}}{X}_{1b}^2(i,j)}\sqrt{\underset{b=1}{\overset{B}{\Σ}}{X}_{2b}^2(i,j)}).$

5. remote sensing image variation detection method as claimed in claim 1, is characterized in that, described according to diverse vector amplitude X _mutilize Bayes principle and EM algorithm to calculate optimum segmentation threshold value T, specifically comprise:

EM algorithm is adopted to estimate X _mimage does not change class ω _naverage m _nand variances sigma _n, change class ω _caverage m _cbe σ with variance _c, wherein,

$m_n^{t+1}=\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}X(i,j)\right\}/\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}\right\}$

${\left({\mathrm{\σ}}_n^2\right)}^{t+1}=\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}\[X(i,j)-m_n^t\]\}/\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}\right\}$

$m_c^{t+1}=\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_c\left)\right)}{p^t\left(X\right(i,j\left)\right)}X(i,j)\right\}/\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_c\left)\right)}{p^t\left(X\right(i,j\left)\right)}\right\}$

${\left({\mathrm{\σ}}_c^2\right)}^{t+1}=\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_c\left)\right)}{p^t\left(X\right(i,j\left)\right)}\[X(i,j)-m_c^t\]\}/\left\{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_c\left)\right)}{p^t\left(X\right(i,j\left)\right)}\right\}$

In formula, t represents iterations, and t subscript represents value during the t time iteration of Current Content,

$p^{t+1}\left({\mathrm{\ω}}_n\right)=\frac{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_n\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_n\left)\right)}{p^t\left(X\right(i,j\left)\right)}}{IJ},p^{t+1}\left({\mathrm{\ω}}_c\right)=\frac{\underset{X(i,j)\∈X_M}{\Σ}\frac{p^t\left({\mathrm{\ω}}_c\right)p^t\left(X\right(i,j|{\mathrm{\ω}}_c\left)\right)}{p^t\left(X\right(i,j\left)\right)}}{IJ},$

I and J represents line number and the columns of image respectively, represent X _mimage does not change class ω _nthe Gaussian distribution of obeying, represent X _mimage changes class ω _cthe Gaussian distribution of obeying;

According to Bayes minimum error principle, solution formula $({\mathrm{\σ}}_n^2-{\mathrm{\σ}}_c^2)T^2+2(m_n{\mathrm{\σ}}_c^2-m_c{\mathrm{\σ}}_n^2)T+m_c^2{\mathrm{\σ}}_n^2-m_n^2{\mathrm{\σ}}_c^2-2{\mathrm{\σ}}_n^2{\mathrm{\σ}}_c^{2}ln\[\frac{{\mathrm{\σ}}_{c}p\left({\mathrm{\ω}}_n\right)}{{\mathrm{\σ}}_{n}p\left({\mathrm{\ω}}_c\right)}\]=0,$ Obtain optimum segmentation threshold value T.

6. remote sensing image variation detection method as claimed in claim 1, is characterized in that, described according to optimum segmentation threshold value T and diverse vector amplitude X _mselect pseudo-training sample region, specifically comprise:

According to optimum segmentation threshold value T and diverse vector amplitude X _mselect not change the pseudo-training set sample of class and be

According to optimum segmentation threshold value T and diverse vector amplitude X _mselect to change the pseudo-training set sample of class and be

Wherein, δ is X _m15% of dynamic range.

7. remote sensing image variation detection method as claimed in claim 1, is characterized in that, described by X _mand X _sAcombine the input as core FCM, according to described pseudo-training sample region, in conjunction with space neighborhood information model, optimization model Selecting parameter carried out to core FCM, specifically comprise:

By X _mand X _sAcombine the input as core FCM, building core FCM in conjunction with space neighborhood information model is:

$J_m=\underset{j=1}{\overset{C}{\Σ}}\underset{k=1}{\overset{N}{\Σ}}{u}_{jk}^m(1-K(X_{MS}\left(k\right),v_j\left)\right)+\mathrm{\α}\underset{j=1}{\overset{C}{\Σ}}\underset{k=1}{\overset{N}{\Σ}}{u}_{jk}^m(1-K({\stackrel{\‾}{X}}_{MS}\left(k\right),v_j\left)\right),$ In formula, C is clusters number, and N is the sum of sample, represent the fuzzy membership of kth sample for jth class cluster centre, m is the weighted index of degree of membership, u _jk∈ [0,1] and parameter alpha controls punishment effect, for X _mlocal mean value image and X _sAthe combination of local mean value image,

Setup parameter α and nuclear parameter g span, utilize pseudo-training sample set, search variability index C _indexfor the value of α and g time minimum is as optimization model parameter value; Wherein, variability index C _index=D _index/ k _t, k _trepresent the Kappa coefficient of model parameter on pseudo-training sample set, N _n(α, g) represents the non-changing number of pixels of the whole image utilizing the minimization of object function to obtain when given α and g; N _c(α, g) represents when given α and g, the change number of pixels of whole image; TN _n(α, g) represents when given α and g, the non-changing number of pixels that pseudo-training sample is concentrated; TN _c(α, g) represents when given α and g, the change number of pixels that pseudo-training sample is concentrated.

8. remote sensing image variation detection method as claimed in claim 7, it is characterized in that the described optimization model parameter value according to selecting adopts core FCM in conjunction with the method for space neighborhood information, determine region of variation and the non-changing region of high-resolution optical remote sensing image, specifically comprise:

Setting core FCM, in conjunction with the clusters number C=2 in space neighborhood information model, as the initial center not changing class and change class, selects and diverse vector amplitude X _mthe vector that minimum value is corresponding with maximal value; If Weighting exponent m=2 of degree of membership, ε be greater than 0 constant, the value of parameter alpha and nuclear parameter g is selected described optimization model parameter value;

Calculate X _m, X _sAlocal window average, window size is set to 3 × 3;

Employing formula $u_{jk}=\frac{{\left(\right(1-K\left(X_{MS}\right(k),v_j))+\mathrm{\α}(1-K\left({\stackrel{\‾}{X}}_{MS}\right(k),v_j)\left)\right)}^{-1/(m-1)}}{\underset{j=1}{\overset{C}{\Σ}}{\left(\right(1-K\left(X_{MS}\right(k),v_j))+\mathrm{\α}(1-K\left({\stackrel{\‾}{X}}_{MS}\right(k),v_j)\left)\right)}^{-1/(m-1)}}$ Upgrade fuzzy partition matrix;

Employing formula $v_j=\frac{\underset{k=1}{\overset{N}{\Σ}}{u}_{jk}^m\left(K\right(X_{MS}\left(k\right),v_j)X_{MS}\left(k\right)+\mathrm{\α}K\left({\stackrel{\‾}{X}}_{MS}\right(k),v_j){\stackrel{\‾}{X}}_{MS}\left(k\right))}{\underset{k=1}{\overset{N}{\Σ}}{u}_{jk}^m\left(K\right(X_{MS}\left(k\right),v_j)+\mathrm{\α}K({\stackrel{\‾}{X}}_{MS}\left(k\right),v_j\left)\right)}$ Upgrade cluster centre;

Repeat to upgrade fuzzy partition matrix and cluster centre until the cluster centre cluster of adjacent twice iteration is less than ε;

According to fuzzy partition matrix u _jkdetermine final change detection figure, obtain region of variation and the non-changing region of optical remote sensing image.