CN102163333A - Change detection method for synthetic aperture radar (SAR) images of spectral clustering - Google Patents
Change detection method for synthetic aperture radar (SAR) images of spectral clustering Download PDFInfo
- Publication number
- CN102163333A CN102163333A CN 201110082948 CN201110082948A CN102163333A CN 102163333 A CN102163333 A CN 102163333A CN 201110082948 CN201110082948 CN 201110082948 CN 201110082948 A CN201110082948 A CN 201110082948A CN 102163333 A CN102163333 A CN 102163333A
- Authority
- CN
- China
- Prior art keywords
- difference
- pixel
- subgraph
- matrix
- map
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention discloses a change detection method for synthetic aperture radar (SAR) images of spectral clustering, and belongs to the technical field of image processing. The change detection method comprises the following steps of: (1) selecting two SAR images T1 and T2 with same size and different time intervals as test images; (2) building a difference molecular map D1 of the test image T1 and the test image T2 by a different value method; (3) building a difference molecular map D2 of the test image T1 and the test image T2 by a field ratio method; (4) merging the difference molecular map D1 with the difference molecular map D2 by a product transformation fusion method to generate a difference map D; and (5) clustering pixels of the difference map D to obtain a change detection result map by a spectral clustering method. By the method, speckle noise of the difference map is suppressed better, speckles in the change detection result map are effectively removed, and more accurate change detection result map is obtained. Moreover, the method can be applied to detection and processing of remote sensing image change.
Description
Technical field
The invention belongs to image processing field, relate to a kind of change detecting method of spectral clustering.This method can be used for from the multi-temporal remote sensing image of the same geographic area that different time obtains, and is qualitative or analyze quantitatively and definite face of land variation characteristic and process, obtains the problem of our needed atural object change information.
Background technology
The develop rapidly of modern Remote Sensing Technical provides a kind of fast way for change-detection, and remotely-sensed data becomes the general data source of change-detection, and this has just proposed new challenge to existing Remote Sensing Imagery Change Detection technology.Synthetic-aperture radar SAR is a kind of radar that remote sensing detects, and he has the advantage that round-the-clock detects landforms, and is not subjected to weather effect.In recent years evict from and showed many kinds of multidate SAR change detecting methods, the most frequently used SAR Image Change Detection technology was made up of two steps: at first to multidate SAR image configuration difference diagram, extract region of variation then on the difference diagram basis.Wherein:
To multidate SAR image configuration difference diagram is a key issue of change-detection, the classic method of structure difference diagram comprises image differential technique, image ratioing technigue and change vector analytic approach etc. at present, the deficiency of these methods is: do not consider the correlativity between the multidate image, adding the SAR image influenced by speckle noise, therefore can not the good restraining speckle noise, thus the accuracy rate of on the basis of difference diagram, extracting region of variation reduced.
Extracting region of variation on the basis of difference diagram is another key issue of change-detection, and traditional algorithm that extracts region of variation on the basis of difference diagram is to determine that on difference diagram change threshold extracts the region of variation of difference diagram.Yet how change threshold determines never good method solve, have now and determine that on the difference diagram basis threshold value extracts the algorithm of region of variation, comprise Nakagami-Ratio GKIT, Log-Normal GKIT and Weibull-Ratio GKIT; These three kinds of methods are utilized Nakagami-Gamma respectively, and three kinds of models of Weibull and log-normal are chosen the optimum variation threshold value, make the overall error rate of change-detection reach minimum in theory; The defective of these three kinds of methods is it all is the approximate Gaussian distributed of supposition difference diagram, but because the complicacy of actual spectral characteristic of ground with single Gaussian distribution model, can produce bigger deviation, obtains wrong change-detection result.
Summary of the invention
The objective of the invention is to overcome the deficiency of above-mentioned prior art, propose a kind of change detecting method of spectral clustering,, improve the accuracy rate of change-detection to suppress speckle noise.
Technical scheme of the present invention is to select one group of SAR view data that is usually used in the precision evaluation of SAR Image Change Detection as experimental data earlier, and concrete steps comprise as follows:
(1) size is identical in the selection Same Scene, two width of cloth SAR image T that the period is different
1And T
2As test pattern;
(2) use difference approach to construct two width of cloth test pattern T
1And T
2Difference subgraph D
1:
D
1=255-|T
1-T
2|,
(3) use the field ratioing technigue to construct two width of cloth test pattern T
1And T
2Difference subgraph D
2:
3a) obtain the pixel value neighborhood S set of two width of cloth test patterns on same position x respectively
1(x) and S
2(x), its size is M * M, M ∈ 3,5};
3b) compare two neighborhood S set
1(x) and S
2(x) similarity obtains difference subgraph D
2Grey scale pixel value D on the x of position
2(x):
Wherein, S
1(x
i) and S
2(x
i) represent test pattern T respectively
1And T
2Neighborhood S set on the x of position
1(x) and S
2(x) i element, wherein 0≤i≤M * M;
3c) to test pattern T
1And T
2Each pixel from left to right, repeating step 3a from top to bottom) and 3b), obtain difference subgraph D
2
(4) use the product of transformation fusion method to difference subgraph D
1With difference subgraph D
2Merge, generate difference diagram D;
(5) use the spectral clustering algorithm,, obtain change-detection figure as a result the pixel cluster of difference diagram D.
The present invention has following advantage compared with prior art:
1, the present invention is owing to adopt the neighborhood ratioing technigue to construct test pattern T
1And T
2Difference subgraph D
2, can farthest suppress background, strengthen region of variation, effectively reduced and be in the influence of the zone that changes between class and the non-variation class the change-detection result.
2, the present invention has been merged difference subgraph D owing to adopt the product of transformation fusion method
1And D
2, obtained difference image D, simultaneously owing to similarity and neighborhood information between the pixel that when merging, has taken into full account difference diagram, thus the speckle noise of good restraining difference diagram.
3, simulation result shows, Differential Image building method and the spectral clustering method and the ratio operator of the product of transformation fusion method that the present invention adopts, logarithm ratio operator structure difference diagram method and Nakagami-Ratio GKIT, Log-Normal GKIT compares with Weibull-Ratio GKIT method, correct verification and measurement ratio height, false retrieval and omission number are low.
Description of drawings
Fig. 1 is a main FB(flow block) of the present invention;
Fig. 2 is first group of SAR image that emulation of the present invention is used;
Fig. 3 is the difference image of Fig. 2 being constructed with the present invention and existing method;
Fig. 4 is the simulation result figure that the difference diagram of Fig. 3 is carried out cluster respectively with the spectral clustering algorithm;
Fig. 5 is second group of SAR image that emulation of the present invention is used;
Fig. 6 is with spectral clustering method and Nakagami-Ratio GKIT, and three kinds of methods of Log-Normal GKIT and Weibull-Ratio GKIT are extracted change-detection that region of variation obtains figure as a result to Fig. 5.
Embodiment
With reference to Fig. 1, SAR image change detection method of the present invention comprises the steps:
Step 1. be chosen in same radar take in the areal scene down big or small identical, two width of cloth SAR image T that the period is different
1And T
2As test pattern.
Step 2. uses difference approach to construct two width of cloth test pattern T
1And T
2Difference subgraph D
1:
D
1=255-|T
1-T
2|,
If test pattern T
1And T
2Do not change at some locational pixel values, then difference subgraph D
1At this locational pixel value is 255, if test pattern T
1And T
2The intensity of variation that takes place at some locational pixel values is big more, then difference subgraph D
1At this position pixel value more near 0.
Step 3. uses the field ratioing technigue to construct two width of cloth test pattern T
1And T
2Difference subgraph D
2:
3a) obtain two width of cloth test pattern T respectively
1And T
2Neighborhood of pixel points S set on same position x
1(x) and S
2(x), its size is M * M, M ∈ 3,5}, wherein neighborhood S set
1(x) with test pattern T
1Position x be the window that a M * M size is got at the center, the neighborhood S set
2(x) with test pattern T
2Position x be the window that a M * M size is got at the center;
3b) compare two neighborhood S set
1(x) and S
2(x) similarity obtains difference subgraph D
2Pixel value D on the x of position
2(x):
Wherein, S
1(x
i) and S
2(x
i) represent test pattern T respectively
1And T
2Neighborhood S set on the x of position
1(x) and S
2(x) i element, wherein 0≤i≤M * M; D
2(x) value is more little, and test pattern T is described
1And T
2Difference on the x of position is big more, and then to belong to the possibility of region of variation also big more for the pixel on the x of position, otherwise the possibility that this position pixel belongs to non-region of variation is big more;
3c) to test pattern T
1And T
2Each pixel from left to right, repeating step 3a from top to bottom) and 3b), obtain difference subgraph D
2
Step 4. uses the product of transformation fusion method to difference subgraph D
1With difference subgraph D
2Merge, generate difference diagram D:
4a) to size be the difference subgraph D of H * W
1Get i neighborhood of pixel points set N respectively
iWith j neighborhood of pixel points set N
j, 1≤i≤H * W, 1≤j≤H * W; Wherein neighborhood is gathered N
iWith position i is the window that a P * P size is got at the center, neighborhood set N
jWith position j is the window that a P * P size is got at the center, P ∈ 3,5};
4b) poor Molecular Graphs D
1Two neighborhood set N
iAnd N
jSimilarity, obtain difference subgraph D
1The similarity u of i pixel and j pixel (i, j):
Wherein
d
1(N
i) be illustrated in difference subgraph D
1(i) in, be the window at center with i, N
iThe vector that interior all gray values of pixel points constitute successively, all windows are got same surely direction;
Euler's distance of expression cum rights, a>0, it is the standard variance of gaussian kernel, h is the parameter of a control attenuation degree, h ≠ 0; U (i, j) satisfy: 0≤u (i, j)≤1, ∑
j(i, j)=1, two the similarity between a pixel i and the j is by pixel value vector d (N for u
i) and d (N
j) between similarity decision, weigh by the Euclidean distance of Gauss's weighting, similarity is high more, (i, value j) is big more for u.
4c) utilize following formula to obtain the pixel value D (i) of difference diagram D on the i of position:
D(i)=∑u(i,j)d
2(j)
D wherein
2(j) expression difference subgraph D
2The value of j element;
4d) to two difference subgraph D
1And D
2Each pixel from left to right, repeating step 4a from top to bottom)-4c), obtain difference diagram D.
Step 5. is used the spectral clustering algorithm, with the pixel cluster of difference diagram D, obtains change-detection figure as a result:
Be that each pixel of H * W difference diagram D is regarded a last point of data set X, X=(x as with size 5a)
1, x
2..., x
n) ∈ R
N * 1, n=H * W, last each point of X is corresponding with the pixel of difference diagram D;
5b) the similarity matrix of construction data set X: G=(g
Ij) ∈ R
N * n, when i ≠ j, g
Ij=exp (| x
i-x
j|
2/ σ
2), when i=j, g
Ij=0,1≤i≤n, 1≤j≤n, wherein x
iBe i the point of data acquisition X, x
jBe j point of data acquisition X, σ is the parameter of a control attenuation degree, σ ≠ 0;
5c) according to similarity matrix G, structure granny rag Lars matrix: L=A
-1/2GA
-1/2, wherein A is a diagonal matrix: A=(a
Ij) ∈ R
N * n, when i ≠ j, a
Ij=0, when i=j,
1≤i≤n, 1≤j≤n;
5d) according to granny rag Lars matrix L, structural attitude matrix V=[v
1, v
2] ∈ R
N * 2, v wherein
1, v
2Be two pairing proper vectors of eigenvalue of maximum of granny rag Lars matrix L, v
1, v
2It all is column vector;
5e) according to eigenmatrix V, structure canonical matrix Y:
1. the data Y that the capable j of i that utilizes following formula to obtain canonical matrix Y lists
Ij:
V wherein
IjThe data that the capable j of i of representation feature matrix V lists;
2. to each data point of eigenmatrix V from left to right, 1. repeating step obtains canonical matrix Y from top to bottom.
5f) regard each row of canonical matrix Y as R
2A bit, using the k average that canonical matrix Y is gathered is 2 classes on the space; If the i line data of canonical matrix Y is the b class, then with former data point x
iAlso divide the b class into, wherein { 0,1} obtains change-detection figure as a result to b ∈.
Effect of the present invention can further specify by following experiment:
1, simulated conditions:
At CPU is to use MATLAB7.0 to carry out emulation in core2 2.4HZ, internal memory 2G, the WINDOWS XP system.
First group of SAR test pattern that emulation of the present invention is used is the Ottaw group image among Fig. 2, and wherein the shooting time of Fig. 2 (a) and Fig. 2 (b) is respectively 1997.05 and 1997.08, and size is 290 * 350, and Fig. 2 (c) is a reference diagram.
Second group of SAR test pattern that emulation of the present invention is used is the Berma group image among Fig. 5, and the Berma picture group is that two width of cloth sizes that the ERS-2 satellite is taken are 301 * 301 SAR image, Fig. 5 mistake! Do not find Reference source.(a) and the shooting time of Fig. 5 (b) be respectively 1999.4 and 1999.5, Fig. 5 (c) is a reference diagram.
2, simulation parameter
For experiment simulation figure with reference diagram, can carry out quantitative change-detection interpretation of result, the index of estimating the change-detection result has 4:
1. calculate the omission number: change among the figure as a result number of pixels in zone of statistical experiment, compare with the number of pixels of region of variation in the reference diagram, detect changing in the reference diagram but in the experimental result picture and be unchanged number of pixels, be called omission number FN;
2. calculate the false retrieval number: do not change among the figure as a result number of pixels in zone of statistical experiment, compare with the number of pixels of non-region of variation in the reference diagram, detect number of pixels not changing in the reference diagram but in the experimental result picture, be called false retrieval number FP for changing;
3. total wrong number: omission number FN and false retrieval number FP sum are total wrong number TE;
4. the correct verification and measurement ratio of experiment with computing testing result figure:
Correct verification and measurement ratio PCC is defined as: the ratio of the number of targets that correctly records and total number of pixels of experimental data figure, as formula 1) shown in; The number of targets that wherein correctly records equal to change in the reference diagram and experimental result picture in detect for do not change in the number of pixels TP that changes and the reference diagram and experimental result picture in detect for unchanged number of pixels TN with;
PCC=((TP+TN)/(TP+FP+TN+FN)) 1)
These four indexs have been embodied a concentrated reflection of the detection performance to region of variation, have very strong specific aim.
3, emulation content
3.1) with the inventive method and existing ratio operator, logarithm ratio operator method, Fig. 2 is carried out difference image emulation structure, simulation result as shown in Figure 3, wherein Fig. 3 (a) is the difference image with ratio operator structure, Fig. 3 (b) is the difference image with the inventive method structure for the difference image with logarithm ratio operator structure, Fig. 3 (c).
3.2) difference diagram of Fig. 3 is carried out cluster with the spectral clustering algorithm respectively, its cluster result as shown in Figure 4, wherein Fig. 4 (a) is to the simulation result figure after Fig. 3 (a) cluster, and Fig. 4 (b) is to the simulation result figure after Fig. 3 (b) cluster, and Fig. 4 (c) is to the simulation result figure after Fig. 3 (c) cluster.
3.3) respectively the described second group of SAR image of Fig. 5 carried out change-detection with spectral clustering algorithm, Nakagami-Ratio GKIT algorithm, Log-Normal GKIT algorithm and Weibull-Ratio GKIT algorithm, figure is as shown in Figure 6 as a result for change-detection, wherein Fig. 6 (a) is the simulation result figure with Log-Normal GKIT method, Fig. 6 (b) is the simulation result figure with Nakagami-Ratio GKIT method, Fig. 6 (c) is the simulation result figure with Weibull-Ratio GKIT method, and Fig. 6 (d) is the simulation result figure with the spectral clustering method.
4, The simulation experiment result and analysis
As can be seen from Figure 3, the speckle noise of the simulation result figure of the inventive method is less, and speckle noise is had good restraining;
As can be seen from Figure 4, the simulation result figure of the inventive method is assorted, and point is less, and change-detection is relatively good.As shown in table 1 to the quantitative test as a result of Fig. 4 change-detection.
Table 1 pair Fig. 4 change-detection quantitative test as a result
As can be seen from Table 1, the inventive method is on the total wrong number of change-detection, all lack than other method of contrast, and on flase drop number and omission number relative all compare balanced, can find out that with the contrast of figure (2c) reference picture the present invention is better than other two methods to the inhibition effect of speckle noise simultaneously.
As seen from Figure 6, the simulation result figure of the inventive method is assorted, and point is less, and change-detection is relatively good, and very good to the testing result of fine edge part, can effectively detect fine edge.Change-detection quantitative test as a result to Fig. 6 is as shown in table 2.
The change-detection quantitative test as a result of table 2 couple Fig. 6
As can be seen from Table 2, the inventive method, the omission number is fewer, and the total wrong number of change-detection all lacks than additive method, and the inventive method is more superior than other three kinds of methods.
Claims (3)
1. the SAR image change detection method of a spectral clustering comprises the steps:
(1) size is identical in the selection Same Scene, two width of cloth SAR image T that the period is different
1And T
2As test pattern;
(2) use difference approach to construct two width of cloth test pattern T
1And T
2Difference subgraph D
1:
D
1=255-|T
1-T
2|,
(3) use the field ratioing technigue to construct two width of cloth test pattern T
1And T
2Difference subgraph D
2:
3a) obtain the pixel value neighborhood S set of two width of cloth test patterns on same position x respectively
1(x) and S
2(x), its size is M * M, M ∈ 3,5};
3b) compare two neighborhood S set
1(x) and S
2(x) similarity obtains difference subgraph D
2Grey scale pixel value D on the x of position
2(x):
Wherein, S
1(x
i) and S
2(x
i) represent test pattern T respectively
1And T
2Neighborhood S set on the x of position
1(x) and S
2(x) i element, wherein 0≤i≤M * M;
3c) to test pattern T
1And T
2Each pixel from left to right, repeating step 3a from top to bottom) and 3b), obtain difference subgraph D
2
(4) use the product of transformation fusion method to difference subgraph D
1With difference subgraph D
2Merge, generate difference diagram D;
(5) use the spectral clustering algorithm,, obtain change-detection figure as a result the pixel cluster of difference diagram D.
2. SAR image change detection method according to claim 1, wherein the described use product of transformation of step (4) fusion method is to difference subgraph D
1With difference subgraph D
2Merge, generate difference diagram D, carry out as follows:
4a) to size be the difference subgraph D of H * W
1Get i pixel value neighborhood set N respectively
iWith j pixel value neighborhood set N
j, its size is P * P, P ∈ 3,5}, 1≤i≤H * W, 1≤j≤H * W;
4b) poor Molecular Graphs D
1Two neighborhood set N
iAnd N
jSimilarity, obtain difference subgraph D
1The similarity u of i pixel and j pixel (i, j):
Wherein
d
1(N
i) be illustrated in difference subgraph D
1(i) in, be the window at center with i, N
iThe vector that interior all gray values of pixel points constitute successively, all windows are got same surely direction,
Euler's distance of expression cum rights, a>0, it is the standard variance of gaussian kernel, h is the parameter of a control attenuation degree, h ≠ 0;
4c) utilize following formula to obtain the grey scale pixel value D (i) of difference diagram D on the i of position:
D(i)=∑u(i,j)d
2(j)
D wherein
2(j) expression difference subgraph D
2The value of j element;
4d) to two difference subgraph D
1And D
2Each pixel from left to right, repeating step 4a from top to bottom)-4c), obtain difference diagram D.
According to and the described SAR image change detection method of claim 1, the described use spectral clustering of step 5 algorithm wherein, with the pixel cluster of difference diagram D, carry out as follows:
Be that each pixel of H * W difference diagram D is regarded a last point of data set X, X=(x as with size 5a)
1, x
2..., x
n) ∈ R
N * 1, n=H * W, wherein upward each point is corresponding with the pixel of difference diagram D for X;
5b) to X structure similarity matrix G=(g
Ij) ∈ R
N * n, when i ≠ j, g
Ij=exp (| x
i~x
j|
2/ σ
2), when i=j, g
Ij=0,1≤i≤n * n, 1≤j≤n * n, wherein x
iBe i the point of data acquisition X, x
jBe j point of data acquisition X, σ is the parameter of a control attenuation degree, σ ≠ 0;
5c) according to similarity matrix G, structure granny rag Lars matrix: L=A
-1/2GA
-1/2, wherein A is a diagonal matrix: A=(a
Ij) ∈ R
N * n, when i ≠ j, a
Ij=0, when i=j,
1≤i≤n * n, 1≤j≤n * n;
5d) according to granny rag Lars matrix L, structural attitude matrix V=[v
1, v
2] ∈ R
N * 2, v wherein
1, v
2Be two pairing proper vectors of eigenvalue of maximum of granny rag Lars L, v
1, v
2It all is column vector;
5e) according to eigenmatrix V, structure canonical matrix Y:
1. the data Y that the capable j of i that utilizes following formula to obtain canonical matrix Y lists
Ij:
V wherein
IjThe data that the capable j of i of representation feature matrix V lists;
2. to each data point of eigenmatrix V from left to right, 1. repeating step obtains canonical matrix Y from top to bottom.
5f) regard each row of canonical matrix Y as R
2A bit, using the k average that it is gathered to canonical matrix Y is 2 classes on the space; If the i line data of canonical matrix Y is the b class, then with former data point x
iAlso divide the b class into, wherein { 0,1} obtains change-detection figure as a result to b ∈.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201110082948A CN102163333B (en) | 2011-04-02 | 2011-04-02 | Change detection method for synthetic aperture radar (SAR) images of spectral clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201110082948A CN102163333B (en) | 2011-04-02 | 2011-04-02 | Change detection method for synthetic aperture radar (SAR) images of spectral clustering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102163333A true CN102163333A (en) | 2011-08-24 |
CN102163333B CN102163333B (en) | 2012-10-24 |
Family
ID=44464544
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201110082948A Expired - Fee Related CN102163333B (en) | 2011-04-02 | 2011-04-02 | Change detection method for synthetic aperture radar (SAR) images of spectral clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102163333B (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102509288A (en) * | 2011-10-19 | 2012-06-20 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image change detection method based on Memetic kernel clustering |
CN102509297A (en) * | 2011-11-14 | 2012-06-20 | 西安电子科技大学 | Clonal selection-based method for detecting change of remote sensing image with optimal entropy threshold |
CN102722892A (en) * | 2012-06-13 | 2012-10-10 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method based on low-rank matrix factorization |
CN102968790A (en) * | 2012-10-25 | 2013-03-13 | 西安电子科技大学 | Remote sensing image change detection method based on image fusion |
CN103810699A (en) * | 2013-12-24 | 2014-05-21 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method based on non-supervision depth nerve network |
CN103839257A (en) * | 2013-12-24 | 2014-06-04 | 西安电子科技大学 | Method for detecting changes of SAR images of generalized Gaussian K&I |
CN103955943A (en) * | 2014-05-21 | 2014-07-30 | 西安电子科技大学 | Non-supervision change detection method based on fuse change detection operators and dimension driving |
CN107392863A (en) * | 2017-06-30 | 2017-11-24 | 西安电子科技大学 | SAR image change detection based on affine matrix fusion Spectral Clustering |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0955557A2 (en) * | 1998-05-08 | 1999-11-10 | Lockheed Martin Corporation | Method and system for detecting/classifying objects |
CN101738607A (en) * | 2009-12-07 | 2010-06-16 | 西安电子科技大学 | Method for detecting SAR image changes of cluster-based higher order cumulant cross entropy |
CN101853491A (en) * | 2010-04-30 | 2010-10-06 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering |
CN101923711A (en) * | 2010-07-16 | 2010-12-22 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image change detection method based on neighborhood similarity and mask enhancement |
-
2011
- 2011-04-02 CN CN201110082948A patent/CN102163333B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0955557A2 (en) * | 1998-05-08 | 1999-11-10 | Lockheed Martin Corporation | Method and system for detecting/classifying objects |
CN101738607A (en) * | 2009-12-07 | 2010-06-16 | 西安电子科技大学 | Method for detecting SAR image changes of cluster-based higher order cumulant cross entropy |
CN101853491A (en) * | 2010-04-30 | 2010-10-06 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering |
CN101923711A (en) * | 2010-07-16 | 2010-12-22 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image change detection method based on neighborhood similarity and mask enhancement |
Non-Patent Citations (1)
Title |
---|
《红外与毫米波学报》 20091231 李金基, 焦李成, 张向荣, 杨咚咚 基于两时相图像联合分类的SAR图像变化检测 第466-471页 1-3 第28卷, 第6期 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102509288A (en) * | 2011-10-19 | 2012-06-20 | 西安电子科技大学 | SAR (Synthetic Aperture Radar) image change detection method based on Memetic kernel clustering |
CN102509297A (en) * | 2011-11-14 | 2012-06-20 | 西安电子科技大学 | Clonal selection-based method for detecting change of remote sensing image with optimal entropy threshold |
CN102722892A (en) * | 2012-06-13 | 2012-10-10 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method based on low-rank matrix factorization |
CN102722892B (en) * | 2012-06-13 | 2014-11-12 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method based on low-rank matrix factorization |
CN102968790A (en) * | 2012-10-25 | 2013-03-13 | 西安电子科技大学 | Remote sensing image change detection method based on image fusion |
CN103810699A (en) * | 2013-12-24 | 2014-05-21 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method based on non-supervision depth nerve network |
CN103839257A (en) * | 2013-12-24 | 2014-06-04 | 西安电子科技大学 | Method for detecting changes of SAR images of generalized Gaussian K&I |
CN103810699B (en) * | 2013-12-24 | 2017-01-11 | 西安电子科技大学 | SAR (synthetic aperture radar) image change detection method based on non-supervision depth nerve network |
CN103839257B (en) * | 2013-12-24 | 2017-01-11 | 西安电子科技大学 | Method for detecting changes of SAR images of generalized Gaussian K&I |
CN103955943A (en) * | 2014-05-21 | 2014-07-30 | 西安电子科技大学 | Non-supervision change detection method based on fuse change detection operators and dimension driving |
CN107392863A (en) * | 2017-06-30 | 2017-11-24 | 西安电子科技大学 | SAR image change detection based on affine matrix fusion Spectral Clustering |
Also Published As
Publication number | Publication date |
---|---|
CN102163333B (en) | 2012-10-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102163333B (en) | Change detection method for synthetic aperture radar (SAR) images of spectral clustering | |
Li et al. | Adaptively constrained dynamic time warping for time series classification and clustering | |
CN103020978B (en) | SAR (synthetic aperture radar) image change detection method combining multi-threshold segmentation with fuzzy clustering | |
CN101673403B (en) | Target following method in complex interference scene | |
Ming et al. | Semivariogram-based spatial bandwidth selection for remote sensing image segmentation with mean-shift algorithm | |
CN108171193B (en) | Polarized SAR (synthetic aperture radar) ship target detection method based on super-pixel local information measurement | |
CN110349160B (en) | SAR image segmentation method based on super-pixel and fuzzy C-means clustering | |
CN101493520B (en) | SAR image variation detecting method based on two-dimension gamma distribution | |
CN103632363A (en) | Object-level high-resolution remote sensing image change detection method based on multi-scale fusion | |
CN103345757A (en) | Optical image and SAR image automatic registration method within multilevel multi-feature constraint | |
CN105389799B (en) | SAR image object detection method based on sketch map and low-rank decomposition | |
CN103456020A (en) | Remote sensing image change detection method based on treelet feature fusion | |
CN102314610B (en) | Object-oriented image clustering method based on probabilistic latent semantic analysis (PLSA) model | |
CN104794729A (en) | SAR image change detection method based on significance guidance | |
CN102779346A (en) | SAR (storage address register) image changing detection method based on improved C-V model | |
CN103839257A (en) | Method for detecting changes of SAR images of generalized Gaussian K&I | |
CN103268496A (en) | Target identification method of SAR (synthetic aperture radar) images | |
CN101794437A (en) | Method for detecting abnormal target in hyperspectral remotely sensed image | |
CN103500453A (en) | SAR(synthetic aperture radar) image significance region detection method based on Gamma distribution and neighborhood information | |
CN102360503A (en) | SAR (Specific Absorption Rate) image change detection method based on space approach degree and pixel similarity | |
CN104268561A (en) | Hyperspectral image mixing eliminating method based on structure prior low rank representation | |
CN103593852A (en) | Hyperspectral image abnormality detection method based on homogeneous patches | |
CN103456029A (en) | Mean Shift tracking method for resisting similar color and illumination variation interference | |
CN105303566A (en) | Target contour clipping-based SAR image target azimuth estimation method | |
CN103473561A (en) | Self-adaptation high-spectrum classification method based on Gaussian distribution |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121024 Termination date: 20200402 |
|
CF01 | Termination of patent right due to non-payment of annual fee |