CN101853491A - SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering - Google Patents

SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering Download PDF

Info

Publication number
CN101853491A
CN101853491A CN201010161497A CN201010161497A CN101853491A CN 101853491 A CN101853491 A CN 101853491A CN 201010161497 A CN201010161497 A CN 201010161497A CN 201010161497 A CN201010161497 A CN 201010161497A CN 101853491 A CN101853491 A CN 101853491A
Authority
CN
China
Prior art keywords
matrix
parallel
sar image
sar
sparse
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
Application number
CN201010161497A
Other languages
Chinese (zh)
Other versions
CN101853491B (en
Inventor
缑水平
王爽
庄雄
焦李成
朱虎明
李阳阳
钟桦
张佳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xidian University
Original Assignee
Xidian University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xidian University filed Critical Xidian University
Priority to CN2010101614977A priority Critical patent/CN101853491B/en
Publication of CN101853491A publication Critical patent/CN101853491A/en
Application granted granted Critical
Publication of CN101853491B publication Critical patent/CN101853491B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/13—Satellite images

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Astronomy & Astrophysics (AREA)
  • Remote Sensing (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering, relating to the technical field of image processing and mainly solving the problem of limitation of segmentation application of large-scale SAR images in the traditional spectral clustering technology. The SAR image segmentation method comprises the steps of: 1, extracting features of an SAR image to be segmented; 2, configuring an MATLAB (matrix laboratory) parallel computing environment; 3, allocating tasks all to processor nodes and computing partitioned sparse similar matrixes; 4, collecting computing results by a parallel task dispatcher and merging into an integral sparse similar matrix; 5, resolving a Laplacian matrix and carrying out feature decomposition; 6, carrying out K-means clustering on a feature vector matrix subjected to normalization; and 7, outputting a segmentation result of the SAR image. The invention can effectively overcome the bottleneck problem in computation and storage space of the traditional spectral clustering technology, has remarkable segmentation effect on large-scale SAR images, and is suitable for SAR image target detection and target identification.

Description

SAR image partition method based on parallel sparse spectral clustering
Technical field
The invention belongs to technical field of image processing, relate to the SAR image segmentation, can be used for Radar Targets'Detection and Target Recognition.
Background technology
Synthetic-aperture radar SAR has round-the-clock, round-the-clock detection and reconnaissance capability.It utilizes pulse compression technique to obtain high range resolution, utilizes the synthetic aperture principle to improve azimuthal resolution, has special advantages thereby compare real aperture radar in the remote sensing field.Understanding and decipher to the SAR image belong to the Flame Image Process category, also relate to numerous subjects such as signal Processing, pattern-recognition and machine learning.The SAR image segmentation is as one of key link of SAR Flame Image Process, just receiving more and more widely concern at national defence and civil area.Existing SAR image partition method roughly can be divided into based on the dividing method in zone with based on the dividing method on border, as the method for Threshold Segmentation, morphology methods, cluster, and the method for random field etc.
Wherein, based on the SAR image partition method of cluster, be that zone similar in the SAR image is divided into a class as far as possible, with dissimilar area dividing in different classifications.There has been the clustering algorithm of a lot of maturations to be used in the SAR image segmentation.As a new branch of science, fully excavated paired some similar characteristic of data in conjunction with the spectral clustering algorithm of spectrogram theory, the Laplce's matrix character by figure decomposes and reaches the dimensionality reduction purpose, makes cluster result more sane.But when data scale n was bigger, the spectral clustering algorithm need calculate size and be the similar matrix of n * n, and calculates the feature decomposition problem of corresponding Laplce's matrix, and the time complexity of its calculating and space complexity are respectively O (n 3) and O (n 2), can not cut apart the SAR image that comprises large-scale data effectively.
Summary of the invention
The objective of the invention is to overcome the deficiency of above-mentioned prior art, a kind of SAR image partition method based on parallel sparse spectral clustering has been proposed, with time and the space complexity that reduces spectral clustering, thereby can cut apart the SAR image that comprises large-scale data effectively.
For achieving the above object, the present invention includes following steps:
(1) SAR image to be split is carried out 3 layers of stationary wavelet conversion, the total number of image slices vegetarian refreshments is n, by following formula each pixel is extracted 10 dimension sub belt energy features, and constituting size is the input sample of data E of n * 10:
E = 1 M × N Σ i = 1 M Σ j = 1 N + | coef ( i , j ) |
Wherein, M * N is a subband size of utilizing moving window to determine, and value is 16 * 16 here, and (i j) is the coefficient value of the capable j row of i in the stationary wavelet subband to coef;
(2) parallel computation environment of configuration MATLAB7.8R2009 (a) version;
(3) carry out parallel task and divide, use the Parallel Task Scheduling device that the input sample of data equilibrium of the total number of pixel as n is divided on p the processor, run into aliquant situation and then remaining data sample is stored on last node;
(4) calculate n/p data sample to the Euclidean distance between other all data samples at distributed earth on each processor node, obtain distance matrix;
(5) adjust the distance each row of matrix sort according to order from small to large, and calculate piecemeal similar matrix S according to following formula:
S = exp ( - | | x i - x j | | 2 2 σ 2 ) , i , j = 1 , . . . , n
Wherein, x 1..., x nRepresent n data sample point, || x i-x j|| 2Distance matrix after the expression ordering, σ represents scale parameter, and similar matrix is only kept each data sample and the value between its most close t data sample on every side, all sparse 0 value that changes in other positions;
(6) similar matrix that each processor node is calculated is collected on the host node by the Parallel Task Scheduling device, is arranged into delegation and pools the sparse similar matrix W that a size is n * n;
(7) calculating diagonal entry according to sparse similar matrix W is
Figure GSA00000086656400022
The degree matrix D, and then calculate Laplce matrix L=D -1/2WD -1/2, utilize the eigs function of MATLAB that Laplce's matrix is carried out feature decomposition;
(8) eigenwert that obtains after the feature decomposition is sorted K eigenvalue of maximum 1=λ before obtaining 1〉=λ 2〉=... 〉=λ K, its characteristic of correspondence vector table is shown as v 1, v 2..., v K
(9) proper vector that obtains is in line, constructs eigenvectors matrix U=[v 1, v 2..., v K], and, obtain the eigenvectors matrix that standardizes to its normalization
Figure GSA00000086656400023
(10) will standardize each row of eigenvectors matrix Y is regarded space as KIn a point, with the K-means clustering algorithm it poly-ly is the K class, and uses the quadrature initialization to seek initial cluster center;
(11) according to the cluster centre that obtains, using Euclidean distance to estimate is divided into all pixels in the SAR image in the different classifications, each pixel obtains a cluster label, these cluster labels are reassembled into the matrix measure-alike with original image, with matrix element as the gray-scale value of segmentation result image and show, thereby obtain final segmentation result.
The present invention has the following advantages compared with prior art:
1) the present invention utilizes the MATLAB parallel computation environment, is designed for the parallel sparse spectral clustering method of SAR image segmentation, can effectively solve the existing excessive problem of spectral clustering technology operand;
2) the present invention utilizes the extensive similar matrix of sparse matrix storage spectral clustering, can effectively solve the existing limitation of spectral clustering technology on storage space;
3) the existing relatively SAR image Segmentation Technology of the present invention, the parallel sparse spectral clustering method of design can fully be excavated the internal information of SAR image, obtains better segmentation effect:
The simulation experiment result shows that the parallel sparse spectral clustering method that the present invention proposes can be effectively applied to the SAR image segmentation, and further is applied to Radar Targets'Detection and Target Recognition.
Description of drawings
Fig. 1 is realization flow figure of the present invention;
Fig. 2 is a MATLAB parallel computation environment synoptic diagram of the present invention;
Fig. 3 is the cluster environment synoptic diagram that emulation of the present invention is used;
Fig. 4 be the present invention and existing K-means clustering algorithm and
Figure GSA00000086656400031
The approximate spectrum clustering algorithm is applied to the contrast and experiment figure of ChinaLake airport SAR image segmentation;
Fig. 5 be the present invention and existing K-means clustering algorithm and
Figure GSA00000086656400032
The approximate spectrum clustering algorithm is applied to the contrast and experiment figure of Gulang Island, Xiamen SAR image segmentation;
Fig. 6 be the present invention and existing K-means clustering algorithm and
Figure GSA00000086656400033
The approximate spectrum clustering algorithm is applied to the contrast and experiment figure of Switzerland's lake region SAR image segmentation;
Fig. 7 is parallel speed-up ratio and the efficient synoptic diagram of the present invention to Switzerland's lake region SAR image segmentation.
Embodiment
With reference to Fig. 1, specific implementation process of the present invention is as follows:
Step 1. is extracted the feature of SAR image to be split.
SAR image not only data volume is big, different atural object has different back to emission and scattering properties in imaging process, thereby have abundant amplitude, phase place, polarization and texture information, and, the intrinsic coherent speckle noise of image directly exerts an influence for segmentation performance, therefore, be necessary before image segmentation, the SAR image to be carried out texture analysis, extract effective textural characteristics and carry out cluster.
On above analysis foundation, earlier SAR image to be split is carried out 3 layers of stationary wavelet conversion, the total number of image slices vegetarian refreshments is n, by following formula each pixel is extracted 10 dimension sub belt energy features, constituting size is the input sample of data E of n * 10:
E = 1 M × N Σ i = 1 M Σ j = 1 N | coef ( i , j ) |
Wherein, M * N is a subband size of utilizing moving window to determine, here value is 16 * 16, coef (i, j) be the coefficient value of the capable j row of i in the stationary wavelet subband, like this, be 256 * 256 SAR image for size, just constitute 65536 * 10 matrix input data as the parallel sparse spectral clustering method.
Step 2. configuration MATLAB parallel computation environment.
The concrete enforcement of this step is as follows with reference to Fig. 2:
2a) the parallel computation tool box and the Distributed Calculation server of installation MATLAB 7.8R2009 (a) version in high-performance 64 blade cluster environment as shown in Figure 3;
2b) on each clustered node, start the Distributed Calculation server with mdce start order;
2c) on the host node of cluster, start the Parallel Task Scheduling device with the startjobmanager order;
2d) on each processor node, start the progress of work with the startworker order.
Step 3. parallel task is divided.
3a) according to the MATLAB parallel computation environment that configures, use the good task dispatcher of findResource order specified configuration;
3b) use this task dispatcher, create operation with the createJob order;
3c) in the operation of being created, be that the input sample of data of n is divided on p the processor with createTask order equilibrium with scale, run into aliquant situation and then remaining data sample is stored on last node;
3d) use submit order submit job, each task is arranged into the execution of waiting for each processor node in the corresponding task queue with the waitForState order then;
3e) each processor node distributed earth calculates n/p data sample to the Euclidean distance between other all data samples, obtain distance matrix, each row of the distance matrix that obtains are sorted according to order from small to large, and convert thereof into piecemeal similar matrix S according to following formula:
S = exp ( - | | x i - x j | | 2 2 σ 2 ) , i，j＝1，?...，n
Wherein, x 1..., x nRepresent n data sample point, || x i-x j|| 2Distance matrix after the expression ordering, σ represents scale parameter, and similar matrix is only kept each data sample and the value between its most close t data sample on every side, all sparse 0 value that changes in other positions.
Step 4. host node task is collected.
With the getAllOutputArguments order result of calculation of p processor node is collected on the host node, and pooled the complete sparse similar matrix W=[S that a size is n * n 1..., S p], S wherein i(i=1 ..., p) p piecemeal similar matrix that processor node calculates of expression.
Step 5. is calculated Laplce's matrix character resolution problem.
Calculating diagonal entry according to this sparse similar matrix W is
Figure GSA00000086656400051
The degree matrix D, and then calculate Laplce matrix L=D -1/2WD -1/2, utilize the eigs function of MATLAB that this extensive sparse Laplce's matrix is carried out feature decomposition, K eigenvalue of maximum 1=λ before obtaining 1〉=λ 2〉=... 〉=λ K, and these eigenwert characteristic of correspondence vector v 1, v 2..., v K
Step 6. pair normalized eigenvectors matrix carries out the K-mean cluster.
The proper vector that obtains is in line, constructs eigenvectors matrix U=[v 1, v 2..., v K], its normalization is obtained the eigenvectors matrix that standardizes
Figure GSA00000086656400052
Regard each row of standardization eigenvectors matrix Y as space KIn a point, with the K-means clustering algorithm it poly-ly is the K class, and uses the quadrature initialization to seek initial cluster center.
Step 7. output segmentation result.
According to the cluster centre that obtains, using Euclidean distance to estimate is divided into all pixels in the SAR image in the different classifications, each pixel obtains a cluster label, these cluster labels are reassembled into the matrix measure-alike with original image, with matrix element as the gray-scale value of segmentation result image and show, thereby obtain final segmentation result.
Effect of the present invention can be verified by following emulation experiment.
Provide parallel sparse spectral clustering method of the present invention among the contrast experiment, and existing K-means clustering algorithm and The segmentation result of approximate spectrum clustering algorithm.Wherein, existing K-mean cluster and
Figure GSA00000086656400054
What the approximate spectrum cluster was chosen is 10 best segmentation results of experiment visual effect, and what parallel sparse spectral clustering method of the present invention was chosen owing to the result is more stable is the segmentation result of 1 experiment.
Figure GSA00000086656400055
The sample point number that the approximate spectrum cluster is chosen is 100, scale parameter s=0.2.The arest neighbors number that parallel sparse spectral clustering is chosen is 100, scale parameter s=0.05.
The contrast and experiment of using above three kinds of dividing methods that 3 width of cloth SAR images are cut apart is as follows:
1) to the segmentation result of Ku wave band SAR image as shown in Figure 4, wherein:
Fig. 4 (a) is the Ku wave band SAR image of the 3 meters resolution in China Lake airport of California, USA, and picture size is 400 * 400, comprises runway, vacant lot and airport building three class atural objects.
Fig. 4 (b) is the segmentation result of existing K-mean cluster method, has a lot of tiny spots to be divided into other two classes on the visible runway by mistake.
Fig. 4 (c) is existing
Figure GSA00000086656400056
The segmentation result of approximation method, although the regional consistance of airfield runway is improved, top-right airport building is then cut apart incoherently, is unfavorable for accurately carrying out Target Recognition in actual applications.
Fig. 4 (d) is the segmentation result of parallel sparse spectral clustering method of the present invention, and it has accurately provided the complete runway and the profile of buildings, and, for the road network that three trouble shape runway central authorities gather, compare other two kinds of algorithms and obtained more complete zone.
2) to the segmentation result of C-wave band SAR image as shown in the figure, wherein:
Fig. 5 (a) is area, Gulang Island, the Chinese Xiamen C-wave band SAR image that a width of cloth RadarSAT-2 satellite is taken, and polarization mode is the HV polarization, and the intercepting image block is of a size of 400 * 400.
Fig. 5 (b) is the segmentation result of existing K-mean cluster method, and it is too big that it is disturbed by topography profile on the island, causes the land part much to divide by mistake, extends between some land in waters and the island to be separated scatteredly.
Fig. 5 (c) is existing The segmentation result of approximation method, although than K-average some improves as a result, but fail to obtain a complete zone.
Fig. 5 (d) is the segmentation result of parallel sparse spectral clustering method of the present invention, can find out clearly that the Gulang Island is the same as with the island, Xiamen the complete land part that is divided into, and also can keep degree of connection preferably with the island part for the less shore line of some and waters contrast difference.
3) to the segmentation result of X-band SAR image as shown in Figure 6, wherein:
Fig. 6 (a) looks X-band SAR image for 3 of Switzerland one lake region in 1994 of Space Radar Laboratory Missions, and picture size is 512 * 512, comprises three class atural objects: lake, city and mountain region.
Fig. 6 (b) is the segmentation result of existing K-mean cluster method, and it divides out with the lake well, still, has also existed serious mistake to be divided into the situation in waters simultaneously in the area, mountain region of the right and left.
Fig. 6 (c) is existing
Figure GSA00000086656400062
The segmentation result of approximation method, it makes moderate progress to the situation of cutting apart in zone, mountain region, but there is a lot of obscuring in city and mountain region.
Fig. 6 (d) is the segmentation result of parallel sparse spectral clustering method of the present invention, it is best in three kinds of algorithms, and it not only accurately extracts the lake bank, does not have at all to be divided into the situation in lake in the zone, mountain region by mistake, and the segmentation result in city has kept regional preferably integrality.Therefore, to be used to cut apart this width of cloth SAR image be very suitable to parallel sparse spectral clustering method of the present invention.
In parallel computation of the present invention, for p processor, T SBe illustrated in the needed from start to end time of algorithm on the series machine, T PBe illustrated on the parallel computer algorithm from start to finish a processor finish the work the needed time speed-up ratio S PWith efficient E PBe defined as S respectively P=T S/ T P, E P=S S/ P.Speed-up ratio is weighed is how many execution speeds of parallel algorithm accelerated doubly with respect to the execution speed of serial algorithm.Efficient is weighed is the ratio that the computing power of single processor is used effectively.
With parallel sparse spectral clustering method of the present invention Switzerland's lake region SAR image shown in Fig. 6 (a) is cut apart, its parallel speed-up ratio curve and parallel efficiency curve are respectively shown in Fig. 7 (a) and Fig. 7 (b).Wherein, used processor number has reached 42 at most.
As can be seen from Figure 7, because parallel sparse spectral clustering method of the present invention adopts the MATLAB Distributed Calculation, except the transmission of the scheduling of task and data sharing, do not need between each processor to communicate, therefore can obtain intimate linear speed-up ratio less than 24 o'clock in the processor number.And when the processor number rises to 42,, but still keeping the trend of rising because extensive sparse matrix transmission and multi-processor task scheduling increase consuming time make speed-up ratio descend to some extent.Aspect efficient, under the parallel situation less than 32 processors, efficient can reach more than 90%, shows that each processor node has obtained utilizing relatively fully.Processor more for a long time, message is blocked and message is delayed time has reduced the utilization factor of single processor.

Claims (3)

1. SAR image partition method based on parallel sparse spectral clustering may further comprise the steps:
(1) SAR image to be split is carried out 3 layers of stationary wavelet conversion, the total number of image slices vegetarian refreshments is n, by following formula each pixel is extracted 10 dimension sub belt energy features, and constituting size is the input sample of data E of n * 10:
E = 1 M × N Σ i = 1 M Σ j = 1 N | coef ( i , j ) |
Wherein, M * N is a subband size of utilizing moving window to determine, and value is 16 * 16 here, and (i j) is the coefficient value of the capable j row of i in the stationary wavelet subband to coef;
(2) parallel computation environment of configuration MATLAB7.8R2009 (a) version;
(3) carry out parallel task and divide, use the Parallel Task Scheduling device that the input sample of data equilibrium of the total number of pixel as n is divided on p the processor, run into aliquant situation and then remaining data sample is stored on last node;
(4) calculate n/p data sample to the Euclidean distance between other all data samples at distributed earth on each processor node, obtain distance matrix;
(5) adjust the distance each row of matrix sort according to order from small to large, and calculate piecemeal similar matrix S according to following formula:
S = exp ( - | | x i - x j | | 2 2 σ 2 ) , i，j＝1，...，n
Wherein, x 1..., x nRepresent n data sample point, || x i-x j|| 2Distance matrix after the expression ordering, σ represents scale parameter, and similar matrix is only kept each data sample and the value between its most close t data sample on every side, all sparse 0 value that changes in other positions;
(6) similar matrix that each processor node is calculated is collected on the host node by the Parallel Task Scheduling device, is arranged into delegation and pools the sparse similar matrix W that a size is n * n;
(7) calculating diagonal entry according to sparse similar matrix W is
Figure FSA00000086656300013
The degree matrix D, and then calculate Laplce matrix L=D -1/2WD -1/2, utilize the eigs function of MATLAB that this Laplce's matrix is carried out feature decomposition;
(8) eigenwert that obtains after the feature decomposition is sorted K eigenvalue of maximum 1=λ before obtaining 1〉=λ 2〉=... 〉=λ K, its characteristic of correspondence vector table is shown as v 1, v 2, v K
(9) proper vector that obtains is in line, constructs eigenvectors matrix U=[v 1, v 2..., v K], and, obtain the eigenvectors matrix that standardizes to its normalization
(10) will standardize each row of eigenvectors matrix Y is regarded space mouth as KIn a point, with the K-means clustering algorithm it poly-ly is the K class, and uses the quadrature initialization to seek initial cluster center;
(11) according to the cluster centre that obtains, using Euclidean distance to estimate is divided into all pixels in the SAR image in the different classifications, each pixel obtains a cluster label, these cluster labels are reassembled into the matrix measure-alike with original image, with matrix element as the gray-scale value of segmentation result image and show, thereby obtain final segmentation result.
2. a kind of SAR image partition method according to claim 1 based on parallel sparse spectral clustering, the parallel computation environment of the described configuration of step (2) MATLAB7.8R2009 (a) version wherein, carry out according to following steps:
2a) the parallel computation tool box and the Distributed Calculation server of MATLAB7.8R2009 (a) version are installed on the High-Performance Computing Cluster environment;
2b) on each clustered node, start the Distributed Calculation server;
2c) on the host node of cluster, start the Parallel Task Scheduling device;
2d) on each processor node, start the progress of work.
3. a kind of SAR image partition method according to claim 1 based on parallel sparse spectral clustering, wherein the described parallel task of step (3) is divided, and carries out according to following steps:
3a) the MATLAB parallel computation environment that configures according to step (2) is specified employed Parallel Task Scheduling device;
3b) use this Parallel Task Scheduling device to create operation;
3c) in the operation of being created, be that the input sample of data equilibrium of n is divided on p the processor with the total number of pixel, run into aliquant situation and then remaining data sample is stored on last node;
3d) submit job, and each task is arranged in the corresponding task queue execution of waiting for each processor node.
CN2010101614977A 2010-04-30 2010-04-30 SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering Expired - Fee Related CN101853491B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2010101614977A CN101853491B (en) 2010-04-30 2010-04-30 SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2010101614977A CN101853491B (en) 2010-04-30 2010-04-30 SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering

Publications (2)

Publication Number Publication Date
CN101853491A true CN101853491A (en) 2010-10-06
CN101853491B CN101853491B (en) 2012-07-25

Family

ID=42804956

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2010101614977A Expired - Fee Related CN101853491B (en) 2010-04-30 2010-04-30 SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering

Country Status (1)

Country Link
CN (1) CN101853491B (en)

Cited By (22)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102096819A (en) * 2011-03-11 2011-06-15 西安电子科技大学 Method for segmenting images by utilizing sparse representation and dictionary learning
CN102129573A (en) * 2011-03-10 2011-07-20 西安电子科技大学 SAR (Synthetic Aperture Radar) image segmentation method based on dictionary learning and sparse representation
CN102163333A (en) * 2011-04-02 2011-08-24 西安电子科技大学 Change detection method for synthetic aperture radar (SAR) images of spectral clustering
CN102842048A (en) * 2011-06-20 2012-12-26 苏州科雷芯电子科技有限公司 Hardware implementation method of related parallel computation of groups in image recognition
CN102867307A (en) * 2012-09-10 2013-01-09 西安电子科技大学 SAR image segmentation method based on feature vector integration spectral clustering
CN102968796A (en) * 2012-11-30 2013-03-13 西安电子科技大学 SAR (Synthetic Aperture Radar) image segmentation method based on sampling learning
CN103440645A (en) * 2013-08-16 2013-12-11 东南大学 Target tracking algorithm based on self-adaptive particle filter and sparse representation
CN103971346A (en) * 2014-05-28 2014-08-06 西安电子科技大学 SAR (Synthetic Aperture Radar) image spot-inhibiting method based on spare domain noise distribution constraint
CN104123372A (en) * 2014-07-25 2014-10-29 浪潮(北京)电子信息产业有限公司 Clustering method and device based on CUDA
CN105279749A (en) * 2014-07-02 2016-01-27 深圳Tcl新技术有限公司 Image matting method and device
CN105787517A (en) * 2016-03-11 2016-07-20 西安电子科技大学 Polarized SAR image classification method base on wavelet sparse auto encoder
CN106204613A (en) * 2016-07-20 2016-12-07 中国科学院自动化研究所 The display foreground object detecting method represented based on low-rank matrix and detecting system
CN106778814A (en) * 2016-11-24 2017-05-31 郑州航空工业管理学院 A kind of method of the removal SAR image spot based on projection spectral clustering
CN107016656A (en) * 2017-04-01 2017-08-04 中国科学院光电技术研究所 The sparse base optimization method of small echo in image reconstruction based on compressed sensing
CN107316308A (en) * 2017-06-27 2017-11-03 苏州大学 A kind of clean robot map dividing method based on improved multi-path spectral clustering algorithm
CN108228844A (en) * 2018-01-09 2018-06-29 美的集团股份有限公司 A kind of picture screening technique and device, storage medium, computer equipment
CN108268891A (en) * 2017-12-29 2018-07-10 安徽中凯信息产业股份有限公司 A kind of data processing method
CN108268320A (en) * 2016-12-31 2018-07-10 英特尔公司 For the hardware accelerator framework and template of network size k mean value clusters
CN109214428A (en) * 2018-08-13 2019-01-15 平安科技（深圳）有限公司 Image partition method, device, computer equipment and computer storage medium
CN111915572A (en) * 2020-07-13 2020-11-10 青岛大学 Self-adaptive gear pitting quantitative detection system and method based on deep learning
CN112633391A (en) * 2020-12-29 2021-04-09 重庆电子工程职业学院 Multi-resolution data clustering analysis method
CN115223050A (en) * 2022-04-28 2022-10-21 湖北工程学院 Polarized satellite image analysis method

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050270285A1 (en) * 2004-06-08 2005-12-08 Microsoft Corporation Stretch-driven mesh parameterization using spectral analysis
US20070226624A1 (en) * 2006-02-23 2007-09-27 Peker Kadir A Content-based video summarization using spectral clustering
US20080243829A1 (en) * 2007-03-29 2008-10-02 Microsoft Corporation Spectral clustering using sequential shrinkage optimization
CN101299243A (en) * 2008-06-27 2008-11-05 西安电子科技大学 Method of image segmentation based on immune spectrum clustering
CN101673398A (en) * 2009-10-16 2010-03-17 西安电子科技大学 Method for splitting images based on clustering of immunity sparse spectrums

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20050270285A1 (en) * 2004-06-08 2005-12-08 Microsoft Corporation Stretch-driven mesh parameterization using spectral analysis
US20070226624A1 (en) * 2006-02-23 2007-09-27 Peker Kadir A Content-based video summarization using spectral clustering
US20080243829A1 (en) * 2007-03-29 2008-10-02 Microsoft Corporation Spectral clustering using sequential shrinkage optimization
CN101299243A (en) * 2008-06-27 2008-11-05 西安电子科技大学 Method of image segmentation based on immune spectrum clustering
CN101673398A (en) * 2009-10-16 2010-03-17 西安电子科技大学 Method for splitting images based on clustering of immunity sparse spectrums

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
《European Conference on Machine Learning and Knowledge Discovery in Databases (ECML/PKDD) 2008》 20080930 Yangqiu Song等 Parallel Spectral Clustering , 2 *

Cited By (34)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102129573A (en) * 2011-03-10 2011-07-20 西安电子科技大学 SAR (Synthetic Aperture Radar) image segmentation method based on dictionary learning and sparse representation
CN102096819A (en) * 2011-03-11 2011-06-15 西安电子科技大学 Method for segmenting images by utilizing sparse representation and dictionary learning
CN102163333A (en) * 2011-04-02 2011-08-24 西安电子科技大学 Change detection method for synthetic aperture radar (SAR) images of spectral clustering
CN102163333B (en) * 2011-04-02 2012-10-24 西安电子科技大学 Change detection method for synthetic aperture radar (SAR) images of spectral clustering
CN102842048A (en) * 2011-06-20 2012-12-26 苏州科雷芯电子科技有限公司 Hardware implementation method of related parallel computation of groups in image recognition
CN102867307A (en) * 2012-09-10 2013-01-09 西安电子科技大学 SAR image segmentation method based on feature vector integration spectral clustering
CN102968796A (en) * 2012-11-30 2013-03-13 西安电子科技大学 SAR (Synthetic Aperture Radar) image segmentation method based on sampling learning
CN103440645B (en) * 2013-08-16 2016-04-27 东南大学 A kind of target tracking algorism based on adaptive particle filter and rarefaction representation
CN103440645A (en) * 2013-08-16 2013-12-11 东南大学 Target tracking algorithm based on self-adaptive particle filter and sparse representation
CN103971346A (en) * 2014-05-28 2014-08-06 西安电子科技大学 SAR (Synthetic Aperture Radar) image spot-inhibiting method based on spare domain noise distribution constraint
CN103971346B (en) * 2014-05-28 2017-01-18 西安电子科技大学 SAR (Synthetic Aperture Radar) image spot-inhibiting method based on spare domain noise distribution constraint
CN105279749A (en) * 2014-07-02 2016-01-27 深圳Tcl新技术有限公司 Image matting method and device
CN104123372A (en) * 2014-07-25 2014-10-29 浪潮(北京)电子信息产业有限公司 Clustering method and device based on CUDA
CN104123372B (en) * 2014-07-25 2018-03-09 浪潮(北京)电子信息产业有限公司 A kind of method and device that cluster is realized based on CUDA
CN105787517A (en) * 2016-03-11 2016-07-20 西安电子科技大学 Polarized SAR image classification method base on wavelet sparse auto encoder
CN105787517B (en) * 2016-03-11 2018-12-14 西安电子科技大学 Classification of Polarimetric SAR Image method based on the sparse self-encoding encoder of small echo
CN106204613B (en) * 2016-07-20 2019-09-24 中国科学院自动化研究所 The display foreground object detecting method and detection system indicated based on low-rank matrix
CN106204613A (en) * 2016-07-20 2016-12-07 中国科学院自动化研究所 The display foreground object detecting method represented based on low-rank matrix and detecting system
CN106778814A (en) * 2016-11-24 2017-05-31 郑州航空工业管理学院 A kind of method of the removal SAR image spot based on projection spectral clustering
CN106778814B (en) * 2016-11-24 2020-06-12 郑州航空工业管理学院 Method for removing SAR image spots based on projection spectral clustering algorithm
CN108268320A (en) * 2016-12-31 2018-07-10 英特尔公司 For the hardware accelerator framework and template of network size k mean value clusters
CN107016656A (en) * 2017-04-01 2017-08-04 中国科学院光电技术研究所 The sparse base optimization method of small echo in image reconstruction based on compressed sensing
CN107316308A (en) * 2017-06-27 2017-11-03 苏州大学 A kind of clean robot map dividing method based on improved multi-path spectral clustering algorithm
CN108268891A (en) * 2017-12-29 2018-07-10 安徽中凯信息产业股份有限公司 A kind of data processing method
CN108228844A (en) * 2018-01-09 2018-06-29 美的集团股份有限公司 A kind of picture screening technique and device, storage medium, computer equipment
CN108228844B (en) * 2018-01-09 2020-10-27 美的集团股份有限公司 Picture screening method and device, storage medium and computer equipment
CN109214428B (en) * 2018-08-13 2023-12-26 平安科技（深圳）有限公司 Image segmentation method, device, computer equipment and computer storage medium
CN109214428A (en) * 2018-08-13 2019-01-15 平安科技（深圳）有限公司 Image partition method, device, computer equipment and computer storage medium
CN111915572A (en) * 2020-07-13 2020-11-10 青岛大学 Self-adaptive gear pitting quantitative detection system and method based on deep learning
CN111915572B (en) * 2020-07-13 2023-04-25 青岛大学 Adaptive gear pitting quantitative detection system and method based on deep learning
CN112633391B (en) * 2020-12-29 2023-09-29 重庆电子工程职业学院 Multi-resolution data clustering analysis method
CN112633391A (en) * 2020-12-29 2021-04-09 重庆电子工程职业学院 Multi-resolution data clustering analysis method
CN115223050A (en) * 2022-04-28 2022-10-21 湖北工程学院 Polarized satellite image analysis method
CN115223050B (en) * 2022-04-28 2023-08-18 湖北工程学院 Polarized satellite image analysis method

Also Published As

Publication number Publication date
CN101853491B (en) 2012-07-25

Similar Documents

Publication Publication Date Title
CN101853491B (en) SAR (Synthetic Aperture Radar) image segmentation method based on parallel sparse spectral clustering
CN102096825B (en) Graph-based semi-supervised high-spectral remote sensing image classification method
CN103593669B (en) A kind of method that polarimetric synthetic aperture radar image four component decomposes
CN104751477A (en) Space domain and frequency domain characteristic based parallel SAR (synthetic aperture radar) image classification method
CN105069796B (en) SAR image segmentation method based on small echo both scatternets
CN103413146B (en) Polarimetric SAR Image sophisticated category method based on Freeman entropys and self study
CN103177458B (en) A kind of visible remote sensing image region of interest area detecting method based on frequency-domain analysis
CN102402685B (en) Method for segmenting three Markov field SAR image based on Gabor characteristic
CN101699513A (en) Level set polarization SAR image segmentation method based on polarization characteristic decomposition
CN103927551A (en) Polarimetric SAR semi-supervised classification method based on superpixel correlation matrix
CN103426175B (en) The polarization SAR image segmentation method of feature based value metric spectral clustering
CN104794730B (en) SAR image segmentation method based on super-pixel
CN102867307A (en) SAR image segmentation method based on feature vector integration spectral clustering
CN104237883A (en) Airborne radar space time self-adaptation processing method with sparse representation
CN101930547A (en) Method for automatically classifying remote sensing image based on object-oriented unsupervised classification
CN101853509A (en) SAR (Synthetic Aperture Radar) image segmentation method based on Treelets and fuzzy C-means clustering
CN102122352A (en) Characteristic value distribution statistical property-based polarized SAR image classification method
CN103258324A (en) Remote sensing image change detection method based on controllable kernel regression and superpixel segmentation
CN107742133A (en) A kind of sorting technique for Polarimetric SAR Image
CN109359525A (en) The Classification of Polarimetric SAR Image method of differentiation spectral clustering based on sparse low-rank
CN102496142B (en) SAR (synthetic aperture radar) image segmentation method based on fuzzy triple markov fields
CN102663724A (en) Method for detecting remote sensing image change based on adaptive difference images
Luo et al. Research on change detection method of high-resolution remote sensing images based on subpixel convolution
CN104008394A (en) Semi-supervision hyperspectral data dimension descending method based on largest neighbor boundary principle
CN103870842A (en) Polarized SAR image classification method combining polarization feature and watershed

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
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20120725

Termination date: 20180430