CN104598922B - Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm - Google Patents
Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm Download PDFInfo
- Publication number
- CN104598922B CN104598922B CN201510007661.1A CN201510007661A CN104598922B CN 104598922 B CN104598922 B CN 104598922B CN 201510007661 A CN201510007661 A CN 201510007661A CN 104598922 B CN104598922 B CN 104598922B
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- fuzzy
- scattering
- class
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Image Processing (AREA)
- Radar Systems Or Details Thereof (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, the problem of ground class border is inaccurate caused by for being fixed in traditional Classification of Polarimetric SAR Image method decomposed based on Cloude Pottier due to segmentation threshold, using the Euclidean distance in Wishart distance substitution traditional fuzzy C mean clusters, and one is introduced apart from the factor, so that classification results are more nearly with real surface, the degree of accuracy is higher.The beneficial effect that the present invention is reached：Traditional Euclidean distance is replaced using Wishart distances based on improved fuzzy C-means clustering method in the present invention, the regularity of distribution of pixel in full polarimetric SAR is more conformed to, while also introducing one apart from the factor to increase indexing between the similar ground class of scattering mechanism.This method is efficiently solved because segmentation threshold fixes the problem of caused ground class border is inaccurate in traditional Classification of Polarimetric SAR Image decomposed based on Cloude Pottier, therefore obtained classification results are more nearly with real surface.
Description
Technical field
The present invention relates to a kind of full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, belong at polarization SAR data
Manage technical field.
Background technology
In recent years, coming into operation successively with borne polarization SAR system, polarization SAR technology is occupied in remote sensing application
More and more important position.Because with four POLARIZATION CHANNELs, the terrestrial object information of full-polarization SAR data acquisition is than single polarization SAR
More enrich.Full-polarization SAR data how are effectively analyzed, the feature one of the scattering object included in full-polarization SAR data is extracted
It is directly the difficult point of polarization research.Many polarization characteristic parameters were once suggested, such as channel strength, channel strength ratio, channel phases
Difference etc., but these characteristic parameters are only applicable to specific application environment, limited by specific experiment and demand.
Cloude-Pottier algorithms have absolutely as a kind of typical Several Kinds of Target Polar Decomposition Methods in above problem is solved
To advantage.The probability distribution state that the algorithm requires no knowledge about data just can reasonably be interpreted to classification results, but
It is due to lack one-to-one relationship between scattering mechanism and real surface, so the classification boundaries of every kind of atural object are relatively obscured.
The content of the invention
To solve the deficiencies in the prior art, it is an object of the invention to provide a kind of full-polarization SAR based on fuzzy C-mean algorithm
Image classification method, for it is traditional based on Cloude-Pottier decompose Classification of Polarimetric SAR Image method in due to segmentation
The problem of ground class border is inaccurate caused by threshold value is fixed, using the Europe in Wishart distance substitution traditional fuzzy C mean clusters
Family name's distance, and one is introduced apart from the factor so that classification results are more nearly with real surface, and the degree of accuracy is higher.
In order to realize above-mentioned target, the present invention is adopted the following technical scheme that：
Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, it is characterised in that：Comprise the following steps：
1) original full-polarization SAR data are pre-processed, eliminated using multiple look processing and Refined Lee filtering methods
The influence of speckle noise in SAR image；
2) Cloude-Pottier polarization decomposings are carried out to pretreated image, obtains scattering entropy H, angle of scattering α, is averaged
Tri- polarization parameters of scattering strength λ；
3) image is divided into by three major types according to the value of average scattering intensity, is respectively：High scattering strength region, medium scattering
Intensity area and low scattering strength region, each major class is further split using H/ α planes, 24 groups are obtained；
4) 24 obtained classes are merged into by n (0 using hierarchical clustering algorithm<The class of n≤24)；
5) distance of each pixel and each class cluster centre is calculated, adjusts each using Fuzzy C-Means Cluster Algorithm
The border of class atural object, until meeting object function minimum, iteration ends and output category result；The fuzzy C-means clustering is calculated
Pixel in method to cluster centre apart from d using Wishart distances and one apart from factor WijRedefine：
Wherein<T>For the coherence matrix of pixel, VmFor the average coherence matrix of each class, i.e. cluster centre, μijRepresent
Data point xiIt is under the jurisdiction of classification j probability；It is described apart from factor WijThe implication of expression is WijIt is bigger, each pixel and cluster centre
VmDistance it is smaller, become from the point of cluster centre closely closer to becoming farther with the point of cluster centre far.
The foregoing full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, it is characterised in that：The step 2) in, dissipate
Penetrate entropy H, angle of scattering α and tri- polarization parameters of average scattering intensity λ using Cloude-Pottier decomposition algorithms to pretreatment after
Data carry out polarization decomposing obtain：
λ=λ1p1+λ2p2+λ3p3
Wherein λ1、λ2、λ3For coherence matrix T characteristic value, and ∞ ＞ λ1≥λ2≥λ3＞ 0, piExpression formula be：αiRepresent the type of i-th kind of scattering mechanism.
The foregoing full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, it is characterised in that：The step 3) in, root
Polarization data is divided into three major types according to average scattering intensity, the determination on the λ borders per class is using the method for taking intermediate value, i.e., according to λ
The dynamic range of value divides the image into uniform three part, and high scattering strength region, medium scattering strength are corresponded to respectively per part
Region and low scattering strength region.
The foregoing full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, it is characterised in that：The step 4) in, profit
Merge the classification of over-segmentation with hierarchical clustering algorithm, formula is as follows:
Wherein TiWith TjRepresent the class center of two classifications, DijFor Wishart distances, represent between classification i and classification j
Separating degree；The truth covered according to selected pilot region atural object, is divided into n (0 by image<The class of n≤24).
The foregoing full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, it is characterised in that：The step 5) in mould
Paste C means clustering algorithms object function be：
Wherein m is any real number more than 1, and C is class number, and N is the number of pixel in image, μijRepresent data point
xiIt is under the jurisdiction of classification j probability, cjFor the class center of jth class.D is distance of the pixel to cluster centre.
The beneficial effect that the present invention is reached：The present invention is on the basis of traditional Cloude-Pottier decomposition, using flat
Initial pictures are divided into three major types by equal scattering entropy, it is to avoid scattering mechanism is similar but it is mixed between atural object that scattering strength is different
Confuse；During fuzzy C-means clustering, replace traditional Euclidean distance with Wishart distances, and in full polarimetric SAR
The distribution of pixel is more identical；One is introduced apart from the factor, between reducing identical category pixel apart from while, increase
The distance of very much not generic pixel.This method efficiently solves traditional polarization SAR decomposed based on Cloude-Pottier
The problem of ground class border is inaccurate caused by being fixed in image classification due to segmentation threshold, thus obtained classification results with it is true
Earth's surface is more nearly, and the classification results of the type of ground objects similar to scattering mechanism are more accurate.
Brief description of the drawings
Fig. 1 is flow chart of the invention；
Fig. 2 decomposes the polarization parameter image extracted for the present invention using Cloude-Pottier；
Fig. 3 is utilizes the present invention to put forward the most termination of sorting technique and existing H/ α/FCM, H/ α/λ/FCM methods
Really.
Embodiment
The invention will be further described below in conjunction with the accompanying drawings.Following examples are only used for clearly illustrating the present invention
Technical scheme, and can not be limited the scope of the invention with this.
Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm as shown in Figure 1, comprises the following steps：
Step 1, original full-polarization SAR data are pre-processed, multiple look processing and Refined Lee filtering is respectively adopted
Deng the influence for eliminating speckle noise in SAR image, strengthen the readability of image.
For the validity of qualitative and quantitative analysis the inventive method, data use L-band ALOS PALSAR complete polarization numbers
It it is 23.858 ° according to, image incident angle, spatial resolution is 9.37m × 3.57m (distance to × orientation).In order to strengthen figure
Influence of the readability and reduction coherent speckle noise of picture to experimental result, is pre-processed to initial data first.It is wherein to regard more
Handle orientation and distance to ratio be 6:The window size of 1, Refined Lee wave filters is 3 × 3.
It should be noted that：The method of the present invention is applicable not only to the ALOS PALSAR data chosen in the experiment, to it
He is spaceborne and on-board data is equally applicable, is the ratio and the window size of filtering process of multiple look processing when pretreatment
Difference is had, it is necessary to be selected for different data sources.
Step 2, Cloude-Pottier polarization decomposings are carried out to pretreated image, obtains scattering entropy H (Fig. 2 b), dissipates
Three polarization parameters such as firing angle α (Fig. 2 c), average scattering intensity λ (Fig. 2 a)；
The formula for carrying out polarization decomposing to pretreated data using Cloude-Pott ier decomposition algorithms is as follows：
λ=λ1p1+λ2p2+λ3p3
Wherein λiFor coherence matrix T characteristic value, and ∞ ＞ λ1≥λ2≥λ3＞ 0, piExpression formula be：αi
Represent the type of i-th kind of scattering mechanism.
Step 3, image is divided into by three major types according to the value of average scattering intensity, the determination on the λ borders per class is using in taking
The method of value, i.e., divide the image into uniform three part according to the dynamic range of λ value, is corresponded to respectively per part：High scattering strength
Region, medium scattering strength region and low scattering strength region, in each major class, are further split using H/ α planes, there are
To 24 groups；
Step 4, in order to avoid there is over-segmentation in previous step, feelings are covered according to the atural object that pilot region is chosen in experiment
24 classes derived above are merged into 7 classes by condition using hierarchical clustering algorithm.It should be noted that：Aggregate into how many classification be by
What the ground mulching type of selected areas was determined, the parameter can be adjusted according to the pilot region of selection.
The calculation formula of hierarchical clustering algorithm is as follows:
Wherein TiWith TjRepresent the class center of two classifications, DijFor Wishart distances, represent between classification i and classification j
Separating degree.It is analyzed according to the progress field investigation of image overlay area and with high-resolution optical satellite image,
Image is divided into 7 classes.
Step 5, the distance of each pixel and each class cluster centre is calculated, is calculated using improved fuzzy C-means clustering
Method adjusts the border of each class atural object, until meeting object function minimum, iteration ends and output category result.
Use the object function of Fuzzy C-Means Cluster Algorithm for：
Wherein m is any real number more than 1, and C is class number, and N is the number of pixel in image, μijRepresent data point
xiIt is under the jurisdiction of classification j probability, cjFor the class center of jth class.D is distance of the pixel to cluster centre.This method is used
Wishart distances and one are apart from factor WijRedefine apart from d, come reduce in traditional fuzzy C means clustering algorithms Euclidean away from
From the defect inaccurate to polarization SAR calculating.
Wherein<T>For the coherence matrix of pixel, VmFor the average coherence matrix of every class, i.e. cluster centre, μijRepresent number
Strong point xiIt is under the jurisdiction of classification j probability.WijIt is bigger, each pixel and class center VmDistance it is smaller, the factor causes from poly-
The point of class center closely becomes closer to becoming farther with the point of cluster centre far.
For the effect (Fig. 3 c) of quantitative analysis full polarimetric SAR sorting technique of the present invention, using overall accuracy (OA),
Producer's precision (PA) and user's precision (UA) evaluate the precision of classification results in the invention, and with H/ α/FCM algorithms (Fig. 3 a)
It is compared (table 1) with H/ α/λ/Fuzzy C-Means Cluster Algorithm (Fig. 3 b).
It should be noted that：Pretreated image is split in H/ α planes in H/ α/FCM methods, then using biography
The FCM Algorithms of system are clustered；In H/ α/λ/FCM methods, average scattering intensity has also assisted in classification, but clusters
Method still uses traditional Fuzzy C-Means Cluster Algorithm.
The ratio of precision of the different Classification of Polarimetric SAR Image methods of table 1 compared with
Can be seen that from table 1 and Fig. 3 has more preferably compared with H/ α/FCM algorithms and H/ α/λ/Fuzzy C-Means Cluster Algorithm
Effect：In the classification results obtained using institute's extracting method of the present invention, the border of atural object is apparent and continuous, and scattering mechanism is similar
Region obtained preferable differentiation, final result is more nearly with real surface.
Described above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, some improvement and deformation can also be made, these improve and deformed
Also it should be regarded as protection scope of the present invention.
Claims (5)
1. the full polarimetric SAR sorting technique based on fuzzy C-mean algorithm, it is characterised in that：Comprise the following steps：
1) original full-polarization SAR data are pre-processed, SAR is eliminated using multiple look processing and Refined Lee filtering methods
The influence of speckle noise in image；
2) Cloude-Pottier polarization decomposings are carried out to pretreated image, obtains scattering entropy H, angle of scattering α, average scattering
Tri- polarization parameters of intensity λ；
3) image is divided into by three major types according to the value of average scattering intensity, is respectively：High scattering strength region, medium scattering strength
Region and low scattering strength region, each major class is further split using H/ α planes, 24 groups are obtained；
4) 24 obtained classes are merged into by n classes using hierarchical clustering algorithm, wherein 0<n≤24；
5) distance of each pixel and each class cluster centre is calculated, using Fuzzy C-Means Cluster Algorithm with adjusting each class
The border of thing, until meeting object function minimum, iteration ends and output category result；In the Fuzzy C-Means Cluster Algorithm
Pixel to cluster centre apart from d using Wishart distances and one apart from factor WijRedefine：
<mrow>
<msup>
<mi>d</mi>
<mn>2</mn>
</msup>
<mrow>
<mo>(</mo>
<mo><</mo>
<mi>T</mi>
<mo>></mo>
<mo>,</mo>
<msub>
<mi>V</mi>
<mi>m</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mrow>
<mo>(</mo>
<mfrac>
<mrow>
<mi>l</mi>
<mi>n</mi>
<mo>|</mo>
<msub>
<mi>V</mi>
<mi>m</mi>
</msub>
<mo>|</mo>
<mo>+</mo>
<mi>T</mi>
<mi>r</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>V</mi>
<mi>m</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mo><</mo>
<mi>T</mi>
<mo>></mo>
<mo>)</mo>
</mrow>
</mrow>
<msub>
<mi>W</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
</mfrac>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
<mrow>
<msub>
<mi>W</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<msub>
<mi>&mu;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<msub>
<mi>&mu;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
</mrow>
</mfrac>
</mrow>
Wherein<T>For the coherence matrix of pixel, VmFor the average coherence matrix of each class, i.e. cluster centre, μijRepresent data
Point xiIt is under the jurisdiction of classification j probability；It is described apart from factor WijThe implication of expression is WijIt is bigger, each pixel and cluster centre Vm's
Apart from smaller, become from the point of cluster centre closely closer to becoming farther with the point of cluster centre far.
2. the full polarimetric SAR sorting technique according to claim 1 based on fuzzy C-mean algorithm, it is characterised in that：
The step 2) in, scattering entropy H, angle of scattering α and tri- polarization parameters of average scattering intensity λ use Cloude-Pottier
Decomposition algorithm carries out polarization decomposing to pretreated data and obtained：
<mrow>
<mi>H</mi>
<mo>=</mo>
<mo>-</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mn>3</mn>
</munderover>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
<msub>
<mi>log</mi>
<mn>3</mn>
</msub>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
</mrow>
<mrow>
<mi>&alpha;</mi>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mn>3</mn>
</munderover>
<msub>
<mi>p</mi>
<mi>i</mi>
</msub>
<msub>
<mi>&alpha;</mi>
<mi>i</mi>
</msub>
</mrow>
λ=λ1p1+λ2p2+λ3p3
Wherein λ1、λ2、λ3For coherence matrix T characteristic value, and ∞ ＞ λ1≥λ2≥λ3＞ 0, piExpression formula be：αi
Represent the type of i-th kind of scattering mechanism.
3. the full polarimetric SAR sorting technique according to claim 1 based on fuzzy C-mean algorithm, it is characterised in that：
The step 3) in, polarization data is divided into by three major types according to average scattering intensity, the determination on the λ borders per class, which is used, to be taken
The method of intermediate value, i.e., divide the image into uniform three part according to the dynamic range of λ value, and high scattering strength is corresponded to respectively per part
Region, medium scattering strength region and low scattering strength region.
4. the full polarimetric SAR sorting technique according to claim 1 based on fuzzy C-mean algorithm, it is characterised in that：
The step 4) in, the classification of over-segmentation is merged using hierarchical clustering algorithm, formula is as follows:
<mrow>
<msub>
<mi>D</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mn>2</mn>
</mfrac>
<mo>{</mo>
<mi>l</mi>
<mi>n</mi>
<mrow>
<mo>(</mo>
<mo>|</mo>
<msub>
<mi>T</mi>
<mi>i</mi>
</msub>
<mo>|</mo>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>l</mi>
<mi>n</mi>
<mrow>
<mo>(</mo>
<mo>|</mo>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
<mo>|</mo>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>T</mi>
<mi>r</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>T</mi>
<mi>i</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<msub>
<mi>T</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<msubsup>
<mi>T</mi>
<mi>j</mi>
<mrow>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<msub>
<mi>T</mi>
<mi>i</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>}</mo>
</mrow>
Wherein TiWith TjRepresent the class center of two classifications, DijFor Wishart distances, point between classification i and classification j is represented
From degree；The truth covered according to selected pilot region atural object, is divided into n classes, wherein 0 by image<n≤24.
5. the full polarimetric SAR sorting technique according to claim 1 based on fuzzy C-mean algorithm, it is characterised in that：
The step 5) in the object function of Fuzzy C-Means Cluster Algorithm be：
<mrow>
<msub>
<mi>J</mi>
<mi>m</mi>
</msub>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>C</mi>
</munderover>
<msup>
<msub>
<mi>&mu;</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mi>m</mi>
</msup>
<msup>
<mi>d</mi>
<mn>2</mn>
</msup>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mi>i</mi>
</msub>
<mo>,</mo>
<msub>
<mi>c</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
Wherein m is any real number more than 1, and C is class number, and N is the number of pixel in image, μijRepresent data point xiIt is subordinate to
Belong to classification j probability, cjFor the class center of jth class, d is distance of the pixel to cluster centre.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510007661.1A CN104598922B (en) | 2015-01-07 | 2015-01-07 | Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510007661.1A CN104598922B (en) | 2015-01-07 | 2015-01-07 | Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104598922A CN104598922A (en) | 2015-05-06 |
| CN104598922B true CN104598922B (en) | 2017-09-05 |
Family
ID=53124694
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510007661.1A Expired - Fee Related CN104598922B (en) | 2015-01-07 | 2015-01-07 | Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104598922B (en) |
Families Citing this family (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105403862B (en) * | 2015-12-04 | 2017-11-21 | 西安电子科技大学 | A kind of radar signal classification method for separating based on evidence C averages |
| CN106295688B (en) * | 2016-08-02 | 2019-10-18 | 浙江工业大学 | A kind of fuzzy clustering method based on sparse mean value |
| CN106408587B (en) * | 2016-09-19 | 2018-11-09 | 辽宁工程技术大学 | Regard SAR image segmentation method and device more |
| CN106683261B (en) * | 2016-12-22 | 2019-07-26 | 深圳怡化电脑股份有限公司 | A kind of false distinguishing method and apparatus of bank note |
| CN108364011A (en) * | 2018-02-05 | 2018-08-03 | 中国民航大学 | PolSAR image multi-stage characteristics extract and unsupervised segmentation method |
| CN109740475B (en) * | 2018-12-25 | 2020-04-21 | 杭州世平信息科技有限公司 | Ground scene classification method for remote sensing image |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101320431A (en) * | 2008-06-13 | 2008-12-10 | 北京航空航天大学 | Self-adapting SAR image classification method based on local standard deviation |
| CN104123563A (en) * | 2014-07-17 | 2014-10-29 | 电子科技大学 | Cloude characteristic decomposition based polarimetric SAR (Synthetic Aperture Radar) image non-supervision classification method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7298866B2 (en) * | 2001-10-15 | 2007-11-20 | Lockheed Martin Corporation | Two dimensional autonomous isotropic detection technique |
-
2015
- 2015-01-07 CN CN201510007661.1A patent/CN104598922B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101320431A (en) * | 2008-06-13 | 2008-12-10 | 北京航空航天大学 | Self-adapting SAR image classification method based on local standard deviation |
| CN104123563A (en) * | 2014-07-17 | 2014-10-29 | 电子科技大学 | Cloude characteristic decomposition based polarimetric SAR (Synthetic Aperture Radar) image non-supervision classification method |
Non-Patent Citations (2)
| Title |
|---|
| 一种利用地物散射特性进行后续类别调整的极化ＳＡＲ影像分类方法;刘利敏等;《光谱学与光谱分析》;20140131;第34卷(第1期);151-156 * |
| 基于Cloude-Pottier目标分解和聚合的层次聚类算法的全极化SAR 数据的非监督分类算法研究;曹芳等;《电子学报》;20080331;第36卷(第3期);543-546 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104598922A (en) | 2015-05-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104598922B (en) | Full polarimetric SAR sorting technique based on fuzzy C-mean algorithm | |
| CN107229917B (en) | A kind of several remote sensing image general character well-marked target detection methods based on iteration cluster | |
| Tang et al. | A multiscale latent Dirichlet allocation model for object-oriented clustering of VHR panchromatic satellite images | |
| CN103955701B (en) | Multi-level-combined multi-look synthetic aperture radar image target recognition method | |
| CN103488968A (en) | Device and method for constituting fining decomposer through mixed pixel materials of remote sensing image | |
| CN103971115A (en) | Automatic extraction method for newly-increased construction land image spots in high-resolution remote sensing images based on NDVI and PanTex index | |
| CN110751612A (en) | Single image rain removing method of multi-channel multi-scale convolution neural network | |
| CN107527023A (en) | Classification of Polarimetric SAR Image method based on super-pixel and topic model | |
| Qi et al. | Global–local 3-D convolutional transformer network for hyperspectral image classification | |
| CN115713694B (en) | Land mapping information management method | |
| CN104657980A (en) | Improved multi-channel image partitioning algorithm based on Meanshift | |
| CN103886334A (en) | Multi-index fused hyperspectral remote sensing image dimensionality reduction method | |
| CN104796582A (en) | Video image denoising and enhancing method and device based on random ejection retinex | |
| CN103839075A (en) | SAR image classification method based on united sparse representation | |
| Xu et al. | COCO-Net: A dual-supervised network with unified ROI-loss for low-resolution ship detection from optical satellite image sequences | |
| Lu et al. | LDRepFM: a real-time end-to-end visible and infrared image fusion model based on layer decomposition and re-parameterization | |
| Ma et al. | Transfer representation learning meets multimodal fusion classification for remote sensing images | |
| CN111626380A (en) | Polarized SAR image classification method based on super-pixels and convolution network | |
| CN108090914B (en) | Color image segmentation method based on statistical modeling and pixel classification | |
| CN106971402B (en) | SAR image change detection method based on optical assistance | |
| CN114067217A (en) | SAR image target identification method based on non-downsampling decomposition converter | |
| Arai | Maximum likelihood classification based on classified result of boundary mixed pixels for high spatial resolution of satellite images | |
| CN112883823A (en) | Land cover category sub-pixel positioning method based on multi-source remote sensing data fusion | |
| CN106101732A (en) | The vector quantization scheme of Fast Compression EO-1 hyperion signal | |
| Chen et al. | Hyperspectral remote sensing IQA via learning multiple kernels from mid-level features |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170905 Termination date: 20210107 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |