CN108764145A - One kind is towards Dragon Wet Soil remote sensing images density peaks clustering method - Google Patents
One kind is towards Dragon Wet Soil remote sensing images density peaks clustering method Download PDFInfo
- Publication number
- CN108764145A CN108764145A CN201810533432.7A CN201810533432A CN108764145A CN 108764145 A CN108764145 A CN 108764145A CN 201810533432 A CN201810533432 A CN 201810533432A CN 108764145 A CN108764145 A CN 108764145A
- Authority
- CN
- China
- Prior art keywords
- pixel
- remote sensing
- sensing images
- density
- images
- 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.)
- Pending
Links
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to image processing fields, disclose one kind and are comprised the following steps towards Dragon Wet Soil remote sensing images density peaks clustering method：(1) high spectrum image of the wetland to obtaining screens, and artificially selects several high spectrum images；(2) several high spectrum images are pre-processed to obtain PC1 with median filter and Principal Component Analysis；(3) change of scale is carried out to PC1；(4) similarity calculation between the pixel of PC1 is carried out；(5) clustering processing is carried out with density peaks clustering algorithm；(6) local density's matrix and distance matrix and improved population evolution algorithmic are combined, finds out all pixels point most suitable as cluster centre；(7) it using centered on the pixel as cluster centre, according to local density's matrix and blocks distance remaining pixel is divided；(8) remote sensing images being disposed are exported.It present invention reduces the time complexity of algorithm, solves the problems, such as that cluster result is unstable, has more universality.
Description
Technical field
The invention belongs to image processing fields, more particularly to one kind is towards Dragon Wet Soil remote sensing images density peaks cluster side
Method.
Background technology
Remote sensing technology is born in the eighties of last century sixties, is one based on geography and physics, combines meter
Calculation machine, signal processing, the interdisciplinary study of the science such as space monitoring.It is recorded by multiple sensors on satellite or aircraft
The picture of various atural object electromagnetic waves is used for environmental monitoring, resource exploration, the fields such as military surveillance.With imaging spectral technology
Development becomes the main imaging mode of remote sensing images with maturation, high spectrum image.High spectrum image is obtaining earth's surface image letter
While breath, its spectral information is also obtained, has been truly realized the combination of spectrum and image for the first time.With multi-spectrum remote sensing image phase
Than being not only greatly improved in terms of abundant information degree, on treatment technology, more being closed to such spectroscopic data
Reason, effective analyzing processing provide possibility.While providing more abundant spectral information, high spectrum image also results in
Phenomena such as " the different spectrum of jljl, same object different images ", for our remote sensing image processings in later stage bring difficulty.
Remote Sensing Image Clustering is as mainly to Hyperspectral imagery processing and then obtaining effective hand of more detailed terrestrial object information
One of section, has obtained scholar and has widely paid close attention to all the time.Especially there is larger promotion image resolution ratio is compared with the past
In the case of, the details that the appearance of an outstanding image processing method further obtains us atural object seem more
It is important.Different from classification problem, cluster is a kind of unsupervised study.For remote sensing image classification, need to extract part in advance
Target image is as training sample, and each classification has one's own label in training set, and one is obtained in training completion
Target image is handled again after the disaggregated model of a maturation.Although improving accuracy, time complexity also carries
It is much higher, and due to the influence of training set and priori, cause algorithm unstable.And clustering algorithm is due to time complexity
It is relatively low, and the advance training pattern of training set is not used to eliminate the factors such as human interference, it can effectively reduce to content
The loss of less object spectrum, the multiple image between different-waveband high spectrum image.
The method of Remote Sensing Image Clustering includes mainly following several：
1.K-means Remote Sensing Image Clustering algorithms：
K-means algorithms have as a kind of unsupervised clustering algorithm it is simple, quickly, it is efficient the advantages that.But it is necessary
Clusters number is preassigned, while to determine initial cluster center point.After central point determines, by calculate each pixel with
Cluster centre obtains distance and updates cluster centre with the minimum criterion of standard of all pixels point distance.Pass through constantly iteration mistake
Journey is finally completed the category division of all pixels point.During cluster, class number is pre-set, therefore meeting
Cause final cluster result inaccurate, the case where high spectrum image pixel mistake point leakage divides occurs.Secondly, the determination of central point is complete
It is to be manually set entirely, choosing different central points can cause final result relatively large deviation occur, and algorithm shakiness is caused to be set to subsequently
Obstruction is brought to the further extraction of image information.When Characteristics of The Remote Sensing Images differs greatly, which must first be divided into several
A class can just obtain final result by arranging to merge.When the Spectral Characteristic in image is more complicated, the algorithm is poly-
A large amount of interpretation work is needed just to can determine that in the sum number purpose selection of class center.Wetland geomorphic feature is complicated, has cities and towns, grass
Ground, lake, ten several geomorphic features such as farmland, at this time K-means algorithms seem no longer applicable.
2. fuzzy C-mean algorithm Remote Sensing Image Clustering algorithm：
FCM Algorithms are a kind of image processing techniques combining Unsupervised clustering and fuzzy set concept, in the machine of calculation
Vision, the fields such as image procossing have extensive use.The algorithm can only belong to the division of pixel with each pixel is required
Hard plot clustering algorithm in class cluster center is different, by the application to fuzzy set, builds pixel and cluster centre
Subordinated-degree matrix is clustered.It can be obtained than common clustering algorithm more on certain unsharp data sets of data point classification
Outstanding result.But equally there is also following defects for the algorithm.It is easily by initial cluster center and initial subordinated-degree matrix
It influences, is likely to be converging on local minimum, affects the effect of segmentation；In the case of mass data collection, such as remote sensing figure
Picture, the iterating of Fuzzy C-means will result in algorithm take it is long.And continuously improving with sensor, Modern remote figure
The pixel and resolution ratio of picture are also higher and higher, and FCM Algorithms need the operation that iterates in cluster process, more delay
The convergence speed of the algorithm.
From the point of view of the above analysis, both existing clustering algorithms are not suitable for the special geographical environment of wetland at present
Analysis.Therefore, the research for the Remote Sensing Image Clustering algorithm of wetland seems very significant.
Invention content
It is an object of the invention to the good one kind of the universality of the open cluster result stabilization suitable for wetland towards bundle dragon
Wetland remote sensing images density peaks clustering method.
The object of the present invention is achieved like this：
One kind is comprised the following steps towards Dragon Wet Soil remote sensing images density peaks clustering method：
Step (1)：The high spectrum image of wetland first to obtaining screens, and it is clear artificially to select atural object, landforms
Feature is apparent, and resolution ratio is higher and does not have when shooting or several high spectrum images of only less cloud cover；
Step (2)：Several high spectrum images are pre-processed to obtain PC1 with median filter and Principal Component Analysis：
Several high spectrum images are handled using median filter；To in median filter treated several high spectrum images
Each Zhang Caiyong Principal Component Analysis carry out dimension-reduction treatment, as possible preserve raw information while removal with analysis indexes without
Variable that is closing and repeating.Choose sample variance instructs sample to choose as overall target function, from more after dimension-reduction treatment
It is maximum as PC1 that variance is chosen in panel height spectrum picture；
Step (3)：Change of scale is carried out to PC1：
Change of scale is carried out by the way of weak sampling, and pending sub-graph size is transformed to A1xB1 by AxB, wherein
A1=A/ksizeAnd B1=B/ksize；ksizeFor change of scale index；
Step (4)：Carry out the similarity calculation between the pixel of PC1：
Using the upper left corner of the PC1 after change of scale as origin, horizontal line is X-axis, and vertical line is that Y-axis establishes rectangular co-ordinate
System；
Assuming that include a M rows N row pixels by step (3) treated subgraph, establish containing n=M*N data
Pixel data set；Individual data includes pixel R under RGB color space, G, the pixel value on B component direction, Yi Jiwei
The characteristic informations such as confidence breath；
Assuming that pixel p1, p2 belong to the pixel data set, then pixel p1, the similarity d of p2sim(p1,p2)：
dsim(p1, p2)=β drgb+(1-β)dxy；
In above formula, r1 indicates that pixel values of the pixel p1 under rgb space on R component direction, g1 indicate that pixel p1 exists
Pixel value under rgb space on G component directions, b1 indicate pixel values of the pixel p1 under rgb space on B component direction；r2
Indicate that pixel values of the pixel p2 under rgb space on R component direction, g2 indicate pixel p2 G component directions under rgb space
On pixel value, b2 indicates pixel values of the pixel p2 under rgb space on B component direction；(x1, y1) is the seat of pixel p1
Mark, (x2, y2) are the coordinate of pixel p2；β is weight factor, β ∈ (0,1)；
Step (5)：Clustering processing is carried out with density peaks clustering algorithm：
The local density of each point is calculated, local density matrix { ρ is createdi}：
The set of pixels of imageIndex set is Is=1,2 ... and M*N }, dijIndicate pixel i and pixel
The similarity of point j, dcDistance is blocked for artificial settings；
The distance value with the pixel than the pixel with higher local density between each pixel is calculated, distance is created
Matrix
IfIt indicatesA descending arrangement, i.e.,：ρq1≥ρq2≥...≥ρqn；
Define δqi：
When i >=2,
As i=1,
Step (6)：Local density's matrix and distance matrix and improved population evolution algorithmic are combined, found out most suitable
Cooperation is all pixels point of cluster centre：
Entire population is initialized in solution space, each pixel is regarded to the individual in particle cluster algorithm；It assigns each
Particle initial velocity and position, while the optimal location of each particle itself arrival is recorded in search process, and it is entire
The optimum individual of population；
Object function is set as：
Find local density with distance obtain higher value pixel be used as cluster centre, even if object function acquirement most
The solution of small value；
Step (7)：Using centered on the pixel as cluster centre, according to local density's matrix and distance is blocked to it
Remaining pixel is divided；
Step (8)：Export the remote sensing images being disposed.
Beneficial effects of the present invention are：
The present invention reduces the time complexity of entire algorithm on the basis of retaining image information as possible, solves existing
The problem that clustering algorithm selection cluster centre is influenced to cause cluster result unstable by priori；Develop in conjunction with population and calculates
Method converts the problem to searching optimal solution；Density peaks algorithm has more universality；The case where time complexity allows
Under, the problem of present invention has well solved high spectrum image " the different spectrum of jljl, same object different images ".
Description of the drawings
Fig. 1 is one kind towards Dragon Wet Soil remote sensing images density peaks clustering method flow chart；
Fig. 2 is particle cluster algorithm flow chart.
Specific implementation mode
Further describe the present invention below in conjunction with the accompanying drawings：
Such as Fig. 1, one kind is comprised the following steps towards Dragon Wet Soil remote sensing images density peaks clustering method：
Step (1)：The high spectrum image of wetland first to obtaining screens, and high-spectrum seems aircraft or people
Make the electromagnetic wave of the different landforms reflection of passing of satelline sensor collection.Several EO-1 hyperions are due to its shooting time difference, shooting
When weather, environment, weather etc. is affected to remote sensing images, artificially selects that atural object is clear, and geomorphic feature is apparent, differentiates
Rate is higher and does not have when shooting or several high spectrum images of only less cloud cover；
Step (2)：Several high spectrum images are pre-processed to obtain PC1 with median filter and Principal Component Analysis：
For the spiced salt phenomenon in remote sensing images, several high spectrum images are handled using median filter；Median filter
The intermediate value replacement of each point value in a field of the value of any in the digital picture or Serial No. point, major function is to allow
The pixel that the difference of surrounding pixel gray value is bigger changes to take the value close with the pixel value of surrounding, so as to eliminate isolated make an uproar
Sound point, convenient for subsequently to the feature extraction of remote sensing images, dimensionality reduction.
Obtained high spectrum image is that resolution ratio is higher at present, information contained amount compared with horn of plenty remote sensing image,
Being used directly to do the initial data of image clustering processing can cause time complexity higher, or even the case where can not handle.So
Dimension-reduction treatment is carried out to each Zhang Caiyong Principal Component Analysis in median filter treated several high spectrum images, to the greatest extent
Amount removes variable that is unrelated with analysis indexes and repeating while preserving raw information.Sample variance is chosen as synthesis to refer to
Scalar functions instruct sample to choose, and it is maximum as PC1 that variance is chosen from several high spectrum images after dimension-reduction treatment；
Step (3)：Change of scale is carried out to PC1：
Change of scale is carried out by the way of weak sampling, reduces the pixel number in cluster process as possible, while again to the greatest extent
Amount remains the characteristic information of remote sensing images.Pending sub-graph size is transformed to A1xB1, wherein A1=A/k by AxBsize
And B1=B/ksize；ksizeFor change of scale index；
Step (4)：Carry out the similarity calculation between the pixel of PC1：
Using the upper left corner of the PC1 after change of scale as origin, horizontal line is X-axis, and vertical line is that Y-axis establishes rectangular co-ordinate
System；
Assuming that include a M rows N row pixels by step (3) treated subgraph, establish containing n=M*N data
Pixel data set；Individual data includes pixel R under RGB color space, G, the pixel value on B component direction, Yi Jiwei
The characteristic informations such as confidence breath；
Assuming that pixel p1, p2 belong to the pixel data set, then pixel p1, the similarity d of p2sim(p1,p2)：
dsim(p1, p2)=β drgb+(1-β)dxy；
In above formula, r1 indicates that pixel values of the pixel p1 under rgb space on R component direction, g1 indicate that pixel p1 exists
Pixel value under rgb space on G component directions, b1 indicate pixel values of the pixel p1 under rgb space on B component direction；r2
Indicate that pixel values of the pixel p2 under rgb space on R component direction, g2 indicate pixel p2 G component directions under rgb space
On pixel value, b2 indicates pixel values of the pixel p2 under rgb space on B component direction；(x1, y1) is the seat of pixel p1
Mark, (x2, y2) are the coordinate of pixel p2；β is weight factor, β ∈ (0,1)；
Step (5)：Clustering processing is carried out with density peaks clustering algorithm：
The local density of each point is calculated, local density matrix { ρ is createdi}：
The set of pixels of imageIndex set is Is=1,2 ... and M*N }, dijIndicate pixel i and pixel
The similarity of point j, dcDistance is blocked for artificial settings；
The distance value with the pixel than the pixel with higher local density between each pixel is calculated, distance is created
Matrix
IfIt indicatesA descending arrangement, i.e.,：ρq1≥ρq2≥...≥ρqn；
Define δqi：
When i >=2,
As i=1,
Step (6)：Local density's matrix and distance matrix and improved population evolution algorithmic are combined, found out most suitable
Cooperation is all pixels point of cluster centre：
Such as Fig. 2, entire population is initialized in solution space, each pixel is regarded to the individual in particle cluster algorithm；It assigns
Each particle initial velocity and position are given, while recording the optimal location of each particle itself arrival in search process, with
And the optimum individual of entire population；
Object function is set as：
The pixel that local density obtains higher value with distance is suitble to the most as cluster centre, in order to ensure that part is close
Degree is in the same order of magnitude with distance, and operation first is normalized to local density and distance, find local density with apart from equal
The pixel of higher value is obtained as cluster centre, even if object function obtains the solution of minimum value；In order to increase the optimizing of particle
Ability avoids being absorbed in locally optimal solution, in addition to the inertia weight that conventional particle group's algorithm uses is used for keeping particle at the beginning of optimizing
Phase has higher speed in order to avoid being absorbed in other than local optimum, and Cauchy's operator is added in particle searching process to ensure that particle has
There is a degree of variability, it is enable to have certain probability to jump out local optimum while finding optimal solution.
Step (7)：Using centered on the pixel as cluster centre, according to local density's matrix and distance is blocked to it
Remaining pixel is divided；
After finding in the image most suitable as the pixel of cluster centre, centered on these pixels, according to office
Portion's density matrix and block distance etc. rest of pixels point is divided.It is each due to being computed in before the step of
The local density of the distance between pixel and each pixel itself does not need to carry out successive ignition in cluster process
It calculates, all pixels point being enough according to existing parameter in the remote sensing images to entire wetland carries out clustering, very great Cheng
Reduce the time complexity of entire algorithm on degree.After cluster process is completed, the remote sensing images being disposed are exported.
Step (8)：Export the remote sensing images being disposed.
The present invention reduces the time complexity of entire algorithm on the basis of retaining image information as possible, solves existing
The problem that clustering algorithm selection cluster centre is influenced to cause cluster result unstable by priori；Develop in conjunction with population and calculates
Method converts the problem to searching optimal solution；Density peaks algorithm has more universality；The case where time complexity allows
Under, the problem of present invention has well solved high spectrum image " the different spectrum of jljl, same object different images ".
The above is not intended to restrict the invention, and for those skilled in the art, the present invention can have various
Change and variation.All within the spirits and principles of the present invention, any modification, equivalent replacement, improvement and so on should all include
Within protection scope of the present invention.
Claims (6)
1. one kind is towards Dragon Wet Soil remote sensing images density peaks clustering method, it is characterised in that：It comprises the following steps：
Step (1)：The high spectrum image of wetland first to obtaining screens, and it is clear artificially to select atural object, geomorphic feature
Obviously, resolution ratio is higher and does not have when shooting or several high spectrum images of only less cloud cover；
Step (2)：Several high spectrum images are pre-processed to obtain PC1 with median filter and Principal Component Analysis；
Step (3)：Change of scale is carried out to PC1；
Step (4)：Carry out the similarity calculation between the pixel of PC1；
Step (5)：Clustering processing is carried out with density peaks clustering algorithm；
Step (6)：Local density's matrix and distance matrix and improved population evolution algorithmic are combined, most suitable work is found out
For all pixels point of cluster centre；
Step (7)：Using centered on the pixel as cluster centre, according to local density's matrix and distance is blocked to remaining
Pixel is divided；
Step (8)：Export the remote sensing images being disposed.
2. one kind according to claim 1 is towards Dragon Wet Soil remote sensing images density peaks clustering method, it is characterised in that：
The step (2) is specially：
Several high spectrum images are handled using median filter；To every in median filter treated several high spectrum images
One carries out dimension-reduction treatment using Principal Component Analysis, and removal is unrelated with analysis indexes while preserving raw information as possible
And the variable repeated；Choose sample variance instructs sample to choose as overall target function, from several height after dimension-reduction treatment
It is maximum as PC1 that variance is chosen in spectrum picture.
3. one kind according to claim 1 is towards Dragon Wet Soil remote sensing images density peaks clustering method, it is characterised in that：
The step (3) is specially：
Change of scale is carried out by the way of weak sampling, and pending sub-graph size is transformed to A1xB1, wherein A1=by AxB
A/ksizeAnd B1=B/ksize；ksizeFor change of scale index.
4. one kind according to claim 1 is towards Dragon Wet Soil remote sensing images density peaks clustering method, it is characterised in that：
The step (4) is specially：
Using the upper left corner of the PC1 after change of scale as origin, horizontal line is X-axis, and vertical line is that Y-axis establishes rectangular coordinate system；
Assuming that including a M rows N row pixels by step (3) treated subgraph, the pixel containing n=M*N data is established
Data set；Individual data includes pixel R under RGB color space, G, the pixel value on B component direction and position letter
The characteristic informations such as breath；
Assuming that pixel p1, p2 belong to the pixel data set, then pixel p1, the similarity d of p2sim(p1,p2)：
dsim(p1, p2)=β drgb+(1-β)dxy；
In above formula, r1 indicates that pixel values of the pixel p1 under rgb space on R component direction, g1 indicate pixel p1 in RGB skies
Between pixel value on lower G component directions, b1 indicates pixel values of the pixel p1 under rgb space on B component direction；R2 indicates picture
Pixel values of the vegetarian refreshments p2 under rgb space on R component direction, g2 indicate pictures of the pixel p2 under rgb space on G component directions
Element value, b2 indicate pixel values of the pixel p2 under rgb space on B component direction；(x1, y1) is the coordinate of pixel p1,
(x2, y2) is the coordinate of pixel p2；β is weight factor, β ∈ (0,1).
5. one kind according to claim 1 is towards Dragon Wet Soil remote sensing images density peaks clustering method, it is characterised in that：
The step (5) is specially：
The local density of each point is calculated, local density matrix { ρ is createdi}：
The set of pixels of imageIndex set is Is=1,2 ... and M*N }, dijIndicate pixel i and pixel j
Similarity, dcDistance is blocked for artificial settings；
The distance value with the pixel than the pixel with higher local density between each pixel is calculated, distance matrix is created
IfIt indicatesA descending arrangement, i.e.,：ρq1≥ρq2≥...≥ρqn；
Define δqi：
When i >=2,
As i=1,
6. one kind according to claim 1 is towards Dragon Wet Soil remote sensing images density peaks clustering method, it is characterised in that：
The step (6) is specially：
Entire population is initialized in solution space, each pixel is regarded to the individual in particle cluster algorithm；Assign each particle
Initial velocity and position, while the optimal location of each particle itself arrival and entire population are recorded in search process
Optimum individual；
Object function is set as：
Find local density with distance obtain higher value pixel be used as cluster centre, even if object function acquirement minimum value
Solution.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810375568 | 2018-04-25 | ||
| CN201810375568X | 2018-04-25 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108764145A true CN108764145A (en) | 2018-11-06 |
Family
ID=64003882
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810533432.7A Pending CN108764145A (en) | 2018-04-25 | 2018-05-29 | One kind is towards Dragon Wet Soil remote sensing images density peaks clustering method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108764145A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276374A (en) * | 2019-05-10 | 2019-09-24 | 浙江工业大学 | User behavior log density peak value clustering method for automatically determining clustering center |
| CN117237646A (en) * | 2023-11-15 | 2023-12-15 | 深圳市润海电子有限公司 | PET high-temperature flame-retardant adhesive tape flaw extraction method and system based on image segmentation |
| CN117953252A (en) * | 2024-03-26 | 2024-04-30 | 贵州道坦坦科技股份有限公司 | Automatic acquisition method and system for highway asset data |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103295030A (en) * | 2013-06-21 | 2013-09-11 | 中国科学院对地观测与数字地球科学中心 | Classification method and device based on hyperspectral remote sensing images |
| CN103914705A (en) * | 2014-03-20 | 2014-07-09 | 西安电子科技大学 | Hyperspectral image classification and wave band selection method based on multi-target immune cloning |
-
2018
- 2018-05-29 CN CN201810533432.7A patent/CN108764145A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103295030A (en) * | 2013-06-21 | 2013-09-11 | 中国科学院对地观测与数字地球科学中心 | Classification method and device based on hyperspectral remote sensing images |
| CN103914705A (en) * | 2014-03-20 | 2014-07-09 | 西安电子科技大学 | Hyperspectral image classification and wave band selection method based on multi-target immune cloning |
Non-Patent Citations (2)
| Title |
|---|
| 盛东慧: ""基于矿产遥感图像的信息提取方法研究"", 《中国优秀硕士学位论文全文数据库（信息科技辑）》 * |
| 陈晋音等: ""基于密度的聚类中心自动确定的混合属性数据聚类算法研究"", 《自动化学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110276374A (en) * | 2019-05-10 | 2019-09-24 | 浙江工业大学 | User behavior log density peak value clustering method for automatically determining clustering center |
| CN117237646A (en) * | 2023-11-15 | 2023-12-15 | 深圳市润海电子有限公司 | PET high-temperature flame-retardant adhesive tape flaw extraction method and system based on image segmentation |
| CN117237646B (en) * | 2023-11-15 | 2024-01-30 | 深圳市润海电子有限公司 | PET high-temperature flame-retardant adhesive tape flaw extraction method and system based on image segmentation |
| CN117953252A (en) * | 2024-03-26 | 2024-04-30 | 贵州道坦坦科技股份有限公司 | Automatic acquisition method and system for highway asset data |
| CN117953252B (en) * | 2024-03-26 | 2024-05-31 | 贵州道坦坦科技股份有限公司 | Automatic acquisition method and system for highway asset data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108573276B (en) | Change detection method based on high-resolution remote sensing image | |
| Wang et al. | Optimal segmentation of high-resolution remote sensing image by combining superpixels with the minimum spanning tree | |
| Huang et al. | A new building extraction postprocessing framework for high-spatial-resolution remote-sensing imagery | |
| Li et al. | A review of remote sensing image classification techniques: The role of spatio-contextual information | |
| Shi et al. | Incorporating spatial information in spectral unmixing: A review | |
| Coburn et al. | A multiscale texture analysis procedure for improved forest stand classification | |
| CN103971115B (en) | Automatic extraction method for newly-increased construction land image spots based on NDVI and PanTex index | |
| Blaschke et al. | Image segmentation methods for object-based analysis and classification | |
| Zhang et al. | Boundary-constrained multi-scale segmentation method for remote sensing images | |
| CN110309781B (en) | House damage remote sensing identification method based on multi-scale spectrum texture self-adaptive fusion | |
| Khazai et al. | An approach for subpixel anomaly detection in hyperspectral images | |
| CN105138970B (en) | Classification of Polarimetric SAR Image method based on spatial information | |
| CN102982338B (en) | Classification of Polarimetric SAR Image method based on spectral clustering | |
| CN109711446A (en) | A kind of terrain classification method and device based on multispectral image and SAR image | |
| CN102214298A (en) | Method for detecting and identifying airport target by using remote sensing image based on selective visual attention mechanism | |
| CN111898688A (en) | Airborne LiDAR data tree species classification method based on three-dimensional deep learning | |
| CN108764145A (en) | One kind is towards Dragon Wet Soil remote sensing images density peaks clustering method | |
| Moussa et al. | A new object based method for automated extraction of urban objects from airborne sensors data | |
| CN111680579B (en) | Remote sensing image classification method for self-adaptive weight multi-view measurement learning | |
| CN113838064B (en) | Cloud removal method based on branch GAN using multi-temporal remote sensing data | |
| CN104657980A (en) | Improved multi-channel image partitioning algorithm based on Meanshift | |
| CN106529472B (en) | Object detection method and device based on large scale high-resolution high spectrum image | |
| CN113628234B (en) | Method for detecting salient polarization SAR image change based on comprehensive neighborhood information | |
| CN110070545A (en) | A kind of method that textural characteristics density in cities and towns automatically extracts cities and towns built-up areas | |
| Wu et al. | ALS data based forest stand delineation with a coarse-to-fine segmentation approach |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181106 |
|
| RJ01 | Rejection of invention patent application after publication |