CN106204596A - A kind of panchromatic wave-band remote sensing image cloud detection method of optic estimated with fuzzy hybrid based on Gauss curve fitting function - Google Patents
A kind of panchromatic wave-band remote sensing image cloud detection method of optic estimated with fuzzy hybrid based on Gauss curve fitting function Download PDFInfo
- Publication number
- CN106204596A CN106204596A CN201610550554.8A CN201610550554A CN106204596A CN 106204596 A CN106204596 A CN 106204596A CN 201610550554 A CN201610550554 A CN 201610550554A CN 106204596 A CN106204596 A CN 106204596A
- Authority
- CN
- China
- Prior art keywords
- curve fitting
- threshold value
- fitting function
- gauss curve
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10032—Satellite or aerial image; Remote sensing
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The present invention proposes a kind of panchromatic wave-band remote sensing image cloud detection method of optic estimated based on Gauss curve fitting function with fuzzy hybrid, after being judged by image self-adaptive two-value, realize the initial markers of the highlighted and low dark two class pixel datas of gray scale, then the method utilizing blur estimation in two class labelling pixels calculates new Gauss curve fitting function parameter and carries out gray scale binary segmentation threshold estimation, weighted calculation two class binary segmentation threshold value, obtain the iteration Closing Binary Marker that image is overall, iteration labelling re-starts the blur estimation of Gauss curve fitting function, two class data set binary segmentation threshold values are made gradually to converge to close, finally meet threshold value limit poor, obtain final highlighted figure speckle intensity slicing threshold value, realize panchromatic wave-band remote sensing image cloud detection.It is an advantage of the invention that and can be classified by the adaptivenon-uniform sampling of panchromatic wave-band image greyscale, in conjunction with Gauss curve fitting function and fuzzy hybrid method of estimation, constantly by highlighted figure speckle gray scale iteration convergence distant for two classes to consistent state, thus calculate the accurate cloud detection threshold value of estimation, significantly improve the precision of detection, breadth and depth.
Description
Technical field
The invention belongs to Surveying Science and Technology field, relate to a kind of estimating based on Gauss curve fitting function and fuzzy hybrid
Panchromatic wave-band remote sensing image cloud detection method of optic, is mainly used in the fields such as optical satellite Remote Sensing Data Processing and application.
Background technology
For a long time, land resources optical satellite image is limited to the impact of cloud, and large-scale earth observation exists bottleneck.
The existence in territory, cloud sector not only covers terrestrial object information, processes also to the registration of image, fusion etc. and causes many impacts.The most more
The conventional main textured analytic process of cloud detection algorithm, homomorphic filtering method, multi-spectrum synthesis method etc..Texture analysis is based on statistics side
Method extracts cloud sector and the spatial character in non-cloud sector, can effectively identify large stretch of stratus, but be difficult to the small pieces cirrus that texture is stronger.
It is more effective that homomorphic filtering method processes the thinnest cloud, but is not suitable for spissatus image, and algorithm relates to wave filter and cutoff frequency
The selection of rate, can lose some useful informations during filtering.Multi-spectrum synthesis method utilizes the difference section of object reflectance
Dividing cloud and clear sky, but it requires that sensor is furnished with multiple Thermal infrared bands, detection wavelength need to contain the absorption of water or carbon dioxide
Band, is used for intermediate-resolution imager and senior very high resolution radiometer etc., and is not properly suited for land resources optics and defends
Star image.
Satellite image would generally exist the cirrocumulus of small pieces texture-rich and the most spissatus, and the high score of main flow simultaneously
Resolution road resource satellite wave band is less but color range is enriched, and typically carries full-color camera and comprises indigo plant, green, red, near-infrared 4 spectrum
The multispectral camera of section, uses 10 ~ 12bits to carry out radiation and quantifies.Wherein, the image data resolution of panchromatic wave-band is higher, is
The key data source of application of satellitic remote sensing, therefore, explores the automatic of a kind of applicable land resources optical satellite panchromatic wave-band image
Cloud detection algorithm has great importance.
Summary of the invention
It is an object of the invention to provide a kind of panchromatic wave-band remote sensing shadow estimated based on Gauss curve fitting function with fuzzy hybrid
As cloud detection method of optic, it can overcome the deficiency of above-mentioned existing cloud detection method of optic technology, meet at single panchromatic wave-band image number
Demand according to the detection of upper high-efficient automatic cloud sector.
The technical scheme is that a kind of panchromatic wave-band remote sensing shadow estimated based on Gauss curve fitting function with fuzzy hybrid
As cloud detection method of optic, comprise the following steps,
Step 1, self-adaptive initial Closing Binary Marker, utilize Otsu algorithm that panchromatic wave-band remote sensing image is carried out self-adaption binaryzation and divide
Cut threshold calculations, and according to binary segmentation, image is divided into highlighted data collection and low dark data set according to gray scale；
Step 2, the fuzzy statistics of Gauss curve fitting function, according to histogrammic statistical result on two class data sets, utilize Gauss
Fitting function carries out blur estimation, determines the binary segmentation threshold value of two class data sets respectively；
Step 3, asks for the overall new Closing Binary Marker threshold value of image, asks for the overall data of weighting according to two binary segmentation threshold values
Collection Closing Binary Marker threshold value, and re-start labelling and the division of two class data sets；
Step 4, iterative process, repeat step 2 and step 3, until the two class numbers that Gauss curve fitting function blur estimation obtains
Difference according to collection threshold value is poor less than limiting, and iteration terminates；
Step 5, final cloud detection threshold value determines, appoints two class data set threshold values of step 4 gained and takes one of them and complete image
Final binary segmentation labelling, obtain final highlighted data collection and low dark data set；
Step 6, region arranges, the mathematical morphology that the highlighted data collection labelling of step 5 gained is expanded successively and corroded
Operation, thus obtain final cloud detection result.
One width is comprised to the satellite image on a large scale of multiple atural object, its histogram distribution is regarded as by multiple simple distribution
The multimodal form being mixed to form, gray scale statistical function in step 2h(x)Can be with a Gauss curve fitting modelg(x)Approximate such as
Following formula,
Wherein,MFor the number of high bass wave in mixed model；It ismGauss of distribution function that individual waveform is corresponding (m =
1,2,……M)；Weight for its correspondence.If image total pixel number isN, thenMeet following formula:
。
Utilize expectation-maximization algorithm can carry out the blur estimation of Gauss curve fitting model.Expectation-maximization algorithm is generally
Finding the maximal possibility estimation of parameter or the alternative manner of MAP estimation in rate model, wherein depend on cannot for probabilistic model
The latent variable of observation.The present invention first automatically detects according to image and determines class numberM, and weight of all categories, average、
Standard deviationInitial value, wherein.Then add a number of random sample, adjusted by iterative computation
The parameter of whole maximum likelihood function, convergence obtains the optimal solution in maximum likelihood meaning.
Step 3 being assumed, the binary segmentation threshold value of two blur estimation is finally intended to convergence consistent, therefore according to European
The distance principle of geometry, the mode of weighting can be selected for meansigma methods weighting, i.e. takes the arithmetical average number as next step of two numbers
According to collection binary segmentation threshold value.
In step 6, it is determined that highlighted data collection be the initial results of cloud detection, but owing to result existing the figure of fritter
Speckle, there is also edge the most scrappy, it is therefore desirable to first remove fritter figure speckle with erosion algorithm, then with expansion algorithm by scrappy edge
Merge.
For the satellite image of panchromatic wave-band, the present invention compensate for traditional texture analytic process, homomorphism filter from different angles
Ripple method, the deficiency of multi-spectrum synthesis method, can be used for automatization's cloud detection of the mass remote sensing image without priori.It is the most excellent
Point is that accuracy of detection is high, judges by accident less, it is not necessary to manual intervention, calculates speed.
Accompanying drawing explanation
Fig. 1 is the panchromatic wave-band remote sensing image cloud estimated with fuzzy hybrid based on Gauss curve fitting function of the embodiment of the present invention
Detection technique flow chart.
Detailed description of the invention
Technical solution of the present invention is described in detail below in conjunction with drawings and Examples.
Seeing Fig. 1, the panchromatic wave-band remote sensing image cloud detection method of optic that the present invention provides, is to be sentenced by image self-adaptive two-value
Have no progeny, it is achieved the initial markers of the highlighted and low dark two class pixel datas of gray scale, then utilize to obscure in two class labelling pixels and estimate
The method of meter calculates new Gauss curve fitting function parameter and carries out gray scale binary segmentation threshold estimation, weighted calculation two class binary segmentation
Threshold value, obtains the iteration Closing Binary Marker that image is overall, re-starts the blur estimation of Gauss curve fitting function, make on iteration labelling
Obtain two class data set binary segmentation threshold values and gradually converge to close, finally meet threshold value limit poor, obtain final highlighted figure speckle gray scale
Segmentation threshold, it is achieved panchromatic wave-band remote sensing image cloud detection
The concrete methods of realizing of embodiment comprises the following steps:
Step 1, self-adaptive initial Closing Binary Marker: utilize Otsu algorithm that panchromatic wave-band remote sensing image is carried out self-adaption binaryzation and divide
Cut threshold calculations, and according to binary segmentation, image is divided into highlighted data collection and low dark data set according to gray scale.
Embodiment without any priori, such as cloud atlas, other data-aided under the conditions of, for a secondary panchromatic wave-band
Satellite remote-sensing image, utilize Otsu algorithm to carry out the binaryzation of image, the location of pixels on image divided according to gray feature
Being two classifications labelling, a class is highlighted data collection, another kind of for low dark data set.
Step 2, the blur estimation of Gauss curve fitting function: according to histogrammic statistical result on two class data sets, utilize
Gauss curve fitting function carries out blur estimation, determines the binary segmentation threshold value of two class data sets respectively.
Embodiment utilizes Gauss curve fitting function that two class data sets carry out grey level histogram matching respectively, and uses expectation
Bigization algorithm carries out the blur estimation of function model parameter, obtains calculated two the binary segmentation threshold values of fitting function.Its
In, the histogram distribution of data set regards the multimodal form being mixed to form by multiple simple distribution, gray-scale statistical function ash(x)Can
With with a Gauss curve fitting modelg(x)Carry out formula approximately as described below,
Wherein,MFor the number of high bass wave in mixed model；It ismGauss of distribution function that individual waveform is corresponding (m =
1,2,……M)；Weight for its correspondence.If image total pixel number isN, thenMeet following formula:
The initial value of Gauss curve fitting function in embodimentObtain according to histogram distribution detection, i.e. arrange one
The local window of sizing, utilizes local maximum method to obtain several histogram peak points, is got rid of by too small peak point, surplus
More than the number of bigger peak point be high bass wave numberM, then between adjacent peak, use local minimum method to obtain paddy
Value point.
Now, noteP m It ismIndividual Gaussian waveform peak point abscissa,V m 、V m+1 It is respectively themTwo valley points about individual waveform
Abscissa.Then initial parameter and can being expressed as:
Wherein,
。
Then, according to initial value and posterior probabilityCarry out blur estimation, obtain new model ginseng
Number valueWherein,x n For samplenObservation value,N
For number of samples, iterative computation makes function model parameter gradually restrain, and the Gauss curve fitting function i.e. obtaining two class data sets is estimated
Meter result, chooses binary segmentation threshold value.
Step 3, asks for the overall new Closing Binary Marker threshold value of image: ask for the entirety of weighting according to two binary segmentation threshold values
Data set binary segmentation threshold value, and re-start labelling and the division of two class data sets.
Embodiment takes the arithmetic mean of instantaneous value new Closing Binary Marker threshold value as image entirety of two binary segmentation threshold values.
Step 4, sets limit poor, iterative process: repeat step 2 and step 3, until Gauss curve fitting function blur estimation
The difference of the two class data set threshold values obtained is poor less than limiting, and iteration terminates.
Embodiment takes and limits difference between threshold value is 3 empirically to limit difference, as step 2, the termination bar of step 3 iterative processing
Part, and then control the result of calculation of final cloud detection threshold value.
Step 5, final cloud detection threshold value determines: appoints two class data set threshold values of step 4 gained and takes one of them and complete
The final binary segmentation labelling of image.
Embodiment selects the segmentation threshold of highlighted data collection as final cloud detection threshold value.
Step 6, region arranges: the mathematics shape that the highlighted data collection labelling of step 5 gained is expanded successively and corroded
State operates, thus obtains final cloud detection result.
Specific embodiment described herein is only to present invention spirit explanation for example.Technology neck belonging to the present invention
Described specific embodiment can be made various amendment or supplements or use similar mode to replace by the technical staff in territory
Generation, but without departing from the spirit of the present invention or surmount scope defined in appended claims.
Claims (3)
1. the panchromatic wave-band remote sensing image cloud detection method of optic estimated with fuzzy hybrid based on Gauss curve fitting function, its feature exists
In: comprise the following steps,
Step 1, self-adaptive initial Closing Binary Marker, utilize Otsu algorithm that panchromatic wave-band remote sensing image is carried out self-adaption binaryzation and divide
Cut threshold calculations, and according to binary segmentation, image is divided into highlighted data collection and low dark data set according to gray scale；
Step 2, the fuzzy statistics of Gauss curve fitting function, according to histogrammic statistical result on two class data sets, utilize Gauss
Fitting function carries out blur estimation, determines the binary segmentation threshold value of two class data sets respectively；
Step 3, asks for the overall new Closing Binary Marker threshold value of image, asks for the overall data of weighting according to two binary segmentation threshold values
Collection Closing Binary Marker threshold value, and re-start labelling and the division of two class data sets；
Step 4, iterative process, repeat step 2 and step 3, until the two class numbers that Gauss curve fitting function blur estimation obtains
Difference according to collection threshold value is poor less than limiting, and iteration terminates；
Step 5, final cloud detection threshold value determines, appoints two class data set threshold values of step 4 gained and takes one of them and complete image
Final binary segmentation labelling, obtain final highlighted data collection and low dark data set；
Step 6, region arranges, the mathematical morphology that the highlighted data collection labelling of step 5 gained is expanded successively and corroded
Operation, thus obtain final cloud detection result.
The panchromatic wave-band remote sensing image cloud detection estimated with fuzzy hybrid based on Gauss curve fitting function the most according to claim 1
Method, it is characterised in that: gray scale statistical function in step 2h(x)With a Gauss curve fitting modelg(x)Carry out formula approximately as described below,
Wherein,MFor the number of high bass wave in mixed model；It ismGauss of distribution function that individual waveform is corresponding (m =
1,2,……M)；For the weight of its correspondence,
If image total pixel number isN, thenMeet following formula:
Utilize expectation-maximization algorithm to carry out the blur estimation of Gauss curve fitting model, in probabilistic model, i.e. find the maximum of parameter
Possibility predication or the alternative manner of MAP estimation, wherein probabilistic model depends on the latent variable that cannot observe, and Gauss intends
Close the initial value of functionObtain according to histogram distribution detection, a certain size local window is i.e. set, profit
Obtain several histogram peak points by local maximum method, too small peak point is got rid of, remain the individual of bigger peak point
Number is high bass wave numberM, then between adjacent peak, use local minimum method to obtain valley point,
Now, noteP m It ismIndividual Gaussian waveform peak point abscissa,V m 、V m+1 It is respectively themTwo valley point abscissa about individual waveform,
Then initial parameter、WithCan be expressed as:
Wherein,
Then, according to initial value and posterior probabilityCarry out blur estimation, obtain new model parameter and take
Value,Wherein,x n For samplenObservation value,NFor
Number of samples, iterative computation makes function model parameter gradually restrain, and i.e. obtains the Gauss curve fitting Function Estimation of two class data sets
As a result, binary segmentation threshold value is chosen.
The panchromatic wave-band remote sensing image cloud estimated with fuzzy hybrid based on Gauss curve fitting function the most according to claim 1 or claim 2
Detection method, it is characterised in that: step 3 being assumed, the binary segmentation threshold value of two blur estimation is finally intended to convergence consistent,
Therefore according to the distance principle of European geometry, the mode of weighting selects meansigma methods to weight, and i.e. takes the arithmetical average conduct of two numbers
Next step data set binary segmentation threshold value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610550554.8A CN106204596B (en) | 2016-07-14 | 2016-07-14 | Panchromatic waveband remote sensing image cloud detection method based on Gaussian fitting function and fuzzy mixed estimation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610550554.8A CN106204596B (en) | 2016-07-14 | 2016-07-14 | Panchromatic waveband remote sensing image cloud detection method based on Gaussian fitting function and fuzzy mixed estimation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106204596A true CN106204596A (en) | 2016-12-07 |
| CN106204596B CN106204596B (en) | 2020-12-29 |
Family
ID=57477971
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610550554.8A Active CN106204596B (en) | 2016-07-14 | 2016-07-14 | Panchromatic waveband remote sensing image cloud detection method based on Gaussian fitting function and fuzzy mixed estimation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106204596B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107103295A (en) * | 2017-04-20 | 2017-08-29 | 苏州中科天启遥感科技有限公司 | Optical remote sensing image cloud detection method of optic |
| CN110046646A (en) * | 2019-03-07 | 2019-07-23 | 深圳先进技术研究院 | Image processing method, calculates equipment and storage medium at system |
| CN115424131A (en) * | 2022-07-19 | 2022-12-02 | 南京航空航天大学 | Remote sensing image cloud detection optimal threshold selection method based on absolute pixels, cloud detection method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101520896A (en) * | 2009-03-30 | 2009-09-02 | 中国电子科技集团公司第十研究所 | Method for automatically detecting cloud interfering naval vessel target by optical remote sensing image |
| US20130079626A1 (en) * | 2011-09-26 | 2013-03-28 | Andriy Shmatukha | Systems and methods for automated dynamic contrast enhancement imaging |
| CN105160306A (en) * | 2015-08-11 | 2015-12-16 | 北京天诚盛业科技有限公司 | Iris image blurring determination method and device |
| CN105354865A (en) * | 2015-10-27 | 2016-02-24 | 武汉大学 | Automatic cloud detection method and system for multi-spectral remote sensing satellite image |
| CN105574502A (en) * | 2015-12-15 | 2016-05-11 | 中海网络科技股份有限公司 | Automatic detection method for violation behaviors of self-service card sender |
-
2016
- 2016-07-14 CN CN201610550554.8A patent/CN106204596B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101520896A (en) * | 2009-03-30 | 2009-09-02 | 中国电子科技集团公司第十研究所 | Method for automatically detecting cloud interfering naval vessel target by optical remote sensing image |
| US20130079626A1 (en) * | 2011-09-26 | 2013-03-28 | Andriy Shmatukha | Systems and methods for automated dynamic contrast enhancement imaging |
| CN105160306A (en) * | 2015-08-11 | 2015-12-16 | 北京天诚盛业科技有限公司 | Iris image blurring determination method and device |
| CN105354865A (en) * | 2015-10-27 | 2016-02-24 | 武汉大学 | Automatic cloud detection method and system for multi-spectral remote sensing satellite image |
| CN105574502A (en) * | 2015-12-15 | 2016-05-11 | 中海网络科技股份有限公司 | Automatic detection method for violation behaviors of self-service card sender |
Non-Patent Citations (3)
| Title |
|---|
| DIHUA GUO等: "Texture-Based Remote-Sensing Image Segmentation", 《2005 IEEE INTERNATIONAL CONFERENCE ON MULTIMEDIA AND EXPO》 * |
| 施鹏: "基于光学遥感图像的舰船目标自动检测技术", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
| 谭凯等: "国产高分辨率遥感卫星影像自动云检测", 《测绘学报》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107103295A (en) * | 2017-04-20 | 2017-08-29 | 苏州中科天启遥感科技有限公司 | Optical remote sensing image cloud detection method of optic |
| CN107103295B (en) * | 2017-04-20 | 2021-01-08 | 苏州中科天启遥感科技有限公司 | Optical remote sensing image cloud detection method |
| CN110046646A (en) * | 2019-03-07 | 2019-07-23 | 深圳先进技术研究院 | Image processing method, calculates equipment and storage medium at system |
| CN110046646B (en) * | 2019-03-07 | 2023-06-30 | 深圳先进技术研究院 | Image processing method, system, computing device and storage medium |
| CN115424131A (en) * | 2022-07-19 | 2022-12-02 | 南京航空航天大学 | Remote sensing image cloud detection optimal threshold selection method based on absolute pixels, cloud detection method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106204596B (en) | 2020-12-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109583425B (en) | Remote sensing image ship integrated recognition method based on deep learning | |
| CN109934200B (en) | RGB color remote sensing image cloud detection method and system based on improved M-Net | |
| CN110909690B (en) | Method for detecting occluded face image based on region generation | |
| CN105335966B (en) | Multiscale morphology image division method based on local homogeney index | |
| CN106023133B (en) | A kind of high-resolution remote sensing image Clean water withdraw method based on multiple features combining processing | |
| CN106295562A (en) | A kind of high-resolution remote sensing image road information extracting method | |
| CN103971115A (en) | Automatic extraction method for newly-increased construction land image spots in high-resolution remote sensing images based on NDVI and PanTex index | |
| CN113239830B (en) | Remote sensing image cloud detection method based on full-scale feature fusion | |
| CN102982538B (en) | Nature color simulating method of resource satellite multi-spectral image | |
| CN102254319A (en) | Method for carrying out change detection on multi-level segmented remote sensing image | |
| CN111967511B (en) | Foundation cloud picture classification method based on heterogeneous feature fusion network | |
| CN105469098A (en) | Precise LINDAR data ground object classification method based on adaptive characteristic weight synthesis | |
| CN107945200A (en) | Image binaryzation dividing method | |
| CN106650812A (en) | City water body extraction method for satellite remote sensing image | |
| CN105139396B (en) | Full-automatic remote sensing image cloud and fog detection method | |
| CN102902956A (en) | Ground-based visible cloud image recognition processing method | |
| CN105608473A (en) | High-precision land cover classification method based on high-resolution satellite image | |
| CN111008664B (en) | Hyperspectral sea ice detection method based on space-spectrum combined characteristics | |
| CN101980294A (en) | Remote sensing image-based method for detecting ice flood of Yellow River | |
| CN105469393A (en) | Shallow water depth multi-temporal remote sensing image inversion method based on decision fusion | |
| CN107992856A (en) | High score remote sensing building effects detection method under City scenarios | |
| CN112200083B (en) | Airborne multispectral LiDAR data segmentation method based on multivariate Gaussian mixture model | |
| CN106204596A (en) | A kind of panchromatic wave-band remote sensing image cloud detection method of optic estimated with fuzzy hybrid based on Gauss curve fitting function | |
| CN104484670A (en) | Remote sensing image cloud detection method based on pseudo color and support vector machine | |
| CN106875407A (en) | A kind of unmanned plane image crown canopy dividing method of combining form and marking of control |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |