CN106327452A - Fragmented remote sensing image synthesis method and device for cloudy and rainy region - Google Patents
Fragmented remote sensing image synthesis method and device for cloudy and rainy region Download PDFInfo
- Publication number
- CN106327452A CN106327452A CN201610665330.1A CN201610665330A CN106327452A CN 106327452 A CN106327452 A CN 106327452A CN 201610665330 A CN201610665330 A CN 201610665330A CN 106327452 A CN106327452 A CN 106327452A
- Authority
- CN
- China
- Prior art keywords
- image
- data
- area
- cloudless
- fragmentation
- 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
- 238000001308 synthesis method Methods 0.000 title abstract description 3
- 238000000034 method Methods 0.000 claims abstract description 59
- 230000005855 radiation Effects 0.000 claims abstract description 51
- 238000012937 correction Methods 0.000 claims abstract description 27
- 238000009826 distribution Methods 0.000 claims abstract description 21
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 20
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 20
- 239000012634 fragment Substances 0.000 claims abstract description 17
- 238000013480 data collection Methods 0.000 claims description 54
- 238000013467 fragmentation Methods 0.000 claims description 54
- 238000006062 fragmentation reaction Methods 0.000 claims description 54
- 238000010606 normalization Methods 0.000 claims description 47
- 230000001568 sexual effect Effects 0.000 claims description 47
- 230000008569 process Effects 0.000 claims description 21
- 230000003595 spectral effect Effects 0.000 claims description 18
- 238000001228 spectrum Methods 0.000 claims description 18
- 238000010189 synthetic method Methods 0.000 claims description 16
- AGVAZMGAQJOSFJ-WZHZPDAFSA-M cobalt(2+);[(2r,3s,4r,5s)-5-(5,6-dimethylbenzimidazol-1-yl)-4-hydroxy-2-(hydroxymethyl)oxolan-3-yl] [(2r)-1-[3-[(1r,2r,3r,4z,7s,9z,12s,13s,14z,17s,18s,19r)-2,13,18-tris(2-amino-2-oxoethyl)-7,12,17-tris(3-amino-3-oxopropyl)-3,5,8,8,13,15,18,19-octamethyl-2 Chemical compound [Co+2].N#[C-].[N-]([C@@H]1[C@H](CC(N)=O)[C@@]2(C)CCC(=O)NC[C@@H](C)OP(O)(=O)O[C@H]3[C@H]([C@H](O[C@@H]3CO)N3C4=CC(C)=C(C)C=C4N=C3)O)\C2=C(C)/C([C@H](C\2(C)C)CCC(N)=O)=N/C/2=C\C([C@H]([C@@]/2(CC(N)=O)C)CCC(N)=O)=N\C\2=C(C)/C2=N[C@]1(C)[C@@](C)(CC(N)=O)[C@@H]2CCC(N)=O AGVAZMGAQJOSFJ-WZHZPDAFSA-M 0.000 claims description 10
- 238000005192 partition Methods 0.000 claims description 8
- 239000012141 concentrate Substances 0.000 claims description 7
- 238000012216 screening Methods 0.000 claims description 7
- 238000007476 Maximum Likelihood Methods 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 6
- 238000012417 linear regression Methods 0.000 claims description 6
- 208000011580 syndromic disease Diseases 0.000 claims description 6
- 238000013139 quantization Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 230000008439 repair process Effects 0.000 claims description 4
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 230000009471 action Effects 0.000 claims description 2
- 239000003623 enhancer Substances 0.000 claims description 2
- 230000010354 integration Effects 0.000 claims description 2
- 239000004744 fabric Substances 0.000 claims 1
- 238000005303 weighing Methods 0.000 claims 1
- 238000012545 processing Methods 0.000 abstract description 2
- 238000001514 detection method Methods 0.000 description 18
- 230000000694 effects Effects 0.000 description 15
- VMXUWOKSQNHOCA-UKTHLTGXSA-N ranitidine Chemical compound [O-][N+](=O)\C=C(/NC)NCCSCC1=CC=C(CN(C)C)O1 VMXUWOKSQNHOCA-UKTHLTGXSA-N 0.000 description 9
- 238000013329 compounding Methods 0.000 description 8
- 238000000746 purification Methods 0.000 description 8
- 230000008901 benefit Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 238000005286 illumination Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 206010008190 Cerebrovascular accident Diseases 0.000 description 3
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 description 3
- 208000006011 Stroke Diseases 0.000 description 3
- 230000004075 alteration Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 3
- 230000007812 deficiency Effects 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000007796 conventional method Methods 0.000 description 2
- 238000013502 data validation Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 238000007670 refining Methods 0.000 description 2
- 241000894007 species Species 0.000 description 2
- 230000002194 synthesizing effect Effects 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 241000005787 Cistanche Species 0.000 description 1
- VAYOSLLFUXYJDT-RDTXWAMCSA-N Lysergic acid diethylamide Chemical compound C1=CC(C=2[C@H](N(C)C[C@@H](C=2)C(=O)N(CC)CC)C2)=C3C2=CNC3=C1 VAYOSLLFUXYJDT-RDTXWAMCSA-N 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000003702 image correction Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000002045 lasting effect Effects 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000877 morphologic effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000002994 raw material Substances 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration using histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a fragmented remote sensing image synthesis method and device for a cloudy and rainy region, and the method comprises the following steps: obtaining a multi-satellite and multi-phase initial image data set of the cloudy and rainy region; carrying out the radiation difference correction of a sensor and an external factor for the initial image data set, so as to obtain a first image data set consistent with the radiation features; obtaining a cloud layer and shadow distribution range of the first image data set, and obtaining a cloudless data fragment set of the cloudy and rainy region through mask processing; selecting a reference data subset from the cloudless data fragment set, repairing a loss region through employing other cloudless data fragment sets, and obtaining the complete cloudless synthetic image of the cloudy and rainy region. The method makes the most of conventional useless data (high cloud images) for synthesis, effectively improves the space-time coverage frequency of the remote sensing observation of the cloudy and rainy region, and provides powerful data support for the remote sensing application based the domestic satellite data especially the application with the high requirements for timeliness.
Description
Technical field
The present invention relates to remote sensing image field, particularly to a kind of fragmentation remote sensing image synthesis towards many sexual intercourses area
Method and device.
Background technology
East China and area, southern province are owing to being affected, often by maritime monsoon or the torrid zone, subtropical climatic condition
Year cloudy rainy day gas, especially in crop (vegetation) Growing season, clear sky weather is the most rare, shows as big face on remote sensing image
Long-pending cloud shadow covers, owing to cloud cover causes earth's surface information cannot be received by remote sensor so that depend on optical satellite
The remote sensing applications such as the resource investigation of remote sensing, ecological environmental protection and crop remote sensing monitoring cannot obtain the cloudless image of crucial phase
Data.Along with the continuous progress of China's satellite technology, the satellite sequential transmissions of multiple different spatial resolutions and temporal resolution
And come into operation, satellite frequency of passing by is greatly improved, and provides potentiality for realizing the lasting observation of earth's surface information.In this background
Under, how from many stars, multidate have cloud image that excavate useful information as far as possible to serve remote sensing application be that current remote sensing should
A difficult problem for solution is needed badly in.
Existing cloud removing technology is based primarily upon single type sensor, such as the multi-temporal remote sensing shadow of Landsat satellite
Picture, select the 1 phase cloudless image of the same area as with reference to base map, by having radiation normalization and the cloud detection of cloud image,
Making cloud image have relatively uniform radiation feature simultaneously with reference to image, calculating cloud distribution, and then utilize
Pixel in cloudless reference image is to there being the cloud masked areas in cloud image to be replaced, thus reaches the purpose of cloud.Technology
Flow process is as it is shown in figure 1, specifically include following steps:
First to having cloud image and carrying out relative radiometric normalization with reference to image, both are made to have the hue and luminance of approximation
Deng radiation feature.Relative radiometric normalization uses iteration weighting Multivariate alteration detection converter technique (Iteration Re-weight
Multivariate Alteration Detection, is called for short IR-MAD), the Scale invariant i.e. utilizing Multivariate alteration detection is special
Property obtain the constant pixel of radiation in the image of same region two (also known as pseudo-invariant features, such as building, road, desert etc., with
Vegetation is different, and these ground classes have relatively stable reflectance, change the most in time)；Based on these pseudo-constant spies
Levying pixel pair, using orthogonal regression method to calculate has cloud image and the regression equation with reference to each wave band of image, then utilizes this time
Return equation to there being cloud image to carry out relative radiometric normalization by wave band.
Then, the method having cloud image to use Threshold segmentation to combine with cluster after radiation normalization is carried out automatic cloud
Detection, determines cloud distribution.Owing to cloud layer is high reflection in each wave band of image, show as highlight regions, with other ground
Thing obvious difference, utilizes this characteristic of cloud layer to carry out cloud detection.Concretely comprise the following steps: 1. come by the similar pixel of Cluster merging
Become image block；2., according to the brightness value distributed area in cloud sector with atural object pixel, utilize threshold value to mark off the kind in cloud sector in image
Subregion and the seed region of atural object, complete preliminary cloud detection.3. based on the seed region of threshold method detection, again with
The thought of cluster, carries out the most Unidentified region refining clustering recognition.
Finally, utilize linear to return the method combined with territory, cloud sector morphological dilations, territory, image cloud sector is carried out
Replacement is filled up, and plays cloud effect.Detailed process is as follows: 1. by cloudless reference image and the target figure of areal different times
Cloudless region corresponding in Xiang carries out linear regression fit, to slacken the SPECTRAL DIVERSITY in two width images further,
To the brightness normalization image with reference to image.2. use flat disc structure is to there being the territory, cloud sector in cloud image to expand, with
Brightness normalization image carries out merging to be replaced, and finally gives Scattered Clouds Or Better image.
Prior art shortcoming is:
1. prior art multi-temporal data merely with single satellite source mostly carries out having the replacement of cloud image to fill up,
Data abundance is difficult to meet the application demand of higher timeliness.
2. during the relative radiometric normalization of multi_temporal images is processed by prior art, by using distinct methods to find image
The pseudo-invariant features point of centering sets up regression equation, whole scape image carries out overall situationization correction, not to different sensors originally
Radiation difference and the external factor difference such as illumination, atmospheric condition of body make a distinction, and also do not account for the spoke between different land types
Penetrate property difference.
3. prior art is based on the cloudless substitute cloud removing with reference to image realization to target image, wherein as reference
The acquisition of cloudless image is the prerequisite of the method.But in southern many sexual intercourses area, cloudless image will be met and cover very
Difficulty, limits effective application of the method.
4. prior art uses cloud detection method of optic based on cluster only to achieve the detection to cloud distribution, and for
For remote sensing image, cloud layer and shadow region thereof are typically the key factor causing data message to lack.The method cannot to by
The shadow region that cloud causes realizes effectively identifying and substitute, and this point also limit the application effect of method.
The information being disclosed in this background section is merely intended to increase the understanding of the general background to the present invention, and should not
When being considered to recognize or imply in any form this information structure prior art well known to persons skilled in the art.
Summary of the invention
It is an object of the invention to provide a kind of fragmentation remote sensing image synthetic method towards many sexual intercourses area and device,
Thus overcome the shortcoming that abundance is relatively low, radiation characteristic is poor of prior art multi-temporal data.
Another object of the present invention is to provide a kind of fragmentation remote sensing image synthetic method towards many sexual intercourses area and
Device, thus overcoming cloudless, the cloud layer difficult with reference to image capturing of prior art and shadow region thereof to cause, that data message lack is scarce
Point.
For achieving the above object, according to an aspect of the present invention, it is provided that a kind of fragmentation remote sensing towards many sexual intercourses area
Image synthesis method, comprises the following steps:
S101: obtain many stars in many sexual intercourses area, the original image data collection of multidate；
S102: described original image data collection is carried out the radiation difference correction of sensor and extrinsicfactor to obtain radiation
The first image data collection that feature is consistent；
S103: obtain cloud layer and the shade distribution thereof of described first image data collection, obtain institute by mask process
State the cloudless fragmentation of data collection in many sexual intercourses area；
S104: choose benchmark data subset from described cloudless fragmentation of data collection, utilizes other described cloudless fragmentation of data collection
Absent region is studied for a second time courses one has flunked, obtains the complete cloudless resultant image in described many sexual intercourses area.
For achieving the above object, according to a further aspect of the invention, it is provided that a kind of fragmentation towards many sexual intercourses area is distant
Sense image synthesizing device, specifically includes:
Data acquisition module, for obtaining many stars in many sexual intercourses area, the original image data collection of multidate；
Data correction module, for carrying out the radiation difference school of sensor and extrinsicfactor to described original image data collection
The first just consistent to obtain radiation feature image data collection；
Data mask module, for obtaining cloud layer and the shade distribution thereof of described first image data collection, by covering
Film processes the cloudless fragmentation of data collection obtaining described many sexual intercourses area；
Repair and integrate module, for choosing benchmark data subset from described cloudless fragmentation of data collection, utilize other described nothing
Cloud fragmentation of data set pair absent region is studied for a second time courses one has flunked, and obtains the complete cloudless resultant image in described many sexual intercourses area.
Compared with prior art, there is advantages that
The present invention, in many stars, multi-temporal data commbined foundations, makes full use of cloud shadow image, especially logical in conventional treatment
The useful pixel region in the middle of high cloud amount image being often rejected, realizes cloudless shadow by the Land use systems of fragmentation valid data
As synthesis, it is effectively improved the remote sensing observations space-time coverage in many sexual intercourses area.
The present invention considers between sensor, external factor and different land types in the middle of relative radiometric normalization processes
Radiation characteristic difference, carries out automanual high accuracy radiant correction by atural object classification.Meanwhile, cloud layer and shade thereof are all had
Effect identification extraction, overcomes the deficiency only detected cloud layer in conventional method, and complete without target area in advance
Cloudless reference image, as substitute data source, improves application effect and the range of application of method, and in order to advance, China is domestic to be defended
Landing of sing data is applied and the remote sensing monitoring application strong data support of offer in south China long-term many sexual intercourses area.
Other features and advantages of the present invention will illustrate in the following description, and, partly become from description
Obtain it is clear that or understand by implementing the present invention.The purpose of the present invention and other advantages can be by the explanations write
Structure specifically noted in book, claims and accompanying drawing realizes and obtains.
Below by drawings and Examples, technical scheme is described in further detail.
Accompanying drawing explanation
Accompanying drawing is for providing a further understanding of the present invention, and constitutes a part for description, with the reality of the present invention
Execute example together for explaining the present invention, be not intended that limitation of the present invention.In the accompanying drawings:
Fig. 1 is the flow chart of existing remote sensing image synthetic method.
Fig. 2 is towards the flow chart of fragmentation remote sensing image synthetic method in many sexual intercourses area according to the present invention.
Fig. 3 is according to multi-source multi_temporal images relative radiometric normalization flow chart of the present invention.
Fig. 4 a is Landsat8-OLI reference sensor clear sky image, and Fig. 4 b is GF1-WFV1 target (waiting to correct) sensor
Clear sky image.
Fig. 5 a is that Fig. 5 b is GF1-WFV1 image to be corrected with reference to Landsat8 image, and Fig. 5 c is that IR method radiates normalizing
Changing image, Fig. 5 d is context of methods radiation normalization image.
Fig. 6 is the flow chart generated with image valid data according to cloud shadow of the present invention detection.
Fig. 7 a and Fig. 7 b is to have cloud image and cloudless image according to areal of the present invention.
Fig. 8 a is cloud layer reinforced effects, and Fig. 8 b is shadow enhancement effect, Fig. 8 c cloud layer region binarization segmentation effect, Fig. 8 d
Shadow region binarization segmentation effect.
Fig. 9 a is raw video, and Fig. 9 b cloud shadow testing result, Fig. 9 c image valid data, in Fig. 9 c, black part is divided into countless
According to region.
Figure 10 is according to present invention Image compounding based on valid data flow chart.
Figure 11 a-Figure 11 h is the cloudy image according to study area of the present invention.
Figure 12 is according to present invention Image compounding based on valid data effect.
Figure 13 is towards the structure chart of fragmentation remote sensing image synthetic method in many sexual intercourses area according to the present invention.
Figure 14 is the structure chart according to Data correction module of the present invention.
Figure 15 is the structure chart obtaining submodule according to coefficient of the present invention.
Figure 16 is the structure chart according to the present invention the first syndrome module.
Figure 17 is the structure chart according to data mask module of the present invention.
Figure 18 is the structure chart repaired according to the present invention and integrate module.
Detailed description of the invention
Below in conjunction with the accompanying drawings, the detailed description of the invention of the present invention is described in detail, it is to be understood that the guarantor of the present invention
Scope of protecting is not limited by detailed description of the invention.
In conventional image handling process, typically the cloud amount image more than 50% is considered as useless image data and is given up,
Thus cause the disappearance of effective image element information in the cloudless region of image.Difficulty is obtained for southern many sexual intercourses area satellite image,
Being difficult to meet the application difficult problem that the most cloudless image covers, the present invention proposes many stars, the united method of multi_temporal images carries
High data cover degree, and use mode based on the cloudless valid data of fragmentation to carry out Image compounding, thus meet many sexual intercourses ground
District's application demand to cloudless image data.
As in figure 2 it is shown, according to a kind of fragmentation remote sensing image towards many sexual intercourses area of the specific embodiment of the invention
Synthetic method, comprises the following steps:
Step S101: obtain many stars in many sexual intercourses area, the original image data collection of multidate；
Step S102: original image data collection carries out the radiation difference correction of sensor and extrinsicfactor, eliminates pass more
The radiance difference caused because of sensor performance, altitude of the sun, Atmospheric Absorption and scattering etc. between sensor and multi_temporal images,
Obtain the first image data collection that radiation feature is consistent；
Step S103: the image data of normalization via radiation is carried out Yun Ying detection, obtains the first image data collection
Cloud layer and shade distribution, by mask process obtain many sexual intercourses area cloudless fragmentation of data collection；
Step S104: according to reality application to sensor of interest and the requirement of center phase, select from cloudless fragmentation of data collection
Take benchmark data subset, utilize other cloudless fragmentation of data set pair absent region to study for a second time courses one has flunked, obtain the complete of many sexual intercourses area
Cloudless resultant image.
The present invention is by relative radiometric normalization, automatization's cloud shadow detection, Image compounding based on fragmentation valid data
Etc. key technology, extract available pixel region from many stars, having cloud image data of multidate, make full use of each pixel
Information, the cloudless image in synthesis certain time window, the remote sensing monitoring application for many sexual intercourses area is provided with force data support.
Above-mentioned steps S102 is as it is shown on figure 3, specifically include:
Step S201: based on clear sky image, the mode using classification to return obtains sensor spectrum normalization coefficient；Specifically
For:
In this step, on the basis of being selected as the respective sensor with reference to image, other sensor is target to be corrected, choosing
Take both and there is the clear sky image of overlapping region and phase close (less than 1 week) to carrying out relative correction, now it is believed that big
Gas and ground mulching change impact are small, and the radiation difference between two images is essentially from sensor itself.The spectrum of sensor is returned
One change process has relative independentability, and therefore clear sky image can be different from image number to be corrected to being chosen in spatial and temporal distributions
According to collection.
First, image being carried out radiation calibration, by DN value, (DN value (Digital Number) is remote sensing image picture element brightness
Value, the gray value of the atural object of record) be converted to spoke brightness, make different images pixel value have identical dimension level, eliminate and pass
Quantization progression (pixel position the is deep) difference impact on fitting precision between sensor.
Secondly, respectively the overlay region image of the clear sky image of other sensor and the clear sky image of reference sensor is carried out
The supervised classification of the big class such as vegetation, settlement place, bare area, water body, and the most automatically choose sufficient amount of at various places apoplexy due to endogenous wind respectively
Sample point；
Finally, according to the sample point set obtained, set up equation of linear regression for the different-waveband in two images and classification,
Ask for regression coefficient i.e. spectral normalization coefficient.
Satellite sensor (usually 1-2) performance within a certain period of time is the most stable, therefore, and the spectrum normalizing of sensor
Change coefficient and can regard constant within this period as, and be used in the form of a lookup table, to improve parameter reusability and applicable model
Enclose.Table 1 is the regression model that GF-WFV1 carries out on the basis of Landsat8-LIO sensor sensor calibration, regression equation
Coefficient is the spectral normalization coefficient of class accordingly.
The table 1 GF-WFV1 regression model on the basis of Landsat8-LIO
Step S202: based on sensor spectrum normalization coefficient, utilize method that sample transmission classifies again to many stars, many time
The original image data collection of phase carries out semi-automatic classification, and sorted image data collection is carried out sensor radiation difference correction；
In this step, in Classification in Remote Sensing Image, sample quality is closely related with nicety of grading, for ensureing the height of sample information
" fidelity ", typically carries out choosing of sample from image to be sorted.When same sample is applied to many scapes image, due to air
The difference of situation, illumination and phase easily causes the different spectrum of jljl, the phenomenon of same object different images, brings bigger uncertainty to classification.
In conjunction with the sequential image feature to the same area Continuous Observation, it is proposed that a kind of semi-automatic classification method based on sample transmission,
The candidate samples position classified as next phase in the classification locus that early stage classification obtains, and for a new phase image again
Carry out the calculating of sample characteristics, optimization and classify again.Therefore for multi-source, multi_temporal images, only one need to be carried out to reference to image
The supervised classification of secondary artificial sampling just can realize the automatic categorizing process of full dataset.
Specific as follows:
First, the magnitude of the image picture element of original image data collection is fixed to carry out radiation with spectral normalization coefficients match
Mark；The calibration results is as the data basis of subsequent processes.
Secondly, choosing the optimum image of the quality of data from calibration image sequence as with reference to image, remaining is then for waiting to entangle
Positive image, and choose and maximum likelihood classification carrying out sample with reference to image, it is thus achieved that with reference to the classification results of image.
Further, on the basis of current reference image and classification results thereof, next is expected that correcting image carries out sample sieve
Choosing and Sample purification are to obtain universal class the most originally；
Finally, utilize universal class the most originally to treat correction image and carry out maximum likelihood classification, and according to the type of sensor, right
The pixel set that image to be corrected of all categories is corresponding carries out sensor spectrum normalization.
The automatic classification transmitted by sample, is effectively increased multi-source, multi_temporal images classification and the place of radiation normalization
Reason efficiency.Wherein, screening sample and Sample purification combine the features of forward and backward two phase images and carry out refining of sample, right
Reduce computation complexity and improve nicety of grading again and play a crucial role.
(1) screening sample
Screening sample, based on when early stage with reference to image and classification results thereof, calculates and obtains each from image overlap district
The Pixel domain position collection that ground class is representative, the candidate samples locus expecting to correct image as next.General and
Speech, similar atural object has similar spectral signature, and the n dimension spectral space constituted at n image wave band is integrated distribution, from classification
The nearest pixel in center has higher representativeness.The method of screening sample is, first obtains current reference image and waits to correct
The overlay region scope of image, retrains using classification chart picture as class scope in overlay region, to when early stage reference image is by classification
Calculate averaged spectrum vector and tie up the class center of spectral space as such at n；Then, with Euclidean distance for tolerance, calculate every
One apoplexy due to endogenous wind all pixels spectral vector is to the distance of class center, and obtains, by sort algorithm, the m that distance class center is nearest
Individual pixel is as such typical pixel sample；Finally pass to wait to entangle by set of spatial locations corresponding for all categories pixel sample
Positive image, as the spatial distribution of candidate samples.The selection of m value to be determined on a case-by-case basis, if same ground, overlay region species
The form of expression of type is complex, and desirable bigger m value is to obtain sufficient amount of sample pixel.M value takes corresponding classification picture herein
The 20% of unit's sum.
In formula, ED is the pixel Euclidean distance to class center, xi, yiRepresent pixel and class center in the i-th wave band respectively
Gray value, n is image wave band number.
(2) Sample purification
Sample purification carries out recalculating of sample characteristics for image to be corrected, and rejects in candidate samples due to ground species
The noise pixel that type change, Yun Ying covering etc. cause, improves sample purity.Variance method of purification is used to carry out Sample purification herein.
First combine image to be corrected and read the pixel spectral vector of candidate samples position by class, ask on this basis and calculate the flat of each class
All spectral vector xi；Secondly, variance var (X) that in solving class by class, the spectral vector of each pixel is vectorial with averaged spectrum is (see public affairs
Formula 2), and all pixel variances in class are sought mean of varianceWherein var (X) describes the different of pixel and generic
Matter degree, is worth the least, shows that this pixel is the lowest with the difference of classification, andThen reflect the average of pixel gray value in classification
Dispersion degree, can be as the reference of the heterogeneous degree of pixel.Finally, threshold value is taken for each ground classBy each apoplexy due to endogenous wind
Var (X) ＞ TvarPixel be considered as heterogeneous pixel and rejected, final pure sample this pixel set obtaining all ground class.
In formula, xiWithBeing respectively pixel spectral vector gray value at i wave band vectorial with averaged spectrum, n is spectral vector
Dimension, i.e. image wave band number.
Step S203: on the basis of classification, the PIFs side of choosing automatically retrained based on NDVI difference value histogram and classification
Method, it is achieved the radiation normalization of the extrinsicfactor of image data collection.
In this step, in conjunction with sensor spectral normalization feature based on classification, it is proposed that based on NDVI difference value histogram
With pseudo-invariant features point (Pseudo Invariant Features, the PIFs) automatically selecting method of classification constraint, basis at this
On build image to be corrected with reference to the equation of linear regression of each wave band in image, it is achieved treat the radiation normalization correcting image
Correction.Shown in the calculating of NDVI and difference thereof such as formula (3), (4).
Δ NDVI=NDVIr-NDVIt (4)
In formula, R_nir and R_red is respectively image near-infrared and infrared band reflectance value；NDVIrAnd NDVItIt is respectively
With reference to image and the NDVI image of image to be corrected.
Image is after sensor relative correction, and the cities and towns more stable for reflectance and bare area pixel are it is believed that it is subject to
Illumination and the entire effect of air and even variation, therefore, the most former image NDVI due to object spectrum various and in unimodal or
Multi-modal, its cities and towns and bare area classification NDVI difference show as relative stable and concentrate, and can approximate and represent with normal distribution.
Stable radiation point is positioned near the histogrammic mean μ of Δ NDVI, and is positioned at scattergram both sides by the point of instability of noise jamming.Will
Being positioned at the PIFs as stable radiation of the point in the range of μ ± c σ, wherein σ is the standard deviation of Δ NDVI, and c is for determining that point of safes is interval
Constant, take c=1 herein.With PIFs as sample point, set up the such as equation of linear regression of formula (5) for each wave band, according to
A young waiter in a wineshop or an inn takes advantage of principle, calculates the optimal coefficient k of each wave bandi、bi, treat correction image and carry out Volatile material.
pri=ki×pti+bi(i=1 ..., n) (5)
In formula, priAnd ptiRepresent the i-th wave band with reference to image and image to be corrected, k respectivelyiAnd biFor fitting coefficient, n is
Wave band number.
Fig. 4 a and Fig. 4 b is respectively Landsat8-OLI and the GF1-WFV1 two sensors in identical imaging time and region
Clear sky image, it can be seen that image contrasts substantially in the radiation characteristic such as contrast, brightness, reflects satellite sensor in response
There is systematic divergence in the aspect such as characteristic, band setting.Also illustrate to be divided into for biography the relative radiometric normalization of image simultaneously
The radiant correction of sensor self and two processes of radiation normalization for external factor such as illumination have necessity.Fig. 5 a and figure
5b is respectively Landsat8-OLI on May 13rd, 2015 and GF-WFV1 on May 14 in the raw video data of the same area, and two
Person's radiation feature obvious difference, chooses the higher Landsat8-OLI data of definition as reference, the latter is carried out relative spoke
Penetrate normalization.Fig. 5 c and Fig. 5 d respectively uses the common image Return Law (Image Regression, IR) with herein
The calibration result of method, it is seen that radiation normalization method of the present invention can be more effectively to different sensors and phase
Radiation difference between image is corrected.
Further, as shown in Figure 6, step S103 specifically includes:
Wherein, cloud layer and shadow region thereof are the principal elements causing image information to be lost, and the most common method is most
Carry out processing for cloud layer region and have ignored the detection to cloud shade.Present invention achieves the automatization to both, effectively examine
Survey, and based on cloud shadow testing result to there being cloud image to carry out validation process, it is thus achieved that the fragmentation only comprising available pixel has
Effect data, particularly as follows:
Step S501: to having the image data of cloud layer and shade thereof, strengthen cloud layer and the moon thereof based on colour space transformation
Shadow, obtains the cloud layer and the partition threshold of shade thereof strengthened；
Cloud layer has the diffuse-reflectance feature that gray average is high, variance is little in remote sensing image, and cloud shadow then shows as local
The low gray value in region and the feature of low variance, therefore can be according to cloud layer and shade and other atural objects ash on remote sensing image
Degree difference identification goes out Bao Yun and shadow region.For adding the intensity contrast of Herba Cistanches shadow and other atural objects further, use based on color
The cloud shadow detection method of spatial alternation, by the RGB band switching of multispectral image to YCbCrSpace, and it is respectively directed to cloud layer and the moon
Shadow carries out Imaging enhanced.The computing formula that colour space transformation and cloud shadow strengthen is as follows:
Shadow region strengthens formula:
Is=(Cb+Cr)/Y (7)
Cloud layer region enhancing formula:
Ih=Y/Is (8)
The thin cloud strengthened and shadow character image are stretched to 0-255 scope respectively, use Otsu method to carry out threshold value choosing
Take.The method, based on maximum between-cluster variance thought, i.e. chooses suitable threshold value T so that following formula acquirement maximum:
σb(T)=ω1(T)ω2(T)[μ1(T)-μ2(T)]2 (9)
Wherein
ω1, μ1Represent percentage ratio and the average of gray value pixel between 0 to T respectively；ω2, μ2Represent gray value respectively
Percentage ratio and average between T to 255 pixel.The segmentation threshold T of two width images after cloud shadow strengthens is calculated respectively according to Otsu methodh
And Ts.Correspondence [0, T in two width imagesh]∩[0,Ts] region be territory, cloudless shadow zone, by (Th, 255] and determine cloud layer region, (Ts,
255] shadow region is determined.In the middle part of Anhui somewhere have cloud image as a example by carried out testing (see figure to cloud shadow detection method
7a and Fig. 7 b), Yun Ying strengthens and extraction effect is as shown in Fig. 8 a-Fig. 8 d.
Step S502: cloud layer and shadow region thereof are carried out respectively Mathematical Morphology expand with vector quantization with obtain cloud layer and
The distribution of shade, completes Yun Ying detection function.
Step S503: based on partition threshold, by the distribution vector of cloud layer and shade thereof and described first image data
Collection carries out mask process to generate cloudless fragmentation data, as Fig. 9 a-Fig. 9 c can be seen that image data validation process is illustrated
Figure.
Finally, as shown in Figure 10, step S104 specifically includes:
In this step, detected by above-mentioned radiation normalization and cloud shadow and data validation processes, define Image compounding
Valid data " raw material ".Based on valid data collection, combining target regional extent and to phase and the industry in key data source
Business demand, is made iteratively Data Synthesis, ultimately generates the cloudless resultant image in certain period internal object district, specific as follows:
Step S601: the blank grid data generating predefined size and wave band number n according to target area scope vector are made
For synthesis base map (all pixels are NULL value)；
Step S602: with target area for constraint, connected applications is to synthesis period and data source requirement, broken from cloudless data
Sheet concentrates the two pieces of image fragments preferentially choosing area accounting maximum in the target area as benchmark data；
Step S603: carry out the regression fit by wave band by all pixels in the overlay region to two pieces of image fragments,
Set up model of fit between the two and be corrected, making brightness between the two keep consistent with tone, it is achieved color balance, enter one
Both are carried out fusion and inlay by step；
Step S604: the blank that fused data is copied in S601 by wave band is synthesized in base map；
Step S605: judge the area without data (pixel is NULL value in n wave band) that base map is the most still had vacant position,
As then completed the synthesis of image without vacancy, otherwise calculate the regional extent of vacancy, and the size of foundation area of absence is successively from institute
State cloudless fragmentation of data and concentrate the image fragment of the maximum area choosing corresponding region；
Step S606: repeat step S603～S605, until target area no data vacancy, finally realize cloudless image
Synthesis.
The mode using overlay region pixel to carry out regression fit in step S603 realizes color balance, and does not use common
The method such as Histogram Matching be the overall situationization color of the view picture image being more suitable for having more class due to Histogram Matching
Equilibrium, and ground class and the area in valid data fragment is less, considers data overshoot normalized, therefore line simultaneously
Property homing method is more suitable for the color balance in local cell territory.Figure 11 is that GF1-WFV has cloud image, and Figure 12 is to utilize side of the present invention
Method has cloud image to be 2.4 ten thousand Km to area based on GF1-WFV2Region carries out the design sketch of Image compounding.
The present invention takes full advantage of and is considered as the high cloud amount image of hash in the past and synthesizes, and is effectively increased many sexual intercourses
The remote sensing observations space-time covering frequence in area, especially requires high answering for remote sensing application based on domestic satellite data to timeliness
With providing powerful data supporting.The relative radiometric normalization of image data is divided into simultaneously the correction for sensor differences with
And two processes of radiant correction for the external factor difference such as illumination, air, consider the radiation characteristic of different land types simultaneously
Difference, is carried out radiation normalization respectively, is effectively increased radiation normalization effect by atural object classification.And achieve simultaneously to cloud layer
And effective detection of cloud shade, overcome the deficiency only cloud layer region being identified in common methods.Can be not required to cloudless in advance
With reference to the support of image, in the way of fragmentation, carry out Image compounding, widened the scope of application of method, reduce application difficulty.
According to the present embodiment on the other hand, as shown in figure 13, it is provided that a kind of fragmentation remote sensing towards many sexual intercourses area
Image synthesizing device, specifically includes:
Data acquisition module 10, for obtaining many stars in many sexual intercourses area, the original image data collection of multidate；
Data correction module 20, for carrying out the radiation difference of sensor and extrinsicfactor to described original image data collection
Correct the first image data collection consistent to obtain radiation feature；
Data mask module 30, for obtaining cloud layer and the shade distribution thereof of described first image data collection, passes through
Mask process obtains the cloudless fragmentation of data collection in described many sexual intercourses area；
Repair and integrate module 40, for choosing benchmark data subset from described cloudless fragmentation of data collection, utilize described in other
Cloudless fragmentation of data set pair absent region is studied for a second time courses one has flunked, and obtains the complete cloudless resultant image in described many sexual intercourses area.
In this embodiment preferably, as shown in figure 14, Data correction module 20 specifically includes:
Coefficient obtains submodule 201, for based on clear sky image, obtains sensor spectrum normalization coefficient；
First syndrome module 202, for based on sensor spectrum normalization coefficient, to many stars, the initial shadow of multidate
As data set is classified, sorted image data collection is carried out sensor radiation difference correction；
Second syndrome module 203, on the basis of classification, retrains based on NDVI difference value histogram and classification
PIFs choosing method, it is achieved the radiation normalization of the extrinsicfactor of image data collection.
In this embodiment preferably, as shown in figure 15, coefficient acquisition submodule 201 specifically includes:
Image scaling unit 2011, for choosing the clear sky image of reference sensor and other sensor, to described clear sky
Image is calibrated；
Sample classification unit 2012, for the weight by the clear sky image of the clear sky image of other sensor and reference sensor
Folded district image is classified, and chooses the sample point set of sorted plot image respectively；
Coefficient acquiring unit 2013, for according to the sample point set obtained, being respectively directed to the clear sky image of other sensor
Equation of linear regression is set up to obtain spectral normalization coefficient with different-waveband and the classification of the clear sky image of reference sensor.
In this embodiment preferably, as shown in figure 16, the first syndrome module 202 specifically includes:
Image scaling unit 2021, for by the magnitude of the image picture element of described original image data collection and spectral normalization
Coefficients match is to carry out radiation calibration；
Image classification unit 2022, is used for choosing with reference to image, and remaining is image to be corrected, and carries out sample to reference to image
Choose and maximum likelihood classification, it is thus achieved that with reference to the classification results of image；
Screening purification unit 2023, on the basis of current reference image and classification results thereof, expects to correct to next
Image carries out screening sample with Sample purification to obtain universal class the most originally；
Sensor calibration unit 2024, is used for utilizing universal class the most originally to treat correction image and carries out maximum likelihood classification, and
According to the type of sensor, the pixel set that image to be corrected of all categories is corresponding is carried out sensor spectrum normalization.
In this embodiment preferably, as shown in figure 17, data mask module 30 specifically includes:
Cloud layer enhancer module 301, has cloud image data for acquisition, strengthens cloud layer and the moon thereof based on colour space transformation
Shadow；
Threshold values obtains submodule 302, for obtaining the cloud layer of enhancing and the partition threshold of shade thereof；
Cloud layer obtains submodule 303, for cloud layer and shadow region thereof are carried out Mathematical Morphology expansion and vector quantization respectively
To obtain the distribution of cloud layer and shade thereof；
Data acquisition submodule 304, for based on partition threshold, by the distribution vector of cloud layer and shade thereof with described
First image data collection carries out mask process to generate cloudless fragmentation data.
In this embodiment preferably, as shown in figure 18, repair integration module 40 to specifically include:
Submodule 401 is selected in region, for generating predefined size and the sky of wave band number n according to target area scope vector
White raster data is as synthesis base map；
Selection of datum submodule 402, for target area for retraining, concentrating from described cloudless fragmentation of data and preferentially choose
Two pieces of image fragments that area accounting is maximum in the target area are as benchmark data；
Merge and inlay submodule 403, for being carried out by ripple by all pixels in the overlay region to two pieces of image fragments
The regression fit of section, sets up model of fit between the two and is corrected, and makes brightness between the two and tone keep one to show reality
Both are carried out fusion and inlay by existing color balance further；
First synthon module 404, for being copied to fused data by wave band in blank synthesis base map；
Second synthon module 405, for judging the area without data that base map is the most still had vacant position, as then completed without vacancy
The synthesis of image, on the contrary calculate the regional extent of vacancy, and the size of foundation area of absence is successively from described cloudless fragmentation of data
Concentrate the image fragment of the maximum area choosing corresponding region；
Wherein, repeat described fusion and inlay submodule, described first synthon module and described second synthon module
Action, until target area no data vacancy, it is achieved the synthesis of cloudless image.
To sum up, the present invention, in many stars, multi-temporal data commbined foundations, makes full use of cloud shadow image, especially at routine
The useful pixel region in the middle of high cloud amount image being generally rejected in reason, is realized by the Land use systems of fragmentation valid data
Cloudless Image compounding, is effectively improved the remote sensing observations space-time coverage in many sexual intercourses area.
The present invention considers between sensor, external factor and different land types in the middle of relative radiometric normalization processes
Radiation characteristic difference, carries out automanual high accuracy radiant correction by atural object classification.Meanwhile, cloud layer and shade thereof are all had
Effect identification extraction, overcomes the deficiency only detected cloud layer in conventional method, and complete without target area in advance
Cloudless reference image, as substitute data source, improves application effect and the range of application of method, and in order to advance, China is domestic to be defended
Landing of sing data is applied and the remote sensing monitoring application strong data support of offer in south China long-term many sexual intercourses area.
The present invention can have the detailed description of the invention of multiple multi-form, combines accompanying drawing to this above as a example by Fig. 1-Figure 18
The explanation for example of bright technical scheme, this is not meant to that the instantiation that the present invention is applied can only be confined to specifically flow
In journey or example structure, those of ordinary skill in the art is it is to be appreciated that specific embodiments presented above is many
Planting some examples in its preferred usage, the embodiment of any embodiment the claims in the present invention all should be in technical solution of the present invention institute
Within the scope of Yao Qiubaohu.
Finally it is noted that the foregoing is only the preferred embodiments of the present invention, it is not limited to the present invention,
Although being described in detail the present invention with reference to previous embodiment, for a person skilled in the art, it still may be used
So that the technical scheme described in foregoing embodiments to be modified, or wherein portion of techniques feature is carried out equivalent.
All within the spirit and principles in the present invention, any modification, equivalent substitution and improvement etc. made, should be included in the present invention's
Within protection domain.
Claims (10)
1. the fragmentation remote sensing image synthetic method towards many sexual intercourses area, it is characterised in that comprise the following steps:
S101: obtain many stars in many sexual intercourses area, the original image data collection of multidate；
S102: described original image data collection is carried out the radiation difference correction of sensor and extrinsicfactor to obtain radiation feature
The first consistent image data collection；
S103: obtain cloud layer and the shade distribution thereof of described first image data collection, obtained by mask process described many
The cloudless fragmentation of data collection in sexual intercourse area；
S104: choose benchmark data subset from described cloudless fragmentation of data collection, utilizes other described cloudless fragmentation of data set pair to lack
Lose region to study for a second time courses one has flunked, obtain the complete cloudless resultant image in described many sexual intercourses area.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 1, it is characterised in that step
Rapid S102 specifically includes:
S201: based on clear sky image, obtains sensor spectrum normalization coefficient；
S202: based on sensor spectrum normalization coefficient, classify the original image data collection of many stars, multidate, to dividing
Image data collection after class carries out sensor radiation difference correction；
S203: on the basis of classification, the PIFs choosing method retrained based on NDVI difference value histogram and classification, it is achieved image number
Radiation normalization according to the extrinsicfactor of collection.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 2, it is characterised in that step
Rapid S201 specifically includes:
S301: choose the clear sky image of reference sensor and other sensor, calibrates both described clear sky images；
S302: the overlay region image of the clear sky image of other sensor and the clear sky image of reference sensor is classified, point
Do not choose the sample point set of sorted plot image；
S303: according to the sample point set obtained, for the clear sky image of the clear sky image of other sensor and reference sensor
Different-waveband and classification set up equation of linear regression to obtain spectral normalization coefficient.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 2, it is characterised in that step
Rapid S202 specifically includes:
S401: the magnitude of the image picture element of described original image data collection is fixed to carry out radiation with spectral normalization coefficients match
Mark；
S402: choosing with reference to image, remaining is image to be corrected, chooses and maximum likelihood classification carrying out sample with reference to image,
Obtain the classification results with reference to image；
To next, S403: on the basis of current reference image and classification results thereof, expects that correcting image carries out screening sample and sample
Originally universal class it is purified to obtain the most originally；
S404: utilize universal class the most originally to treat correction image and carry out maximum likelihood classification, and according to the type of sensor, to all kinds of
The pixel set that image to be corrected is corresponding carries out sensor spectrum normalization.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 1, it is characterised in that step
Rapid S103 specifically includes:
S501: obtain and have cloud image data, strengthens cloud layer and shade thereof based on colour space transformation；
S502: obtain the cloud layer and the partition threshold of shade thereof strengthened；
S503: cloud layer and shadow region thereof are carried out respectively Mathematical Morphology expansion and divides to obtain cloud layer and shade thereof with vector quantization
Cloth scope；
S504: based on partition threshold, the distribution vector of cloud layer and shade thereof is covered with described first image data collection
Film processes to generate cloudless fragmentation data.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 1, it is characterised in that step
Rapid S104 specifically includes:
S601: generate the blank grid data of predefined size and wave band number n as the synthesis end according to target area scope vector
Figure；
S602: with target area for constraint, concentrates from described cloudless fragmentation of data and preferentially chooses area accounting in the target area
Two pieces of maximum image fragments are as benchmark data；
S603: carry out the regression fit by wave band by all pixels in the overlay region to two pieces of image fragments, sets up both
Between model of fit and be corrected, make brightness between the two keep consistent with tone, it is achieved color balance, further will both
Carry out fusion to inlay；
S604: the blank that fused data is copied in S601 by wave band is synthesized in base map；
S605: judge the area without data that base map is the most still had vacant position, as then completed the synthesis of image without vacancy, otherwise calculates sky
The regional extent lacked, and concentrate the maximum choosing corresponding region successively from described cloudless fragmentation of data according to the size of area of absence
The image fragment of area；
S606: repeat step S603～S605, until target area no data vacancy, finally realize the synthesis of cloudless image.
7. the fragmentation remote sensing image synthesizer towards many sexual intercourses area, it is characterised in that specifically include:
Data acquisition module, for obtaining many stars in many sexual intercourses area, the original image data collection of multidate；
Data correction module, for described original image data collection carried out the radiation difference correction of sensor and extrinsicfactor with
Obtain the first image data collection that radiation feature is consistent；
Data mask module, for obtaining cloud layer and the shade distribution thereof of described first image data collection, at mask
Reason obtains the cloudless fragmentation of data collection in described many sexual intercourses area；
Repair and integrate module, for choosing benchmark data subset from described cloudless fragmentation of data collection, utilize other described cloudless number
Study for a second time courses one has flunked according to fragment set pair absent region, obtain the complete cloudless resultant image in described many sexual intercourses area.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 7, it is characterised in that institute
State Data correction module to specifically include:
Coefficient obtains submodule, for based on clear sky image, obtains sensor spectrum normalization coefficient；
First syndrome module, for based on sensor spectrum normalization coefficient, to many stars, the original image data collection of multidate
Classify, sorted image data collection is carried out sensor radiation difference correction；
Second syndrome module, on the basis of classification, the PIFs retrained based on NDVI difference value histogram and classification chooses
Method, it is achieved the radiation normalization of the extrinsicfactor of image data collection.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 7, it is characterised in that institute
State data mask module to specifically include:
Cloud layer enhancer module, has cloud image data for acquisition, strengthens cloud layer and shade thereof based on colour space transformation；
Threshold values obtains submodule, for obtaining the cloud layer of enhancing and the partition threshold of shade thereof；
Cloud layer obtains submodule, expands with vector quantization for cloud layer and shadow region thereof carry out Mathematical Morphology respectively to obtain cloud
Layer and the distribution of shade thereof；
Data acquisition submodule, for based on partition threshold, by the distribution vector of cloud layer and shade thereof and described first shadow
As data set carries out mask process to generate cloudless fragmentation data.
Fragmentation remote sensing image synthetic method towards many sexual intercourses area the most according to claim 7, it is characterised in that
Described integration module of repairing specifically includes:
Submodule is selected in region, for generating predefined size and the blank grid of wave band number n according to target area scope vector
Data are as synthesis base map；
Selection of datum submodule, for target area for retraining, concentrating from described cloudless fragmentation of data and be preferentially chosen at target
Two pieces of image fragments that in region, area accounting is maximum are as benchmark data；
Merge and inlay submodule, for carrying out the recurrence by wave band by all pixels in the overlay region to two pieces of image fragments
Matching, sets up model of fit between the two and is corrected, and makes brightness between the two and tone keep one to show and realizes color and put down
Both are carried out fusion and inlay by weighing apparatus further；
First synthon module, for being copied to fused data by wave band in blank synthesis base map；
Second synthon module, for judging the area without data that base map is the most still had vacant position, as then completed image without vacancy
Synthesis, on the contrary calculate the regional extent of vacancy, and concentrate choosing from described cloudless fragmentation of data successively according to the size of area of absence
Take the image fragment of the maximum area of corresponding region；
Wherein, repeat described fusion and inlay submodule, described first synthon module and the action of described second synthon module,
Until target area no data vacancy, it is achieved the synthesis of cloudless image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610665330.1A CN106327452B (en) | 2016-08-14 | 2016-08-14 | A kind of fragmentation remote sensing image synthetic method and device towards cloudy rain area |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610665330.1A CN106327452B (en) | 2016-08-14 | 2016-08-14 | A kind of fragmentation remote sensing image synthetic method and device towards cloudy rain area |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106327452A true CN106327452A (en) | 2017-01-11 |
| CN106327452B CN106327452B (en) | 2019-05-07 |
Family
ID=57739354
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610665330.1A Active CN106327452B (en) | 2016-08-14 | 2016-08-14 | A kind of fragmentation remote sensing image synthetic method and device towards cloudy rain area |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106327452B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107240084A (en) * | 2017-06-14 | 2017-10-10 | 湘潭大学 | A kind of removing rain based on single image method and device |
| CN109671038A (en) * | 2018-12-27 | 2019-04-23 | 哈尔滨工业大学 | One kind is based on the classified and layered relative radiometric correction method of pseudo- invariant features point |
| CN109859118A (en) * | 2019-01-03 | 2019-06-07 | 武汉大学 | A kind of method and system for effectively inlaying polygon optimization removal cloud covered areas domain based on quaternary tree |
| CN109934788A (en) * | 2019-03-22 | 2019-06-25 | 鲁东大学 | A kind of remote sensing images missing data restorative procedure based on standard remote sensing images |
| CN111402162A (en) * | 2020-03-13 | 2020-07-10 | 北京华云星地通科技有限公司 | Clear sky data set processing method for satellite remote sensing image |
| CN112017177A (en) * | 2020-09-08 | 2020-12-01 | 科大讯飞股份有限公司 | Reference image determining method, device, equipment and storage medium |
| CN112180406A (en) * | 2020-09-17 | 2021-01-05 | 广州地理研究所 | Method, device and equipment for monitoring data integrity of satellite image |
| CN113077458A (en) * | 2021-04-25 | 2021-07-06 | 北京艾尔思时代科技有限公司 | Cloud and shadow detection method and system in remote sensing image |
| CN113223040A (en) * | 2021-05-17 | 2021-08-06 | 中国农业大学 | Remote sensing-based banana yield estimation method and device, electronic equipment and storage medium |
| CN115410096A (en) * | 2022-11-03 | 2022-11-29 | 成都国星宇航科技股份有限公司 | Satellite remote sensing image multi-scale fusion change detection method, medium and electronic device |
| CN116109478A (en) * | 2023-04-10 | 2023-05-12 | 中国科学院空天信息创新研究院 | Remote sensing image mosaic processing method and device based on comparability |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004001659A1 (en) * | 2001-12-17 | 2003-12-31 | Digitalglobe Inc. | System, method, and apparatus for satellite remote sensing |
| CN102708307A (en) * | 2012-06-26 | 2012-10-03 | 上海大学 | Vegetation index construction method applied to city |
-
2016
- 2016-08-14 CN CN201610665330.1A patent/CN106327452B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004001659A1 (en) * | 2001-12-17 | 2003-12-31 | Digitalglobe Inc. | System, method, and apparatus for satellite remote sensing |
| CN102708307A (en) * | 2012-06-26 | 2012-10-03 | 上海大学 | Vegetation index construction method applied to city |
Non-Patent Citations (5)
| Title |
|---|
| XIANGYUN HU 等: "Automatic Recognition of Cloud Images by Using Visual Saliency Features", 《IEEE GEOSCIENCE AND REMOTE SENSING LETTERS》 * |
| 周伟 等: "多光谱遥感影像中云影区域的检测与修复", 《遥感学报》 * |
| 骆剑承 等: "地理时空大数据协同计算技术", 《地球信息科学》 * |
| 黄启厅 等: "多星数据协同的地块尺度作物分类与面积估算方法研究", 《地球信息科学》 * |
| 黄启厅 等: "多源多时相遥感影像相对辐射归一化方法研究", 《地球信息科学》 * |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107240084B (en) * | 2017-06-14 | 2021-04-02 | 湘潭大学 | Method and device for removing rain from single image |
| CN107240084A (en) * | 2017-06-14 | 2017-10-10 | 湘潭大学 | A kind of removing rain based on single image method and device |
| CN109671038B (en) * | 2018-12-27 | 2023-04-28 | 哈尔滨工业大学 | Relative radiation correction method based on pseudo-invariant feature point classification layering |
| CN109671038A (en) * | 2018-12-27 | 2019-04-23 | 哈尔滨工业大学 | One kind is based on the classified and layered relative radiometric correction method of pseudo- invariant features point |
| CN109859118A (en) * | 2019-01-03 | 2019-06-07 | 武汉大学 | A kind of method and system for effectively inlaying polygon optimization removal cloud covered areas domain based on quaternary tree |
| CN109934788A (en) * | 2019-03-22 | 2019-06-25 | 鲁东大学 | A kind of remote sensing images missing data restorative procedure based on standard remote sensing images |
| CN111402162A (en) * | 2020-03-13 | 2020-07-10 | 北京华云星地通科技有限公司 | Clear sky data set processing method for satellite remote sensing image |
| CN111402162B (en) * | 2020-03-13 | 2023-11-07 | 北京华云星地通科技有限公司 | Method for processing clear sky data set of satellite remote sensing image |
| CN112017177A (en) * | 2020-09-08 | 2020-12-01 | 科大讯飞股份有限公司 | Reference image determining method, device, equipment and storage medium |
| CN112017177B (en) * | 2020-09-08 | 2023-12-01 | 科大讯飞股份有限公司 | Reference image determining method, device, equipment and storage medium |
| CN112180406A (en) * | 2020-09-17 | 2021-01-05 | 广州地理研究所 | Method, device and equipment for monitoring data integrity of satellite image |
| CN113077458A (en) * | 2021-04-25 | 2021-07-06 | 北京艾尔思时代科技有限公司 | Cloud and shadow detection method and system in remote sensing image |
| CN113077458B (en) * | 2021-04-25 | 2023-09-19 | 北京艾尔思时代科技有限公司 | Cloud and shadow detection method and system in remote sensing image |
| CN113223040A (en) * | 2021-05-17 | 2021-08-06 | 中国农业大学 | Remote sensing-based banana yield estimation method and device, electronic equipment and storage medium |
| CN113223040B (en) * | 2021-05-17 | 2024-05-14 | 中国农业大学 | Banana estimated yield method and device based on remote sensing, electronic equipment and storage medium |
| CN115410096A (en) * | 2022-11-03 | 2022-11-29 | 成都国星宇航科技股份有限公司 | Satellite remote sensing image multi-scale fusion change detection method, medium and electronic device |
| CN116109478A (en) * | 2023-04-10 | 2023-05-12 | 中国科学院空天信息创新研究院 | Remote sensing image mosaic processing method and device based on comparability |
| CN116109478B (en) * | 2023-04-10 | 2023-07-04 | 中国科学院空天信息创新研究院 | Remote sensing image mosaic processing method and device based on comparability |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106327452B (en) | 2019-05-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106327452B (en) | A kind of fragmentation remote sensing image synthetic method and device towards cloudy rain area | |
| Makarau et al. | Adaptive shadow detection using a blackbody radiator model | |
| CN108681715A (en) | Hainan Tropical wildwood vegetation type group sorting technique | |
| CN107145891A (en) | A kind of Clean water withdraw method and system based on remote sensing image | |
| EP4010687B1 (en) | Automated concrete/asphalt detection based on sensor time delay | |
| Sudharsan et al. | A survey on hyperspectral imaging for mineral exploration using machine learning algorithms | |
| CN110751036A (en) | High spectrum/multi-spectrum image fast fusion method based on sub-band and blocking strategy | |
| CN102231190B (en) | Automatic extraction method for alluvial-proluvial fan information | |
| Csathó et al. | Inclusion of multispectral data into object recognition | |
| CN112329790A (en) | Rapid extraction method for urban impervious surface information | |
| Sah et al. | Semi-automated cloud/shadow removal and land cover change detection using satellite imagery | |
| CN110070513A (en) | The radiation correction method and system of remote sensing image | |
| Joachim et al. | Cloud detection for night-time panchromatic visible and near-infrared satellite imagery | |
| US11250260B2 (en) | Automated process for dynamic material classification in remotely sensed imagery | |
| CN113920407A (en) | Ship target identification method and system based on multiband remote sensing image fusion | |
| Sharma et al. | Saliency based segmentation of satellite images | |
| Harris et al. | Remote Predictive Mapping 6. A Comparison of Different Remotely Sensed Data for Classifying Bedrock Types in Canada's Arctic: Application of the Robust Classification Method and Random Forests. | |
| Sadjadi et al. | Combining Hyperspectral and LiDAR Data for Building Extraction using Machine Learning Technique. | |
| Chaudhri et al. | Synergetic Effect of Complementary Nature of Hyperspectral and LiDAR Data for High Performance LULC Classification | |
| Monier et al. | The BTC40 Survey for Quasars at 4.8< z< 6 | |
| Neteler et al. | Satellite image processing | |
| Kozoderov et al. | Vegetation pattern recognition using hyperspectral air sounding data | |
| Cui et al. | High-Resolution Optical Satellite Images Color Consistency Method Based on External Color References | |
| CN118135412A (en) | Rural black and odorous water body remote sensing identification method | |
| Zekoll | Cloud shadow detection and removal for high spatial resolution optical satellite data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |