CN106780414A - A kind of processing method of multiple-target remote sensing image clouds - Google Patents
A kind of processing method of multiple-target remote sensing image clouds Download PDFInfo
- Publication number
- CN106780414A CN106780414A CN201611134735.9A CN201611134735A CN106780414A CN 106780414 A CN106780414 A CN 106780414A CN 201611134735 A CN201611134735 A CN 201611134735A CN 106780414 A CN106780414 A CN 106780414A
- Authority
- CN
- China
- Prior art keywords
- image
- cloud
- cloudless
- picture
- atlas picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 9
- 239000000284 extract Substances 0.000 claims abstract description 9
- 238000000605 extraction Methods 0.000 claims abstract description 7
- 150000001875 compounds Chemical class 0.000 claims abstract description 5
- 238000009499 grossing Methods 0.000 claims abstract description 5
- 238000000034 method Methods 0.000 abstract description 26
- 238000007781 pre-processing Methods 0.000 abstract 1
- 230000005855 radiation Effects 0.000 description 10
- 238000012937 correction Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 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 3
- 238000005286 illumination Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 239000000443 aerosol Substances 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000002310 reflectometry Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000012876 topography Methods 0.000 description 1
- 230000009466 transformation Effects 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
-
- 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
- G06T2207/10036—Multispectral image; Hyperspectral image
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of processing method of multiple-target remote sensing image clouds, the method by two width times it is close have cloud and cloudless image rigid registrations；Image after registration is carried out into Histogram Matching；Image to there is cloud enters line mask；Extract with respect to that there should be the image that has cloud sector domain in cloud atlas picture in cloudless image, extraction has the image of cloudless part in cloud atlas picture；Image compound operation；Smoothing denoising, generates result figure.It is roughly the same that the method realizes that two gradation of images are distributed by Histogram Matching, so as to reach the cloud layer that can both reject in image, can ensure that original image information is intact again；For some bad points left in image, treatment is easy for smooth algorithm, so as to obtain apparent image；Pre-processing image data is easy and effective, and step operation is simple and clear.
Description
Technical field
Cloud side is gone to the present invention relates to a kind of processing method of multiple-target remote sensing image clouds, more particularly to remote sensing image
Method.
Background technology
With developing rapidly for information technology and sensor technology, Modern remote data are enriched very much, satellite remote-sensing image
Resolution ratio just develops towards spatial resolution higher, spectral resolution and temporal resolution direction, and remote sensing image is extensive
Ground is for multiple fields such as Evaluation of Agriculture, geologic survey, environmental monitorings.But due to weather, obtain completely cloudless distant
Sense image is sometimes relatively difficult, and what most of image all can be more or less when obtaining is projected by cloud and cloud on ground
When shade and aerosol etc. influence.This brings trouble to the application person of many remote sensing images, how from remote sensing image
Remove the influence of cloud, the matter of utmost importance that often many application persons are faced.Common remote sensing images go the cloud method mainly to have：It is many
Cloud method, the image co-registration method of wavelet decomposition, homomorphic filtering method etc. are gone in item formula correction method, digital elevation model correction.
Cloud method is gone in multinomial correction：It is by another width image rectification to benchmark image, to reach cloud using multinomial
Purpose.The main thought of the method is：A cloudless image in identical area is chosen as reference map, then, by two width figures
As with multinomial registration, the topography being hidden by clouds being replaced with cloudless image, the purpose that cloud is removed to target area image is reached.Should
The shortcoming one of method is that the determination of coefficient has certain difficulty；Two be registering atural object reference point quantity depend on system of polynomials
Number, if the cloud distribution of target area is more, multinomial correction goes the image registration workload of cloud just than larger, and a width figure
As that may have multiple polynomial supports.
Cloud method is gone in digital elevation model correction：Be directed to height above sea level higher area's landform lifting effect, in massif overhead shape
Into cloud body and the method that carries the removal cloud of (B), the method is gone to correct and has the distant of cloud using the figure with identical Geographic Reference
Sense image, reaches the purpose of removal cloud.
Homomorphic filtering method：It is that remote sensing images are filtered with treatment using illumination/Reflectivity Model, is usually applied to disclose
The minutia of shadow region, but algorithmically realize more complicated, processing speed is slow.
It is preferable using homomorphic filtering method for the interior image that there is thin cloud on a large scale.Come for relatively thicker cloud layer
Say, the cloud removing based on single scape image cannot.It is therefore possible to use the complementary information of the image of multidate is carried out
Cloud removing, i.e.,：Cloud sector is substituted using the cloudless image of time close areal, so as to reach the effect of cloud
Really.
The content of the invention
Present invention aim at a kind of, there is provided processing method of multiple-target remote sensing image clouds, the method is by two width time phases
Near has cloud and cloudless image rigid registrations；Image after registration is carried out into Histogram Matching；Image to there is cloud is covered
Film；Extract with respect to that there should be the image that has cloud sector domain in cloud atlas picture in cloudless image, extraction has the image of cloudless part in cloud atlas picture；
Image compound operation；Smoothing denoising, generates result figure.The method realizes that two gradation of images are distributed substantially phase by Histogram Matching
Together, so as to reach by reject cloud layer in image, can ensure that intact seamless spliced of original image information goes cloud effect again；Pin
To some bad points left in image, treatment is easy for smooth algorithm, so as to obtain apparent image；Picture number
Data preprocess is easy and effective, and step operation is simple and clear.
The processing method of multiple-target remote sensing image clouds of the present invention, follow these steps to carry out：
A, by two width times it is close have cloud and cloudless image rigid registrations；
B, the image after registration is carried out into Histogram Matching, the histogram distribution of two images is similar after matching；
C, the image to there is cloud enter line mask, generate two width bianry images (A) and (B), there is the region of cloud in image (A)
Value white portion is (1), and cloudless regional value black region is (0)；There is the regional value black region of cloud in image (B) for (1), nothing
Cloud sector thresholding white portion is (1)；
D, extract it is relative in cloudless image should have the image for having cloud sector domain in cloud atlas picture, extracting has cloudless part in cloud atlas picture
Image, carry out computing with image (A) and cloudless image, extract in cloudless image with respect to should have the figure that has cloud sector domain in cloud atlas picture
As (C), computing is carried out with there is cloud atlas picture with image (B), extraction has the image (D) of cloudless part in cloud atlas picture；
E, image compound operation, computing is overlapped by image (C) and (D), forms the complete image gone after cloud of a width
(E)；
Total some noise spots of image formed in f, step e, noise spot is removed by selective smoothing filter, then to step
The image of e carries out bad point reparation, obtains last result figure F.
The processing method of multiple-target remote sensing image clouds of the present invention, due to the state of sensor, posture and the sun
Illumination, Atmospheric Diffusion and absorption can all cause the radiation difference of atural object, have obvious boundary line between the image and map of filling.
Therefore how to eliminate boundary line is a main problem.To realize the non-boundary between image, typically there are two methods.One is straight
Be connected in existing software directly carries out Histogram Matching to two images；Two is pseudo- invariant features method.Realize pseudo- invariant features
Method generally requires 3 steps：
First, in selection two images radiation value without or little target area (referred to as invariant region) for changing；
Then with the average radiation value of invariant region, solve parameter between image in the linear equality (1) of radiation value and, obtain radiation value
Linear relationship；
Finally, according to the linear relationship obtained between radiation value, by the linear transformation to radiation value, relative radiation school is completed
Just, make the radiation value of image to be corrected consistent with reference picture.Therefore, to realize the relative detector calibration of remote sensing images, just
3 problems must be solved：1. the selection of invariant region；2. the solution of linear relation；3. the conversion of gradation of image.
Both approaches are all based on principle of uniformity.The Luminance Distribution that can make two images by Histogram Matching to the greatest extent may be used
Approaching for energy, can so eliminate radiation difference, be that the image after fusion does not have obvious boundary line.Histogram Matching principle is such as
Under：
Matched with the histogram of target image by by the histogram of source images, it is had consistent with target image
Histogram distribution, but reach the purpose of boundary line.
Assuming that the image intensity value of different phases is asked meets linear relationship, this hypothesis is to set up in the case of approximate.
At this moment can by linear equality to describe different when phase images in gray-scale relation.In processing procedure, each wave band is given birth to
Into a gap mask file, data available in image is labeled as 1, and it is 0 that will need the data markers filled.Once really
The position for having cloud sector domain is determined, it is possible to utilize linear histogram matching method, a linear change has been set up between the two images
Change, conversion correcting gain used and biasing can be calculated (B) by the average of view data and standard variance.
For the pixel value X and the pixel value Y for having same point in cloud atlas picture, the linear change of foundation of certain point in cloudless image
Changing can be equal to GX plus B with formula Y, and wherein G is the correcting gain for Histogram Matching, and B is the biasing for Histogram Matching.
This conversion is applied in view picture blank map picture, as histogram matching.G and B are solid during Histogram Matching
Fixed value, therefore have and have cloud part in cloud atlas picture each needs the picture element of filling so as to reach the purpose of cloud.
Brief description of the drawings
Fig. 1 is flow chart of the present invention
Fig. 2 be the two width time of the invention it is close have cloud and cloudless striograph, wherein 1 to there is cloud atlas, 2 is without cloud atlas
Fig. 3 is cloud mask figure (A) of the present invention, background mask figure (B)
Fig. 4 extracts the corresponding figure (C) in cloud sector for the present invention
Fig. 5 is the Background (D) that the present invention is extracted
Fig. 6 is the figure (E) after the present invention is combined
Fig. 7 is the result figure (F) after the present invention is smoothed
Specific embodiment
Embodiment
To realize going the purpose of cloud, the present invention is mainly realized using IDL programming softwares, further retouched with two width striographs
State
A, by two width times it is close have cloud and cloudless image rigid registrations；
Image after registration is carried out Histogram Matching by b, and the histogram distribution of two images is similar after matching；
C, the image to there is cloud enter line mask, generate two width bianry image A and B, the regional value white for having cloud in image A
Region is 1, and cloudless regional value black region is 0；It is 0, cloudless regional value white area to have the regional value black region of cloud in image B
Domain is 1, and in envi softwares, setting up a mask and extracting has cloud administrative division map A, an anti-mask extraction figure B is resettled, by building
A simple expression formula (b1 eq 0) * lor (b1 eq 0) * 0 is stood to realize.1 is changed to when the value of b1 is 0, its value
For 1 when change 0 into, realize black and white overturn.Especially note when mask is set up, suitably to select threshold value to have extracted cloud sector domain (this
The threshold value in cloud sector is set to 60-255 in invention experiment)；
D, extract it is relative in cloudless image should have the image for having cloud sector domain in cloud atlas picture, extracting has cloudless part in cloud atlas picture
Image, computing is carried out with image A and cloudless image, extract in cloudless image with respect to should have the image that has cloud sector domain in cloud atlas picture
C, is gone to carry out computing with there is cloud atlas picture with image B, and extraction has the image D of cloudless part in cloud atlas picture, in envi softwares directly
Corresponding image can be obtained using mask；
E, image compound operation, carry out image C and D computing and are superimposed upon one piece, form the complete figure gone after cloud of a width
As E, this two images C, D is combined by a band math, expression formula is (b1 eq 0) * lor (b1 eq 0) * 0,
When the value of picture dot in wave band b1 is 0, the picture dot that value is 0 is substituted for the pixel value corresponding to wave band b2；When in wave band b1 as
When the value of unit is not 0, do not replace, it is image C that wherein b1 is corresponding, and it is image D that b2 is corresponding；
F. total some noise spots of image for being formed in step e, noise spot is removed by selective smoothing filter, then to step
The image of e carries out bad point reparation, obtains last result figure F, can be completed with an expression formula in envi softwares, expression
Formula：(b1 eq 0) * lor (b1 eq 0) * 0, if the pixel value of the point is more than 200, with 8 adjacent pixel values of its periphery
Average fills the value；It is unchanged if value is less than 200, noise spot can be thus eliminated, reach the effect of needs.
The region that see from the result figure for the treatment of has cloud in artwork is filled up well has gone up corresponding atural object, and disappears
Except boundary line, the improvement of visual effect of needs is reached.
The processing method of multiple-target remote sensing image clouds of the present invention, mainly uses the mutual of different phase image informations
Characteristic is mended to reach the best visual effect to a certain regional aim.The information of artwork can to greatest extent be retained, do not damaged
Recovery of being tried one's best when losing artwork information has the information in cloud sector domain.Realize that two gradation of images are distributed by Histogram Matching roughly the same,
The cloud layer in image is rejected by so as to reach, can ensure that intact seamless spliced of original image information goes cloud effect again；The party
Method is not limited to the image of different phases, and it realizes removing cloud to the multispectral image data that the same area different sensors are obtained
Process, and the method can be realized in existing various image processing softwares.
Claims (1)
1. a kind of processing method of multiple-target remote sensing image clouds, it is characterised in that follow these steps to carry out：
A. by two width times it is close have cloud and cloudless image rigid registrations；
B. the image after registration is carried out into Histogram Matching, the histogram distribution of two images is similar after matching；
C. enter line mask to the image for having cloud, generate two width bianry images (A) and (B), have the regional value of cloud white in image (A)
Color region is (1), and cloudless regional value black region is (0)；There is the regional value black region of cloud in image B for (0), cloudless region
Value white portion (1)；
D. extract with respect to that there should be the image that has cloud sector domain in cloud atlas picture in cloudless image, extraction has the figure of cloudless part in cloud atlas picture
Picture, carries out computing with image (A) and cloudless image, extracts in cloudless image with respect to should have the image that has cloud sector domain in cloud atlas picture
(C) computing, is carried out with there is cloud atlas picture with image (B), extraction has the image (D) of cloudless part in cloud atlas picture；
E. image compound operation, computing is overlapped by image and (C) with (D), forms the complete image gone after cloud of a width
(E)；
F. total some noise spots of image for being formed in step e, noise spot is removed by selective smoothing filter, then to step e's
Image carries out bad point reparation, obtains last result figure (F).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611134735.9A CN106780414A (en) | 2016-12-11 | 2016-12-11 | A kind of processing method of multiple-target remote sensing image clouds |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611134735.9A CN106780414A (en) | 2016-12-11 | 2016-12-11 | A kind of processing method of multiple-target remote sensing image clouds |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106780414A true CN106780414A (en) | 2017-05-31 |
Family
ID=58879794
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611134735.9A Pending CN106780414A (en) | 2016-12-11 | 2016-12-11 | A kind of processing method of multiple-target remote sensing image clouds |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106780414A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112150461A (en) * | 2020-10-19 | 2020-12-29 | 北京百度网讯科技有限公司 | Method and device for evaluating head-tail definition of cell image |
-
2016
- 2016-12-11 CN CN201611134735.9A patent/CN106780414A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112150461A (en) * | 2020-10-19 | 2020-12-29 | 北京百度网讯科技有限公司 | Method and device for evaluating head-tail definition of cell image |
| CN112150461B (en) * | 2020-10-19 | 2024-01-12 | 北京百度网讯科技有限公司 | Method and apparatus for assessing head-to-tail sharpness of a cell image |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101710416A (en) | Processing method for multiple-target remote sensing image clouds | |
| US10366501B2 (en) | Method and apparatus for performing background image registration | |
| CN107563964B (en) | Rapid splicing method for large-area-array sub-meter-level night scene remote sensing images | |
| Zhou et al. | Seamless fusion of LiDAR and aerial imagery for building extraction | |
| US20170358067A1 (en) | Method and Device for Fusing Panchromatic Image and Infrared Image | |
| CN103049898B (en) | Method for fusing multispectral and full-color images with light cloud | |
| CN106504284A (en) | A kind of depth picture capturing method combined with structure light based on Stereo matching | |
| US10325349B2 (en) | Method and apparatus for enhancing 3D model resolution | |
| Valentini et al. | A new video monitoring system in support of Coastal Zone Management at Apulia Region, Italy | |
| CN109767454B (en) | Unmanned aerial vehicle aerial video moving target detection method based on time-space-frequency significance | |
| JP6798860B2 (en) | Boundary line estimation device | |
| CN103942796A (en) | High-precision projector and camera calibration system and method | |
| CN111739031B (en) | Crop canopy segmentation method based on depth information | |
| CN107301420A (en) | A kind of thermal infrared imagery object detection method based on significance analysis | |
| CN114973028B (en) | Aerial video image real-time change detection method and system | |
| CN103544491A (en) | Optical character recognition method and device facing complex background | |
| CN103218821A (en) | Regional-coverage-oriented automatic image mosaic method | |
| CN100589520C (en) | Detection method for characteristic of edge and angle point of color image | |
| CN112598709B (en) | Pedestrian movement speed intelligent sensing method based on video stream | |
| CN103218811A (en) | Statistical distribution-based satellite multi-spectral image waveband registration method | |
| Tatar et al. | High-resolution satellite stereo matching by object-based semiglobal matching and iterative guided edge-preserving filter | |
| CN109598736A (en) | The method for registering and device of depth image and color image | |
| CN112836634A (en) | Multi-sensor information fusion gate trailing prevention method, device, equipment and medium | |
| Hnatushenko et al. | Variational approach for rigid co-registration of optical/SAR satellite images in agricultural areas | |
| Wan et al. | An assessment of shadow enhanced urban remote sensing imagery of a complex city–Hong Kong |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170531 |
|
| WD01 | Invention patent application deemed withdrawn after publication |