CN102682441B - Hyperspectral image super-resolution reconstruction method based on subpixel mapping - Google Patents
Hyperspectral image super-resolution reconstruction method based on subpixel mapping Download PDFInfo
- Publication number
- CN102682441B CN102682441B CN201210051365.8A CN201210051365A CN102682441B CN 102682441 B CN102682441 B CN 102682441B CN 201210051365 A CN201210051365 A CN 201210051365A CN 102682441 B CN102682441 B CN 102682441B
- Authority
- CN
- China
- Prior art keywords
- pixel
- atural object
- image
- sub
- distributes
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000013507 mapping Methods 0.000 title claims abstract description 20
- 238000005457 optimization Methods 0.000 claims abstract description 16
- 238000002922 simulated annealing Methods 0.000 claims abstract description 6
- VMXUWOKSQNHOCA-UKTHLTGXSA-N ranitidine Chemical compound [O-][N+](=O)\C=C(/NC)NCCSCC1=CC=C(CN(C)C)O1 VMXUWOKSQNHOCA-UKTHLTGXSA-N 0.000 claims description 97
- 238000001228 spectrum Methods 0.000 claims description 45
- 238000000354 decomposition reaction Methods 0.000 claims description 15
- 239000000654 additive Substances 0.000 claims description 3
- 230000000996 additive effect Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims description 2
- 230000036961 partial effect Effects 0.000 abstract 2
- 230000006870 function Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 5
- 230000003595 spectral effect Effects 0.000 description 4
- 230000004304 visual acuity Effects 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a hyperspectral image super to resolution reconstruction method based on the subpixel mapping. The method comprises the following steps of: carrying out the subpixel mapping to acquire proportional values of various ground objects in each pixel element in the original hyperspectral remote sensing image; dividing each pixel element, namely a rough pixel element, in the original hyperspectral remote sensing image into nPoundn subpixels, wherein n is a predefined resolution amplifying factor; assigning each subpixel element to some ground object at random according to the abundance value of each known ground object in each pixel element, so as to obtain the initialized ground object distribution image; then obtaining a ground object distribution image of which resolution is n times of that of the original image by using the simulated annealing algorithm according to the iterative optimization rule; and finally, performing the super to resolution reconstruction based on the subpixel mapping. By considering the partial and overall distribution properties of the ground object at the same time, the hyperspectral image super-resolution reconstruction method based on the subpixel mapping avoids the inauthenticity caused by solving the partial property for the ground object distribution, and thus the hyperspectral remote sensing image can be obtained from the ground object distribution image.
Description
Technical field
The invention belongs to technical field of image processing, be applicable to high-spectrum remote sensing and rebuild, be specifically related to a kind of high spectrum image Super-resolution Reconstruction method based on sub-pixel mapping.
Background technology
In recent years, satellite remote sensing technology has obtained development at full speed, the new high-spectrum remote-sensing imaging technique rising can be a lot, obtain earth's surface observation data on very narrow and quasi-continuous spectral band, for people provide abundanter atural object observation information, greatly strengthened the directly ability of identification and analysis atural object situation from remote sensing images.But high light spectrum image-forming technology spatial resolution when obtaining very high spectral resolution is restricted, and the spatial resolution that therefore how to promote high-spectrum remote sensing becomes problem demanding prompt solution.
For with space plane to be imaged at a distance of the space remote sensing platform of R, ignoring under the condition of optical system nonlinear distortion and noise impact, its spatial resolution Δ L can be expressed as
Wherein, W is CCD device array element width, the focal length that f is optical system.So the spatial resolution that wants to improve remote sensing images can be passed through three kinds of modes, a kind of is to reduce remote-sensing flatform flight track height, but can make the lost of life of this platform like this; In the situation that keeping R constant, other two kinds to put forward high-resolution method be increase focal length and dwindle array element width, yet, increasing focal length can make the difficulty of processing of optical element increase, expense increases, and causes the large Heavy Weight of volume of remote sensor, for actual application brings difficulty; The size of CCD array element is subject to the restriction of technique, and due to external blockade on new techniques, it is enough little that the CCD aperture that we can access can not be done.To sum up, how can be in distance R, under the big or small constant prerequisite of focal distance f and CCD aperture W, remote sensing images spatial resolution being got a promotion is problem demanding prompt solution.Technique of Super-resolution Image Construction is the effective way addressing this problem, this technology utilize one or more from different perspectives, the low-resolution image that obtains such as diverse location, different sensors reconstructs one or more high-definition picture.
High light spectrum image-forming mode is subject to the restriction of sensor spatial resolution and the impact of atural object complex distribution and makes the spatial resolution of gained image not high, this mostly is with regard to the numerical information that causes single pixel to record the mixture that multiple type of ground objects spectral signature forms, and is called as mixed pixel.The existence of mixed pixel phenomenon has brought great difficulty to the quantitative decipher of image, also for improving the spatial resolution of high spectrum image, brings difficulty.Mixed pixel decomposition method is to solve one of this difficult effective way, its basic meaning is for to extract by mixed pixel spectrum typical case " pure " object spectrum that is called as end member, estimate each end member shared ratio in mixed pixel, this ratio is called Abundances simultaneously.For every class atural object, its Abundances in each pixel forms a width abundance figure.But this class mixed pixel decomposition method only provides the ratio of every kind of atural object in each pixel, do not provide them and how to distribute in this pixel.The abundance figure that sub-pixel mapping method utilizes mixed pixel decomposition method to obtain estimates the distribution of every kind of atural object in each pixel, thereby can access the terrain classification image higher than original image resolution under some restrictive condition.Existing sub-pixel mapping method is mainly divided into two classes, based on study and based on space continuity, the former collects some known high resolving power atural object distributed images as training set, by being reduced to resolution, they obtain abundance figure, then adopt the means training of machine learning to obtain the relation between high resolving power atural object distributed image and corresponding abundance figure, and by this relational application on pending high spectrum image abundance figure, thereby obtain needed high resolving power atural object distributed image; Mostly the algorithm of the latter based on space continuity is to distribute and to have this feature of local continuity and propose different models for atural object, and the optimum atural object that is then met this model by optimization distributes.
Summary of the invention
In order to overcome above-mentioned the deficiencies in the prior art, the object of the present invention is to provide a kind of high spectrum image Super-resolution Reconstruction method based on sub-pixel mapping, can access the high resolving power high-spectrum remote sensing higher than original image resolution.
To achieve these goals, the technical solution used in the present invention is:
High spectrum image Super-resolution Reconstruction method based on sub-pixel mapping, comprises the following steps:
Step 1, sub-pixel mapping;
Step 1.1, adopts mixed pixel decomposition method to obtain the ratio value of the interior various atural objects of each pixel in original high-spectrum remote sensing, is called Abundances;
Step 1.2, is called thick pixel by each pixel in original high-spectrum remote sensing, and each thick pixel is divided into n ￡ n sub-pixel, and wherein n is predefined resolution enlargement factor;
Step 1.3, the Abundances according to every kind of known atural object in each pixel, is a certain atural object by each sub-pixel Random assignment, obtains initialized atural object distributed image;
Step 1.4, according to iteration optimization criterion, utilizes simulated annealing to carry out iteration optimization to initialized atural object distributed image, until meet stopping criterion for iteration, the overall similarity that the local continuity that described iteration optimization criterion distributes based on atural object and atural object distribute, wherein
The local continuity that atural object distributes, refers near that the atural object of classification distributes than other atural object of variety classes equally in subrange;
The overall similarity that atural object distributes, refers to that the atural object having in global scope in the thick pixel that similar proportion atural object forms distributes and has similarity, and arbitrary pixel can be by representing to its atural object similar pixel that distributes thus;
Following objective function is as described iteration optimization criterion:
Wherein, t is for belonging to the sub-pixel of atural object i in the pending pixel of index and neighborhood thereof, and c refers to atural object class number,
finger belongs to horizontal ordinate and the ordinate of the sub-pixel of atural object classification i, μ
ix, μ
iyrefer to respectively
mean value, η is the regularization factor, q
jbe to have the pixel of similarity with pixel p, it is selected according to being
the proper vector that f (p) is comprised of the Abundances of pixel p and neighborhood thereof, th is a default threshold value, if a certain pixel q
jproper vector belong to the individual vector similar to f (p) of front J (j≤J), so this pixel q
jthe similar pixel that is just considered to pixel p, therefore the sub-pixel distribution in pixel p can be by pixel q
jthe linear combination that interior sub-pixel distributes represents,
α
jer
jnormalization reciprocal.First of this objective function
represent the local continuity that atural object distributes, second
represent the overall similarity that atural object distributes;
Obtaining thus a width resolution is original image n atural object distributed image doubly;
Step 2, the Super-resolution Reconstruction method based on sub-pixel mapping;
It is m ￡ m that the atural object distributed image that step 1 is obtained is divided into a series of sizes; The subimage block of 1m < n, for each subimage block, recalculate the ratio that wherein comprises various atural objects, then with this ratio value, be multiplied by the pure end member object spectrum obtaining in mixed pixel decomposition method, and linear, additive obtains the curve of spectrum of this subimage block, thereby obtain the high-spectrum remote sensing doubly of n=m that resolution is original image.
Mixed pixel decomposition method in described step 1.1 refers to the ratio of the every kind of typical feature that obtains forming mixed pixel, and wherein mixed pixel refers to the pixel that includes dissimilar atural object in the high spectrum image of acquisition.
In described step 1.4, iteration optimization process is by continuous iteration, to change the distribution of atural object, and wherein each iterative process all can make objective function
value reduces, and described end condition is that target function value no longer declines or reached predetermined iterations.
Because the present invention will improve the resolution of high spectrum image, so can claim this to wait to carry high-resolution image, be low-resolution image.
Compared with prior art, advantage of the present invention is:
1) the present invention sets up in the local continuity that atural object is distributed the criterion that has adopted dispersion minimum in the class making between identical atural object on model;
2) the present invention has considered the overall similarity that atural object distributes when considering the local continuity that atural object distributes, the untrue property that the local characteristics of having avoided only distributing for atural object brings while solving;
3) the present invention proposes the super resolution ratio reconstruction method based on sub-pixel mapping, can obtain high-spectrum remote sensing from atural object distributed image.
Accompanying drawing explanation
Fig. 1 is the dividing mode schematic diagram of thick pixel in the present invention.
Fig. 2 is the result schematic diagram that obtains of sub-pixel mapping method.
Embodiment
Below in conjunction with embodiment, the present invention is described in further details.
High spectrum image Super-resolution Reconstruction method based on sub-pixel mapping, comprises the steps:
Step 1, sub-pixel mapping;
Step 1.1, adopts mixed pixel decomposition method to obtain the ratio value of the interior various atural objects of each pixel in original high-spectrum remote sensing, is called Abundances;
Mixed pixel decomposition method is a kind for the treatment of technology conventional in this area, provides a more conventional mixed pixel decomposition method herein.Secondly first the method chooses the pure atural object of suitable composition high spectrum image, is referred to as end member, from image or extract the spectrum of end member in wave spectrum storehouse, then adopts linear mixed model to be used as spectral mixing model, i.e. r=∑
ia
is
i+ w 0≤a
i≤ 1, ∑
ia
i=1, wherein, r is the spectrum of each pixel in high spectrum image, s
ifor the spectrum of each end member, a
ifor the ratio value of each end member in this pixel, i.e. Abundances, 0≤a
i≤ 1 refers to that the ratio value of every class atural object should be between 0 and 1, ∑
ia
i=1 refers to that the ratio value sum of all atural objects in this pixel should be error for 1, w.Thus, according to the principle that makes error w minimum, the Abundances of each atural object in each pixel can obtain by solving aforesaid equation.The present embodiment be take a simple analog image and is illustrated embodiments of the present invention as example, chooses two class atural objects, i.e. target " 0 " and background " 1 " in this example.Pending image size is 9 pixel * 9 pixels, is not that real high-spectrum remote sensing and mixed pixel decomposition method do not belong to content of the present invention, so set the Abundances of two class atural objects in each pixel herein as shown in Table 1 and Table 2 due to what adopt.
Table 1, the Abundances of atural object " 0 " in each pixel
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | |
| 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 2 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 3 | 0 | 0 | 0 | 0.0625 | 0.2500 | 0.0625 | 0 | 0 | 0 |
| 4 | 0 | 0 | 0.1250 | 0.9375 | 1 | 0.9375 | 0.1250 | 0 | 0 |
| 5 | 0 | 0 | 0.5000 | 1 | 1 | 1 | 0.5000 | 0 | 0 |
| 6 | 0 | 0 | 0.3750 | 1 | 1 | 1 | 0.3750 | 0 | 0 |
| 7 | 0 | 0 | 0 | 0.5000 | 0.7500 | 0.5000 | 0 | 0 | 0 |
| 8 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| 9 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
Table 2, the Abundances of atural object " 1 " in each pixel
| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | |
| 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 3 | 1 | 1 | 1 | 0.9375 | 0.7500 | 0.9375 | 1 | 1 | 1 |
| 4 | 1 | 1 | 0.8750 | 0.0625 | 0 | 0.0625 | 0.8750 | 1 | 1 |
| 5 | 1 | 1 | 0.5000 | 0 | 0 | 0 | 0.5000 | 1 | 1 |
| 6 | 1 | 1 | 0.6250 | 0 | 0 | 0 | 0.6250 | 1 | 1 |
| 7 | 1 | 1 | 1 | 0.5000 | 0.2500 | 0.5000 | 1 | 1 | 1 |
| 8 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
| 9 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Step 1.2, is called thick pixel by each pixel in original high-spectrum remote sensing, and each thick pixel is divided into n ￡ n sub-pixel, and wherein n is predefined resolution enlargement factor;
In the present invention, each pixel in pending high spectrum image is called to thick pixel, in table 1,9 * 9 pixels are all called thick pixel.In the present embodiment, setting resolution enlargement factor is 4, and n=4, is then equally divided into 4 * 4 sub-pixels by each thick pixel, as shown in Figure 1.
Step 1.3, the Abundances according to every kind of known atural object in each pixel, is a certain atural object by each sub-pixel Random assignment, obtains initialized atural object distributed image;
The pixel that the Abundances of the third line the 5th row target " 0 " is 0.25 in table 1 of take is example.After by each, thick pixel is divided into 4 * 4 sub-pixels, should there is 4 * 4 * 0.25=4 sub-pixel to belong to target " 0 ", in like manner, there are 12 sub-pixels to belong to background " 1 ".Then in these 4 * 4 sub-pixels, random 4 of pickings are assigned as target " 0 ", and remaining 12 sub-pixel is assigned as background " 1 ".
Step 1.4, according to iteration optimization criterion, utilizes simulated annealing to carry out iteration optimization to initialized atural object distributed image, until meet stopping criterion for iteration, the overall similarity that the local continuity that described iteration optimization criterion distributes based on atural object and atural object distribute, wherein
The local continuity that atural object distributes, refers near that the atural object of classification distributes than other atural object of variety classes equally in subrange;
The overall similarity that atural object distributes, refers to that the atural object having in global scope in the thick pixel that similar proportion atural object forms distributes and has similarity, and arbitrary pixel can be by representing to its atural object similar pixel that distributes thus;
Following objective function is as described iteration optimization criterion:
Wherein, t is for belonging to the sub-pixel of atural object i in the pending pixel of index and neighborhood thereof, and c refers to atural object class number,
finger belongs to horizontal ordinate and the ordinate of the sub-pixel of atural object classification i, μ
ix, μ
iyrefer to respectively
mean value, η is the regularization factor, q
jbe to have the pixel of similarity with pixel p, it is selected according to being
the proper vector that f (p) is comprised of the Abundances of pixel p and neighborhood thereof, th is a default threshold value, if a certain pixel q
jproper vector belong to the individual vector similar to f (p) of front J (j≤J), so this pixel q
jthe similar pixel that is just considered to pixel p, therefore the sub-pixel distribution in pixel p can be by pixel q
jthe linear combination that interior sub-pixel distributes represents,
α
jer
jnormalization reciprocal.First of this objective function
represent the local continuity that atural object distributes, second
represent the overall similarity that atural object distributes;
Simulated annealing is general optimized algorithm, for seeking the optimum solution of objective function.In the present embodiment, one by one thick pixel is optimized to calculating, first calculate the value of objective function under initial situation, then according to simulated annealing, change the atural object classification ownership of this thick pixel Nei Geya pixel, so constantly repeat this iterative step, until target function value no longer declines or reached predetermined iterations.
Obtain thus the atural object distributed image that a width resolution is 4 times of original images, as shown in Figure 2, wherein target " 0 " represents with black, and background 1 use white represents.
Step 2,
In the present embodiment, every 2 * 2 pixels in the high-resolution atural object distributed image that step 1 is obtained are looked as a whole, be m=2, thereby formed a series of subimage block, the ratio of the two class atural objects that can again obtain wherein comprising by simple calculating to each subimage block, then with this ratio value, being multiplied by the pure atural object obtaining in mixed pixel decomposition method is the spectrum of end member, and linear, additive obtains the curve of spectrum of this subimage block,
b wherein
ifor ratio value, s
ifor the end member spectrum obtaining in mixed pixel decomposition method.Because the present embodiment be take analog image as example, so do not provide the high-resolution remote sensing images that finally obtain.
Claims (3)
1. the high spectrum image Super-resolution Reconstruction method based on sub-pixel mapping, is characterized in that, comprises the following steps:
Step 1, sub-pixel mapping;
Step 1.1, adopts mixed pixel decomposition method to obtain the ratio value of the interior various atural objects of each pixel in original high-spectrum remote sensing, is called Abundances;
Step 1.2, is called thick pixel by each pixel in original high-spectrum remote sensing, and each thick pixel is divided into n * n sub-pixel, and wherein n is predefined resolution enlargement factor;
Step 1.3, the Abundances according to every kind of known atural object in each pixel, is a certain atural object by each sub-pixel Random assignment, obtains initialized atural object distributed image;
Step 1.4, according to iteration optimization criterion, utilizes simulated annealing to carry out iteration optimization to initialized atural object distributed image, until meet stopping criterion for iteration, the overall similarity that the local continuity that described iteration optimization criterion distributes based on atural object and atural object distribute, wherein
The local continuity that atural object distributes, refers near that the atural object of classification distributes than other atural object of variety classes equally in subrange;
The overall similarity that atural object distributes, refers to that the atural object having in global scope in the thick pixel that similar proportion atural object forms distributes and has similarity, and arbitrary pixel can be by representing to its atural object similar pixel that distributes thus;
Following objective function is as described iteration optimization criterion:
Wherein, t is for belonging to the sub-pixel of atural object i in the pending pixel of index and neighborhood thereof, and c refers to atural object class number,
finger belongs to horizontal ordinate and the ordinate of the sub-pixel of atural object classification i, μ
ix, μ
iyrefer to respectively
mean value, η is the regularization factor, q
jbe to have the pixel of similarity with pixel p, it is selected according to being
the proper vector that f (p) is comprised of the Abundances of pixel p and neighborhood thereof, th is a default threshold value, if a certain pixel q
jproper vector belong to front J the vector similar to f (p), wherein j≤J, so this pixel q
jthe similar pixel that is just considered to pixel p, therefore the sub-pixel distribution in pixel p can be by pixel q
jthe linear combination that interior sub-pixel distributes represents,
α
jer
jnormalization reciprocal, first of this objective function
represent the local continuity that atural object distributes, second
represent the overall similarity that atural object distributes;
Obtaining thus a width resolution is original image n atural object distributed image doubly;
Step 2, the Super-resolution Reconstruction method based on sub-pixel mapping;
It is m * m that the atural object distributed image that step 1 is obtained is divided into a series of sizes, the subimage block of 1≤m≤n, for each subimage block, recalculate the ratio that wherein comprises various atural objects, then with this ratio value, be multiplied by the pure end member object spectrum obtaining in mixed pixel decomposition method, and linear, additive obtains the curve of spectrum of this subimage block, thereby obtain the high-spectrum remote sensing doubly of n/m that resolution is original image.
2. the high spectrum image Super-resolution Reconstruction method of shining upon based on sub-pixel according to claim 1, it is characterized in that, mixed pixel decomposition method in described step 1.1 refers to the ratio of the every kind of typical feature that obtains forming mixed pixel, and wherein mixed pixel refers to the pixel that includes dissimilar atural object in the high spectrum image of acquisition.
3. the high spectrum image Super-resolution Reconstruction method of shining upon based on sub-pixel according to claim 1, it is characterized in that, in described step 1.4, iteration optimization process is by continuous iteration, to change the distribution of atural object, and wherein each iterative process all can make objective function
value reduces, and described end condition is that target function value no longer declines or reached predetermined iterations.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210051365.8A CN102682441B (en) | 2012-03-01 | 2012-03-01 | Hyperspectral image super-resolution reconstruction method based on subpixel mapping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210051365.8A CN102682441B (en) | 2012-03-01 | 2012-03-01 | Hyperspectral image super-resolution reconstruction method based on subpixel mapping |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102682441A CN102682441A (en) | 2012-09-19 |
| CN102682441B true CN102682441B (en) | 2014-11-26 |
Family
ID=46814308
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210051365.8A Expired - Fee Related CN102682441B (en) | 2012-03-01 | 2012-03-01 | Hyperspectral image super-resolution reconstruction method based on subpixel mapping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102682441B (en) |
Families Citing this family (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103559506B (en) * | 2013-11-19 | 2015-04-15 | 中国科学院地理科学与资源研究所 | Sub-pixel drawing method based on vector boundaries |
| CN104182953B (en) * | 2013-12-30 | 2017-02-01 | 北京大学 | Normalized multi-end-member decomposition hyperspectral reconstruction method in pixel unmixing inverse process |
| CN106157240B (en) * | 2015-04-22 | 2020-06-26 | 南京理工大学 | Remote sensing image super-resolution method based on dictionary learning |
| CN106548450A (en) * | 2016-09-30 | 2017-03-29 | 西安电子科技大学 | A kind of high spectrum image ultra-resolution method, device and computer-readable medium |
| CN106558036B (en) * | 2016-10-27 | 2019-08-02 | 中国科学院光电研究院 | A kind of spaceborne super-resolution imaging design method |
| CN107392208B (en) * | 2017-05-23 | 2020-05-22 | 三亚中科遥感研究所 | Object spectral feature extraction method based on spectral space mapping and purification |
| CN108985154B (en) * | 2018-06-06 | 2020-10-27 | 中国农业科学院农业资源与农业区划研究所 | Small-size ground object sub-pixel positioning method and system based on image concentration |
| CN108921035B (en) * | 2018-06-06 | 2020-08-11 | 中国农业科学院农业资源与农业区划研究所 | Sub-pixel positioning method and system based on spatial gravitation and pixel aggregation |
| CN110211042B (en) * | 2019-05-10 | 2021-03-12 | 北京航空航天大学 | Sub-pixel positioning method and device for enhancing spectral image spatial resolution |
| CN113624691B (en) * | 2020-05-07 | 2022-10-04 | 南京航空航天大学 | Spectral image super-resolution mapping method based on space-spectrum correlation |
| CN111899300B (en) * | 2020-07-30 | 2022-05-31 | 北京航空航天大学 | Abundance correction method and device for light field spectral data subpixel positioning |
| CN113034637A (en) * | 2021-03-11 | 2021-06-25 | 郑州轻工业大学 | Multi-scale rapid simulated annealing modeling method based on two-dimensional core structure |
| CN113409193B (en) * | 2021-06-18 | 2023-07-04 | 北京印刷学院 | Super-resolution reconstruction method and device for hyperspectral image |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1874499A (en) * | 2006-05-12 | 2006-12-06 | 北京理工大学 | High dynamic equipment for reconstructing image in high resolution |
| EP1906357A1 (en) * | 2006-09-28 | 2008-04-02 | Matsushita Electric Industrial Co., Ltd. | Combined optimization of motion estimator and super-resolution |
| CN102096920A (en) * | 2010-12-31 | 2011-06-15 | 清华大学 | Target image-based sub-pixel registering method |
| CN102136144A (en) * | 2011-04-11 | 2011-07-27 | 北京大学 | Image registration reliability model and reconstruction method of super-resolution image |
-
2012
- 2012-03-01 CN CN201210051365.8A patent/CN102682441B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1874499A (en) * | 2006-05-12 | 2006-12-06 | 北京理工大学 | High dynamic equipment for reconstructing image in high resolution |
| EP1906357A1 (en) * | 2006-09-28 | 2008-04-02 | Matsushita Electric Industrial Co., Ltd. | Combined optimization of motion estimator and super-resolution |
| CN102096920A (en) * | 2010-12-31 | 2011-06-15 | 清华大学 | Target image-based sub-pixel registering method |
| CN102136144A (en) * | 2011-04-11 | 2011-07-27 | 北京大学 | Image registration reliability model and reconstruction method of super-resolution image |
Non-Patent Citations (2)
| Title |
|---|
| Super resolution of remote sensing image based on structure similarity in CS frame;Zongxu Pan;《Multispectral Image Acquistion Processing and Analysis 2011》;20111208;1-8 * |
| Zongxu Pan.Super resolution of remote sensing image based on structure similarity in CS frame.《Multispectral Image Acquistion Processing and Analysis 2011》.2011,1-8. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102682441A (en) | 2012-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102682441B (en) | Hyperspectral image super-resolution reconstruction method based on subpixel mapping | |
| Long et al. | RXDNFuse: A aggregated residual dense network for infrared and visible image fusion | |
| CN103488968B (en) | The mixed pixel material of remote sensing images constitutes decomposer and the method for becoming more meticulous | |
| CN105261000B (en) | A kind of hyperspectral image fusion method mixed based on Endmember extraction and spectrum solution | |
| CN102324089B (en) | Maximum posteriori reconstruction method of PET (positron emission tomography) image based on generalized entropy and MR (magnetic resonance) prior | |
| CN105976310A (en) | Partition-based VCA end-member extraction method | |
| CN102968762A (en) | Polyethylene glycol terephthalate (PET) reconstruction method based on sparsification and Poisson model | |
| CN101794440A (en) | Weighted adaptive super-resolution reconstructing method for image sequence | |
| CN107301372A (en) | Hyperspectral image super-resolution method based on transfer learning | |
| CN103413292B (en) | Based on the hyperspectral image nonlinear abundance estimation method of constraint least square | |
| CN102446278B (en) | Multitemporal remote sensing image-based subpixel positioning method | |
| CN110765912B (en) | SAR image ship target detection method based on statistical constraint and Mask R-CNN | |
| CN111598786B (en) | Hyperspectral image unmixing method based on depth denoising self-coding network | |
| CN104851098A (en) | Objective evaluation method for quality of three-dimensional image based on improved structural similarity | |
| CN102789639A (en) | Method for fusing high-spectrum image and visible light image based on nonnegative matrix decomposition | |
| CN105447452A (en) | Remote sensing sub-pixel mapping method based on spatial distribution characteristics of features | |
| CN102789631A (en) | Remote sensing image resolution improving and processing method based on image segmentation and gravity model | |
| CN111583330B (en) | Multi-scale space-time Markov remote sensing image sub-pixel positioning method and system | |
| CN104200513A (en) | Matrix row-column sampling based multi-light-source rendering method | |
| CN106778530A (en) | A kind of hyperspectral image nonlinear solution mixing method based on bilinearity mixed model | |
| CN105678698A (en) | Non-local regularization remote sensing image super-resolution reconstruction method | |
| CN105760857B (en) | A kind of target in hyperspectral remotely sensed image object detection method | |
| CN104318046A (en) | System and method for incrementally converting high dimensional data into low dimensional data | |
| CN103679648B (en) | A kind of match by moment satellite image Strip noise removal method based on space segmentation | |
| CN102831621A (en) | Video significance processing method based on spectral analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20141126 Termination date: 20180301 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |