CN109447898A - A kind of compressed sensing based EO-1 hyperion super-resolution calculating imaging system - Google Patents
A kind of compressed sensing based EO-1 hyperion super-resolution calculating imaging system Download PDFInfo
- Publication number
- CN109447898A CN109447898A CN201811095176.4A CN201811095176A CN109447898A CN 109447898 A CN109447898 A CN 109447898A CN 201811095176 A CN201811095176 A CN 201811095176A CN 109447898 A CN109447898 A CN 109447898A
- Authority
- CN
- China
- Prior art keywords
- resolution
- image
- super
- spectrum image
- module
- 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
- 238000003384 imaging method Methods 0.000 title claims abstract description 16
- 238000001228 spectrum Methods 0.000 claims abstract description 70
- 230000006835 compression Effects 0.000 claims abstract description 9
- 238000007906 compression Methods 0.000 claims abstract description 9
- 239000004973 liquid crystal related substance Substances 0.000 claims abstract description 8
- 238000005457 optimization Methods 0.000 claims description 23
- 239000011159 matrix material Substances 0.000 claims description 13
- 238000000034 method Methods 0.000 claims description 12
- 238000000513 principal component analysis Methods 0.000 claims description 11
- 230000003595 spectral effect Effects 0.000 claims description 5
- 238000003064 k means clustering Methods 0.000 claims description 4
- 239000002131 composite material Substances 0.000 claims description 3
- 230000003287 optical effect Effects 0.000 abstract description 8
- 238000011084 recovery Methods 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 16
- 238000005070 sampling Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000012271 agricultural production Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000000701 chemical imaging Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- ONCZDRURRATYFI-QTCHDTBASA-N methyl (2z)-2-methoxyimino-2-[2-[[(e)-1-[3-(trifluoromethyl)phenyl]ethylideneamino]oxymethyl]phenyl]acetate Chemical compound CO\N=C(/C(=O)OC)C1=CC=CC=C1CO\N=C(/C)C1=CC=CC(C(F)(F)F)=C1 ONCZDRURRATYFI-QTCHDTBASA-N 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
- G06T3/4076—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution using the original low-resolution images to iteratively correct the high-resolution images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2135—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on approximation criteria, e.g. principal component analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2136—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on sparsity criteria, e.g. with an overcomplete basis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
-
- 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
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Probability & Statistics with Applications (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of compressed sensing based EO-1 hyperion super-resolution to calculate imaging system, comprising: liquid crystal tunable filter, space encoding module, planar array detector, compression reconfiguration module, super-resolution module；Original image successively after liquid crystal tunable filter, space encoding module, is detected by planar array detector, obtains space, spectrum ties up the high-spectral data compressed；Compression reconfiguration module is reconstructed the high-spectral data, obtains the high spectrum image of the low resolution of recovery for the restructing algorithm using compressed sensing；Super-resolution module, for, only from the high spectrum image of the low resolution, recovering high-resolution high spectrum image in the case where not needing to assist high-resolution RGB image using the non local self-similarity of high spectrum image.It can reduce data acquisition end pressure using the present invention, while the system does not need to increase the super-resolution reconstruct that additional optical path achieves that high spectrum image.
Description
Technical field
The invention belongs to spectral imaging technology field, in particular to a kind of compressed sensing based EO-1 hyperion super-resolution calculates
Imaging system.
Background technique
Hyperspectral imager can provide tens of to hundreds of continuous narrow-band information for each pixel of earth's surface object,
Spectral resolution has reached the order of magnitude of nanometer, contains type of ground objects information abundant.The appearance of Hyperspectral imager and
The ability that people observed by remote sensing technology and recognized things greatly improved in development, receives the common concern of world community,
Geologic survey, vegetation study, atmosphere observation, agricultural production, in terms of play an important role.
High-spectrum seems to be made of the 3 d image data of different spectral coverage under same scene, referred to as " spectrum cube
Body ", it comprises space dimension information and tens of to hundreds of spectrum to tie up information, therefore the data volume of high-spectral data is huge, gives
Data collection terminal brings very big pressure.With the appearance of compressive sensing theory, quick obtaining high-spectral data is broken through,
Imaging system completes the acquisition of the high-spectral data with high spectral resolution while significantly reducing data volume.Relative to
Traditional signal acquisition and treatment process, using compressive sensing theory, signal sampling rate depends no longer on the bandwidth of signal, and
It is the structure and content depending on information in signal, the sampling of such sensor and calculating cost will substantially reduce, and signal is extensive
Multiple process is the process of an optimal reconfiguration.
The basic principle of compressive sensing theory is to find some orthogonal basis Ψ for the signal x that length is N, make it at this
By transformation be sparse (i.e. in coefficient only have a small amount of nonzero element) under orthogonal basis, design another with orthogonal basis Ψ not phase
The observing matrix Φ of pass, projection x obtain observation signal y, original signal x, mathematical formulae are reconstructed by solving optimization problem are as follows:
But existing compressed sensing based EO-1 hyperion calculates imaging method and is difficult to have the high-spectral data obtained
Relatively high spatial resolution, so recovering high-resolution bloom from low resolution high spectrum image using signal processing technology
Spectrogram is as significant, value is far-reaching.And traditional compression EO-1 hyperion super-resolution imaging system usually requires to increase panchromatic observation
Optical path introduces auxiliary high-resolution RGB image information realization super-resolution reconstruct.Such system structure is complicated, and volume increases, cost
Go up.
Summary of the invention
In view of this, the present invention is for traditional Hyperspectral imager data storage, transmission, processing pressure is big and obtains
The problem that image spatial resolution is lower is taken, a kind of compressed sensing based EO-1 hyperion super-resolution calculating imaging system is proposed,
To reduce data acquisition end pressure, while the system does not need to increase the super-resolution weight that additional optical path achieves that high spectrum image
Structure.Furthermore the task of denoising, deblurring can also be completed by the selection of further matrix.
In order to solve the above-mentioned technical problem, the present invention is implemented as follows:
A kind of compressed sensing based EO-1 hyperion super-resolution calculating imaging system, comprising: liquid crystal tunable filter, space are compiled
Code module, planar array detector, compression reconfiguration module, super-resolution module；
Original image successively after liquid crystal tunable filter, space encoding module, is detected by planar array detector, is obtained empty
Between, spectrum tie up the high-spectral data that compresses；
Compression reconfiguration module is reconstructed the high-spectral data for the restructing algorithm using compressed sensing, obtains
The high spectrum image of the low resolution of recovery；
Super-resolution module, for using high spectrum image non local self-similarity, do not need to assist it is high-resolution
In the case where RGB image, only from the high spectrum image of the low resolution, high-resolution high spectrum image is recovered.
Preferably, super-resolution module includes:
Self-adapting dictionary learns submodule, for the high spectrum image of the low resolution to be divided into the image cube of overlapping
These image cubic blocks are divided into K cluster using K-means clustering method by block, learn each class cluster by principal component analysis PCA
Dictionary, K PCA dictionary may finally constitute a big super complete study dictionary；The study word formed by cluster
The non local self-similarity of the overall situation of image is utilized in allusion quotation, and ensure that local sparsity；
Optimization object function submodule is rebuild, for being based on the study dictionary, is introducing the non local sparse item of concentration
On the basis of, construct rebuilding spectrum optimization object function:
Wherein, | | αi,j-βi,j||1To utilize the described non local of the high spectrum image of adjacent non local self-similarity building
Concentrate sparse item；
The high spectrum image for remembering high-resolution and low resolution is respectively X ∈ RM×N×L、Y∈Rm×n×L, M, N represent high-resolution
Rate picture size, m, n represent low-resolution image size, and M > m, N > n, and L represents spectral band number；Matrix H indicate it is fuzzy and
The composite operator of down-sampling；Vector xi,jIndicate the space center position extracted from X in the image cubic block of (i, j)；
It is xi,jEstimated value, αi,j、βi,jIt is x respectivelyi,jWithRarefaction representation；α indicates all αi,jSet,It is the estimation of α
Value；It is the estimated value of X；As cubic block xi,jWhen belonging to k-th of class cluster, it is denoted as Ck；ΦkFor the dictionary of corresponding k-th of class cluster；λ
It is regularization parameter with η；
Submodule is resolved, for being directed to the optimization object function, using alternating minimization scheme, alternating more new variables
αi,j、βi,jAnd X, several times until algorithmic statement, the high-resolution high spectrum image being restored after iteration
Preferably, the space encoding module is using digital microlens array DMD.
Preferably, the planar array detector uses COMS planar array detector.
Preferably, when realizing high spectrum image denoising, matrix H chooses unit matrix；When realization high spectrum image removes mould
When paste, matrix H chooses fuzzy operator.
The utility model has the advantages that
(1) compressive sensing theory is utilized, the pressure for reducing data collection terminal is increased, reduces hardware requirement；
(2) super-resolution of high spectrum image may be implemented in imaging system, takes full advantage of the sparse characteristic of high spectrum image
And architectural characteristic, image can be restored more accurately；
(3) it is not required to the second optical path of additional, structure is simple, reduces volume, save the cost；
(4) identical super resolution algorithm Optimized model is utilized, by selecting different operators, can also realize high-spectrum
As denoising, deblurring.
Detailed description of the invention
Fig. 1 is that imaging system construction drawing is calculated the present invention is based on the EO-1 hyperion super-resolution of compressed sensing.
Specific embodiment
The present invention will now be described in detail with reference to the accompanying drawings and examples.
The present invention provides a kind of compressed sensing based EO-1 hyperion super-resolution to calculate imaging system, and basic thought is:
Building does not include the imaging system of panchromatic observation optical path, utilizes the sparse characteristic (non-local sparse constraint) and knot of high spectrum image
Structure characteristic (non local self-similarity), restores image more accurately.
Fig. 1 shows system of the invention and constitutes, as shown in Figure 1, the system includes realizing the hardware of compressed sensing imaging
Module 1, and improve the software module 2 of high spectrum image resolution ratio.
The hardware module 1 for realizing compressed sensing imaging includes: liquid crystal tunable filter (LCTF, Liquid Crystal
Tunable Filter), space encoding module and planar array detector.Wherein, space encoding module is using digital microlens array
(DMD, Digital Micromirror Device), planar array detector use COMS planar array detector.Certain hardware components are also
Need to be arranged necessary optical system, including preposition optical system, lens group, collimation lens.
Preposition optical system converges to the light of target scene on LCTF, and LCTF makes selected a series of in transmitted light
The light of central wavelength passes through, and completes the compressed encoding of spectrum dimension；DMD is reached by lens group later, to each central wavelength
Image is encoded, and the compressed encoding of space dimension is completed；Finally by collimation lens by modulated comprising two-dimensional space information
Light carry out shrink beam and collimation, obtain collimated light beam, be converted to modulation image by the reception of COMS planar array detector and deposited
Storage.What is obtained at this time is that space, spectrum tie up the high-spectral data that compresses.
The software module for improving high spectrum image resolution ratio includes: compression reconfiguration module, super-resolution module and output module.
Compression reconfiguration module utilizes the restructing algorithm of compressed sensing, and the high-spectral data of storage is reconstructed, and obtains extensive
The high spectrum image of multiple low resolution inputs super-resolution module.
Restructuring procedure is routine techniques, comprising: sparse basis selection, Optimization Solution, inverse transformation reconstruct three parts.It chooses first
High spectrum image is projected to the sparse basis in sparse domain, sparse transformation is carried out, obtained sparse coefficient, sparse basis is optimized
It solves, the algorithm of common Optimization Solution has GPSR, TWIST, SOMP etc., and the sparse coefficient after output optimization carries out sparse inverse
The high spectrum image being restored is converted, but spatial resolution at this time is lower.
Super-resolution module, for using high spectrum image non local self-similarity, do not need to assist it is high-resolution
In the case where RGB image, only from the high spectrum image of the low resolution, high-resolution high spectrum image is recovered.Wherein,
Non local self-similarity includes global non local self-similarity and adjacent non local self-similarity.
Super-resolution module includes following part:
Self-adapting dictionary learns submodule, first by system acquisition to the high spectrum image of low resolution be divided into overlapping
These image cubic blocks are divided into K cluster using K-means clustering method, pass through PCA (Principal by image cubic block
Component Analysis, principal component analysis) study each class cluster dictionary, K PCA dictionary finally constitute one greatly
Super complete study dictionary gives reconstruction optimization object function submodule.Here, image is utilized by the study dictionary that cluster is formed
The non local self-similarity of the overall situation, and ensure that local sparsity.
Optimization object function submodule is rebuild, for being based on the study dictionary, is introducing the non local sparse item of concentration
On the basis of, construct rebuilding spectrum optimization object function.The building mode of the optimization object function can hereafter be illustrated.It is described non-
The sparse item of concentration of local is constructed using adjacent non local self-similarity.
Submodule is resolved, for being directed to the optimization object function, using alternating minimization scheme by optimization object function
It is converted into simpler subproblem, alternately more new variables, several times until algorithmic statement, the high-resolution being restored after iteration
High spectrum image.High spectrum image denoising, deblurring are also able to achieve using the model simultaneously.
Output module, the high-resolution high spectrum image for going out super-resolution module recovery export.
Below for the building, optimization object function building and the mistake for rebuilding high-resolution high spectrum image of study dictionary
Journey is described in detail.
1, empty spectrum dictionary learning
The high-spectral data Y ∈ R for the low resolution that system acquisition is arrivedm×n×L(m, n representative image size, L represent spectrum
Wave band number) input super-resolution module.The size for being divided into overlapping by self-adapting dictionary study submodule is a × a × L image cube
Block, wherein a < m, n.Two spaces dimension is considered as an entirety, cubic block of the center at spatial position (i, j) is denoted asEach element representation in P is Pi,j[b, l], wherein b=1,2 ..., a2, l=1,2 ..., L.
These image cubic blocks are divided into K cluster using K-means clustering method, so collect image block mutually similar
A general dictionary is replaced at a local dictionary by each class fasciation.Then pass through PCA (Principal Component
Analysis, principal component analysis) dictionary of each class cluster of study describes all possible partial structurtes of high spectrum image, note the
The PCA dictionary of k class cluster is
2, the building of rebuilding spectrum optimization object function
Remember that high-resolution high-spectral data is X ∈ RM×N×L(M, N representative image size, and M > m, N > n；L represents spectrum
Wave band number), it is assumed that meet following linear relationship Y=HX between low resolution high spectrum image Y and high-resolution high-spectral data X
+ n, wherein H indicates that fuzzy and down-sampling composite operator, n indicate additive noise.
Enable vectorIt indicates Pi,jOne-dimensional vector (the x being stacked intoi,jAnd Pi,jCentered on spatial position (i,
J) cubic block, only expresses dimension difference, which is embodied as routine techniques), and have: xi,j=Ti,jX, whereinIt is to realize to extract block x by Xi,jMatrix.By extracting space block in two-dimensional spatial location movement to obtain G
A space block, wherein G=(M-a+1) (N-a+1).
If cubic block xi,jBelong to k-th of class cluster CkWhen, it can use learnt dictionary ΦkCan by its sparse coding at
Φkαi,j.Since the block of cubic block and periphery in high spectrum image has self-similarity abundant, in this way, cubic block can be effective
Ground describes its periphery block.It can use this neighbouring non local self-similarity so to complete the building of optimization object function.
Enable xi,jEstimated value beIt isRarefaction representation.Due toIt is similar to xi,j,'s
Sparse coding βi,jX should be similar toi,jSparse coding αi,j, and the difference of the two should be made small as far as possible.Based on natural image packet
Containing a large amount of non local redundancy, we may search for non local similar cubic block.For xi,j, similar with its piece of set
It is denoted as Ωi,j, and set omegai,jOther interior blocks are denoted as xi,j,c,d.Then βi,jIt can be obtained by weighted calculation.
Wherein, wi,j,c,dIt is corresponding weight.It can be arranged toW is to return
One changes the factor, and q is a predetermined scalar.
Therefore, non local concentration sparse constraint is introduced | | αi,j-βi,j||1, high-resolution high spectrum image X can be by
Optimization object function shown in following formula restores.
Wherein, α indicates all αi,jCascade,It is the estimated value of α.
The optimization object function utilizes neighbouring non local self-similarity to introduce non-local sparse bound term, and used
Self-adapting dictionary learning method considers the non local self-similarity of the overall situation of image, and ensure that local sparsity, that is to say, that
||αi,j||1Be it is sufficiently small, can not consider；
The more commonly used super-resolution optimization object function based on sparse constraint is as follows:
Wherein, S is the high resolution R GB image for needing to be additionally provided, D1、D2It is the corresponding dictionary of Y, S.
The algorithm proposed has sufficiently excavated high spectrum image information, can not have to be additionally provided high-definition picture, directly
It connects and super-resolution is realized by low resolution high spectrum image.
3, high-resolution high spectrum image is rebuild
Simpler subproblem is converted for optimization object function using alternating minimization scheme, i.e., is alternately updated non local
(β indicates all β to self-similarity βi,jSet), target high spectrum image X and sparse coding α.βi,jInitial valueIt is set as
0, X initial value X0It is set as the bicubic interpolation filter of Y, sparse coding α initial value α0It can be byIt obtains.
In an iterative process, the accuracy of sparse coding α is improved, and so also improves non local self-similarity β's
Accuracy.Dictionary ΦkEqually also updated according to the sparse coding α updated.Several times until algorithmic statement, obtains after iteration
Last X, that is, required high-resolution high spectrum image.
If matrix H is unit matrix, high spectrum image denoising may be implemented；If matrix H is fuzzy operator, can
To realize high spectrum image deblurring.
In conclusion the above is merely preferred embodiments of the present invention, being not intended to limit the scope of the present invention.
All within the spirits and principles of the present invention, any modification, equivalent replacement, improvement and so on should be included in of the invention
Within protection scope.
Claims (5)
1. a kind of compressed sensing based EO-1 hyperion super-resolution calculates imaging system characterized by comprising liquid crystal tunable filters
Device, space encoding module, planar array detector, compression reconfiguration module, super-resolution module；
Original image successively after liquid crystal tunable filter, space encoding module, is detected by planar array detector, obtains space, light
The high-spectral data that spectrum dimension is compressed；
Compression reconfiguration module is reconstructed the high-spectral data, is restored for the restructing algorithm using compressed sensing
Low resolution high spectrum image；
Super-resolution module for the non local self-similarity using high spectrum image is not needing that high-resolution RGB is assisted to scheme
As in the case where, only from the high spectrum image of the low resolution, high-resolution high spectrum image is recovered.
2. the system as claimed in claim 1, which is characterized in that super-resolution module includes:
Self-adapting dictionary learns submodule, for the high spectrum image of the low resolution to be divided into the image cubic block of overlapping,
These image cubic blocks are divided into K cluster using K-means clustering method, each class cluster is learnt by principal component analysis PCA
Dictionary, K PCA dictionary may finally constitute a big super complete study dictionary；The study dictionary formed by cluster
The non local self-similarity of the overall situation of image is utilized, and ensure that local sparsity；
Optimization object function submodule is rebuild, for being based on the study dictionary, is introducing the non local basis for concentrating sparse item
On, construct rebuilding spectrum optimization object function:
Wherein, | | αi,j-βi,j||1To utilize the non local concentration of the high spectrum image of adjacent non local self-similarity building
Sparse item；
The high spectrum image for remembering high-resolution and low resolution is respectively X ∈ RM×N×L、Y∈Rm×n×L, M, N represent high resolution graphics
As size, m, n represent low-resolution image size, and M > m, N > n, and L represents spectral band number；Matrix H indicate it is fuzzy and under adopt
The composite operator of sample；Vector xi,jIndicate the space center position extracted from X in the image cubic block of (i, j)；It is xi,j
Estimated value, αi,j、βi,jIt is x respectivelyi,jWithRarefaction representation；α indicates all αi,jSet,It is the estimated value of α；
It is the estimated value of X；As cubic block xi,jWhen belonging to k-th of class cluster, it is denoted as Ck；ΦkFor the dictionary of corresponding k-th of class cluster；λ and η are
Regularization parameter；
Submodule is resolved, for being directed to the optimization object function, using alternating minimization scheme, alternately updates variable αi,j、
βi,jAnd X, several times until algorithmic statement, the high-resolution high spectrum image being restored after iteration
3. the system as claimed in claim 1, which is characterized in that the space encoding module is using digital microlens array DMD.
4. the system as claimed in claim 1, which is characterized in that the planar array detector uses COMS planar array detector.
5. the system as claimed in claim 1, which is characterized in that when realizing high spectrum image denoising, matrix H chooses unit square
Battle array；When realizing high spectrum image deblurring, matrix H chooses fuzzy operator.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811095176.4A CN109447898B (en) | 2018-09-19 | 2018-09-19 | Hyperspectral super-resolution calculation imaging system based on compressed sensing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811095176.4A CN109447898B (en) | 2018-09-19 | 2018-09-19 | Hyperspectral super-resolution calculation imaging system based on compressed sensing |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109447898A true CN109447898A (en) | 2019-03-08 |
| CN109447898B CN109447898B (en) | 2023-02-17 |
Family
ID=65530594
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811095176.4A Active CN109447898B (en) | 2018-09-19 | 2018-09-19 | Hyperspectral super-resolution calculation imaging system based on compressed sensing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109447898B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110081977A (en) * | 2019-05-22 | 2019-08-02 | 北京理工大学 | A kind of compressed sensing based tunable optical filter type hyperspectral imager and method |
| CN111157114A (en) * | 2019-12-26 | 2020-05-15 | 西安电子科技大学 | Long-wave infrared multispectral imaging method and device based on wavelength conversion |
| CN112785662A (en) * | 2021-01-28 | 2021-05-11 | 北京理工大学重庆创新中心 | Self-adaptive coding method based on low-resolution priori information |
| CN112927149A (en) * | 2021-02-18 | 2021-06-08 | 北京印刷学院 | Spatial resolution enhancement method and device for hyperspectral image and electronic equipment |
| CN113139903A (en) * | 2021-04-27 | 2021-07-20 | 西安交通大学 | Method for improving infrared spectrum resolution based on compressed sensing theory |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105469360A (en) * | 2015-12-25 | 2016-04-06 | 西北工业大学 | Non local joint sparse representation based hyperspectral image super-resolution reconstruction method |
| US20160241797A1 (en) * | 2015-02-17 | 2016-08-18 | Canon Kabushiki Kaisha | Devices, systems, and methods for single-shot high-resolution multispectral image acquisition |
-
2018
- 2018-09-19 CN CN201811095176.4A patent/CN109447898B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160241797A1 (en) * | 2015-02-17 | 2016-08-18 | Canon Kabushiki Kaisha | Devices, systems, and methods for single-shot high-resolution multispectral image acquisition |
| CN105469360A (en) * | 2015-12-25 | 2016-04-06 | 西北工业大学 | Non local joint sparse representation based hyperspectral image super-resolution reconstruction method |
Non-Patent Citations (2)
| Title |
|---|
| XI WANG等，: "Compressive spectral imaging system based on liquid crystal tunable filter", 《OPTICS EXPRESS》 * |
| 潘宗序等: "基于光谱相似性的高光谱图像超分辨率算法", 《自动化学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110081977A (en) * | 2019-05-22 | 2019-08-02 | 北京理工大学 | A kind of compressed sensing based tunable optical filter type hyperspectral imager and method |
| CN110081977B (en) * | 2019-05-22 | 2020-06-19 | 北京理工大学 | Adjustable filter type hyperspectral imager based on compressed sensing and method |
| CN111157114A (en) * | 2019-12-26 | 2020-05-15 | 西安电子科技大学 | Long-wave infrared multispectral imaging method and device based on wavelength conversion |
| CN111157114B (en) * | 2019-12-26 | 2022-01-28 | 西安电子科技大学 | Long-wave infrared multispectral imaging method and device based on wavelength conversion |
| CN112785662A (en) * | 2021-01-28 | 2021-05-11 | 北京理工大学重庆创新中心 | Self-adaptive coding method based on low-resolution priori information |
| CN112927149A (en) * | 2021-02-18 | 2021-06-08 | 北京印刷学院 | Spatial resolution enhancement method and device for hyperspectral image and electronic equipment |
| CN112927149B (en) * | 2021-02-18 | 2023-07-04 | 北京印刷学院 | Method and device for enhancing spatial resolution of hyperspectral image and electronic equipment |
| CN113139903A (en) * | 2021-04-27 | 2021-07-20 | 西安交通大学 | Method for improving infrared spectrum resolution based on compressed sensing theory |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109447898B (en) | 2023-02-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109447898A (en) | A kind of compressed sensing based EO-1 hyperion super-resolution calculating imaging system | |
| CN110119780B (en) | Hyper-spectral image super-resolution reconstruction method based on generation countermeasure network | |
| Meng et al. | Review of the pansharpening methods for remote sensing images based on the idea of meta-analysis: Practical discussion and challenges | |
| Zhou et al. | Pyramid fully convolutional network for hyperspectral and multispectral image fusion | |
| CN111932457B (en) | High space-time fusion processing algorithm and device for remote sensing image | |
| CN109727207B (en) | Hyperspectral image sharpening method based on spectrum prediction residual convolution neural network | |
| CN114119444B (en) | Multi-source remote sensing image fusion method based on deep neural network | |
| CN106920214B (en) | Super-resolution reconstruction method for space target image | |
| CN111260576A (en) | Hyperspectral unmixing algorithm based on de-noising three-dimensional convolution self-coding network | |
| Sdraka et al. | Deep learning for downscaling remote sensing images: Fusion and super-resolution | |
| CN110544212B (en) | Convolutional neural network hyperspectral image sharpening method based on hierarchical feature fusion | |
| CN106204450B (en) | Based on the sparse multiple dimensioned geometry remote sensing image fusion method from coding of depth | |
| CN109003239B (en) | Multispectral image sharpening method based on transfer learning neural network | |
| CN112488978A (en) | Multi-spectral image fusion imaging method and system based on fuzzy kernel estimation | |
| CN114723631B (en) | Image denoising method, system and device based on depth context prior and multi-scale reconstruction sub-network | |
| CN104794681B (en) | Remote sensing image fusion method based on more redundant dictionaries and sparse reconstruct | |
| CN106097253A (en) | A kind of based on block rotation and the single image super resolution ratio reconstruction method of definition | |
| CN114841856A (en) | Image super-pixel reconstruction method of dense connection network based on depth residual channel space attention | |
| CN109146787A (en) | A kind of real-time reconstruction method of the double camera spectrum imaging system based on interpolation | |
| Long et al. | Dual self-attention Swin transformer for hyperspectral image super-resolution | |
| CN118212127A (en) | Misregistration-based physical instruction generation type hyperspectral super-resolution countermeasure method | |
| Shen et al. | Deep-learning-based super-resolution of video satellite imagery by the coupling of multiframe and single-frame models | |
| CN117350923A (en) | Panchromatic and multispectral remote sensing image fusion method based on GAN and transducer | |
| CN113112591B (en) | Multi-temporal remote sensing image space-time spectrum fusion method based on coupling sparse tensor decomposition | |
| CN118212129A (en) | Hyper-spectral remote sensing image multi-source fusion super-division method based on bilinear unmixing |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |