CN103440675A - Overall situation reconstitution optimization model construction method for image block compressed sensing - Google Patents
Overall situation reconstitution optimization model construction method for image block compressed sensing Download PDFInfo
- Publication number
- CN103440675A CN103440675A CN2013103245082A CN201310324508A CN103440675A CN 103440675 A CN103440675 A CN 103440675A CN 2013103245082 A CN2013103245082 A CN 2013103245082A CN 201310324508 A CN201310324508 A CN 201310324508A CN 103440675 A CN103440675 A CN 103440675A
- Authority
- CN
- China
- Prior art keywords
- image
- reconstitution
- block
- compressed sensing
- reconstruction
- 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
Images
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses an overall situation reconstitution optimization model construction method for image block compressed sensing. The procedures of a collecting end include that firstly, an image x is divided into n B*B small blocks xi, wherein x and xi are pulled to be column vectors in a raster scanning manner; secondly, an independent identically distributed gauss random matrix phi B with the size of MB*B2 is generated; thirdly, incoherent measuring is performed on each block xi to obtain observed value vectors yi which is equal to phi B xi; fourthly, the observed value vectors yi and a seed for generating the gauss random matrix are sent to a reconstruction end. The procedures of the reconstruction end include that firstly, the received observed value vectors yi of all the blocks are accumulated to be y=[yi; y2;..., yn] in columns; secondly, an overall situation reconstitution measurement operator theta ( ) is constructed, wherein the input of the overall situation reconstitution measurement operator is an image x, the corresponding output of the overall situation reconstitution measurement operator is y, and the overall situation reconstitution measurement operator is composed of a block measurement matrix set phi and a ranking operator P ( ); thirdly, an overall optimization reconstitution model is set up, and the image is recovered with a corresponding compressed sensing reconstitution algorithm. The overall situation reconstitution optimization model construction method for image block compressed sensing can effectively eliminate the block effect in the prior art, and strong robustness on variation of the block size B is achieved.
Description
Technical field
The invention belongs to technical field of image processing, particularly image block compressed sensing reconstruction and optimization model construction method, can be used for natural image is reconstructed.
Background technology
In traditional imaging system, natural image first is converted into digital picture with higher sampling rate, and recycling JPEG or JPEG2000 carry out compressed encoding to it, to reduce the memory space of digital picture.But, because the computation complexity of this type of compaction coding method is high, thus they and be not suitable for the imaging device (for example, in Internet of Things, gathering the sensor of image) of low energy consumption, low resolution, low computation complexity.In recent years, compressed sensing (Compressed Sensing, CS) is theoretical proposes: direct compressed signal in sampling.This theory is based upon Candes et al. and Donoho has on breakthrough work, point out meeting under certain condition, and lower than Nyquist (Nyquist) speed sampled signal, still can undistorted release signal.Compressive sensing theory provides possibility for the existence of the digital collection equipment of low sampling rate, low energy consumption, low computation complexity.The also potential huge more practical value just because of it, at a few years in the time, in sphere of learning or all take to this theoretical research work at industrial circle.
In the image field, the CS coding strategy based on different exists two class Image Reconstruction models: the Image Reconstruction model based on Global CS and the Image Reconstruction model based on Block CS.Image Reconstruction model based on Global CS takes to reorganize the strategy of whole solution: at coding side, measure entire image; In decoding end reconstruct entire image.Due to the scale of image excessive (usually can reach mega pixel), this class Image Reconstruction model makes that the required memory space of random measurement matrix is excessive, Measuring Time is long, is difficult in actual applications realize.Candes et al. is at document " Robust signal recovery from incomplete observations ", in Proceedings of the International Conference on Image Processing, Atlanta, CA, 2006, a kind of structuring stochastic matrix SFE (Scrambled Fourier Ensemble) has been proposed in pp.1281-1284, the required memory space of this matrix is little, meanwhile also can realize Quick Measurement, but it is randomly ordered that the process of structural matrix relates to, fast orthogonal transforms, the operations such as random down-sampling, make the scrambler more complicated, be not easy to realize.Lu Gan is at document " Block compressed sensing of natural images ", in Proceedings of the International Conference on Digital Signal Processing, Cardiff, UK, July 2007, the strategy that the Image Reconstruction model based on Block CS proposed in pp.403-406 takes lacing to decompose: at coding side, the macro block that image is divided into to same size, adopt identical random measurement matrix respectively these macro blocks to be measured; In decoding end, first reconstruct each macro block, then these macro blocks are merged into to entire image.The random measurement matrix stores amount of this class Image Reconstruction model is little, simple structure, and block-based measuring method is well suited for the realization of real-time system.But the Image Reconstruction model based on Block CS is Shortcomings still: the mode of independent each piece of reconstruct has been destroyed characteristics of image, cause blocking effect to occur.
In view of this, be necessary to provide a kind of overall reconstruction and optimization model construction method of image block compressed sensing, to address the above problem.
Summary of the invention
The objective of the invention is: the blocking effect problem existed in order to overcome Block CS reconstruction model, the building method of the overall reconstruction and optimization model of the image block compressed sensing that proposition can a reconstruct entire image.
The technical solution adopted in the present invention is: a kind of overall reconstruction and optimization model construction method of image block compressed sensing, it is characterized in that, and the method comprises the steps:
(1) piece image
x be divided into
nindividual
b*
bfritter
x i, wherein
x with
x iall by grating scanning mode, draw as column vector;
(2) generate and be of a size of
m b*
b 2independent same distribution gaussian random matrix
Φ b;
(3) to every
x imake the incoherent observation vector of measuring to obtain
y i=
Φ b.
x i;
(4) by observation vector
y iwith the seed that generates the gaussian random matrix
seedbe sent to the reconstruct end;
(5) each piece observation vector that will receive
y ipile up by row and be
y =[
y 1;
y 2; ...;
y n];
(6) utilize seed
seedre-construct out the gaussian random matrix
Φ b, and generate piece measurement set of matrices
Φ= diag (
Φ b,
Φ b...,
Φ b), then itself and sequence operator P () combination being obtained to overall reconfigurable measurement operator Θ (), it is input as image
x , correspondence is output as
y ;
(7) set up the global optimization reconstruction model, and adopt corresponding compressed sensing restructing algorithm restored image.
The overall reconstruction and optimization model construction method of image block compressed sensing as above, is characterized in that, in the step of described image overall reconstruction model (6), sequence operator P () is by input picture
x sequence is all image blocks
x i(
i=1 ...,
n) the row stacked arrangement
x p=[
x 1;
x 2; ...,
x n]; By the gaussian random matrix
Φ barrange by the diagonal line order piece formed and measure set of matrices
Φ , with
x pmultiply each other and can obtain
y , i.e. original image
x with the piece observation vector, pile up
y between pass be
y =
Φ p (
x )=Θ (
x ).
The overall reconstruction and optimization model construction method of image block compressed sensing as above, is characterized in that described original image
x with the piece observation vector, pile up
y between relation known, the global optimization reconstruction model can reconstruct original image by a view picture, and not independently each piece of reconstruct remerge as entire image, overall reconstruction model is:
Wherein λ is the regularization factor, and Pior () means for the functional of priori.
The present invention compared with prior art has following advantage: the present invention can overcome the following defect of existing Block CS reconstruction model method:
1. block sparsity is inhomogeneous.Picture breakdown is measure-alike macro block, and each macro block comprises different characteristics of image, for example, can be divided into flat block, boundary block, texture block.In decoding end, these macro blocks are done to identical conversion, but its degree of rarefication at transform domain is not identical, like this when piecemeal reconstruct, the piece that degree of rarefication is little, reconstruction quality is good, the piece that degree of rarefication is large, reconstruction quality is poor.Therefore, the recovery level of piece and piece differs, and makes the image reconstructed have a large amount of blocking effects.Yet, for overall reconstruct, in decoding end, be that entire image is converted, there do not is the inhomogeneous problem of this degree of rarefication, so there is not blocking effect in overall reconstruct.
2. spectrum leakage phenomenon.At reconstruct end piecemeal, do after conversion is equivalent to the image windowing to remake conversion, the windowing meeting is leaked frequency spectrum, increases the degree of rarefication of transform domain, thereby affects reconstruction quality.Yet, for overall reconstruct, do not have the windowing process, thereby its picture quality reconstructed is better than the picture quality that piecemeal reconstructs.
3. block size
bcan not be too small, can not be excessive.
btoo small, obtaining less measurement matrix stores amount and faster in measuring speed, the reconstruction quality of image will descend;
bexcessive, when obtaining better Image Reconstruction quality, the quantitative change of measurement matrix stores is large, and measuring speed is slack-off.Therefore, block size
bsize need the balance.Yet, for overall reconstruct, do not need to consider this balance.
The present invention, owing at the reconstruct end, carrying out overall reconstruct, takes full advantage of the overall degree of rarefication of image, has effectively overcome above-mentioned three problems that exist in image block reconstruct, therefore can obtain better Image Reconstruction quality.
The accompanying drawing explanation
Fig. 1 is model framework chart of the present invention;
Fig. 2 is the reconstruction result comparison diagram of Block CS reconstruction model and direct overall reconstruction model;
Fig. 3 is the construction process that 4 * 4 images (being divided into 42 * 2) are measured operator;
Fig. 4 is the present invention and the Block CS reconstruction model reconstruction result comparison diagram to the Lena image;
Fig. 5 is to be 0.3 o'clock at measured rate, the Lena image reconstructed
pSNRvalue is with the piece size
bthe comparison diagram changed.
Embodiment
In order to understand better the present invention, further illustrate content of the present invention below in conjunction with embodiment, but content of the present invention not only is confined to the following examples.Those skilled in the art can make various changes or modifications the present invention, and these equivalent form of values are equally within the listed claims limited range of the application.
With reference to Fig. 1, the specific embodiment of the invention process is as follows:
Step 5, at reconstruct end overall situation reconstructed image, must construct the measurement operator Θ () of entire image.Can consider, at the reconstruct end according to following formula:
The piece that constructs entire image is measured set of matrices
Φ , recycle following Image Reconstruction model:
Release signal.Yet what carry out due to collection terminal is that piecemeal is measured, what utilize the actual measurement of this measurement matrix is the signal of arranging according to the piece measuring sequence, not original image signal itself.The data structure of sort signal is upset the locus of original image pixels, and it is done to global change, the inevitable degree of rarefication lower than original image of the degree of rarefication of transform domain.Fig. 2 (a) is 256 * 256 Peppers figure that use Block CS model reconstruction to go out, and Fig. 2 (b) is direct use
Φ reconstruct Peppers figure out can obviously find out that the picture quality that the picture quality directly reconstructed will reconstruct far below piecemeal, so piece is measured set of matrices
Φ can not be directly used in overall reconstruct.According to above analysis, the known matrix of the measurement for overall reconstruct must possess two characteristics: the first, and measured signal must be original image signal; The second, must be adapted to the piecemeal measurement data.In order to illustrate in decoding end how to construct the measurement matrix that meets these two characteristics, Fig. 3 has listed the construction process of 4 * 4 images (being divided into 42 * 2) measurement matrix.Yet, because the dimension of picture signal is millions of up to ten million, can't construct corresponding measurement matrix
Θ , but can use the operator Θ () that can complete identical function to realize.
If use the priori of image in the sparse property of wavelet transformed domain, can use this model of GPSR Algorithm for Solving; If use the priori of image gradient territory flatness, can use this model of min-TV Algorithm for Solving.
Advantage of the present invention is further illustrated by data and the image of following emulation.
1. simulated conditions
The present invention's (whole solution of lacing): adopt piecemeal CS to measure at collection terminal, at the reconstruct end, utilize respectively GPSR algorithm, min-TV algorithm to be reconstructed entire image.
Reference model 1(Block CS, lacing is decomposed): adopt the Image Reconstruction model based on Block CS, at collection terminal, adopt piecemeal CS to measure, at the reconstruct end, utilize respectively GPSR algorithm, min-TV algorithm piecemeal to reconstruct each macro block, then these macro blocks are merged into to entire image.
Reference model 2(Global CS, reorganize whole solution): adopt the Image Reconstruction model based on Global CS, use structuring stochastic matrix SFE (Scrambled Fourier Ensemble), at collection terminal, entire image is carried out to the CS measurement, at the reconstruct end, utilize respectively GPSR algorithm, min-TV algorithm to reconstruct entire image.
For GPSR algorithm, transformation matrix
Ψ formed block size by the Daubechies-4 Orthogonal Wavelets
b=32, adopt the high frequency details gray level image Lena of 3 256 * 256 from low to high, Peppers, Mandrill tests respectively above-mentioned model
pSNRvalue.
2. emulation content and result
As can be seen from Table 1, the image PSNR value that the model reconstruction that the present invention proposes goes out is not apparently higher than the model based on Block CS, what reason was the utilization of min-TV algorithm is the sparse property of image gradient, and for gradient, the gradient degree of rarefication of each piece inhomogeneous although (flat block gradient degree of rarefication is little, and texture block is large), the gradient degree of rarefication that piecemeal can't have influence on image block rises, so utilize overall degree of rarefication reconstructed image, can significantly not promote the PSNR value.As can be seen from Table 2, the PSNR value of model reconstruction image is proposed than based on Block CS model, having improved 1 ~ 2dB.This be because the utilization of GPSR algorithm be that image is at transform domain
Ψ in sparse property, and piecemeal is equivalent to windowing, when piecemeal reconstruct, spectrum leakage occurs, the degree of rarefication of image macro raises, and has caused reconstruction quality decline.Associative list 1 and table 2 can find out, aspect the reconstructed image quality, the model that this paper proposes is better than the model based on Block CS, but is inferior to the model based on Global CS.Obviously, the model based on Global CS, no matter at collection terminal, has still all utilized the sparse property of entire image at the reconstruct end, and its Image Reconstruction quality must be better than other two models.
Each model of table 1 adopts the PSNR value (unit: dB) of min-TV algorithm reconstructed image
Each model of table 2 adopts the PSNR value (unit: dB) of GPSR algorithm reconstructed image
Fig. 4 has shown that employing GPSR algorithm reconstructs the subjective visual quality do of 256 * 256 Lena images with measured rate 0.3, and Fig. 4 (a) is for adopting the restructuring graph of Block CS model, and Fig. 4 (b) is for adopting the restructuring graph of model proposed by the invention.Can find out that proposed model eliminated the blocking effect existed based in Block CS model, this be because what propose the model utilization is the overall degree of rarefication reconstructed image of image, can not exist based on the existing block sparsity problem of non-uniform of Block CS model.Fig. 5 has shown in the different masses size
bunder (getting successively 32,16,8), model more proposed by the invention and Block CS model reconstruct image
pSNRvalue, can find out and no matter adopt the min-TV restructing algorithm (Fig. 5 (a)), still adopts GPSR restructing algorithm (Fig. 5 (b)), along with the piece size
bminimizing, the quality degradation amplitude of the model reconstruction image that proposes is little, and the quality degradation of Block CS reconstructed image is larger.The model that this explanation this paper proposes has higher robustness to the variation of piece size.
In sum, the image gone out with model reconstruction proposed by the invention has been eliminated a large amount of blocking effects, and especially, in low measured rate situation, the reconstructed image quality is apparently higher than the Image Reconstruction model based on Block CS.In addition, diminishing of piece size can not produce significantly impact to the restructuring graph image quality, in this, is better than the Image Reconstruction model based on Block CS yet.
The content be not described in detail in this instructions belongs to the known prior art of professional and technical personnel in the field.
Claims (3)
1. the overall reconstruction and optimization model construction method of an image block compressed sensing, is characterized in that, the method comprises the steps:
(1) piece image
x be divided into
nindividual
b*
bfritter
x i, wherein
x with
x iall by grating scanning mode, draw as column vector;
(2) generate and be of a size of
m b*
b 2independent same distribution gaussian random matrix
Φ b;
(3) to every
x imake the incoherent observation vector of measuring to obtain
y i=
Φ b.
x i;
(4) by observation vector
y iwith the seed that generates the gaussian random matrix
seedbe sent to the reconstruct end;
(5) each piece observation vector that will receive
y ipile up by row and be
y =[
y 1;
y 2; ...;
y n];
(6) utilize seed
seedre-construct out the gaussian random matrix
Φ b, and generate piece measurement set of matrices
Φ= diag (
Φ b,
Φ b...,
Φ b), then itself and sequence operator P () combination being obtained to overall reconfigurable measurement operator Θ (), it is input as image
x , correspondence is output as
y ;
(7) set up the global optimization reconstruction model, and adopt corresponding compressed sensing restructing algorithm restored image.
2. the overall reconstruction and optimization model construction method of image block compressed sensing according to claim 1, is characterized in that, in the step of described image overall reconstruction model (6), sequence operator P () is by input picture
x sequence is all image blocks
x i(
i=1 ...,
n) the row stacked arrangement
x p=[
x 1;
x 2; ...,
x n]; By the gaussian random matrix
Φ barrange by the diagonal line order piece formed and measure set of matrices
Φ , with
x pmultiply each other and can obtain
y , i.e. original image
x with the piece observation vector, pile up
y between pass be
y =
Φ p (
x )=Θ (
x ).
3. the overall reconstruction and optimization model construction method of image block compressed sensing according to claim 2, is characterized in that described original image
x with the piece observation vector, pile up
y between relation known, the global optimization reconstruction model can reconstruct original image by a view picture, and not independently each piece of reconstruct remerge as entire image, overall reconstruction model is:
Wherein λ is the regularization factor, and Pior () means for the functional of priori.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013103245082A CN103440675A (en) | 2013-07-30 | 2013-07-30 | Overall situation reconstitution optimization model construction method for image block compressed sensing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013103245082A CN103440675A (en) | 2013-07-30 | 2013-07-30 | Overall situation reconstitution optimization model construction method for image block compressed sensing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103440675A true CN103440675A (en) | 2013-12-11 |
Family
ID=49694368
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013103245082A Pending CN103440675A (en) | 2013-07-30 | 2013-07-30 | Overall situation reconstitution optimization model construction method for image block compressed sensing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103440675A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104778665A (en) * | 2015-04-14 | 2015-07-15 | 清华大学 | Compressed ghost imaging reconstruction method based on natural image block prior driving and system |
| CN104952039A (en) * | 2015-06-14 | 2015-09-30 | 中国电子科技集团公司第十研究所 | Distributed compressive sensing reconstruction method for images |
| CN107154061A (en) * | 2017-05-09 | 2017-09-12 | 北京航宇天穹科技有限公司 | The regularization coding/decoding method that a kind of splits' positions are perceived |
| CN107169968A (en) * | 2017-03-23 | 2017-09-15 | 天津理工大学 | A kind of image block and processing method based on raster scanning principle |
| CN107301630A (en) * | 2017-06-28 | 2017-10-27 | 重庆大学 | A kind of CS MRI image reconstructing methods constrained based on ordering structure group non-convex |
| CN108573509A (en) * | 2017-03-13 | 2018-09-25 | 信阳师范学院 | Self-adapting compressing perceptual image coding method based on the notable detection model of measurement field |
| CN108734191A (en) * | 2017-05-25 | 2018-11-02 | 湖北工业大学 | Deep learning is applied to the data training method that compressed sensing is rebuild |
| CN109448065A (en) * | 2018-10-11 | 2019-03-08 | 北京理工大学 | A kind of compression sensing method based on the measurement of gradient block adaptive |
| CN110675331A (en) * | 2019-08-13 | 2020-01-10 | 南京人工智能高等研究院有限公司 | Image denoising method and device, computer readable storage medium and electronic device |
| CN111147853A (en) * | 2019-12-20 | 2020-05-12 | 东华大学 | Priori subblock generation method based on zoom pre-observation |
| CN111311752A (en) * | 2020-02-14 | 2020-06-19 | 福州大学 | LiDAR data random sampling and reconstruction method based on mapping chart |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090274356A1 (en) * | 2008-05-02 | 2009-11-05 | The Uwm Research Foundation, Inc. | Toeplitz random encoding for reduced acquisition in compressed sensing magnetic resonance imaging |
| CN102253388A (en) * | 2011-06-23 | 2011-11-23 | 哈尔滨工程大学 | Collaborative detection method for multiple unmanned underwater vehicles on basis of compressed sensing |
-
2013
- 2013-07-30 CN CN2013103245082A patent/CN103440675A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090274356A1 (en) * | 2008-05-02 | 2009-11-05 | The Uwm Research Foundation, Inc. | Toeplitz random encoding for reduced acquisition in compressed sensing magnetic resonance imaging |
| CN102253388A (en) * | 2011-06-23 | 2011-11-23 | 哈尔滨工程大学 | Collaborative detection method for multiple unmanned underwater vehicles on basis of compressed sensing |
Non-Patent Citations (1)
| Title |
|---|
| 李然等: "基于分块压缩感知的图像全局重构模型", 《信号处理》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104778665A (en) * | 2015-04-14 | 2015-07-15 | 清华大学 | Compressed ghost imaging reconstruction method based on natural image block prior driving and system |
| CN104778665B (en) * | 2015-04-14 | 2018-12-11 | 清华大学 | Compression ghost image reconstruction method and system based on the driving of natural image block priori |
| CN104952039B (en) * | 2015-06-14 | 2017-10-10 | 中国电子科技集团公司第十研究所 | Distributed image compressed sensing method for reconstructing |
| CN104952039A (en) * | 2015-06-14 | 2015-09-30 | 中国电子科技集团公司第十研究所 | Distributed compressive sensing reconstruction method for images |
| CN108573509A (en) * | 2017-03-13 | 2018-09-25 | 信阳师范学院 | Self-adapting compressing perceptual image coding method based on the notable detection model of measurement field |
| CN108573509B (en) * | 2017-03-13 | 2021-12-07 | 信阳师范学院 | Self-adaptive compressed sensing image coding method based on measurement domain significance detection model |
| CN107169968A (en) * | 2017-03-23 | 2017-09-15 | 天津理工大学 | A kind of image block and processing method based on raster scanning principle |
| CN107154061A (en) * | 2017-05-09 | 2017-09-12 | 北京航宇天穹科技有限公司 | The regularization coding/decoding method that a kind of splits' positions are perceived |
| CN108734191B (en) * | 2017-05-25 | 2020-11-06 | 湖北工业大学 | Data training method for applying deep learning to compressed sensing reconstruction |
| CN108734191A (en) * | 2017-05-25 | 2018-11-02 | 湖北工业大学 | Deep learning is applied to the data training method that compressed sensing is rebuild |
| CN107301630A (en) * | 2017-06-28 | 2017-10-27 | 重庆大学 | A kind of CS MRI image reconstructing methods constrained based on ordering structure group non-convex |
| CN109448065A (en) * | 2018-10-11 | 2019-03-08 | 北京理工大学 | A kind of compression sensing method based on the measurement of gradient block adaptive |
| CN109448065B (en) * | 2018-10-11 | 2023-07-25 | 北京理工大学 | Compressed sensing method based on gradient blocking self-adaptive measurement |
| CN110675331A (en) * | 2019-08-13 | 2020-01-10 | 南京人工智能高等研究院有限公司 | Image denoising method and device, computer readable storage medium and electronic device |
| CN111147853A (en) * | 2019-12-20 | 2020-05-12 | 东华大学 | Priori subblock generation method based on zoom pre-observation |
| CN111147853B (en) * | 2019-12-20 | 2021-05-25 | 东华大学 | Priori subblock generation method based on zoom pre-observation |
| CN111311752A (en) * | 2020-02-14 | 2020-06-19 | 福州大学 | LiDAR data random sampling and reconstruction method based on mapping chart |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103440675A (en) | Overall situation reconstitution optimization model construction method for image block compressed sensing | |
| Yang et al. | Novel super resolution restoration of remote sensing images based on compressive sensing and example patches-aided dictionary learning | |
| CN108419083B (en) | Image multilevel wavelet full subband compressed sensing coding method | |
| Zhuoran et al. | An improved Hadamard measurement matrix based on Walsh code for compressive sensing | |
| CN109584319A (en) | A kind of compression of images sensing reconstructing algorithm based on non-local low rank and full variation | |
| CN103142228A (en) | Compressed sensing magnetic resonance fast imaging method | |
| CN102075749B (en) | Image compression reconstruction method under compressed sensing frame based on non-convex model | |
| Zhao et al. | Image compressive-sensing recovery using structured laplacian sparsity in DCT domain and multi-hypothesis prediction | |
| CN105678699B (en) | Compressed sensing image reconstructing method based on the detection of measurement field piecemeal conspicuousness | |
| CN103823133A (en) | On-line power quality monitoring system based on compression sensing | |
| Wan et al. | An improved complex multi-task Bayesian compressive sensing approach for compression and reconstruction of SHM data | |
| CN103714534A (en) | Material surface defect detection method based on compressed sensing | |
| Zeng et al. | Image reconstruction of IoT based on parallel CNN | |
| CN107154061B (en) | Regularized decoding method for block compressed sensing | |
| CN103985100A (en) | Partitioned compression sensing method based on self-adaptive observation combinational optimization | |
| Xian-chuan et al. | Remote sensing image fusion based on integer wavelet transformation and ordered nonnegative independent component analysis | |
| Wang et al. | Image reconstruction from patch compressive sensing measurements | |
| Sha et al. | Holographic display based on compressive sensing | |
| Li et al. | Image reconstruction for compressed sensing based on joint sparse bases and adaptive sampling | |
| CN107689797B (en) | Compressed sensing signal reconstruction method based on block Hadamard measurement matrix with arbitrary dimension | |
| Zhang et al. | Reweighted minimization model for MR image reconstruction with split Bregman method | |
| CN109115350B (en) | Wavefront detection system based on compressed sensing | |
| Zhou et al. | Fast compression and reconstruction of astronomical images based on compressed sensing | |
| Hao et al. | Compressed sensing remote sensing image reconstruction based on wavelet tree and nonlocal total variation | |
| Liu et al. | Undersampled hyperspectral image reconstruction based on surfacelet transform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20131211 |