CN112785662B - Self-adaptive coding method based on low-resolution first-pass information - Google Patents

Self-adaptive coding method based on low-resolution first-pass information Download PDF

Info

Publication number
CN112785662B
CN112785662B CN202110117326.2A CN202110117326A CN112785662B CN 112785662 B CN112785662 B CN 112785662B CN 202110117326 A CN202110117326 A CN 202110117326A CN 112785662 B CN112785662 B CN 112785662B
Authority
CN
China
Prior art keywords
image
low
resolution
matrix
information
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.)
Active
Application number
CN202110117326.2A
Other languages
Chinese (zh)
Other versions
CN112785662A (en
Inventor
许廷发
张宇寒
秦庆旺
王茜
张一博
胡如康
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing Innovation Center of Beijing University of Technology
Original Assignee
Chongqing Innovation Center of Beijing University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chongqing Innovation Center of Beijing University of Technology filed Critical Chongqing Innovation Center of Beijing University of Technology
Priority to CN202110117326.2A priority Critical patent/CN112785662B/en
Publication of CN112785662A publication Critical patent/CN112785662A/en
Application granted granted Critical
Publication of CN112785662B publication Critical patent/CN112785662B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T9/00—Image coding
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Abstract

The invention discloses a self-adaptive coding method based on low-resolution first-pass information, which comprises the following steps: estimating the mean value and the variance of each image block corresponding to the original spectrum image by using the integral gray value and the variance of the low-resolution image, wherein the image blocks are scene information subregions corresponding to each point of the low-resolution image respectively; and calculating the threshold distribution of the image blocks of the original spectrum image scene information according to the mean value and the variance of each image block. And constructing an approximate image of the original spectrum image by using the low-resolution image. Based on the approximate image of the original spectrum image and the corresponding threshold distribution, the self-adaptive coding matrix is generated by using a dithering method with the aim of maximizing the correlation between the observation matrix and the sparse matrix. The invention can generate the self-adaptive coding matrix by utilizing the low-resolution spectrum information which can be acquired in the compressed spectrum imaging system without the need of a reconstruction process to provide high-resolution first-pass information, and without the need of adding additional detection processes and detection devices.

Description

Self-adaptive coding method based on low-resolution first-pass information
Technical Field
The invention relates to the technical field of compressed spectrum imaging, in particular to a self-adaptive coding method based on low-resolution first-pass information.
Background
Spectral imaging techniques can acquire not only two-dimensional spatial information of a scene, but also spectral features of the scene at the same time, and these data containing spatial information and spectral information are referred to as spectral data cubes. In recent years, spectral imaging techniques have been widely used in the fields of security monitoring, food safety detection, geology research, medical research, and the like. The conventional spectral imager has the following drawbacks: 1) Limited by the traditional imaging theory, only single information (gray response) of a two-dimensional image can be recorded in a single exposure time, and light field information (such as polarization, depth, phase information and the like) cannot be obtained; 2) Image data with wide field of view, high spatial-spectral resolution, high temporal resolution, and small-volume, low-power imaging systems are not available due to device performance limitations.
With the development of optical computing imaging technology, the independent design thought of the imaging process, the computing process and the device in the traditional spectrum imaging instrument design is broken, and the design integrating illumination, media, targets, imaging devices, platforms and data processing can be realized. The optical computing imaging technology can be mainly divided into the following: 1) Wavefront shaping and optical memory effect-based scattering medium imaging techniques; 2) Novel polarization imaging technologies such as underwater polarization imaging, polarization haze transmission imaging and the like; 3) Bionic optical imaging technology; 4) Calculating a detector technology; 5) Three-dimensional computational imaging techniques represented by three-dimensional binocular reconstruction, structured light three-dimensional imaging, holographic three-dimensional imaging; 6) Computational imaging techniques based on compressed sensing. The computational imaging technology based on compressed sensing utilizes sparsity of a scene in certain information dimension, and the information is restored through reconstruction by sampling times far lower than those of the traditional method, so that the spatial resolution and the spectral resolution are improved.
The adaptive coding method based on compressed sensing is a method for designing a coding matrix by using prior information such as space, spectrum, time and the like of a scene, so that the quality of a reconstructed image is improved. Researchers have the ability to obtain more accurate a priori information by adding additional detection devices; there are also compressed observations of the system to obtain an approximate reconstruction of the scene and use this as a priori information. These approaches come at the cost of increasing the complexity of the structure and reconstruction time of the compression imaging system.
Disclosure of Invention
The invention aims at: in order to solve the problems, an adaptive coding method based on low-resolution prior information is provided to solve the defect that the conventional adaptive coding method generally needs high-resolution prior information under a typical adaptive coding framework.
The technical scheme adopted by the invention is as follows:
an adaptive coding method based on low-resolution first-pass information, comprising:
A. estimating the mean value and the variance of each image block corresponding to the original spectrum image by using the integral gray value and the variance of the low-resolution image, wherein the image blocks are scene information subregions corresponding to each point of the low-resolution image respectively, and the sizes of the image blocks correspond to the compression ratio of the original spectrum image; calculating threshold distribution of the image blocks of the original spectrum image scene information according to the mean value and the variance of each image block;
B. constructing an approximate image of the original spectrum image by utilizing the low-resolution image;
C. based on the approximate image of the original spectrum image and the corresponding threshold distribution, the self-adaptive coding matrix is generated by using a dithering method with the aim of maximizing the correlation between the observation matrix and the sparse matrix.
According to the method, an additional detection process and a detection device are not required to be added, the self-adaptive coding matrix can be generated by utilizing low-resolution spectrum information which can be acquired in a spectrum imaging system based on the compressed sensing principle, and a high-quality hyperspectral image can be reconstructed without high-resolution priori information.
Further, the low resolution image is a spatial downsampling of the original spectral image by equal proportion compression in the x-dimension and the y-dimension. For example, compression with a compression ratio R is performed in the x dimension and the y dimension, and the size of the corresponding image block is r×r.
Further, the spatial downsampling is uniform sampling.
Further, in the step B, an approximate image of the original scene is constructed from the low resolution image by using a kernel-based interpolation method.
Further, the kernel-based interpolation method is a Lanczos kernel-based interpolation method.
Further, the step C includes:
the method comprises the steps of adaptively generating a coding matrix frame by combining a dithering method with threshold distribution with the aim of maximizing correlation between an observation matrix and a sparse matrix;
and evaluating an observation matrix of the scene information of the low-resolution image by using an encoding matrix frame, wherein in the evaluation process, the value of the approximate image is used for replacing the observation value of the scene information in the encoding matrix frame.
Further, the adaptively generating the coding matrix framework by using the dithering method in combination with the threshold distribution for the design purpose of maximizing the correlation between the observation matrix and the sparse matrix includes:
let the observation matrix adopted for the image block in the Kth observation beWherein, the liquid crystal display device comprises a liquid crystal display device,for observing the column vector of the matrix, a sparse matrix of +.>Column vectors that are sparse bases;
let the sparse matrix share S spa A non-zero column vector set ofWherein l (i) represents the position of the non-zero column vector; the correlation between the observation matrix and the sparse matrix is expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,is the complement of Y;
by maximizing xi Y And (3) withThe differences between the two are used as design targets of the coding matrix framework, and the coding matrix framework is designed by combining a dithering method with threshold distribution.
Further, the evaluating the observation matrix of the scene information of the low resolution image by using the encoding matrix frame, wherein in the evaluating process, the value of the approximate image is used to replace the observation value of the scene information in the encoding matrix frame, includes:
using the value f of the approximation image low_inp (R·m x +i,R·m y +j,λ i ) Substituting the observed value of the scene information in the coding matrix frame, and distributing according to the threshold value of the scene information areaEstimating a threshold value of any point, and then the elements of the j-th column in the i-th row in the adaptive coding matrix corresponding to the j-th column in the i-th row in the scene information are as follows:
wherein R is the compression ratio of the original spectrum image to the low resolution image in the x dimension or the y dimension, (m) x ,m y ) Representing points on the low resolution image, λ represents spectral dimensional coordinates, sgn (·) represents a sign operand.
Further, the elements of the j-th column in the i-th row in the adaptive coding matrix corresponding to the j-th column in the i-th row in the scene information are:
the design aims at the consideration that a common coding device cannot directly realize a coding matrix containing a negative value in an imaging system based on a compressed sensing principle, and replaces the formula, so that the imaging system based on the compressed sensing principle can apply the method.
In summary, due to the adoption of the technical scheme, the beneficial effects of the invention are as follows:
1. according to the invention, an additional detection process and a detection device are not required to be added, and the self-adaptive coding matrix can be generated by utilizing low-resolution spectrum information which can be acquired in a spectrum imaging system based on the compressed sensing principle, so that the reconstruction quality of a spectrum image is finally improved.
2. The invention can generate the self-adaptive coding matrix by utilizing the low-resolution spectrum information which can be acquired in the compressed spectrum imaging system without providing high-resolution first-pass information in the reconstruction process, improves the efficiency of acquiring the compressed scene information of the scene, and has the potential of realizing real-time acquisition.
3. Compared with the traditional coding method which is not suitable for self-adaptive coding, the method improves the quality of the reconstructed spectrum image.
Drawings
The invention will now be described by way of example and with reference to the accompanying drawings in which:
fig. 1 is a flow chart of an adaptive coding method based on low resolution anterior information.
Fig. 2 is a graph comparing the implementation effect (PSNR, SSIM) of the method of the present invention with that of the conventional random encoding method.
Detailed Description
All of the features disclosed in this specification, or all of the steps in a method or process disclosed, may be combined in any combination, except for mutually exclusive features and/or steps.
Any feature disclosed in this specification (including any accompanying claims, abstract) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. That is, each feature is one example only of a generic series of equivalent or similar features, unless expressly stated otherwise.
Noun interpretation that the invention may relate to:
dithering (dither) is a method of simulating high resolution chromatography using low resolution chromatography. Common dithering methods are classified into spatial dithering and temporal dithering. Spatial dithering manifests more chromatograms by a regular recurrence of limited chromatograms. The time jitter is realized by adding a time dimension and utilizing the visual residual effect of human eyes, so that the superposition of different chromatograms on the same pixel is realized, and the fineness of chromatographic variation is increased.
Basic knowledge related to the invention:
in a compressed spectrum imaging system, a low-resolution spatial detector is often used to receive compressed original scene information. Common compressed spectral imaging systems naturally have the ability to acquire the original scene spatial information in the form of a low spatial resolution (i.e., the scene information corresponding to the low resolution image).
Example 1
As shown in fig. 1, this embodiment discloses an adaptive coding method based on low-resolution first-pass information, which includes:
A. and estimating the mean value and the variance of each image block corresponding to the original spectrum image by using the gray value and the variance of the whole low-resolution image, wherein the image blocks are scene information subregions of the original spectrum image corresponding to each point of the low-resolution image respectively. And calculating the threshold distribution of the image blocks of the original spectrum image scene information according to the mean value and the variance of each image block.
Assuming that the spatial spectrum information of the original scene is F, the spatial dimension coordinate is (x, y), and the scene information at the spectral dimension coordinate is lambda is F (x, y, lambda) i ). The process of spatial compression encoding of scene information by the encoding device and then acceptance by the detector can be regarded as a spatial downsampling process. Ideal low resolution sampling of scene information is F low_ideal (m x ,m y ,λ i ) The low-resolution space image actually obtained by the detector is F low (m x ,m y ,λ i ) The noise of the detector is omega-N (0, sigma) s 2 ) Assuming that the spatial downsampling is uniform sampling, there is the following relationship:
F low_ideal (m x ,m y ,λ i )＝S left_s ·F(x,y,λ i )·S right_s (1)
F low (m x ,m y ,λ i )＝F low_ideal (m x ,m y ,λ i )+ω (2)
wherein, the liquid crystal display device comprises a liquid crystal display device,the ons represents a matrix of elements all 1, where M x ,M y Representing the size of the low resolution scene information in the x and y dimensions, respectively, where N x ,N y Representing the size of the original scene information in the x and y dimensions, respectively.
Assuming that the aerial image is equally proportionally compressed in the x-dimension and the y-dimension by a ratio R,low resolution image F low Can be regarded as being subjected to the spatial compression ratio gamma by the scene image information F spa ＝R 2 Downsampling and taking into account the results of the detector noise effects. The scene information image size is equal to the low resolution image size, corresponding to the low resolution scene information f low (m x ,m y ,λ i ) Is F sub (m x ,m y ,λ i ). Suppose F sub (m x ,m y ,λ i ) Mean value of (1)Standard deviation isAnalyzing the image mean relation of the two in the space dimension:
it can be seen that the low resolution image points (m x ,m y ) The spectral response intensity f of (2) low (m x ,m y ,λ i ) I.e. scene information image block F sub (m x ,m y ,λ i ) Is a mean value of (c). Because the corresponding low resolution image points do not contain changes within the scene information image block, i.e. no effective information can be provided to calculate the image blockThus, the whole standard deviation of the low resolution image +.>Approximately estimate +.>That is, it is assumed that the change condition inside each scene information image block coincides with the change condition of the whole low resolution image. Such that sub-region F of scene space information sub (m x ,m y ,λ i ) Is +.>
B. An approximation of the original spectral image is constructed using the low resolution image. In the embodiment of the invention, an approximate image of the original scene is constructed from the low-resolution image by adopting a kernel-based interpolation method.
This example illustrates how an approximation of the original spectral image can be constructed by interpolation, using a typical Lanczos kernel interpolation method.
First, an interpolation image is initialized:
wherein F is low_inp (x,y,λ i ) Representing a low resolution interpolated image. Then selecting a proper window template, and calculating the weight coefficient of each position in the window, wherein the weight coefficient of the x-axis is as follows:
where a is a coefficient. Similarly, a weight coefficient Lan (y) in the y-axis direction can be calculated, and the weight coefficient according to the two-dimensional interpolation template is the product of one-dimensional weight coefficients, so that the method can be obtained:
Lan(x,y)＝Lan(x)·Lan(y) (6)
according to the initialized interpolation image, carrying out weighted average on points in the template:
in this way, an interpolated image F can be obtained from the low resolution anterior information low_inp . In some embodiments, the coefficient a takes 3, i.e., lanczos-3 kernel is employed.
C. Based on the approximate image of the original spectrum image and the corresponding threshold distribution, the self-adaptive coding matrix is generated by using a dithering method with the aim of maximizing the correlation between the observation matrix and the sparse matrix.
The method comprises the steps of firstly, adaptively generating a coding matrix framework by using a dithering method and combining threshold distribution with the aim of maximizing correlation between an observation matrix and a sparse matrix.
Suppose that sub-image block F is targeted in the Kth observation sub (m x ,m y ,λ i ) The observation matrix adopted isWherein (1)>Is the column vector of the observation matrix. The sparse basis adopted isIs a sparse-based column vector. The correlation between the observation matrix and the sparse basis can be estimated by the following formula:
ξ＝max{|<φ i ,ψj>| 2 },i＝1,2,...K and j＝1,2,...R 2 (8)
when a low resolution image is known, it is assumed that the positions and number of non-zero column vectors in the sparse basis corresponding to the image block can be calculated. S is shared in sparse basis corresponding to image block spa A non-zero column vector set ofWhere l (i) represents the position of the non-zero column vector. At this time, according to the orthogonal principle, the problem of recovering the scene information can be converted into the problem of making the vector of the existing projection matrix array equal to the ideal sparse basis, then only S is needed spa A secondary observation is sufficient to accurately recover the scene information. Under this assumption, the calculation formula of the cross correlation in formula (8) is divided into two parts:
wherein, the liquid crystal display device comprises a liquid crystal display device,is the complement of Y, phi i Sum phi j The energy normalization process has been performed. By maximizing xi Y And->The difference between them is a design goal. Assume that the noise omega distribution of a detector employed to acquire scene information follows omega-N (0, sigma) s 2 ) Is a distribution of (a). The observed value of the acquired scene information is as follows: />Wherein->And->Respectively corresponding to the low resolution image points (m x ,m y ) And when the scene information is acquired by the detector, the information acquired by the detector and the original scene information are acquired by the detector.
An observation matrix is designed by using a dithering method and combining a threshold operation, and the elements of the adaptive observation matrix of the j th column in the i th row are as follows:
where sgn (·) represents the sign operand,representing the element of the j-th column in the i-th row in the scene information, the threshold Λ ij Is a compliance->A random threshold of distribution.
And evaluating an observation matrix of the scene information of the low-resolution image by using an encoding matrix frame, wherein in the evaluation process, the value of the approximate image is used for replacing the observation value of the scene information in the encoding matrix frame.
Based on the above designed coding matrix framework, the threshold distribution generated in the step A is recombinedAnd an approximation image of the original spectral image produced in step B +.>Under the low-resolution adaptive coding framework designed by the process, an adaptive coding matrix of the low-resolution image scene information is obtained. In the process of evaluating the observation matrix of the scene information of the low-resolution image by using the low-resolution adaptive coding framework designed by the process, when the spectrum band is lambda i When the value f of the approximate image is adopted low_inp (R·m x +i,R·m y +j,λ i ) New observations of the scene in the alternative coding framework, < >>According to->Threshold distribution of regions->A threshold value for any point is estimated. Thus (S)>The elements of the j-th column in the i-th row in the corresponding adaptive coding matrix are:
in an imaging system based on the compressed sensing principle, a common encoding device cannot directly realize an encoding matrix containing a negative value, so the following equivalent conversion is performed on elements of the adaptive encoding matrix in the formula (11) to adapt to the realization in a practical device:
example two
According to the embodiment, the effectiveness and superiority of the scheme are proved by practical tests, and the random coding method which is wide in application range and has public confidence in the field of compression spectrum imaging is selected for further comparison through simulation experiments. The present example selects spectral data disclosed by university of telawamori in the united states as experimental data, which has 24 spectral bands with a spatial resolution of 256×256. Matlab 2020a is selected as a simulation experiment platform. The algorithm was evaluated for its merits in terms of peak signal-to-noise ratio (Peak Signal to Noise Ratio, PSNR) and structural similarity (Structural Similarity, SSIM).
Comparing the implementation effect of the method provided by the invention with that of the traditional random coding method, as shown in figure 2, a sub-graph (1) in figure 2 shows the advantages and disadvantages of the two methods under PSNR evaluation indexes in different spectrum bands, and the method provided by the invention has obvious advantages compared with the random coding method; the sub-graph (2) in fig. 2 shows the advantages and disadvantages of the two methods under the SSIM evaluation index in different spectrum bands, and the method provided by the invention has obvious advantages compared with the random coding method.
The invention is not limited to the specific embodiments described above. The invention extends to any novel one, or any novel combination, of the features disclosed in this specification, as well as to any novel one, or any novel combination, of the steps of the method or process disclosed.

Claims (6)

1. An adaptive coding method based on low-resolution first-pass information, comprising:
A. estimating the mean value and the variance of each image block corresponding to the original spectrum image by using the integral gray value and the variance of the low-resolution image, wherein the image blocks are scene information subregions corresponding to each point of the low-resolution image respectively; calculating threshold distribution of the image blocks of the original spectrum image scene information according to the mean value and the variance of each image block;
B. constructing an approximate image of the original spectrum image by utilizing the low-resolution image;
C. based on the approximate image of the original spectrum image and the corresponding threshold distribution, generating a self-adaptive coding matrix by using a dithering method with the aim of maximizing the correlation between the observation matrix and the sparse matrix; the step C comprises the following steps:
with the design objective of maximizing the correlation between the observation matrix and the sparse matrix, the adaptive generation of the coding matrix framework using dithering methods in combination with threshold distribution, comprising:
let the observation matrix adopted for the image block in the Kth observation beWherein, the liquid crystal display device comprises a liquid crystal display device,for observing the column vector of the matrix, a sparse matrix of +.> Column vectors that are sparse bases;
let the sparse matrix share S spa A non-zero column vector set ofWherein l (i) represents the position of the non-zero column vector; the correlation between the observation matrix and the sparse matrix is expressed as:
wherein, the liquid crystal display device comprises a liquid crystal display device,is the complement of Y;
by maximizing xi Y And (3) withThe difference between the two is used as a design target of a coding matrix frame, and the coding matrix frame is designed by combining a dithering method with threshold distribution;
evaluating an observation matrix of the scene information of the low resolution image using an encoding matrix framework, wherein in the evaluating process, the value of the approximate image is used for replacing the observation value of the scene information in the encoding matrix framework, comprising: using the value f of the approximation image low_inp (R·m x +i,R·m y +j,λ i ) Substituting the observed value of the scene information in the coding matrix frame, and distributing according to the threshold value of the scene information areaEstimating a threshold value of any point, and then the elements of the j-th column in the i-th row in the adaptive coding matrix corresponding to the j-th column in the i-th row in the scene information are as follows:
wherein R is the compression ratio of the original spectrum image to the low resolution image in the x dimension or the y dimension, (m) x ,m y ) Representing points on the low resolution image, λ represents spectral dimensional coordinates, sgn (·) represents a sign operand.
2. The adaptive encoding method based on low-resolution anterior information of claim 1, wherein the low-resolution image is a spatial downsampling of an original spectral image by equal proportion in x-dimension and y-dimension.
3. The adaptive encoding method based on low-resolution anterior information of claim 2, wherein said spatial downsampling is uniform sampling.
4. A low-resolution first-pass information-based adaptive coding method according to any one of claims 1 to 3, wherein in said step B, an approximate image of the original scene is constructed from the low-resolution image by using a kernel-based interpolation method.
5. The adaptive encoding method based on low-resolution anterior information of claim 4, wherein said kernel-based interpolation method is a Lanczos kernel-based interpolation method.
6. The adaptive coding method based on low-resolution anterior information of claim 1, wherein the elements of the j-th column in the i-th row in the adaptive coding matrix corresponding to the j-th column in the i-th row in the scene information are:
CN202110117326.2A 2021-01-28 2021-01-28 Self-adaptive coding method based on low-resolution first-pass information Active CN112785662B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110117326.2A CN112785662B (en) 2021-01-28 2021-01-28 Self-adaptive coding method based on low-resolution first-pass information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110117326.2A CN112785662B (en) 2021-01-28 2021-01-28 Self-adaptive coding method based on low-resolution first-pass information

Publications (2)

Publication Number Publication Date
CN112785662A CN112785662A (en) 2021-05-11
CN112785662B true CN112785662B (en) 2023-07-25

Family

ID=75759330

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110117326.2A Active CN112785662B (en) 2021-01-28 2021-01-28 Self-adaptive coding method based on low-resolution first-pass information

Country Status (1)

Country Link
CN (1) CN112785662B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115499658B (en) * 2022-09-20 2024-05-07 支付宝(杭州)信息技术有限公司 Virtual world data transmission method and device

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5917952A (en) * 1995-09-20 1999-06-29 Hewlett-Packard Company Compression of randomly dithered bi-level images
US20140063024A1 (en) * 2012-12-19 2014-03-06 Iowa State University Research Foundation, Inc. Three-dimensional range data compression using computer graphics rendering pipeline
CN106780342A (en) * 2016-12-28 2017-05-31 深圳市华星光电技术有限公司 Single-frame image super-resolution reconstruction method and device based on the reconstruct of sparse domain
CN108955882B (en) * 2018-07-10 2020-06-19 北京理工大学 Three-dimensional data reconstruction method based on liquid crystal hyperspectral calculation imaging system
CN109146787B (en) * 2018-08-15 2022-09-06 北京理工大学 Real-time reconstruction method of dual-camera spectral imaging system based on interpolation
CN109447898B (en) * 2018-09-19 2023-02-17 北京理工大学 Hyperspectral super-resolution calculation imaging system based on compressed sensing
CN111343376B (en) * 2018-12-19 2021-09-10 南京理工大学 Transmission-type double-slit-based aperture coding imaging system and super-resolution method thereof
CN109682476B (en) * 2019-02-01 2020-06-19 北京理工大学 Method for performing compressed hyperspectral imaging by using self-adaptive coding aperture

Also Published As

Publication number Publication date
CN112785662A (en) 2021-05-11

Similar Documents

Publication Publication Date Title
He et al. A new pansharpening method based on spatial and spectral sparsity priors
Xie et al. Hyperspectral image super-resolution using deep feature matrix factorization
Qian et al. Hyperspectral imagery restoration using nonlocal spectral-spatial structured sparse representation with noise estimation
CN110675347B (en) Image blind restoration method based on group sparse representation
Lahav et al. Wiener reconstruction of galaxy surveys in spherical harmonics
CN108416723B (en) Lens-free imaging fast reconstruction method based on total variation regularization and variable splitting
US20130011051A1 (en) Coded aperture imaging
CN109584319A (en) A kind of compression of images sensing reconstructing algorithm based on non-local low rank and full variation
CN107742133A (en) A kind of sorting technique for Polarimetric SAR Image
Shen et al. Coupling model-and data-driven methods for remote sensing image restoration and fusion: Improving physical interpretability
CN113870132A (en) Noise elimination method and system in ghost imaging sampling calculation process and related components
CN112785662B (en) Self-adaptive coding method based on low-resolution first-pass information
Harmany et al. Spatio-temporal compressed sensing with coded apertures and keyed exposures
Zhang et al. Polarization image demosaicking via nonlocal sparse tensor factorization
CN109087262B (en) Multi-view spectral image reconstruction method and storage medium
CN116739958B (en) Dual-spectrum polarization super-resolution fusion detection method and system
CN112784747A (en) Multi-scale eigen decomposition method for hyperspectral remote sensing image
Marcia et al. Compressive optical imaging: Architectures and algorithms
Moustafa et al. Acceleration of super-resolution for multispectral images using self-example learning and sparse representation
CN110926611A (en) Noise suppression method applied to compressed sensing spectral imaging system
Solanki et al. An efficient satellite image super resolution technique for shift-variant images using improved new edge directed interpolation
KR100633555B1 (en) Signal processing
CN112989593B (en) High-spectrum low-rank tensor fusion calculation imaging method based on double cameras
CN114240756A (en) RGB image spectral information reconstruction method based on dictionary atom embedding
Li et al. Research on ghost imaging method based on wavelet transform

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