CN107888915A - A kind of perception compression method of combination dictionary learning and image block - Google Patents
A kind of perception compression method of combination dictionary learning and image block Download PDFInfo
- Publication number
- CN107888915A CN107888915A CN201711086754.3A CN201711086754A CN107888915A CN 107888915 A CN107888915 A CN 107888915A CN 201711086754 A CN201711086754 A CN 201711086754A CN 107888915 A CN107888915 A CN 107888915A
- Authority
- CN
- China
- Prior art keywords
- dictionary
- image
- piecemeal
- carried out
- data
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/70—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/91—Entropy coding, e.g. variable length coding [VLC] or arithmetic coding
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/97—Matching pursuit coding
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
Abstract
The invention discloses a kind of combination dictionary learning and the perception compression method of image block, training image is chosen first, it is determined that training parsing dictionary in the case of dictionary atom number.Then re-encode end and read in image to be compressed, rarefaction representation is carried out to image using dictionary is parsed, size selection further according to element in sparse coefficient matrix optimizes coefficient while the corresponding atomic space for optimizing coefficient is extracted into the dictionary as current data reconstruction end, then the label information and positional information of coefficient are obtained, is transferred to decoding end.Finally it is reconstructed in decoding end, recovers original image.The present invention shortens reconstitution time than congenic method, and improves reconstruction accuracy, there is stronger practicality.
Description
Technical field
The invention belongs to Image Compression field, is related to a kind of method for compressing image, and in particular to one kind combines dictionary
Study and the perception compression method of image block.
Background technology
With the continuous development of computer vision technique, image is closed as the carrier of visual information by numerous scholars
Note.But the development of high-resolution, Hyperspectral imaging technology, serious challenge is brought to image transmitting and preservation.In recent years, press
Contracting perception theory gradually by the extensive concern of domestic and foreign scholars, point out if being sparse in image transform domain by compressive sensing theory
Or it is compressible when then signal can lead to too small amount of observation sample data and realize signal by solving optimal method
High precision reconstruction.
Compression of images sensor model mainly has two kinds of collective model and analytic modell analytical model.Collective model is by the process of rarefaction representation
It is defined as X=DS, s.t. | | si||0≤ k, wherein D were complete dictionary, and S is sparse table of the image under current excessively complete dictionary
Show coefficient matrix, siFor image local data x in image XiCorresponding coefficient vector.During solving optimal rarefaction representation
By limiting each element s in SiNonzero element number be less than k, to realize k sparse tables in image under excessively complete dictionary D
Show.Although collective model rarefaction representation has developed for a long time, it is still in the continuous development perfect stage.It is another
Model, analytic modell analytical model had attracted substantial amounts of focus of attention in recent years, its sparse analytic modell analytical model (Cosparse altogether that is otherwise known as
analysis model).The process of rarefaction representation is defined as S=Φ X, s.t. by sparse analytic modell analytical model altogether | | si||0≤ p-l, its
Middle X is the image set signal for treating rarefaction representation, and Φ is current parsing dictionary, and S is then image set signal X in currently parsing dictionary
Rarefaction representation coefficient matrix under Ω, siFor the column vector in sparse coefficient matrix, p siLine number, l siIn zero number
Degree of rarefication is also known as total to, then by limiting siL0Norm is less than p-l so as to realize the rarefaction representation process of signal.
By X=DS, s.t. | | si||0≤ k and S=Φ X, s.t. | | si||0≤ p-l is understood, when D, Ω are all square formation,
Then both can realize equivalencing Φ=D this moment-1, D-1Represent D pseudoinverse;When D, Φ were all complete dictionary, then D ∈ Rm ×n(m < n), Φ ∈ Rp×q(p > q) is no longer simple mutual inverse process so as to both, entirely different between both.By X=
DS,s.t.||si||0≤ k and S=Φ X, s.t. | | si||0Knowable to≤p-l when obtaining the sparse coefficient matrix of identical dimensional,
Analytic modell analytical model includes more subspace quantity, there is more abundant flexible expression performance.While analytic modell analytical model rarefaction representation
Process with signal obtain by way of matrix multiple seeks inner product dictionary, i.e. S=Φ X, s.t. | | si||0≤ p-l, so
Reduce substantial amounts of calculating process with respect to collective model, so as to improve computational efficiency, therefore analytic modell analytical model is in the sparse table of image
Show and compression process in will more advantage.Increasingly attracted attention in image procossing based on this reasons analysis model.
Rubinstein etc. introduces Analysis K-SVD analytic modell analytical model dictionary learning algorithms on the basis of K-SVD dictionary learnings,
The step of algorithm utilizes sparse coding and dictionary updating alternately study parsing dictionary, and demonstrate algorithm and answered in image denoising
Advantage.Simon Hawe etc. are converted into the manifold optimization problem of matrix by the problem of dictionary obtains is parsed, in lpNorm is most
Propose geometry conjugate gradient decent on the basis of small solution optimization problem constraints, and the algorithm by experimental verification
In the advantage and potentiality of image processing field.Some documents gradient decline on the basis of using it is sparse observation determinant value as
Optimal conditions are introduced during dictionary learning, advantage of the verification algorithm in dictionary learning on experiment basis.Martin
Kiechle proposes bimodal sparse model altogether, and joint gradient descent method is proposed on the basis of matrix flow pattern and obtains parsing word
Allusion quotation, at the same by experimental verification algorithm image reconstruction advantage.With being continuously increased for research, parsing dictionary learning is in image
Advantage in processing is also more and more prominent.
In summary, there is problems with current compression of images:
1. using fixed dictionary, matching degree is low, and quality reconstruction is poor;
2. carrying out rarefaction representation using collective model, computationally intensive, time and memory consumption are big.
The content of the invention
In order to solve the above-mentioned technical problem, the present invention learns dictionary in rarefaction representation, image procossing based on analytic modell analytical model
Advantage, while use for reference block image processing advantage, it is proposed that it is a kind of based on parsing dictionary block image compression method, will
The reconstruction accuracy that sparse analytic modell analytical model is applied in the compression process of image to improve the treatment effeciency of image and improve image altogether.
The technical solution adopted in the present invention is:A kind of perception compression method of combination dictionary learning and image block, its
It is characterised by, comprises the following steps:
Step 1:Define S=Φ X, s.t. | | Si||0≤p-l;By limiting siL0Norm is less than p-l so as to realize letter
Number rarefaction representation process;
Wherein X is the image set signal for treating rarefaction representation, and Φ is current parsing dictionary, and S is then working as image set signal X
Rarefaction representation coefficient matrix under preceding parsing dictionary Φ, siFor the column vector in sparse coefficient matrix, p siLine number, l si
In zero number also known as altogether degree of rarefication;
Step 2:Rarefaction representation is carried out in the case where parsing dictionary to test image;
Step 3:Quantization entropy code is based on to sparse coefficient and is compressed coding;
Step 4:By the information transfer after coding to decoding end.
The present invention is a kind of method applied to compression of images, compared with prior art with advantages below:
(1) present invention will parse dictionary on the basis of piecemeal perceives compression from the time-consuming angle of rarefaction representation
Study is incorporated into piecemeal and perceived in compression process.
(2) present invention based on rarefaction representation under analytic modell analytical model is obtained using the inner product of dictionary and signal, therefore its
To be taken calculating on time-consuming far fewer than the rarefaction representation of other models, so as to improve the efficiency that compression is rebuild.
(3) present invention belongs to study acquisition using parsing dictionary and make it that better than other are based on conversion on rarefaction representation for it
The rarefaction representation of model.
Brief description of the drawings
Fig. 1 is the compression of images model of the embodiment of the present invention.
Fig. 2 is the compression reconfiguration result (sample rate is respectively 0.21,0.28) of the embodiment of the present invention, wherein (a), (c) are
Raw image data;(b), (d) is the view data rebuild after compressing.
Embodiment
Understand for the ease of those of ordinary skill in the art and implement the present invention, below in conjunction with the accompanying drawings and embodiment is to this hair
It is bright to be described in further detail, it will be appreciated that implementation example described herein is merely to illustrate and explain the present invention, not
For limiting the present invention.
See Fig. 1, the perception compression method of a kind of combination dictionary learning and image block provided by the invention, including it is following
Step:
Step 1:The image for the premise input that the present invention realizes can carry out rarefaction representation to it, and model definition is:S=Φ
X,s.t.||Si||0≤ p-l, wherein X are the image set signal for treating rarefaction representation, and Φ is current parsing dictionary, and S is then image
Rarefaction representation coefficient matrixes of the signal collection X under currently parsing dictionary Ω, siFor the column vector in sparse coefficient matrix, p si's
Line number, l siIn zero number also known as altogether degree of rarefication, then by limiting siL0Norm is less than p-l so as to realize the dilute of signal
Dredge expression process;l0For norm, i.e. siThe number of middle nonzero element;
Wherein parsing the acquisition of dictionary includes following sub-step:
Step 1.1:Training parsing dictionary;
Piecemeal processing is carried out to training image data, piecemeal is carried out to image using B*B equal in magnitude formwork, obtained
Data X after piecemealΓ∈Rd×L, wherein, d=B*B, L are the number after image block, and its value is image size of population divided by divided
Block size obtains;
Step 1.2:The dictionary needed for parsing dictionary algorithm acquisition;Specific implementation includes following sub-step:
Step 1.2.1:Initiation parameter;
Iterations t=1, maximum iteration K=50, redundant error ε, initialization dictionary Φ0, make Φ1=Φ0;
Step 1.2.2:Calculate rarefaction representation coefficient of the data under current dictionary:St=ΦtXΓStS transposition is represented,
ΦtRepresent Φ transposition;
Step 1.2.3:Dictionary is obtained using MOD algorithms:Φt+1←Φt;
Step 1.2.4:Calculation error factor values r=| | S- Φ X | |2+||Φ-1Φ-I||2, Φ-1Represent Φ pseudoinverse, I
Represent unit matrix;
Step 1.2.5:Judge;
If r >=ε and iterations t≤K, t=t+1, and turn round and perform step 1.2.2- steps 1.2.4;
Otherwise iterative process is terminated;
Step 1.2.6:Obtain final parsing dictionary Φ=Φt。
Step 2:Rarefaction representation is carried out in the case where parsing dictionary to test image;
Step 3:Quantization entropy code is based on to sparse coefficient and is compressed coding;
Specific implementation includes following sub-step:
Step 3.1:Piecemeal processing is carried out to test image data, image carried out using B*B equal in magnitude formwork
Piecemeal, obtain the data after piecemealWherein, d=B*B, L1For the number after image block, its value is that image is overall
Size divided by piecemeal size obtain;
Step 3.2:Rarefaction representation is carried out to dataObtain rarefaction representation coefficient matrix S;
Step 3.3:Optimization selection is carried out to sparse coefficient, selection gist isObtain most
Optimize sparse coefficient matrix
Step 3.4:The number N for optimizing non-zero item in sparse coefficient matrix is calculated, then compression ratio R=N/ picture sizes.
Step 4:By the information transfer after coding to decoding end;
Specific implementation includes following sub-step:
Step 4.1:Using coding side optimal sparse coefficient matrix byObtain the block data of reconstruct, ΦΛ
For ΦΛPseudoinverse, ΦΛIt is by selecting to optimize coefficient in step 3.3, the corresponding atomic space for optimizing coefficient is proposed
The dictionary at end is out rebuild as current data;
Step 4.2:Piecemeal is carried out to block data to recover to obtain final reconstruction image
See Fig. 2, the present embodiment is using 400 × 400 image as training image, and 256 × 256 image is as test chart
Picture;
Wherein training parsing dictionary:
(1) image for 400 × 400, piecemeal processing is carried out to view data, utilizes 8*8 equal in magnitude formwork
Piecemeal is carried out to image, obtains the data X after piecemealΓ∈R64×2500;
(2) atom number K=64 (atom number is equal to piecemeal size) is chosen, is obtained using block sparse dictionary learning algorithm
Training dictionary Φ, object function are:
Wherein X ∈ Rm×NFor given training dataset, Φ ∈ Rp×mTo be compiled by training dataset dictionary to be learned
Code side pressure contract drawing picture, | | S:,j||0=p-l represents the nonzero element number that jth arranges in S, and l is then when the common degree of rarefication in forefront.
Wherein coding side compression image:
(1) image for 256 × 256, piecemeal is carried out to image using 8*8 equal in magnitude formwork, obtains piecemeal
Data afterwards
(2) rarefaction representation is carried outObtain sparse coefficient S.
(3) optimization selection is carried out to sparse coefficient, selection gist isIt is dilute to obtain optimization
Sparse coefficient matrix
(4) the number N for optimizing non-zero item in sparse coefficient matrix, compression ratio R=N/ picture sizes, R in this example are calculated
=0.28.
Wherein decoding end recovers image:
(1) using coding side optimal sparse coefficient matrix byObtain the block data of reconstruct, ΦΛFor ΦΛ
Pseudoinverse, ΦΛIt is that coefficient is optimized by selection in coding side compression image step (3), the corresponding atom of coefficient will be optimized
Space proposes out to rebuild the dictionary at end as current data.
(3) piecemeal is carried out to block data to recover to obtain final reconstruction image
It should be appreciated that the part that this specification does not elaborate belongs to prior art.
It should be appreciated that the above-mentioned description for preferred embodiment is more detailed, therefore can not be considered to this
The limitation of invention patent protection scope, one of ordinary skill in the art are not departing from power of the present invention under the enlightenment of the present invention
Profit is required under protected ambit, can also be made replacement or deformation, be each fallen within protection scope of the present invention, this hair
It is bright scope is claimed to be determined by the appended claims.
Claims (5)
1. the perception compression method of a kind of combination dictionary learning and image block, it is characterised in that comprise the following steps:
Step 1:Define S=Φ X, s.t. | | Si||0≤p-l;By limiting siL0Norm is less than p-l so as to realize the dilute of signal
Dredge expression process;
Wherein X is the image set signal for treating rarefaction representation, and Φ is current parsing dictionary, and S then solves for image set signal X currently
Analyse the rarefaction representation coefficient matrix under dictionary Φ, siFor the column vector in sparse coefficient matrix, p siLine number, l siIn zero
Number also known as altogether degree of rarefication;l0For norm, i.e. siThe number of middle nonzero element;
Step 2:Rarefaction representation is carried out in the case where parsing dictionary to test image;
Step 3:Quantization entropy code is based on to sparse coefficient and is compressed coding;
Step 4:By the information transfer after coding to decoding end.
2. the perception compression method of combination dictionary learning according to claim 1 and image block, it is characterised in that step
The acquisition of dictionary is parsed in 1 includes following sub-step:
Step 1.1:Training parsing dictionary;
Piecemeal processing is carried out to training image data, piecemeal is carried out to image using B*B equal in magnitude formwork, obtains piecemeal
Data X Γ ∈ R afterwardsd×L, wherein, d=B*B, L are the number after image block, and its value is that image size of population divided by piecemeal are big
Small acquisition;
Step 1.2:The dictionary needed for parsing dictionary algorithm acquisition.
3. the perception compression method of combination dictionary learning according to claim 2 and image block, it is characterised in that step
1.2 specific implementation includes following sub-step:
Step 1.2.1:Initiation parameter;
Iterations t=1, maximum iteration K=50, redundant error ε, initialization dictionary Φ0, make Φ1=Φ0;
Step 1.2.2:Calculate rarefaction representation coefficient of the data under current dictionary:St=ΦtXΓ;StRepresent S transposition, ΦtTable
Show Φ transposition;
Step 1.2.3:Dictionary is obtained using MOD algorithms:Φt+1←Φt;
Step 1.2.4:Calculation error factor values r=| | S- Φ X | |2+||Φ-1Φ-I||2, Φ-1Φ pseudoinverse is represented, I is represented
Unit matrix;
Step 1.2.5:Judge;
If r >=ε and iterations t≤K, t=t+1, and turn round and perform step 1.2.2- steps 1.2.4;
Otherwise iterative process is terminated;
Step 1.2.6:Obtain final parsing dictionary Φ=Φt。
4. the perception compression method of combination dictionary learning according to claim 1 and image block, it is characterised in that step
3 specific implementation includes following sub-step:
Step 3.1:Piecemeal processing is carried out to test image data, piecemeal is carried out to image using B*B equal in magnitude formwork,
Obtain the data after piecemealWherein, d=B*B, L1For the number after image block, its value is image size of population
Divided by piecemeal size obtains;
Step 3.2:Rarefaction representation is carried out to dataObtain rarefaction representation coefficient matrix S;
Step 3.3:Optimization selection is carried out to sparse coefficient, selection gist isOptimized
Sparse coefficient matrix
Step 3.4:The number N for optimizing non-zero item in sparse coefficient matrix is calculated, then compression ratio R=N/ picture sizes.
5. the perception compression method of combination dictionary learning according to claim 4 and image block, it is characterised in that step
4 specific implementation includes following sub-step:
Step 4.1:Using coding side optimal sparse coefficient matrix byObtain the block data of reconstruct, ΦΛFor ΦΛ
Pseudoinverse, ΦΛIt is by selecting to optimize coefficient in step 3.3, the corresponding atomic space for optimizing coefficient being proposed out
The dictionary at end is rebuild as current data;
Step 4.2:Piecemeal is carried out to block data to recover to obtain final reconstruction image
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711086754.3A CN107888915A (en) | 2017-11-07 | 2017-11-07 | A kind of perception compression method of combination dictionary learning and image block |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711086754.3A CN107888915A (en) | 2017-11-07 | 2017-11-07 | A kind of perception compression method of combination dictionary learning and image block |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107888915A true CN107888915A (en) | 2018-04-06 |
Family
ID=61779240
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711086754.3A Pending CN107888915A (en) | 2017-11-07 | 2017-11-07 | A kind of perception compression method of combination dictionary learning and image block |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107888915A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109801367A (en) * | 2019-02-25 | 2019-05-24 | 广西大学 | A kind of grid model feature edit method based on compression manifold mode |
CN113139918A (en) * | 2021-04-23 | 2021-07-20 | 大连大学 | Image reconstruction method based on decision-making gray wolf optimization dictionary learning |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102609681A (en) * | 2012-01-12 | 2012-07-25 | 北京大学 | Face recognition method based on dictionary learning models |
CN102708576A (en) * | 2012-05-18 | 2012-10-03 | 西安电子科技大学 | Method for reconstructing partitioned images by compressive sensing on the basis of structural dictionaries |
US20130151924A1 (en) * | 2011-12-08 | 2013-06-13 | Harris Corporation, Corporation Of The State Of Delaware | Data system for interfacing with a remote data storage facility using compressive sensing and associated methods |
CN103886625A (en) * | 2014-01-09 | 2014-06-25 | 北京工业大学 | Point cloud data sparse representation method based on compressed sensing |
CN105242245A (en) * | 2015-10-09 | 2016-01-13 | 中国科学院大学 | Noise inhibition method based on low rank and sparsity of polar region ice-penetrating radar data |
CN105741252A (en) * | 2015-11-17 | 2016-07-06 | 西安电子科技大学 | Sparse representation and dictionary learning-based video image layered reconstruction method |
CN106295689A (en) * | 2016-08-01 | 2017-01-04 | 广东工业大学 | A kind of sparse signal representation method and device |
CN107169936A (en) * | 2017-05-12 | 2017-09-15 | 攀枝花学院 | Ancient wall image repair method based on rarefaction representation |
-
2017
- 2017-11-07 CN CN201711086754.3A patent/CN107888915A/en active Pending
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20130151924A1 (en) * | 2011-12-08 | 2013-06-13 | Harris Corporation, Corporation Of The State Of Delaware | Data system for interfacing with a remote data storage facility using compressive sensing and associated methods |
CN102609681A (en) * | 2012-01-12 | 2012-07-25 | 北京大学 | Face recognition method based on dictionary learning models |
CN102708576A (en) * | 2012-05-18 | 2012-10-03 | 西安电子科技大学 | Method for reconstructing partitioned images by compressive sensing on the basis of structural dictionaries |
CN103886625A (en) * | 2014-01-09 | 2014-06-25 | 北京工业大学 | Point cloud data sparse representation method based on compressed sensing |
CN105242245A (en) * | 2015-10-09 | 2016-01-13 | 中国科学院大学 | Noise inhibition method based on low rank and sparsity of polar region ice-penetrating radar data |
CN105741252A (en) * | 2015-11-17 | 2016-07-06 | 西安电子科技大学 | Sparse representation and dictionary learning-based video image layered reconstruction method |
CN106295689A (en) * | 2016-08-01 | 2017-01-04 | 广东工业大学 | A kind of sparse signal representation method and device |
CN107169936A (en) * | 2017-05-12 | 2017-09-15 | 攀枝花学院 | Ancient wall image repair method based on rarefaction representation |
Non-Patent Citations (4)
Title |
---|
ZONGWEI FENG等: "Image Compression Based on Analysis Dictionary", 《INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTING,ICIC 2016: INTELLIGENT COMPUTING THEORIES AND APPLICATION 》 * |
施帅: "《基于稀疏分解的分块字典图像压缩算法》", 《中国公共安全(综合版)》 * |
练秋生等: "字典学习模型、算法及其应用研究进展", 《自动化学报》 * |
邹建成: "《一种基于MOD 字典学习的图像超分辨率重建新算法》", 《图学学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109801367A (en) * | 2019-02-25 | 2019-05-24 | 广西大学 | A kind of grid model feature edit method based on compression manifold mode |
CN109801367B (en) * | 2019-02-25 | 2023-01-13 | 广西大学 | Grid model characteristic editing method based on compressed manifold mode |
CN113139918A (en) * | 2021-04-23 | 2021-07-20 | 大连大学 | Image reconstruction method based on decision-making gray wolf optimization dictionary learning |
CN113139918B (en) * | 2021-04-23 | 2023-11-10 | 大连大学 | Image reconstruction method based on decision-making gray wolf optimization dictionary learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104574336B (en) | Super-resolution image reconstruction system based on adaptive sub- mould dictionary selection | |
EP3678059B1 (en) | Image processing method, image processing apparatus, and a neural network training method | |
CN106204449B (en) | A kind of single image super resolution ratio reconstruction method based on symmetrical depth network | |
CN102842115B (en) | Based on the compressed sensing image super-resolution rebuilding method of double dictionary study | |
CN105430416B (en) | A kind of Method of Fingerprint Image Compression based on adaptive sparse domain coding | |
CN109447921A (en) | A kind of image measurement matrix optimizing method based on reconstructed error | |
CN107845117B (en) | hyperspectral image compression method based on block sparse expression mode and structural dictionary | |
WO2018107414A1 (en) | Apparatus, equipment and method for compressing/decompressing neural network model | |
CN113516601A (en) | Image restoration technology based on deep convolutional neural network and compressed sensing | |
CN104199627B (en) | Gradable video encoding system based on multiple dimensioned online dictionary learning | |
CN104732566B (en) | Compression of hyperspectral images cognitive method based on non-separation sparse prior | |
CN110751265A (en) | Lightweight neural network construction method and system and electronic equipment | |
CN113222834A (en) | Visual data tensor completion method based on smooth constraint and matrix decomposition | |
CN107301630A (en) | A kind of CS MRI image reconstructing methods constrained based on ordering structure group non-convex | |
CN109887050A (en) | A kind of code aperture spectrum imaging method based on self-adapting dictionary study | |
Song et al. | Robust tensor completion using transformed tensor svd | |
CN107888915A (en) | A kind of perception compression method of combination dictionary learning and image block | |
Xu et al. | Singular vector sparse reconstruction for image compression | |
Otair et al. | Improved near-lossless technique using the Huffman coding for enhancing the quality of image compression | |
CN104200439B (en) | Image super-resolution method based on adaptive filtering and regularization constraint | |
Li et al. | Nonconvex nonsmooth low-rank minimization for generalized image compressed sensing via group sparse representation | |
CN105590296A (en) | Dual-dictionary learning-based single-frame image super-resolution reconstruction method | |
Liu et al. | Image super-resolution via hierarchical and collaborative sparse representation | |
Qin et al. | Multi-resolution space-attended residual dense network for single image super-resolution | |
Wang et al. | An improved coupled dictionary and multi-norm constraint fusion method for CT/MR medical images |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180406 |
|
RJ01 | Rejection of invention patent application after publication |