CN106447610A - Image reconstruction method and image reconstruction device - Google Patents
Image reconstruction method and image reconstruction device Download PDFInfo
- Publication number
- CN106447610A CN106447610A CN201610789828.9A CN201610789828A CN106447610A CN 106447610 A CN106447610 A CN 106447610A CN 201610789828 A CN201610789828 A CN 201610789828A CN 106447610 A CN106447610 A CN 106447610A
- Authority
- CN
- China
- Prior art keywords
- image
- reconstructed
- image block
- dictionary
- resolution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000012549 training Methods 0.000 claims abstract description 99
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 17
- 230000001143 conditioned effect Effects 0.000 claims description 24
- 230000003321 amplification Effects 0.000 claims description 15
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 15
- 230000015572 biosynthetic process Effects 0.000 claims description 7
- 238000003786 synthesis reaction Methods 0.000 claims description 7
- 239000000203 mixture Substances 0.000 claims description 6
- 230000002194 synthesizing effect Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 abstract description 22
- 230000000694 effects Effects 0.000 abstract description 17
- 239000002355 dual-layer Substances 0.000 abstract 2
- 230000006870 function Effects 0.000 description 11
- 238000004891 communication Methods 0.000 description 9
- RKMGAJGJIURJSJ-UHFFFAOYSA-N 2,2,6,6-Tetramethylpiperidine Substances CC1(C)CCCC(C)(C)N1 RKMGAJGJIURJSJ-UHFFFAOYSA-N 0.000 description 5
- 238000003384 imaging method Methods 0.000 description 5
- 230000004087 circulation Effects 0.000 description 4
- 230000005236 sound signal Effects 0.000 description 4
- RINRSJBJOGCGBE-UHFFFAOYSA-N 3,3,5,6-tetramethyl-2h-pyrazine Chemical compound CC1=NCC(C)(C)N=C1C RINRSJBJOGCGBE-UHFFFAOYSA-N 0.000 description 3
- 238000001816 cooling Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 230000007850 degeneration Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- KLDZYURQCUYZBL-UHFFFAOYSA-N 2-[3-[(2-hydroxyphenyl)methylideneamino]propyliminomethyl]phenol Chemical compound OC1=CC=CC=C1C=NCCCN=CC1=CC=CC=C1O KLDZYURQCUYZBL-UHFFFAOYSA-N 0.000 description 1
- 230000003416 augmentation Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 201000001098 delayed sleep phase syndrome Diseases 0.000 description 1
- 208000033921 delayed sleep phase type circadian rhythm sleep disease Diseases 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 239000012634 fragment Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008929 regeneration Effects 0.000 description 1
- 238000011069 regeneration method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/77—Retouching; Inpainting; Scratch removal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an image reconstruction method and an image reconstruction device. The image reconstruction method comprises steps that an to-be-reconstructed image is acquired; the to-be-reconstructed image is divided into at least one to-be-reconstructed image block according to preset sizes; gray change conditions of the at least one to-be-reconstructed image block are determined; when the gray change conditions of the to-be-reconstructed image block satisfy first pre-set conditions, the to-be-reconstructed image block is reconstructed according to a dual-layer dictionary after dictionary training; when the gray change conditions of the to-be-reconstructed image block satisfy second pre-set conditions, the o-be-reconstructed image block is reconstructed through an iteration algorithm based on singular value decomposition. The method is advantaged in that when high resolution reconstruction for the to-be-reconstructed image is carried out, for the to-be-reconstructed image block with smooth gray change, the dual-layer dictionary after training is employed to acquire a projection matrix, for the to-be-reconstructed image block with relatively rich edge details, the iteration algorithm based on SVD decomposition is employed to reconstruct the corresponding high resolution image block, so the relatively good reconstruction effect is acquired, and the reconstruction time can be saved.
Description
Technical field
It relates to image procossing, in particular it relates to a kind of image rebuilding method and device.
Background technology
Image describes means as a kind of visual information, is one of main source that people obtain information.And weigh
The key index of piece image is exactly image resolution ratio.Image resolution ratio refers to the detailed information contained by piece image, and details is believed
Breath is more, and resolution is higher, and the information now obtaining from image is also more.But due to being subject in the actual image obtaining
The restriction of imaging device and the impact of surrounding, can make to actually obtain image presence degeneration it is impossible to meet the need of practical application
Ask.
The super-resolution reconstruction of image is a kind of technology of the resolution improving image.In correlation technique, image high score
Resolution is rebuild mainly using the method based on frequency domain or the method based on spatial domain.However, carrying out image high score using these methods
When resolution is rebuild, not only reconstruction effect is bad, and reconstruction time is longer.
Therefore, prior art existing defects, need to improve.
Content of the invention
The purpose of the disclosure is to provide a kind of image rebuilding method and device, efficiently to realize the high-resolution weight of image
Build.
To achieve these goals, in a first aspect, the disclosure provides a kind of image rebuilding method, including:
Obtain image to be reconstructed;
Described image to be reconstructed is divided at least one image block to be reconstructed according to default size;
Determine the grey scale change situation of described image block to be reconstructed;
When described image block to be reconstructed grey scale change situation meet first pre-conditioned when, according to double-deck dictionary training
Dictionary is rebuild to described image block to be reconstructed;
When the grey scale change situation of described image block to be reconstructed meet second pre-conditioned when, based on singular value decomposition repeatedly
For algorithm, described image block to be reconstructed is rebuild.
Optionally, the step of the described grey scale change situation determining described image block to be reconstructed includes:
Obtain the variance yields of the gray value of described image block to be reconstructed;
Described variance yields and predetermined threshold value are compared, to determine described grey scale change situation;
When described variance yields are less than described predetermined threshold value, the grey scale change situation satisfaction of described image block to be reconstructed is described
First is pre-conditioned;
When described variance yields are more than described predetermined threshold value, the grey scale change situation satisfaction of described image block to be reconstructed is described
Second is pre-conditioned.
Optionally, the step bag that the described dictionary according to double-deck dictionary training is rebuild to described image block to be reconstructed
Include:
Determine training storehouse, obtain training data;Described training storehouse includes:The original high-resolution image of predetermined number and according to
The low resolution image that described original high-resolution image obtains;
According to described training data, training obtains main dictionary;
According to described main dictionary, described training data and described low-resolution image, obtain residual dictionary;
According to described main dictionary, first stage reconstruction is carried out to described image block to be reconstructed;
The image being obtained according to described residual dictionary and the reconstruction of described first stage, carries out second stage and rebuilds to obtain
State the high-definition picture of image block to be reconstructed.
Optionally, described training data includes:The high-frequency information composition being extracted using described original high-resolution image
The low resolution interpolation amplification image that described low-resolution image carries out obtaining after interpolation amplification is entered by image block characteristics vector sum
The characteristics of image block eigenvector obtaining after row filtering.
Optionally, methods described also includes:
The high-definition picture obtaining after each image block reconstruction to be reconstructed is carried out synthesizing and obtains described image to be reconstructed
High-definition picture.
Second aspect, provides a kind of equipment for reconstructing image, including:
Acquisition module, for obtaining image to be reconstructed;
Image block module, for being divided at least one image to be reconstructed by described image to be reconstructed according to default size
Block;
Grey scale change determining module, for determining the grey scale change situation of described image block to be reconstructed;
First reconstruction module, for when described image block to be reconstructed grey scale change situation meet first pre-conditioned when,
Dictionary according to double-deck dictionary training is rebuild to described image block to be reconstructed;
Second reconstruction module, for when described image block to be reconstructed grey scale change situation meet second pre-conditioned when,
Iterative algorithm based on singular value decomposition is rebuild to described image block to be reconstructed.
Optionally, described grey scale change determining module includes:
Variance acquisition submodule, for obtaining the variance yields of the gray value of described image block to be reconstructed;
Comparison sub-module, for being compared described variance yields and predetermined threshold value, to determine described grey scale change situation;
When described variance yields are less than described predetermined threshold value, the grey scale change situation satisfaction of described image block to be reconstructed is described
First is pre-conditioned;
When described variance yields are more than described predetermined threshold value, the grey scale change situation satisfaction of described image block to be reconstructed is described
Second is pre-conditioned.
Optionally, described first reconstruction module includes:
Training data acquisition submodule, for determining training storehouse, obtains training data;Described training storehouse includes:Present count
The original high-resolution image of amount and the low resolution image being obtained according to described original high-resolution image;
Main dictionary acquisition submodule, for according to described training data, training obtains main dictionary;
Residual dictionary acquisition submodule, for according to described main dictionary, described training data and described low-resolution image,
Obtain residual dictionary;
First stage rebuilds submodule, for according to described main dictionary, carrying out the first stage to described image block to be reconstructed
Rebuild;
Second stage rebuilds submodule, for the image being obtained according to described residual dictionary and the reconstruction of described first stage,
Carry out second stage to rebuild to obtain the high-definition picture of described image block to be reconstructed.
Optionally, described training data includes:The high-frequency information composition being extracted using described original high-resolution image
The low resolution interpolation amplification image that described low-resolution image carries out obtaining after interpolation amplification is entered by image block characteristics vector sum
The characteristics of image block eigenvector obtaining after row filtering.
Optionally, described device also includes:
Synthesis module, the high-definition picture for obtaining after rebuilding each image block to be reconstructed carry out synthesis obtain described
The high-definition picture of image to be reconstructed.
By technique scheme, when treating reconstruction image and carrying out super-resolution reconstruction, gentle for grey scale change
The projection matrix that image block to be reconstructed is tried to achieve using the double-deck dictionary of training, to project to high-definition picture space, thus
Obtain corresponding high-definition picture block, reduce image reconstruction times;For edge details, more rich image block to be reconstructed, adopts
Rebuild corresponding high-definition picture block with the iterative algorithm decomposing based on SVD, reach preferable reconstruction effect.Both can obtain relatively
Good reconstruction effect, also can save reconstruction time.
Other feature and advantage of the disclosure will be described in detail in subsequent specific embodiment part.
Brief description
Accompanying drawing is used to provide further understanding of the disclosure, and constitutes the part of description, with following tool
Body embodiment is used for explaining the disclosure together, but does not constitute restriction of this disclosure.In the accompanying drawings:
Fig. 1 is the schematic flow sheet of the image rebuilding method of the disclosure one embodiment;
Fig. 2 is the high-definition picture of the dictionary acquisition image block by double-deck dictionary training of the disclosure one embodiment
Schematic flow sheet;
Fig. 3 is the double-deck dictionary training schematic flow sheet of the disclosure one embodiment;
Fig. 4 is the double-deck dictionary reconstruction image schematic flow sheet of the disclosure one embodiment;
Fig. 5 is the schematic flow sheet of the solution p norm minimum of the disclosure one embodiment;
Fig. 6 is the schematic flow sheet of the cooling algorithm of the disclosure one embodiment;
Fig. 7 is the image rebuilding method schematic flow sheet of the disclosure one embodiment;
Fig. 8 a- Fig. 8 c is to be illustrated using the effect that the image rebuilding method of the embodiment of the present disclosure carries out full resolution pricture reconstruction
Figure;
Fig. 9 is the structural representation of the equipment for reconstructing image that the disclosure one is implemented;
Figure 10 is a kind of block diagram of equipment for reconstructing image for terminal according to an exemplary embodiment for the common disclosure.
Specific embodiment
It is described in detail below in conjunction with accompanying drawing specific embodiment of this disclosure.It should be appreciated that this place is retouched
The specific embodiment stated is merely to illustrate and explains the disclosure, is not limited to the disclosure.
Schematic flow sheet referring to the image rebuilding method for the disclosure one embodiment for the Fig. 1.
In step slo, obtain image to be reconstructed.
In embodiment of the disclosure, image to be reconstructed can for storage image (for example, after shooting storage image, or
By image of Network Capture etc.), shoot image obtaining etc. by capture apparatus in real time.
In step s 11, image to be reconstructed is divided at least one image block to be reconstructed according to default size.
The default big I of image block pre-sets.In one embodiment, the reconstruction effect according to image determines figure
Size as piecemeal.In practice, the too small image block that can result in the need for of image block is excessive, leads to the reconstruction quality of image to drop
Low.Image block is excessive, then the gray scale situation according to image block different method for reconstructing can not be selected to carry out image reconstruction well, will
Accuracy can be reduced.Thus, in one embodiment, can by image block default be sized to 32 pixel × 32 pixels or
64 pixel × 64 pixels.
In step s 12, determine the grey scale change situation of image block to be reconstructed.The grey scale change situation of image block to be reconstructed
Can be determined according to the variance yields of the gray scale of image block to be reconstructed.Also can be by the side of the gray value of pixel in statistical picture block
Formula determines.
In step s 13, when image block to be reconstructed grey scale change situation meet first pre-conditioned when, according to double-deck word
The dictionary of allusion quotation training is treated reconstruction image block and is rebuild;
When image block to be reconstructed grey scale change situation meet second pre-conditioned when, based on singular value decomposition iteration calculate
Method is treated reconstruction image block and is rebuild.
The embodiment of the present disclosure, according to the grey scale change situation of image block, obtains the high-resolution of image block in different ways
Image.The grey scale change of image block is gentle, for example, when the edge detail information of image block is less, its grey scale change situation meets
First is pre-conditioned;The grey scale change of image block is fast, for example, when the edge detail information of image block is relatively enriched, then its gray scale becomes
It is pre-conditioned that change situation meets second.
In one embodiment, the variance yields of the gray value according to image block and predetermined threshold value σ2Determine the ash of this image block
Degree situation of change is that to meet first pre-conditioned or second pre-conditioned.Preset when the variance yields of the gray value of image block are less than
Threshold value σ2When it is determined that the gray scale of this image block smooth, change slow, this image block is obtained according to the dictionary of double-deck dictionary training
High-definition picture.When the variance yields of the gray value of image block are more than predetermined threshold value σ2When it is determined that the details of this image block
Information is abundanter, carries out weight based on the iterative algorithm of singular value decomposition (Singular Value Decomposition, SVD)
Build, obtain the high-definition picture of this image block.
Obtained according to the dictionary of double-deck dictionary training this image block high-definition picture when, image block to be reconstructed is multiplied by
Projection matrix, image block to be reconstructed is projected to high-definition picture space, obtains corresponding high-definition picture block.
In one embodiment, the image rebuilding method of the embodiment of the present disclosure also includes:Each image block to be reconstructed is rebuild
The high-definition picture obtaining afterwards carries out synthesizing the high-definition picture obtaining described image to be reconstructed.
According to the grey scale change situation of each image block to be reconstructed, obtain the height of each image block to be reconstructed in different ways
Then each high-definition picture block is carried out synthesizing the high-definition picture that can get image to be reconstructed by image in different resolution.
Here synthesis, and the piecemeal of step S11 is corresponding.For example, carry out piecemeal by the way of 32 pixel × 32 pixels, then
The full resolution pricture of the image block of these 32 pixel × 32 pixels is sequentially synthesized the high-resolution that can get image to be reconstructed
Image.
Thus, by the image rebuilding method of the embodiment of the present disclosure, when treating reconstruction image and carrying out super-resolution reconstruction,
The projection matrix tried to achieve using the double-deck dictionary of training for the gentle image block to be reconstructed of grey scale change, to project to high-resolution
In rate image space, thus obtaining corresponding high-definition picture block, reduce image reconstruction times;Abundanter for edge details
Image block to be reconstructed, using based on SVD decompose iterative algorithm rebuild corresponding high-definition picture block, reach preferably heavy
Build effect.Both preferable reconstruction effect can have been obtained, also can save reconstruction time.
In embodiment of the disclosure, in the embodiment of the present disclosure, when the grey scale change situation of the image block in image to be reconstructed
Meet second pre-conditioned when, obtain the high-definition picture of image block by the dictionary of double-deck dictionary training.Referring to Fig. 2, it is
The dictionary by double-deck dictionary training of the disclosure one embodiment obtains the schematic flow sheet of the high-definition picture of image block.
Included by the high-definition picture that the dictionary of double-deck dictionary training obtains image block:Training stage and reconstruction rank
Section.Following steps S21- step S23 belongs to the training stage, and step S24- step S25 belongs to phase of regeneration.
In the step s 21, determine training storehouse, obtain training data.
Referring to Fig. 3, storehouse is trained to include:One or more image.These images include original high-resolution image HORGWith low
Image in different resolution LLF, for example, training storehouse may include the original high-resolution image that 4 resolution are 512*512 and a resolution
Rate is the low-resolution image of 256*256.
Wherein, low-resolution image LLFCan be by high-definition picture HORGCarry out down-sampling and (for example, be spaced a point
Take a pixel to reduce the resolution of image), fuzzy (for example, fuzzy operation carried out using Gauss operator, changes pixel
Gray value to be to reduce the resolution of image) etc. degeneration obtain.
In one embodiment, in order to improve accuracy, the image in training storehouse includes the more rich image of marginal information.
The more rich image of marginal information, i.e. the big image of the grey scale change of image.
The acquisition of training data:
Respectively feature extraction is carried out to the image in training storehouse, and piecemeal processes and obtains training data
Wherein,It is the image block characteristics vector of the high-frequency information composition being extracted using original high-resolution image.It is by low resolution
The low resolution interpolation amplification image that rate image obtains after carrying out interpolation amplification, using the firstorder filter as shown in following formula (1)
Extract the image block characteristics vector that feature obtains with second order filter.
f1=[0,0,1,0,0, -1], f2=f1 T
f3=[1,0,0, -2,0,0,1], f4=f3 T(1)
In one embodiment, can by original high-resolution image withSubtract each other and obtain
I.e. training data includes:The image block characteristics vector of the high-frequency information composition being extracted using high-definition pictureWith
Low-resolution image is carried out the low resolution interpolation amplification image obtaining after interpolation amplification, using the single order filter shown in formula (1)
Ripple device and second order filter extract the image block characteristics vector that feature obtainsMain dictionary training high-definition picture in Fig. 3
HHFIt isMain dictionary training low-resolution image HLFIt is
In step S22, according to training data, training obtains main dictionary.Main dictionary includes main low frequency dictionary LMDWith main height
Frequency dictionary HMD.
Wherein, for the image block characteristics vector in training dataDictionary training is carried out using K-SVD algorithm, obtains
To main low frequency dictionary LMD, that is,:
In formula (2), { qk}kIt is vectorial for rarefaction representation, | | | |0Represent L0Norm.
Based on the image block characteristics vector in training dataDue toThen main high frequency dictionary can pass through
Optimize following formula (3) to obtain.
Wherein, matrix PhRepresent respectively with Q{ qk}kSet, that is,Q={ qk}k, using generalized inverse
The method of matrix solves.After optimization, main high frequency dictionary HMDAs shown in formula (4).
HMD=PhQ+=PhQT(QQT)-1(4)
In step S23, according to main dictionary, training data and low-resolution image, obtain residual dictionary.
The reconstruction effect of high-definition picture in embodiment of the disclosure, can be strengthened by remaining the training of dictionary.
Using main dictionary { LMD,HMDTo main dictionary training high-definition picture HHF(i.e.) carry out image reconstruction, obtain weight
Build main high frequency imaging HMHF, and by the image obtaining HMHFRemain the input of dictionary training as the second layer, with main dictionary training
Journey is similar to, and obtains remaining dictionary { LRD,HRD}.Referring to Fig. 3, by image HMHFWith main dictionary training high-definition picture HHF(i.e.)
Subtract each other and obtain remaining dictionary training high-definition picture HRHF.By image HMHFWith low-resolution image LLFAddition obtains remaining dictionary
Training low-resolution image HTMP.
Thus, according to residual dictionary training high-definition picture HRHFWith residual dictionary training low-resolution image HTMPCarry out
Training obtains remaining dictionary { LRD,HRD}.It should be understood that can be trained using the method shown in formula 2 and formula 3 during training.
In step s 24, according to main dictionary, treat reconstruction image and carry out first stage reconstruction.
Referring to Fig. 4, image to be reconstructed is low-resolution image LINPUT.When carrying out first stage image reconstruction, to input
Low-resolution image to be reconstructed carry out interpolation amplification and be filtered using wave filter, and extract the feature of overlapping image block
Set(i.e. main dictionary training low-resolution image H in Fig. 6LF).Wave filter can be using the wave filter shown in formula (1).Carry
Take the characteristic set of overlapping image block, be the error in order to prevent edge pixel.General overlaid pixel is The more the better, but examines
Consider operand, overlapping pixel can be arranged in certain scope, for example, overlapping block of pixels is set to 9*9 picture
Element.
Using main dictionary { LMD,HMD, according to formula (5), to main dictionary training low-resolution image HLFCarry out collaborative expression,
To obtain rebuilding main high frequency imaging HMHF.
Wherein, DlFor low-resolution dictionary, it is the L in this main dictionaryMD, parameter lambda be used for equation of equilibrium emulation item and
Sparse item, y is main dictionary training low-resolution image HLF.
Can be obtained according to formula (5):
α=(Dl TDl+λI)-1Dl Ty (6)
In one embodiment, sparse coefficient and high-resolution dictionary directly can be mapped to height by high-definition picture block
In resolution space, that is,:
X=Dhα (7)
Wherein, x is the high-definition picture of output, DhIt is in low-resolution dictionary corresponding high-resolution dictionary, as
Corresponding H in main dictionaryMD.
Be can get according to formula (6) and (7), the high-definition picture x of output can be represented by formula (8).
X=Dh(Dl TDl+λI)-1Dl Ty (8)
Referring to Fig. 4, according to the main high frequency imaging H of reconstructionMHFWith main dictionary training low-resolution image HLFAddition obtains high-resolution
Rate image, it rebuilds the high-definition picture of output for the first stage.
Low-resolution image be can get to the projection matrix P of high-definition picture according to formula (8)G, it is by formula (9) table
Show.
PG=Dh(Dl TDl+λI)-1Dl T(9)
Wherein, PGFor projection matrix.H in main dictionaryMDAnd LMDIt is that training obtains, I is unit matrix.Thus, project
Matrix can be with calculated off line.Thus, directly by low resolution characteristic vector pl kIt is multiplied by this projection matrix, obtain corresponding high-resolution
Rate image block, thus greatly reduce the high resolution image reconstruction time.
In step s 25, the full resolution pricture being obtained according to residual dictionary and first stage reconstruction, treats reconstruction image and enters
Row second stage is rebuild.
Referring to Fig. 4, according to the main high frequency imaging H of reconstructionMHFWith main dictionary training low-resolution image HLFAddition obtains remaining word
Allusion quotation trains low-resolution image HTMP.
First stage is rebuild the input that the image obtaining is rebuild as second stage.Second stage process of reconstruction and first
Stage process of reconstruction is similar to, you can arrives the formula shown in formula 8 according to formula 5, is only that main dictionary is changed to residual dictionary, by main dictionary
Training low-resolution image HLFReplace with residual dictionary training low-resolution image HTMP.Referring to Fig. 4, according to residual dictionary { LRD,
HRDAnd residual dictionary training low-resolution image HTMPObtain remaining dictionary training high-definition picture HRHF, then will remain word
Allusion quotation trains high-definition picture HRHFWith residual dictionary training low-resolution image HTMPAddition obtains full resolution pricture HEST, this height
Image in different resolution HESTIt is the full resolution pricture of image to be reconstructed.
Projection matrix P asked for by the dictionary (main dictionary and residual dictionary) being obtained using the learning method training of double-deck dictionaryG,
Low-resolution image to be reconstructed is projected directly into high resolution graphics image space according to projection matrix and obtains high-definition picture,
There is reconstruction time short, the simple effect of process of reconstruction.
In embodiment of the disclosure, in the embodiment of the present disclosure, when the grey scale change situation of the image block in image to be reconstructed
Meet second pre-conditioned when, based on SVD decompose iterative algorithm carry out image reconstruction.
Singular value decomposition (SVD), as a kind of important matrix disassembling method, can be compressed using the non-full rank of image
Mass data.Treat reconstruction image x to represent with matrix X, its SVD decomposition is defined as:
X=USXV′ (10)
Wherein, U and V is normal orthogonal basic matrix, SXFor a pair of diagonal matrices, formula (10) be one effectively dilute to X
Thin method for expressing, can be by giving up the relatively small singular value of some numerical value Lai approximate.Optimal problem can be exchanged into be made to treat weight
The order building image reaches minimum.
minimize:rank(X)s.t.||y-Fx||2< ε (11)
Wherein, X is the matrix form of image x to be reconstructed, and F is then the observing matrix of image x to be reconstructed.Observing matrix,
Can be obtained by treating reconstruction image and carrying out the operations such as fuzzy or down-sampling.Formula (11) is represented by:
min||X||*s.t.||y-Fx||2≤ε (12)
Wherein, | | X | |*Nuclear norm for X, is defined as singular value sum, due to usual, lpNorm (0<p<1) l can be compared1
Norm obtains preferable effect.Similar, the embodiment of the present disclosure adopts lpNorm is replacing l1Norm is estimated, that is,:
Now, the optimization problem of formula (13) is converted to:
min||X||p, 0 < p < 1s.t. | | y-Fx | |2≤ε (14)
Object function is set as J (x) during solution:
J (x)=min | | y-Fx | |2+λ||X||p(15)
Wherein, parameter lambda is sparse item and fidelity item in balancing objective function.J (x) is minimized and then can obtain optimal solution:
Iteration threshold algorithm is a kind of simple and effective rarefaction representation image rebuilding method, and this algorithm can directly eliminate
The artifact brought by K space down-sampling, has reconstruction effect good, the few feature of free parameter.
Referring to Fig. 5 for the solution p norm minimum of the disclosure one embodiment schematic flow sheet.The embodiment of the present disclosure adopts
A kind of First Order Iterative algorithm being decomposed based on SVD of Majumdar, solves p norm minimum to formula (14), it includes following
Step:
In step s 51, if xk=xk-1+FT(y-Fxk-1).
In step S52, to matrix xkCarry out changing and be modelled as matrix X to be rebuildk.
In step S53, to XkCarry out SVD decomposition, i.e. Xk=U ∑ VT.
In step S54, carry out soft-threshold using the singular value that step S653 is decomposed and solve the singular value updating gained
In step S55, according to the singular value of step S54 acquisition, solve Xk+1, Xk+1=U ∑ VT, by Xk+1Vector turns to
xk+1.
In step S56, update iterationses, even k=k+1, and return to step S51.
For solving the optimization problem of formula (14), the embodiment of the present disclosure adopts Cooling algorithm.This algorithm is parameter ε and λ builds
Stand contact, this algorithm includes two circulations, and the target of interior circulation is when λ is fixing, solves formula (15) minima, its termination
Function is:
When terminating function less than iteration ends value, interior circulation terminates.Outer circulation operation is then mainly according to decay factor
DecFac, reduces the value of λ.As no longer satisfaction | | y-Fx | |2> ε, outer loop end.
Referring to Fig. 6, the cooling algorithm of the embodiment of the present disclosure comprises the following steps:
In step S61, initialized.Sparse coefficient vector is made to be x0=0, λ < max (FTX), iterationses t=1,
Set iteration ends value Tol and ε respectively, and decay factor DecFac.
The object function of current iteration in step S62, is calculated using formula (15)And
According to the above-mentioned First Order Iterative algorithmic minimizing J based on SVD decompositionk.
In step S63, it is that next iteration calculates new object function
In step S64, update iterationses t=t+1.
In step S65, compare the size terminating function and iteration ends value Tol, if function is more than or equal to Tol
When, it is back to step S63, otherwise continue executing with.
In step S66, according to initialized decay factor DecFac, reduce the size of λ.
In step S67, judge | | y-Fx | |2With the size of ε, if | | y-Fx | |2> ε, return to step S63, otherwise stop
Only iteration.
Good image reconstruction effect is had for undersized image augmentation based on the iterative algorithm that SVD decomposes.
Referring to Fig. 7, the image rebuilding method of the embodiment of the present disclosure, to low-resolution image to be reconstructed according to the figure setting
As block piecemeal size carries out piecemeal process, try to achieve the Variance feature value of each image block.Variance by each image block tried to achieve
Value is made a decision with the image variance threshold value setting.If variance yields are less than threshold value, judge that this image block gray scale smooths, change is delayed
Slowly, directly low-resolution image block to be reconstructed is multiplied by projection matrix and projects and obtain corresponding high score to high-definition picture space
Resolution image block (adopts double-deck dictionary method), conversely, variance yields are more than threshold value then it is assumed that this image block detailed information is richer
Richness, using the iterative algorithm reconstruction image based on singular value decomposition (Singular Value Decomposition, SVD).By
Fast in the reconstruction time of double-deck dictionary method, and the method reconstruction effect based on singular value decomposition is good, therefore, the embodiment of the present disclosure
Image rebuilding method can reach reduce algorithm required time purpose can the Equilibrium fitting time and rebuild effect, make image high-resolution
The performance that rate is rebuild reaches the result of optimum.
It is the effect carrying out full resolution pricture reconstruction using the image rebuilding method of the embodiment of the present disclosure referring to Fig. 8 a- Fig. 8 c
Schematic diagram.Wherein, Fig. 8 a is original high-resolution image (as training storehouse), and Fig. 8 b is low-resolution image, and Fig. 8 c is to rebuild
High-definition picture afterwards.
The structural representation of the equipment for reconstructing image implemented for the disclosure referring to Fig. 9.This equipment for reconstructing image 900 includes:
Acquisition module 901, for obtaining image to be reconstructed;
Image block module 902, to be reconstructed for described image to be reconstructed is divided at least one according to default size
Image block;
Grey scale change determining module 903 is used for determining the grey scale change situation of described image block to be reconstructed;
First reconstruction module 904, pre-conditioned for the grey scale change situation satisfaction first when described image block to be reconstructed
When, the dictionary according to double-deck dictionary training is rebuild to described image block to be reconstructed;
Second reconstruction module 905, pre-conditioned for the grey scale change situation satisfaction second when described image block to be reconstructed
When, the iterative algorithm based on singular value decomposition is rebuild to described image block to be reconstructed.
In one embodiment, grey scale change determining module 903 includes:
Variance acquisition submodule 9031, for obtaining the variance yields of the gray value of described image block to be reconstructed;
Comparison sub-module 9032, for being compared described variance yields and predetermined threshold value, to determine described grey scale change
Situation;
When described variance yields are less than described predetermined threshold value, the grey scale change situation satisfaction of described image block to be reconstructed is described
First is pre-conditioned;
When described variance yields are more than described predetermined threshold value, the grey scale change situation satisfaction of described image block to be reconstructed is described
Second is pre-conditioned.
In one embodiment, the first reconstruction module 904 includes:
Training data acquisition submodule 9041, for determining training storehouse, obtains training data;Described training storehouse includes:In advance
If the original high-resolution image of quantity and the low resolution image being obtained according to described original high-resolution image;
Main dictionary acquisition submodule 9042, for according to described training data, training obtains main dictionary;
Residual dictionary acquisition submodule 9043, for according to described main dictionary, described training data and described low resolution
Image, obtains residual dictionary;
First stage rebuilds submodule 9044, for according to described main dictionary, carrying out first to described image block to be reconstructed
Stage rebuilds;
Second stage rebuilds submodule 9045, for the figure being obtained according to described residual dictionary and the reconstruction of described first stage
Picture, carries out second stage and rebuilds to obtain the high-definition picture of described image block to be reconstructed.
In one embodiment, training data includes:The high-frequency information group extracted using described original high-resolution image
Described low-resolution image is carried out the low resolution interpolation amplification figure obtaining after interpolation amplification by the image block characteristics vector sum becoming
As the characteristics of image block eigenvector obtaining after being filtered.
In one embodiment, device 900 also includes:
Synthesis module 906, the high-definition picture for obtaining after rebuilding each image block to be reconstructed carries out synthesis and obtains
The high-definition picture of described image to be reconstructed.
With regard to the device in above-described embodiment, wherein the concrete mode of modules execution operation is in relevant the method
Embodiment in be described in detail, explanation will be not set forth in detail herein.
Figure 10 is a kind of block diagram of the equipment for reconstructing image 100 for terminal according to an exemplary embodiment, should
Device 100 can be mobile terminal.As shown in Figure 10, this device 100 can include:Processor 1001, memorizer 1002, many matchmakers
Body assembly 1003, input/output (I/O) interface 1004, communication component 1005 and video capture assembly 1006.
Wherein, processor 1001 is used for controlling the integrated operation of this device 100, to complete the above-mentioned image for terminal
All or part of step in method for reconstructing.Memorizer 1002 is used for storing various types of data to support in this device 100
Operation, these data for example can include on this device 100 operation any application program or method instruction,
And the data that application program is related, such as contact data, the message of transmitting-receiving, picture, audio frequency, video etc..This memorizer
1002 can be realized by any kind of volatibility or non-volatile memory device or combinations thereof, and for example static random is deposited
Access to memory (Static Random Access Memory, abbreviation SRAM), Electrically Erasable Read Only Memory
(Electrically Erasable Programmable Read-Only Memory, abbreviation EEPROM), erasable programmable
Read only memory (Erasable Programmable Read-Only Memory, abbreviation EPROM), programmable read only memory
(Programmable Read-Only Memory, abbreviation PROM), and read only memory (Read-Only Memory, referred to as
ROM), magnetic memory, flash memory, disk or CD.
Multimedia groupware 1003 can include screen and audio-frequency assembly.Wherein screen can be for example touch screen, audio group
Part is used for output and/or input audio signal.For example, audio-frequency assembly can include a mike, and mike is used for receiving outward
Portion's audio signal.The audio signal being received can be further stored in memorizer 1002 or pass through communication component 1005
Send.Audio-frequency assembly also includes at least one speaker, for exports audio signal.I/O interface 1004 be processor 1001 and its
Interface is provided, other interface modules above-mentioned can be keyboard, mouse, button etc. between his interface module.These buttons can be
Virtual push button or entity button.
Communication component 1005 is used for carrying out wired or wireless communication between this device 100 and other equipment.Radio communication, example
As Wi-Fi, bluetooth, near-field communication (Near Field Communication, abbreviation NFC), 2G, 3G or 4G, or in them
The combination of one or more, this communication component 1005 therefore corresponding can include:Wi-Fi module, bluetooth module, NFC module.
Video capture assembly 1006 may include the modules such as photographic head, signal processing, for gathering image.
In embodiment of the disclosure, image to be reconstructed can be the image by video capture assembly 1006 shooting, collecting,
It is alternatively the image passing through that communication component 1005 obtains at the webserver or other terminal unit.
In one exemplary embodiment, device 100 can be by one or more application specific integrated circuits
(Application Specific Integrated Circuit, abbreviation ASIC), digital signal processor (Digital
Signal Processor, abbreviation DSP), digital signal processing appts (Digital Signal Processing Device,
Abbreviation DSPD), PLD (Programmable Logic Device, abbreviation PLD), field programmable gate array
(Field Programmable Gate Array, abbreviation FPGA), controller, microcontroller, microprocessor or other electronics unit
Part is realized, for executing the above-mentioned image rebuilding method for terminal.
In a further exemplary embodiment, additionally provide a kind of non-transitory computer-readable storage medium including instruction
Matter, for example, include the memorizer 1002 instructing, above-mentioned instruction can be executed above-mentioned to complete by the processor 10101 of device 100
Image rebuilding method for terminal.Illustratively, this non-transitorycomputer readable storage medium can be ROM, random access memory
Memorizer (Random Access Memory, abbreviation RAM), CD-ROM, tape, floppy disk and optical data storage devices etc..
The image rebuilding method of the embodiment of the present disclosure and device, according to the variance yields of image block, different grey scale change are not
With selecting different image rebuilding methods, the performance indications of the reconstruction time of synthetic image and image reconstruction effect, ensureing weight
In the case of building high-definition picture quality, can largely reduce the time needed for reconstruction image.
Any process described otherwise above or method description in flow chart or in embodiment of the disclosure can be by
It is interpreted as, represent the code of the executable instruction including one or more steps for realizing specific logical function or process
Module, fragment or part, and the scope of disclosure embodiment includes other realization, wherein can not press shown or
Discuss order, including according to involved function by substantially simultaneously in the way of or in the opposite order, carry out perform function, this should
Described in embodiment of the disclosure, those skilled in the art understand.
Describe the preferred implementation of the disclosure above in association with accompanying drawing in detail, but, the disclosure is not limited to above-mentioned reality
Apply the detail in mode, in the range of the technology design of the disclosure, multiple letters can be carried out with technical scheme of this disclosure
Monotropic type, these simple variant belong to the protection domain of the disclosure.
It is further to note that each particular technique feature described in above-mentioned specific embodiment, in not lance
In the case of shield, can be combined by any suitable means, in order to avoid unnecessary repetition, the disclosure to various can
The compound mode of energy no longer separately illustrates.
Additionally, combination in any can also be carried out between the various different embodiment of the disclosure, as long as it is without prejudice to this
Disclosed thought, it equally should be considered as disclosure disclosure of that.
Claims (10)
1. a kind of image rebuilding method is it is characterised in that include:
Obtain image to be reconstructed;
Described image to be reconstructed is divided at least one image block to be reconstructed according to default size;
Determine the grey scale change situation of described image block to be reconstructed;
When described image block to be reconstructed grey scale change situation meet first pre-conditioned when, according to the dictionary of double-deck dictionary training
Described image block to be reconstructed is rebuild;
When described image block to be reconstructed grey scale change situation meet second pre-conditioned when, based on singular value decomposition iteration calculate
Method is rebuild to described image block to be reconstructed.
2. method according to claim 1 is it is characterised in that the grey scale change feelings of the described image block to be reconstructed of described determination
The step of condition includes:
Obtain the variance yields of the gray value of described image block to be reconstructed;
Described variance yields and predetermined threshold value are compared, to determine described grey scale change situation;
When described variance yields are less than described predetermined threshold value, the grey scale change situation of described image block to be reconstructed meets described first
Pre-conditioned;
When described variance yields are more than described predetermined threshold value, the grey scale change situation of described image block to be reconstructed meets described second
Pre-conditioned.
3. method according to claim 1 is it is characterised in that the described dictionary according to double-deck dictionary training treats weight to described
Build the step that image block rebuild to include:
Determine training storehouse, obtain training data;Described training storehouse includes:The original high-resolution image of predetermined number and according to described
The low resolution image that original high-resolution image obtains;
According to described training data, training obtains main dictionary;
According to described main dictionary, described training data and described low-resolution image, obtain residual dictionary;
According to described main dictionary, first stage reconstruction is carried out to described image block to be reconstructed;
According to described residual dictionary and the image that obtains of reconstruction of described first stage, carry out second stage and rebuild to treat described in obtaining
The high-definition picture of reconstruction image block.
4. method according to claim 1 is it is characterised in that described training data includes:Using described original high-resolution
Described low-resolution image is carried out obtaining after interpolation amplification by the image block characteristics vector sum of the high-frequency information composition of rate image zooming-out
To low resolution interpolation amplification image be filtered after the characteristics of image block eigenvector that obtains.
5. method according to claim 1 is it is characterised in that methods described also includes:
The high-definition picture obtaining after each image block to be reconstructed is rebuild carries out synthesizing the high score obtaining described image to be reconstructed
Resolution image.
6. a kind of equipment for reconstructing image is it is characterised in that include:
Acquisition module, for obtaining image to be reconstructed;
Image block module, for being divided at least one image block to be reconstructed by described image to be reconstructed according to default size;
Grey scale change determining module, for determining the grey scale change situation of described image block to be reconstructed;
First reconstruction module, for when described image block to be reconstructed grey scale change situation meet first pre-conditioned when, according to
The dictionary of double-deck dictionary training is rebuild to described image block to be reconstructed;
Second reconstruction module, for when described image block to be reconstructed grey scale change situation meet second pre-conditioned when, be based on
The iterative algorithm of singular value decomposition is rebuild to described image block to be reconstructed.
7. device according to claim 6 is it is characterised in that described grey scale change determining module includes:
Variance acquisition submodule, for obtaining the variance yields of the gray value of described image block to be reconstructed;
Comparison sub-module, for being compared described variance yields and predetermined threshold value, to determine described grey scale change situation;
When described variance yields are less than described predetermined threshold value, the grey scale change situation of described image block to be reconstructed meets described first
Pre-conditioned;
When described variance yields are more than described predetermined threshold value, the grey scale change situation of described image block to be reconstructed meets described second
Pre-conditioned.
8. device according to claim 6 is it is characterised in that described first reconstruction module includes:
Training data acquisition submodule, for determining training storehouse, obtains training data;Described training storehouse includes:Predetermined number
Original high-resolution image and the low resolution image being obtained according to described original high-resolution image;
Main dictionary acquisition submodule, for according to described training data, training obtains main dictionary;
Residual dictionary acquisition submodule, for according to described main dictionary, described training data and described low-resolution image, obtaining
Residual dictionary;
First stage rebuilds submodule, for according to described main dictionary, carrying out first stage reconstruction to described image block to be reconstructed;
Second stage rebuilds submodule, for the image being obtained according to described residual dictionary and the reconstruction of described first stage, carries out
Second stage is rebuild to obtain the high-definition picture of described image block to be reconstructed.
9. device according to claim 6 is it is characterised in that described training data includes:Using described original high-resolution
Described low-resolution image is carried out obtaining after interpolation amplification by the image block characteristics vector sum of the high-frequency information composition of rate image zooming-out
To low resolution interpolation amplification image be filtered after the characteristics of image block eigenvector that obtains.
10. device according to claim 6 is it is characterised in that described device also includes:
Synthesis module, the high-definition picture for obtaining after rebuilding each image block to be reconstructed carry out synthesis obtain described in treat weight
Build the high-definition picture of image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610789828.9A CN106447610B (en) | 2016-08-31 | 2016-08-31 | Image rebuilding method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610789828.9A CN106447610B (en) | 2016-08-31 | 2016-08-31 | Image rebuilding method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106447610A true CN106447610A (en) | 2017-02-22 |
CN106447610B CN106447610B (en) | 2019-07-23 |
Family
ID=58164042
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610789828.9A Expired - Fee Related CN106447610B (en) | 2016-08-31 | 2016-08-31 | Image rebuilding method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106447610B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107123097A (en) * | 2017-04-26 | 2017-09-01 | 东北大学 | A kind of imaging method of the calculation matrix based on optimization |
CN109712209A (en) * | 2018-12-14 | 2019-05-03 | 深圳先进技术研究院 | The method for reconstructing of PET image, computer storage medium, computer equipment |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6304605B1 (en) * | 1994-09-13 | 2001-10-16 | Nokia Mobile Phones Ltd. | Video compressing method wherein the direction and location of contours within image blocks are defined using a binary picture of the block |
CN101980291A (en) * | 2010-11-03 | 2011-02-23 | 天津大学 | Random micro-displacement-based super-resolution image reconstruction method |
CN102222357A (en) * | 2010-04-15 | 2011-10-19 | 温州大学 | Foot-shaped three-dimensional surface reconstruction method based on image segmentation and grid subdivision |
CN103366340A (en) * | 2012-04-06 | 2013-10-23 | 索尼公司 | Image processing device and method and electronic device using image processing device and method |
CN105844590A (en) * | 2016-03-23 | 2016-08-10 | 武汉理工大学 | Image super-resolution reconstruction method and system based on sparse representation |
-
2016
- 2016-08-31 CN CN201610789828.9A patent/CN106447610B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6304605B1 (en) * | 1994-09-13 | 2001-10-16 | Nokia Mobile Phones Ltd. | Video compressing method wherein the direction and location of contours within image blocks are defined using a binary picture of the block |
CN102222357A (en) * | 2010-04-15 | 2011-10-19 | 温州大学 | Foot-shaped three-dimensional surface reconstruction method based on image segmentation and grid subdivision |
CN101980291A (en) * | 2010-11-03 | 2011-02-23 | 天津大学 | Random micro-displacement-based super-resolution image reconstruction method |
CN103366340A (en) * | 2012-04-06 | 2013-10-23 | 索尼公司 | Image processing device and method and electronic device using image processing device and method |
CN105844590A (en) * | 2016-03-23 | 2016-08-10 | 武汉理工大学 | Image super-resolution reconstruction method and system based on sparse representation |
Non-Patent Citations (4)
Title |
---|
YONG YIN,ET AL: "An Improved Super-resolution Image Reconstruction Algorithm", 《INTERNATIONAL JOURNAL OF SIGNAL PROCESSING,IMAGE PROCESSING AND PATTERN RECOGNITION》 * |
印勇等: "图像插值的改进自适应核回归方法", 《光电工程》 * |
印勇等: "采用CDD模型的自适应图像修复算法", 《重庆大学学报》 * |
印勇等: "采用曲率驱动的自适应图像修复算法", 《光电子·激光》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107123097A (en) * | 2017-04-26 | 2017-09-01 | 东北大学 | A kind of imaging method of the calculation matrix based on optimization |
CN107123097B (en) * | 2017-04-26 | 2019-08-16 | 东北大学 | A kind of imaging method of the calculation matrix based on optimization |
CN109712209A (en) * | 2018-12-14 | 2019-05-03 | 深圳先进技术研究院 | The method for reconstructing of PET image, computer storage medium, computer equipment |
CN109712209B (en) * | 2018-12-14 | 2022-09-20 | 深圳先进技术研究院 | PET image reconstruction method, computer storage medium, and computer device |
Also Published As
Publication number | Publication date |
---|---|
CN106447610B (en) | 2019-07-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US8805120B2 (en) | Image processing apparatus and method which generates reference data according to learning images | |
Jiang et al. | Unsupervised decomposition and correction network for low-light image enhancement | |
CN104376542B (en) | A kind of image enchancing method | |
US11153575B2 (en) | Electronic apparatus and control method thereof | |
CN109525859A (en) | Model training, image transmission, image processing method and relevant apparatus equipment | |
CN110324664A (en) | A kind of video neural network based mends the training method of frame method and its model | |
US11620480B2 (en) | Learning method, computer program, classifier, and generator | |
US20160070979A1 (en) | Method and Apparatus for Generating Sharp Image Based on Blurry Image | |
CN109409503A (en) | Training method, image conversion method, device, equipment and the medium of neural network | |
JPH05502534A (en) | Digital image denoising system and method | |
CN110222758A (en) | A kind of image processing method, device, equipment and storage medium | |
WO2020119581A1 (en) | Magnetic resonance parameter imaging method and apparatus, device and storage medium | |
US9569684B2 (en) | Image enhancement using self-examples and external examples | |
CN107464217B (en) | Image processing method and device | |
JP2011512086A (en) | Reduction of noise and / or flicker in video sequences using spatial and temporal processing | |
JP2018527687A (en) | Image processing system for reducing an image using a perceptual reduction method | |
CN103985085A (en) | Image super-resolution amplifying method and device | |
CN107784628A (en) | A kind of super-resolution implementation method based on reconstruction optimization and deep neural network | |
EP4075373A1 (en) | Image processing method and apparatus | |
CN110443764A (en) | Video repairing method, device and server | |
Chira et al. | Image super-resolution with deep variational autoencoders | |
CN106447610A (en) | Image reconstruction method and image reconstruction device | |
CN113222855A (en) | Image recovery method, device and equipment | |
US20080303951A1 (en) | Image processing apparatus, image processing method, and program | |
KR20210019835A (en) | Apparatus and method for generating super resolution inmage using orientation adaptive parallel neural networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190723 Termination date: 20200831 |
|
CF01 | Termination of patent right due to non-payment of annual fee |