CN109919871A - Fuzzy core estimation method based on image and fuzzy core mixed constraints - Google Patents
Fuzzy core estimation method based on image and fuzzy core mixed constraints Download PDFInfo
- Publication number
- CN109919871A CN109919871A CN201910164437.1A CN201910164437A CN109919871A CN 109919871 A CN109919871 A CN 109919871A CN 201910164437 A CN201910164437 A CN 201910164437A CN 109919871 A CN109919871 A CN 109919871A
- Authority
- CN
- China
- Prior art keywords
- image
- fuzzy core
- fuzzy
- dark
- core
- 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
- 238000000034 method Methods 0.000 title claims abstract description 48
- 239000011159 matrix material Substances 0.000 claims description 4
- 230000002401 inhibitory effect Effects 0.000 claims description 3
- 238000012804 iterative process Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 238000006467 substitution reaction Methods 0.000 claims description 2
- 230000017105 transposition Effects 0.000 claims description 2
- 241000208340 Araliaceae Species 0.000 claims 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims 1
- 235000003140 Panax quinquefolius Nutrition 0.000 claims 1
- 235000008434 ginseng Nutrition 0.000 claims 1
- 238000007796 conventional method Methods 0.000 abstract description 2
- 238000011084 recovery Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000015572 biosynthetic process Effects 0.000 description 2
- 238000012512 characterization method Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000003786 synthesis reaction Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Landscapes
- Image Processing (AREA)
Abstract
The present invention proposes a kind of fuzzy core estimation method based on image and fuzzy core mixed constraints, including using image L in fuzzy kernel estimates model0Mixed constraints and fuzzy core L2Mixed constraints utilize image gradient L0Regular terms protects image border, image gradient dark L as global priori0Regular terms protects clear image dark sparsity as local priori；L is applied to fuzzy core and its gradient2Regular terms protects the sparsity and continuity of fuzzy core；Then convexity approximation and linear approximation are introduced in solution procedure to solve L0Regular terms and dark constrain the non-convex nonlinear problem of bring, are solved based on fuzzy kernel estimates model realization of the half secondary split algorithm to proposition.The present invention can estimate accurate fuzzy core for blurred picture, especially when handling abundant Scene Blur image, the fuzzy core estimated by the method for the present invention it is resilient go out the image more relatively sharp than conventional method.
Description
Technical field
The invention belongs to image processing methods, in particular to fuzzy core estimation method.
Background technique
Image restoration technology is widely used in various fields.According to fuzzy core whether it is known that can be by image restoration point
For two kinds of fundamental types: image known to fuzzy core is non-blind to restore and the unknown blindly restoring image of fuzzy core.It is non-compared to image
Blind recovery, blindly restoring image is more complicated, is more of practical significance.Using prior blur identification frame, i.e., by blindly restoring image process point
Two stages are restored for fuzzy kernel estimates and clear image, it is possible to reduce the calculation amount of image restoration process is easy to answer in practice
With obtaining the extensive concern of researcher in recent years.Under priori framework of identification, fuzzy kernel estimates are the passes of blindly restoring image
Whether key, estimation accurately determine whether image restoration effect is good.Regular terms is reasonably selected, before being accurate ambiguous estimation core
It mentions.
Most methods are conceived to the construction of image regular terms at present, focus on the figures such as strong edge is extracted, image border is protected
As global priori, the recovery to single Scene Blur image is preferably realized.For fuzzy core bound term, most methods are used
Single its sparsity of fuzzy core confining guard.And scene image is enriched as natural image, facial image, text image,
It is each provided with different priori knowledges --- natural image gradient has heavytailed distribution feature, face blurred picture blur margin
Clear, text image has double-colored tonality.Global prior-constrained item is only used only, ignores image local feature, is unfavorable for abundant
Scene Blur image restoration.Meanwhile method lacks to the successional constraint of fuzzy core at present, the fuzzy core of estimation is ineffective,
To influence the result of Subsequent rehabilitation image.
Pass through analysis, when handling abundant Scene Blur image, existing processing technique are as follows: 1) to image apply pixel domain and
The global restriction of gradient field ignores local priori, leads to the fuzzy kernel estimates of inaccuracy；2) single constraint is applied to fuzzy core,
Only consider fuzzy core sparsity and ignore continuity, leads to the fuzzy kernel estimates of inaccuracy.
Summary of the invention
For the above the deficiencies in the prior art, the present invention proposes a kind of fuzzy core based on image and fuzzy core mixed constraints
Estimation method, main purpose are, using image overall and local priori is combined, to consider fuzzy core for abundant Scene Blur image
The fuzzy kernel estimates model of sparsity and continuity priori, accurately estimates fuzzy core from blindly restoring image.
Technical scheme is as follows:
A kind of fuzzy core estimation method based on image and fuzzy core mixed constraints, comprising: introduce image gradient and its dark
The sparse prior and fuzzy core in channel mix prior information, construct corresponding image bound term and fuzzy core bound term；Using L0
Norm describes image sparse priori, L2Norm describes the sparse and continuous priori of fuzzy core；According to the image L of construction0Bound term
With fuzzy core L2Bound term establishes non-convex fuzzy kernel estimates model, using multiple dimensioned horizontal frame, using blurred picture gradient as
Initial pictures substitute into model and obtain the fuzzy core and intermediate clear image of lowest scale horizontal estimated；According to the fuzzy core of estimation,
The intermediate clear image of previous horizontal estimated is restored again, and then obtains the intermediate clear image of more next scale level
And the fuzzy core of estimation.
Specifically include following methods step:
Step 1: excavating image and fuzzy core priori knowledge, fuzzy kernel estimates model is established；It is constrained and is protected using image single order
Protect image border sparsity, image gradient dark confining guard image local feature sparsity；Sparsity is applied to fuzzy core
And continuity constraint, guarantee the accuracy of ambiguous estimation core.
Step 2: implementation model solves using convexity approximation and linear approximation during fuzzy kernel estimates model solution,
Complete fuzzy kernel estimates.
Fuzzy kernel estimates model of the present invention is the calculating process that an iteration updates, specifically, based on image and
The fuzzy kernel estimates model of fuzzy core mixed constraints:
Wherein, k is fuzzy core to be estimated,For blurred picture (the intermediate clear image of previous scale level estimation)
Gradient,BxAnd ByRespectively first difference of the blurred picture f on the direction x and the direction y point；For
Clear image (the intermediate clear image of scale level estimation to be estimated) gradient,IxAnd IyIt is respectively intermediate clear
First difference of the clear image I on the direction x and the direction y point；For image gradient dark, indicate in image gradient block
Minimum pixel value, P (x) indicate center pixel be x image block, c indicate image channel；ω and σ controls image canonical respectively
Relative weighting inside item and fuzzy core regular terms, γ is image regular terms parameter, and λ is fuzzy core regular terms parameter.
Specifically, image gradient dark is as follows:
Wherein, P (x) is an image block centered on pixel x；C is image channel, and r, g, b are colored triple channel.
The study found that clear image gradient dark is more sparse than blurred picture dark, using L0Norm characterization is sparse, obtains figure
As gradient dark constrains:
When solution, due to L0Norm | | u | |0Constrained with dark | | D (u) | |0Introducing, cause cost function in non-convex
It is non-linear.Therefore, in second step, the solution to model can be realized using half secondary split algorithm and linear approximation.
In the present invention, the intermediate clear image is the relatively clear image obtained in iterative process, among this
Clear image obtains more accurate clear image gradient information, so that the accuracy for improving fuzzy kernel estimates is not to obtain
Final restored image.
The fuzzy core that will be estimated using the above method is used in non-blind restoration method, i.e., resilient clear image out.
The present invention is firstly introduced into the L of image gradient dark0Sparse prior, to image gradient in fuzzy kernel estimates model
And its dark carries out mixing L0Regularization constraint utilizes image gradient L0The sparsity of image border, figure are protected in constraint well
As gradient dark L0Confining guard image local feature sparsity, to obtain part and global image constraint.Synthesis is examined
The sparsity and continuity for considering fuzzy core, use L2Norm constrains fuzzy core and its gradient, advantageously reduces estimation mould
Paste the noise in core.During model solution, in order to solve L0What non-convex problem and the dark constraint that regular terms introduces introduced
Nonlinear problem, has been respectively adopted convexity approximation and linear approximation is solved.Therefore, the present invention relates to abundant scene images to go
When fuzzy problem, while considering part and global priori and the fuzzy core sparsity and continuity of image, can obtain accurate
Fuzzy core, which is used under non-blind recovery frame, image relatively sharp out can be effectively restored.
Detailed description of the invention
Fig. 1 is the basic framework figure of the method for the present invention；
Fig. 2 is that image gradient dark proposed by the present invention indicates as a result, (a) and (d) is respectively clear image and obscures
Image is (b) its corresponding gradient image with (e), and (c) and (f) is that clear image gradient dark and blurred picture gradient are dark
Channel.
Clear image and 8 realistic blur cores of the Fig. 3 for 8 width standards used in the experiment of the method for the present invention compliance test result, figure
As including nature, face and text image.
Fig. 4 be using the method for the present invention for the wherein width blur size in 64 width blurred pictures be 27 × 27 it is fuzzy
The restoration result of image,.
Fig. 5 is the result restored using the method for the present invention to abundant scene realistic blur image, wherein (a) (c)
(e) figure is respectively nature blurred picture, face blurred picture and text blurred picture, and (b) (d) (f) is its corresponding recovery knot
Fruit.
Specific embodiment
The specific implementation and effect of this method that the following is further explained with reference to the attached drawings.
A kind of fuzzy core estimation method based on image and fuzzy core mixed constraints, comprising the following steps:
Step 1: excavating image and fuzzy core priori knowledge, fuzzy kernel estimates model is established；It is constrained and is protected using image single order
Protect image border sparsity, image gradient dark confining guard image local feature sparsity；Sparsity is applied to fuzzy core
And continuity constraint, guarantee the accuracy of ambiguous estimation core.
Step 2: carrying out convexity approximation and linear approximation during fuzzy kernel estimates model solution, implementation model is solved,
Complete fuzzy kernel estimates.
It as figure 1 shows the process that image restoration is carried out using the above method.It is in more rulers for original blurred picture
Implement the estimation of accurate fuzzy core on degree horizontal frame, multiple dimensioned frame is the multi-layer image pyramid by resolution ratio from low to high
Model composition, if first level, the second level ... the 5th in figure are horizontal, pyramid model can effectively avoid part
Optimal solution guarantees that finally obtained solution converges on globally optimal solution, especially in the case where fog-level is more serious.Specifically
Realize operation are as follows: pre-process to original blurred picture, obtain its gradient image and carry out down-sampling, subsequently enter pyramid
The first scale level of model uses model proposed by the present invention in the scale level, to image gradient and its dark in model
It is applied with L0Sparse constraint applies L to fuzzy core and its gradient2Constraint, then solves model, obtains the first scale water
Fuzzy core result and intermediate image under flat carry out successive iterations as a result, using the result as the input value of next scale level.
Until acquiring the fuzzy core finally estimated under last scale level.The fuzzy core that will be estimated is used in non-blind restoration method,
I.e. resilient clear image out.
Intermediate clear image of the invention is not obtained final restored image.The fuzzy kernel estimates that the present invention is mentioned are one
The process that a iteration updates, which is the relatively clear image obtained in iterative process, clear with the centre
Image obtains more accurate clear image gradient information, thus the accuracy for improving fuzzy kernel estimates not obtain it is final
Restored image.
Completely fuzzy kernel estimates model proposed by the present invention is as follows:
Wherein, k is fuzzy core to be estimated,For blurred picture (the intermediate clear image of previous scale level estimation)
Gradient,BxAnd ByRespectively first difference of the blurred picture f on the direction x and the direction y point；For
Clear image (the intermediate clear image of scale level estimation to be estimated) gradient,IxAnd IyIt is respectively intermediate clear
First difference of the clear image I on the direction x and the direction y point；For image gradient dark, indicate in image gradient block
Minimum pixel value, P (x) indicates that center pixel is the image block of x, and c indicates different channels；ω and σ controls image canonical respectively
Relative weighting inside item and fuzzy core regular terms, γ is image regular terms parameter, and λ is fuzzy core regular terms parameter.
Wherein, P (x) is an image block centered on pixel x；C is channel, and r, g, b are colored triple channel.Clearly
Image gradient dark is more sparse than blurred picture dark, using L0Norm characterization is sparse, obtains image gradient dark
Constraint:
A kind of iterative algorithm by alternating minimization is expanded based on half secondary split algorithm, dark is constrained and is carried out
Linear approximation, to solve the fuzzy kernel estimates model of proposition.
Auxiliary variable u is introduced to replaceF is replacedThe fuzzy kernel estimates model of proposition is rewritten are as follows:
Above formula is the non-convex nonlinear problem of height, in order to optimize to it, it usually needs initial from one
U and k start alternating iteration and update them, and introduce dark constraint approximate substitution.
Here is the specific solution procedure of each iteration u and k:
1. u subproblem
In the more new stage of intermediate clear image gradient u, the k that last iterative estimate obtains is immobilized, then u
Problem is converted into following minimization problem:
Due to containing L in above formula0Norm | | u | |0With | | D (u) | |0, therefore be a non-convex Nonlinear Optimization Problem,
It can not be solved with traditional gradient descent method, and brute-force searching algorithm is too time-consuming.
Present invention introduces two auxiliary variable g=(gh,gv)TIt is respectively intended to indicate with dWith D (u), ghAnd gvFor auxiliary
For variable g in first difference both horizontally and vertically point, T indicates transposition operation.Above formula can be rewritten are as follows:
Wherein, α and β is two positive punishment parameters, they constantly change during entire Optimization Solution.Pass through fixation
Other two variable alternately to solve u, g and d respectively, i.e., fixed u and g solves d, and fixed u and d solves g, and fixed g and d solves u.
Above formula is split into two mutually independent cost functions to solve:
It is respectively as follows: by the solution that approximate evaluation obtains
G and d that last iterative estimate obtains are immobilized, u subproblem can simplify are as follows:
By nonlinear operation D (u) approximate transform be linear matrix operations M (x, z) U, M be sparse matrix, is defined as:
Wherein M (x, z) indicates that center pixel is the image block of x, minimum pixel value z.
Obviously, simplified u subproblem is a least square problem, its closing can be obtained by Fast Fourier Transform (FFT)
The solution of form.
2. k subproblem
In the more new stage of fuzzy core k, the u that last iterative estimate obtains is immobilized, k subproblem is converted into as follows
Minimization problem:
Equally, above formula is a least square problem, and the solution of closing form can quickly be asked by Fourier transformation
?.
After estimating fuzzy core k, apply normalization constraint and dynamic threshold constraint to it, to inhibit noise jamming
While protect fuzzy core intrinsic characteristic not to be destroyed.
∫ k (p) dxdy=1
Wherein, k (p) is the gray value in fuzzy core k at pixel p, and max (k) indicates all pixels point in fuzzy core k
Gray scale maximum value, δ is a smaller positive number, for inhibiting the noise in fuzzy core.
Fig. 2 is that image gradient dark proposed by the present invention indicates as a result, (a) and (d) is respectively clear image and obscures
Image is (b) its corresponding gradient image with (e), and (c) and (f) is that clear image gradient dark and blurred picture gradient are dark
Channel.As can be seen from the figure clear image gradient dark is more sparse than blurred picture gradient dark, shows as dark
It include more black picture element blocks in figure.
Clear image and 8 realistic blur cores of the Fig. 3 for 8 width standards used in the experiment of the method for the present invention compliance test result, figure
As including nature, face and text image.
Fig. 4 be using the method for the present invention for the wherein width blur size in 64 width blurred pictures be 27 × 27 it is fuzzy
The restoration result of image, it can be seen that it is sparse and continuous by the fuzzy core that the method for the present invention estimates, restore image out
Subjective effect is preferable.
Fig. 5 is the result restored using the method for the present invention to abundant scene realistic blur image, wherein (a) (c)
(e) figure is respectively nature blurred picture, face blurred picture and text blurred picture, and (b) (d) (f) is its corresponding recovery knot
Fruit.There it can be seen that the method for the present invention is still estimated that accurate fuzzy core to different types of blurred picture,
Guarantee its sparsity and continuity, to restore clearly image out.
It is carried out by the synthesis blurred picture and realistic blur image in the above abundant scene (such as nature, face, text)
Experiments have shown that: the fuzzy core estimation method of proposition be it is effective, restore image out compared with classical way in recent years, it is subjective
It visual effect and objectively evaluates index and is all significantly improved.
As it can be seen that the present invention can estimate accurate fuzzy core for blurred picture, especially when the abundant field of processing
When scape blurred picture, the fuzzy core estimated by the method for the present invention it is resilient go out the image more relatively sharp than conventional method.
Claims (8)
1. a kind of fuzzy core estimation method based on image and fuzzy core mixed constraints, which is characterized in that the method includes with
Lower step:
Step 1: excavating image and fuzzy core priori knowledge, fuzzy kernel estimates model is established；Utilize image single order confining guard figure
As edge sparsity, image gradient dark confining guard image local feature sparsity；Sparsity and company are applied to fuzzy core
Continuous property constraint, guarantees the accuracy of ambiguous estimation core；
The fuzzy kernel estimates model are as follows:
Wherein, k is fuzzy core to be estimated, and ▽ B is the intermediate clear image gradient of blurred picture, that is, previous scale level estimation,BxAnd ByRespectively first difference of the blurred picture f on the direction x and the direction y point；▽ I is clear
The intermediate clear image gradient of image, that is, scale level to be estimated estimation,IxAnd IyRespectively intermediate clear figure
As first difference of the I on the direction x and the direction y point；D (▽ I) is image gradient dark, is indicated in image gradient block most
Small pixel value, ω and σ control the relative weighting inside image regular terms and fuzzy core regular terms respectively, and γ is image regular terms ginseng
Number, λ is fuzzy core regular terms parameter；
Step 2: implementation model solves using convexity approximation and linear approximation during fuzzy kernel estimates model solution, complete
Fuzzy kernel estimates.
2. the fuzzy core estimation method according to claim 1 based on image and fuzzy core mixed constraints, which is characterized in that
Image gradient dark D (▽ I) is as follows in the fuzzy kernel estimates model:
Wherein, P (x) indicates that center pixel is the image block of x, and c indicates image channel, and r, g, b are colored triple channel, clear image
Gradient dark is more sparse than blurred picture dark, using L0Norm characterizes sparsity, obtains image gradient dark about
Beam:
Ed(I)=| | D (▽ I) | |0。
3. fuzzy core estimation method according to claim 1 or 2, which is characterized in that in second step, using convexity approximation and
Linear approximation implementation model solves: expanding a kind of iteration calculation by alternating minimization based on half secondary split algorithm
Method constrains dark and carries out linear approximation, to solve the fuzzy kernel estimates model of proposition.
4. fuzzy core estimation method according to claim 3, which is characterized in that introduce auxiliary variable u and ▽ I, f is replaced to replace
▽ B rewrites the fuzzy kernel estimates model of proposition are as follows:
Then, it is optimized, i.e., alternating iteration updates since initial a u and k, and introduces dark constraint
Approximate substitution.
5. fuzzy core estimation method according to claim 4, which is characterized in that the specific solution of u:
In the more new stage of intermediate clear image gradient u, the k that last iterative estimate obtains is immobilized, u subproblem is turned
Turn to following minimization problem:
Introduce two auxiliary variable g=(gh,gv)TIt is respectively intended to indicate ▽ u and D (u), g with dhAnd gvIt is auxiliary variable g in level
With the first difference point of vertical direction, T indicates transposition operation；Above formula is rewritten are as follows:
Wherein, α and β is two positive punishment parameters, they constantly change during entire Optimization Solution；By fixing remaining
Two variables alternately to solve u, g and d respectively, i.e., fixed u and g solves d, and fixed u and d solves g, and fixed g and d solves u；
Above formula is split into two mutually independent cost functions to solve:
It is respectively as follows: by the solution that approximate evaluation obtains
G and d that last iterative estimate obtains are immobilized, u subproblem can simplify are as follows:
By nonlinear operation D (u) approximate transform be linear matrix operations M (x, z) U, M be sparse matrix, is defined as:
Wherein M (x, z) indicates that center pixel is the image block of x, minimum pixel value z.
6. fuzzy core estimation method according to claim 4, which is characterized in that the specific solution of k:
In the more new stage of fuzzy core k, the u that last iterative estimate obtains is immobilized, k subproblem is converted into following minimum
Change problem:
7. fuzzy core estimation method according to claim 5, which is characterized in that after estimating fuzzy core k, apply to it
Normalization constraint and dynamic threshold constraint, to protect fuzzy core intrinsic characteristic not to be destroyed while inhibiting noise jamming
∫ k (p) dxdy=1
Wherein, k (p) is the gray value in fuzzy core k at pixel p, and max (k) indicates the gray scale of all pixels point in fuzzy core k
Maximum value, δ is a smaller positive number, for inhibiting the noise in fuzzy core.
8. -7 described in any item fuzzy core estimation methods according to claim 1, which is characterized in that clearly scheme the centre
As being the relatively clear image obtained in iterative process, more accurate clear image ladder is obtained with the intermediate clear image
Information is spent, so that the accuracy for improving fuzzy kernel estimates is not obtained final restored image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910164437.1A CN109919871A (en) | 2019-03-05 | 2019-03-05 | Fuzzy core estimation method based on image and fuzzy core mixed constraints |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910164437.1A CN109919871A (en) | 2019-03-05 | 2019-03-05 | Fuzzy core estimation method based on image and fuzzy core mixed constraints |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109919871A true CN109919871A (en) | 2019-06-21 |
Family
ID=66963390
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910164437.1A Pending CN109919871A (en) | 2019-03-05 | 2019-03-05 | Fuzzy core estimation method based on image and fuzzy core mixed constraints |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109919871A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110675347A (en) * | 2019-09-30 | 2020-01-10 | 北京工业大学 | Image blind restoration method based on group sparse representation |
| CN110969589A (en) * | 2019-12-03 | 2020-04-07 | 重庆大学 | Dynamic scene fuzzy image blind restoration method based on multi-stream attention countermeasure network |
| CN112819723A (en) * | 2021-02-05 | 2021-05-18 | 河海大学 | High-energy X-ray image blind restoration method and system |
| CN113177889A (en) * | 2021-04-27 | 2021-07-27 | 深圳市慧鲤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN113538374A (en) * | 2021-07-15 | 2021-10-22 | 中国科学院上海技术物理研究所 | Infrared image blur correction method for high-speed moving object |
| CN114663304A (en) * | 2022-03-16 | 2022-06-24 | 中国科学院光电技术研究所 | Sparse prior-based optical synthetic aperture system imaging enhancement method |
| CN114897734A (en) * | 2022-05-18 | 2022-08-12 | 北京化工大学 | Detected target image restoration method based on gradient direction prior |
| WO2023050026A1 (en) * | 2021-09-28 | 2023-04-06 | 华为技术有限公司 | Image processing method and related device |
| WO2023115801A1 (en) * | 2021-12-21 | 2023-06-29 | 武汉工程大学 | Point-by-point correction and restoration method and system for large field-of-view degraded image having aero-optical effect |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105741243A (en) * | 2016-01-27 | 2016-07-06 | 北京航空航天大学 | Blurred image restoration method |
| CN106204502A (en) * | 2016-08-17 | 2016-12-07 | 重庆大学 | Based on mixing rank L0regularization fuzzy core method of estimation |
-
2019
- 2019-03-05 CN CN201910164437.1A patent/CN109919871A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105741243A (en) * | 2016-01-27 | 2016-07-06 | 北京航空航天大学 | Blurred image restoration method |
| CN106204502A (en) * | 2016-08-17 | 2016-12-07 | 重庆大学 | Based on mixing rank L0regularization fuzzy core method of estimation |
Non-Patent Citations (1)
| Title |
|---|
| LIAO Y: ""Blur Kernel Estimation Model with Combined Constraints for Blind Image Deblurring"", 《DIGITAL IMAGE COMPUTING: TECHNIQUES AND APPLICATIONS (DICTA)》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110675347A (en) * | 2019-09-30 | 2020-01-10 | 北京工业大学 | Image blind restoration method based on group sparse representation |
| CN110675347B (en) * | 2019-09-30 | 2022-05-06 | 北京工业大学 | Image blind restoration method based on group sparse representation |
| CN110969589A (en) * | 2019-12-03 | 2020-04-07 | 重庆大学 | Dynamic scene fuzzy image blind restoration method based on multi-stream attention countermeasure network |
| CN110969589B (en) * | 2019-12-03 | 2023-06-02 | 重庆大学 | Dynamic scene blurred image blind restoration method based on multi-stream annotating countermeasure network |
| CN112819723A (en) * | 2021-02-05 | 2021-05-18 | 河海大学 | High-energy X-ray image blind restoration method and system |
| CN112819723B (en) * | 2021-02-05 | 2022-07-19 | 河海大学 | High-energy X-ray image blind restoration method and system |
| CN113177889A (en) * | 2021-04-27 | 2021-07-27 | 深圳市慧鲤科技有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN113538374A (en) * | 2021-07-15 | 2021-10-22 | 中国科学院上海技术物理研究所 | Infrared image blur correction method for high-speed moving object |
| CN113538374B (en) * | 2021-07-15 | 2023-11-28 | 中国科学院上海技术物理研究所 | Infrared image blurring correction method for high-speed moving object |
| WO2023050026A1 (en) * | 2021-09-28 | 2023-04-06 | 华为技术有限公司 | Image processing method and related device |
| WO2023115801A1 (en) * | 2021-12-21 | 2023-06-29 | 武汉工程大学 | Point-by-point correction and restoration method and system for large field-of-view degraded image having aero-optical effect |
| CN114663304B (en) * | 2022-03-16 | 2023-09-19 | 中国科学院光电技术研究所 | Sparse prior-based optical synthetic aperture system imaging enhancement method |
| CN114663304A (en) * | 2022-03-16 | 2022-06-24 | 中国科学院光电技术研究所 | Sparse prior-based optical synthetic aperture system imaging enhancement method |
| CN114897734A (en) * | 2022-05-18 | 2022-08-12 | 北京化工大学 | Detected target image restoration method based on gradient direction prior |
| CN114897734B (en) * | 2022-05-18 | 2024-05-28 | 北京化工大学 | Measured target image restoration method based on gradient direction priori |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109919871A (en) | Fuzzy core estimation method based on image and fuzzy core mixed constraints | |
| CN105913419B (en) | TFT-LCD mura defect inspection methods based on ICA study and Multichannel fusion | |
| WO2021022929A1 (en) | Single-frame image super-resolution reconstruction method | |
| Xu et al. | Image inpainting by patch propagation using patch sparsity | |
| CN105243670B (en) | A kind of sparse and accurate extracting method of video foreground object of low-rank Combined expression | |
| CN105631807B (en) | The single-frame image super-resolution reconstruction method chosen based on sparse domain | |
| CN109671023A (en) | A kind of secondary method for reconstructing of face image super-resolution | |
| CN109934154A (en) | A kind of remote sensing image variation detection method and detection device | |
| CN105046672A (en) | Method for image super-resolution reconstruction | |
| CN106127688A (en) | A kind of super-resolution image reconstruction method and system thereof | |
| CN116682120A (en) | Multilingual mosaic image text recognition method based on deep learning | |
| CN110830043B (en) | Image compressed sensing reconstruction method based on mixed weighted total variation and non-local low rank | |
| CN101561932A (en) | Method and device for detecting real-time movement target under dynamic and complicated background | |
| CN106339996A (en) | Image blind defuzzification method based on hyper-Laplacian prior | |
| CN106204502B (en) | Based on mixing rank L0Regularization fuzzy core estimation method | |
| CN112581378A (en) | Image blind deblurring method and device based on significance intensity and gradient prior | |
| CN111260585A (en) | Image recovery method based on similar convex set projection algorithm | |
| CN112862838A (en) | Natural image matting method based on real-time click interaction of user | |
| CN109272539A (en) | The decomposition method of image texture and structure based on guidance figure Total Variation | |
| Zhang et al. | Crafting training degradation distribution for the accuracy-generalization trade-off in real-world super-resolution | |
| CN113781375B (en) | Vehicle-mounted vision enhancement method based on multi-exposure fusion | |
| CN113487512B (en) | Digital image restoration method and device based on edge information guidance | |
| CN116523985B (en) | Structure and texture feature guided double-encoder image restoration method | |
| CN116563133A (en) | Low-illumination color image enhancement method based on simulated exposure and multi-scale fusion | |
| CN106375773B (en) | Altering detecting method is pasted in frame duplication based on dynamic threshold |
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: 20190621 |
|
| RJ01 | Rejection of invention patent application after publication |