CN104867110A - Lattice Boltzmann model-based video image defect repairing method - Google Patents
Lattice Boltzmann model-based video image defect repairing method Download PDFInfo
- Publication number
- CN104867110A CN104867110A CN201410733849.XA CN201410733849A CN104867110A CN 104867110 A CN104867110 A CN 104867110A CN 201410733849 A CN201410733849 A CN 201410733849A CN 104867110 A CN104867110 A CN 104867110A
- Authority
- CN
- China
- Prior art keywords
- image
- video
- model
- lattice boltzmann
- boltzmann method
- 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
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a lattice Boltzmann model-based video image defect repairing method. The method comprises operation steps: 1) a to-be-repaired object is inputted; 2) a format of the repaired object is judged, and if the repaired object is an image, a fifth step is carried out for repairing, or otherwise, a third step is carried out; 3) a format of inputted video signals is judged, a digital video is firstly converted via a video acquisition card in the case of an analog video, and a fourth step is then carried out, or otherwise, the fourth step is directly carried out; 4) the video is decomposed into frames; 5) the mentioned lattice Boltzmann method is used for repairing each decomposed frame image; and 6) the needed output format is judged, a video is synthesized in the case of need, or otherwise, the image is directly outputted. According to the lattice Boltzmann repairing method, particle density of each node in the model is updated during a collision and transfer process, and output is carried out if iterations are reached or a threshold condition is met. The repairing method is simple in realization and good in parallelism, and the repairing efficiency can be improved while the repairing quality is good.
Description
Technical field
The invention belongs to field of video image processing, propose a kind of video image flaw restorative procedure based on grid wave pattern.
Background technology
Along with the development of digitizing technique, preservation and the reparation of digital video have become old archives, older picture, the important means that old television data etc. are preserved and repaired.Namely the reparation of digital video utilizes the continuity between the spatial structural form of frame of video or each frame of video to realize specifying in video reparation and the filling of damaged area, video is made to have spatial smoothness and time continuity, improve the appreciative value of video, be widely used in multiple situation, as removed the reparation of spot or cut in unnecessary target, old television data in scene, and solve the problem etc. that old film produces flicker in playing process.
Video recovery technique is the extension to Digital Image Inpainting, video recovery technique main at present and image repair technology type are seemingly, and mainly adopt both at home and abroad based on the method for partial differential equation (Partial Differential Equation, PDE) and the method based on textures synthesis for image repair problem.
Restorative procedure based on partial differential equation is copy artificial principle of repairing the earliest, the mode of the peripheral information of image damaged area along isophote direction gradually to damaged area diffusion is utilized to reach level and smooth effect of repairing, the method, for compared with small scale and the breakage image without complex texture structure, can obtain good repairing effect.Chan and Shen proposes Total Variation (TV model) for repairing, (Chan T, Shen is models for local non-texture inpainting [J] .SIAM Journal on Applied Mathematics J.2002.Mathematical, 62 (3): 1019-1043.) TV restorative procedure is by minimizing to energy functional.
TV model reparation compared with there will be discontinuous problem during large regions, destroy visual communication principle, Curvature-driven diffusion model (Curvature Driven Diffusion, CDD) (T. Chan time for repairing, J. Shen. Non-Texture Inpaintings by Curvature-Driven Diffusions [C] .JVCIR, 2001,12 (4): 436-449) there will not be discontinuous problem, curvature is embedded in coefficient of diffusion.CDD model can realize the reparation of larger damaged area, when restoring area is in marginal portion, can obtain good repairing effect.TV model and CDD model all can realize good repairing effect, but from counting yield, solve for partial differential equation the calculating needing numerical value, and algorithm realization is more complicated, and institute's spended time is longer, and implementation efficiency is low.
Utilize the information beyond border to be repaired based on the image repair mode of textures synthesis and image completion technology, utilize the mode of search match block to copy gradually to border, damaged area and fill border, damaged area (Zhang Hongying, Peng Qicong. Digital Image Inpainting summary [J]. Chinese image graphics journal, 2007,12 (1): 1-10).
This restorative procedure can realize the reparation compared with large regions breakage, and can obtain good texture features by coupling filling.But these class methods are in the inner inconsistency easily producing obvious color and structure of restoring area, and repairing effect is not ideal enough, and the calculating of itself is more consuming time, causes the inefficiency of image repair.
And Lattice Boltzmann method model has physical thought clearly, the advantage of simple boundary treatment and fast parallel calculating, and it is discrete spatial model, is particularly suitable for Digital Image Processing.From the microvisual model of Lattice Boltzmann method model, design Lattice Boltzmann method model evolution equation, finally can be met macroscopical partial differential equation of image processing requirements.Therefore Lattice Boltzmann method model method be realize image procossing rapidly and efficiently provide realization means with accuracy.
Summary of the invention
The object of the invention is to there is flaw and the problem of distortion and propose a kind of novelty reparation implementation based on Lattice Boltzmann method model cross-platform fast for what exist in image, discontinuous to solve restoring area in prior art, the problem that image repair efficiency is not high, the method can obtain good remediation efficiency while being intended to keep better repairing effect, can be used for repairing the various flaws in image, as the removal etc. of spot, cut and object.And with this restorative procedure for core, develop the repair system that can run on multiple soft or hard platform, make the method obtain applying more widely.
To achieve these goals, the present invention adopts following technical scheme:
The present invention proposes a kind of reparation implementation based on Lattice Boltzmann method model, can at the hardware computing platform comprising PC and server, and the various software operating system comprising Android, Linux, Mac, Windows is run.Its key step is as follows:
1), input is containing image defective
;
2), the form repairing object is judged, if image repair then goes to step 5) repair, otherwise enter step 3);
3), judge input be digital video or analog video, if analog video is then first converted to digital video by video frequency collection card, then enter step 4), otherwise, directly enter step 4);
4), video is decomposed framing as 25 frames/second;
5), the Lattice Boltzmann method model proposed is used to repair to the every two field picture decomposed;
6), required output format is judged, if required output is video, then synthetic video, otherwise direct output image
.
Repair for Lattice Boltzmann method model mentioned above, the present invention proposes a kind of anisotropy parameter Lattice Boltzmann method model (D2Q9 model), and utilize this model realization flaw reparation to be utilize the geological informations such as the gradient of area to be repaired and peripheral information or curvature to design coefficient of diffusion, control the diffusion Evolution Rates of particle according to the geometrical property of image itself thus realize level and smooth process of repairing.Because all kinds of video is all decomposed into image, without loss of generality, the present invention is to repair spot and the cut image of bmp form, and the PC that Windows operating system is housed is tested, and reparation the results are shown in
fig. 9, 10, data record is shown in
fig. 12,
fig. 13.The main flow of repairing
figure is as Fig. 3shown in, the present invention is in conjunction with the coefficient of diffusion feature in TV model and CDD model Macroscopic Equation, the coefficient of diffusion in Lattice Boltzmann method model for the spot occurred in image and video and cut flaw design ap-plication, thus remediation efficiency is improved while the better repairing effect of maintenance.Image defective is contained for what read in
, its key step is as follows:
1), judge to detect damaged area
, and expansion is carried out to the region detected obtain
, pre-service obtains image
;
2), set up Lattice Boltzmann method model, initial equilibrium state distribution function is set
;
3) iterations of Lattice Boltzmann method model evolution equation, is established
, and stopping criterion for iteration and iteration ends threshold value
;
4) coefficient of diffusion of Lattice Boltzmann method model evolution equation, is designed
;
5), according to the D2Q9 model in Lattice Boltzmann method model, the equilibrium distribution function in Lattice Boltzmann method model evolution equation is upgraded
;
6) transition process in Lattice Boltzmann method model, is calculated:
;
7) collision process in Lattice Boltzmann method model, is calculated:
8), each Nodes particle density of Renewal model
, and the peripheral information of area to be repaired;
9), judge whether to meet stopping criterion for iteration, if do not meet iteration stopping condition, enter step 10), otherwise repair end;
10), judge whether iterations reaches N, if do not reach iterations N, go to step 4), and repeat step 4)-9), until terminate after reaching iterations N to repair;
Above-mentioned steps 3) in, propose to utilize stopping criterion for iteration to judge whether the iteration stopping Lattice Boltzmann method model evolution equation, conventional stopping criterion for iteration is:
Wherein,
for the threshold value preset, and
for the damaged area number of pixels of input picture.The present invention is optimized above-mentioned stopping criterion for iteration, and the absolute value of area to be repaired pixel value difference before and after each iteration is divided into 5 sections, and distributes a weight for each section, is called difference weight here,
as Fig. 5shown in.
Before and after each iteration, the chance of the larger required continuation iteration of pixel value difference is larger, so difference weight
tableshow the importance that pixel value difference place pixel develops in evolutionary process.Here difference weight is set to
, and the stopping criterion for iteration that the present invention proposes is:
Wherein,
according to different images to be repaired and can pre-set the requirement of repairing quality.
Above-mentioned steps 4) in, propose to design the coefficient of diffusion of Lattice Boltzmann method model evolution equation.In TV model and CDD model, its coefficient of diffusion is respectively:
with
Wherein,
for the gradient of image to be repaired,
for curvature function.In spot distorted image, spot mostly comparatively is sub-circular or ellipse, and speck area is less, and in cut distorted image, cut mostly is strip, both can not cause the loss of most of texture.Meanwhile, in order to improve remediation efficiency, the design of coefficient of diffusion only comprises gradient information.Coefficient of diffusion is
, wherein
for non-negative monotone decreasing, and
with
.
The present invention proposes three kinds of different Diffusion Coefficient Models and their combinations, and coefficient of diffusion is as follows:
，
，
Wherein,
for the gradient of image to be repaired,
for the threshold value preset,
for on the occasion of.The curve of above-mentioned three kinds of coefficient of diffusion is analyzed,
as Fig. 6shown in, and to three kinds of coefficient of diffusion carry out combination can obtain different coefficient of diffusion:
Wherein,
get different values respectively by coefficient of diffusion different for correspondence, different impacts can be produced on the reparation of image.
The present invention compared with prior art, there is following apparent outstanding substantive distinguishing features and remarkable technical progress: because Lattice Boltzmann method model has concurrency structure, be easy to realize concurrent operation, compared to textures synthesis method, improve the efficiency that spot is repaired, the reparation of spot distortion in old television data can better be applied to.And compared with classical PDE repairing model, there is good repairing effect, overcome the uncontinuity of restoring area, also can improve the efficiency of reparation simultaneously.
Accompanying drawing explanation
fig. 1: based on each direction vector of the Lattice Boltzmann method model of D2Q9;
fig. 2: repair operating process next time based on the video image of Lattice Boltzmann method model
figure;
fig. 3: for the flaw in image or video, the flow process of Lattice Boltzmann method model restorative procedure
figure;
fig. 4: the two-dimensional lattice Boltzmann model D2Q9 model structure signal that discretize network is formed
figure;
fig. 5: the segmentation of area to be repaired pixel value difference, and the difference weight distribution of each section;
fig. 6: the curve of three kinds of coefficient of diffusion that this invention proposes and combination thereof
figure, wherein
in
value is all set to 10, in build-up curve
be set to
;
fig. 7: spot distortion gray level image,
in figurestain be spot;
fig. 8: the image of older picture cut distortion, white strip is cut;
fig. 9: the reparation result of spot distortion gray level image, wherein
figure be corresponding in turn to and in coefficient of diffusion be
lattice Boltzmann method model repair result;
fig. 10: the reparation result of older picture cut, wherein
figure be corresponding in turn to and in coefficient of diffusion be
lattice Boltzmann method model repair result;
fig. 11: repair running software and show
figure;
fig. 12 are
table 1: for the spot distortion of gray level image, the Lattice Boltzmann method model of different coefficient of diffusion repairs required iterations and repair time;
fig. 13 are
table 2: for the cut distortion of older picture, the Lattice Boltzmann method model of different coefficient of diffusion repairs required iterations and repair time.
Embodiment
Below with reference to
accompanying drawingand the present invention is described in detail in conjunction with the embodiments:
Embodiment one: see
fig. 1with
fig. 2, this is based on Lattice Boltzmann Method operation steps: 1), input containing image defective
; 2), judge the form repairing object, if image repair then proceeds to step 5) reparation, otherwise enter step 3); 3), judge input be digital video or analog video, if analog video is then first converted into digital video by video frequency collection card, then proceed to step 4), otherwise, directly enter step 4); 4), video is decomposed framing; 5), the Lattice Boltzmann Method proposed is used to repair to the every two field picture decomposed; 6), required output format is judged, if required output is video, then synthetic video, otherwise direct output image
.
Described video packets contains numeral or the analog video of MPEG-1, MPEG-4, AVI, RM, ASF, WMV, MOV or MKV form; Described image comprises the analog or digital image that all kinds of remotely sensed image or medical instrument imaging obtain, and its form comprises bmp, jpg, gif or png; Described flaw source comprises the color distortion that imaging device fault causes, the fading of older picture, folding line, the aging broadcasting flicker caused of film, the spot of old television data, cut.Described Lattice Boltzmann method model: solve diffusion equation in the relaxation factor of Lattice Boltzmann method model be achieved by the edge of image cut-off function is embedded into, conventional two-dimensional lattice Boltzmann model comprises D2Q5---two dimension 5 direction diffusions, D2Q9, three-dimensional Lattice Boltzmann method model comprises D3Q7, D3Q15.The described video image flaw reparation based on Lattice Boltzmann method model can at the hardware computing platform comprising PC or server, and runs in the operation system of software comprising Android, iOS, Linux or Windows.
The concrete operation step of Lattice Boltzmann method restorative procedure is as follows:
1), judge to detect damaged area
, and expansion is carried out to the region detected obtain
, pre-service obtains image
2), set up LB model, initial equilibrium state distribution function is set
;
3) iterations of LB EVOLUTION EQUATION, is established
, and stopping criterion for iteration and iteration ends threshold value
;
4) coefficient of diffusion of LB EVOLUTION EQUATION, is designed
;
5), according to the D2Q9 model in LB model, the equilibrium distribution function in LB EVOLUTION EQUATION is upgraded
;
6) transition process in LB model, is calculated:
;
7) mechanism in LB model, is calculated:
；
8), each Nodes particle density of Renewal model
, and the peripheral information of area to be repaired;
9), judge whether to meet stopping criterion for iteration, if do not meet iteration stopping condition, enter step 10), otherwise repair end;
10), judge whether iterations reaches N, if do not reach iterations N, go to step 4), and repeat step 4)-9), until terminate after reaching iterations N to repair;
Described edge cut-off function is non-negative monotonic decreasing function, image gradient information and curvature information can be utilized, slow down in boundary rate of propagation, at level and smooth place, rate of propagation is accelerated, and makes all have good performance to rate of propagation and edge hold facility in diffuse images process.Described relaxation factor controls rate of diffusion, realizes nonlinear diffusion, to reach protection image border characteristic, the object of restraint speckle; The design of equilibrium distribution function realizes anisotropy parameter or isotropic diffusion.
Embodiment two: the present embodiment is implemented premised on technical scheme of the present invention, and implement to repair to spot distorted image and older picture cut distorted image respectively,
as Fig. 7, shown in 8.Following present detailed embodiment.
Described Lattice Boltzmann method model D2Q9 model is made up of uniform grid, and wherein the node of grid is counted as a cellular, corresponding to the pixel in image
.And namely the gray-scale value at pixel place corresponds to the particle density in cellular
.Meanwhile,
describe at direction vector
on particle density distribution function, the wherein direction vector of Nodes
tableshow
as Fig. 1shown in:
D2Q9 model has 9 discrete speed, and in model, microscopic particle diffusion EVOLUTION EQUATION is:
Wherein,
tablebe shown in node
the Particle diffusion coefficients at place, with the relaxation factor in Lattice Boltzmann method model
between pass be:
for equilibrium distribution function, close between the density of equilibrium distribution function and each Nodes particle here and be:
Embodiment 1: the gray level image repairing spottiness flaw
For the Lattice Boltzmann method model reparation of the video or image that there is distortion, for having obtained image to be repaired in advance
, because the objective for implementation chosen is image, so the present invention directly adopts
as Fig. 3shown reparation flow process, the method mainly comprises the steps:
1), damaged area detection is carried out to input picture and obtain area to be repaired
, and expansion is carried out to the region detected obtain
, pre-service obtains image
;
Wherein,
for the Zone Full of input picture;
2), set up Lattice Boltzmann method model, initial equilibrium state distribution function is set
;
3) iterations of L model evolution equation, is preset
=500 and iteration ends threshold value
=0.03;
4), the coefficient of diffusion of design Lattice Boltzmann method model evolution equation,
with
, wherein
.When
time, calculate the combination of three kinds of coefficient of diffusion
, respectively above-mentioned four kinds of coefficient of diffusion are applied to Lattice Boltzmann method model evolution equation, and repairing effect are compared;
5), according to the D2Q9 model in Lattice Boltzmann method model,
as Fig. 4shown in, upgrade the equilibrium distribution function in Lattice Boltzmann method model evolution equation
;
6) transition process in Lattice Boltzmann method model, is calculated:
;
7) collision process in Lattice Boltzmann method model, is calculated:
8), each Nodes particle density of Renewal model
, and upgrade the peripheral information of area to be repaired, obtaining the image after upgrading is:
9), judge whether to meet iteration threshold condition, if do not meet stopping criterion for iteration, enter step 10), otherwise repair end;
10), judge whether iterations reaches 500, if do not reached, go to step 4), and repeat step 4)-9), until terminate after reaching iterations 500 to repair;
To be repaired complete after, export repairing effect,
as Fig. 9shown in.The hardware condition of above-mentioned enforcement is processor Intel Core i5-2450M, dominant frequency 2.50GHz, internal memory 4G; Software merit rating is Window7 system and matlab2012b; For the experiment of the spot distortion of gray level image, iterations and iteration time needed for four kinds of different coefficient of diffusion
as Fig. 1shown in 2.
Embodiment 2: repair the older picture having cut flaw
fig. 8for the image of cut distortion, implementation step is consistent with above-mentioned, the reparation result utilizing above-mentioned four kinds of coefficient of diffusion to carry out the reparation of Lattice Boltzmann method model to obtain
as Fig. 1shown in 0,
fig. 1different coefficient of diffusion to be completed the iterations needed for reparation and compares repair time by 3.
Experiment display, for the reparation of image and video flaw, adopt the recovery scenario based on Lattice Boltzmann method model can obtain good repairing effect, and improve remediation efficiency, the method can better be applied in engineer applied.
Claims (7)
1. the video image flaw restorative procedure based on Lattice Boltzmann method model novel cross-platform fast, is characterized in that the operation steps that video image flaw is repaired is as follows:
1), input is containing image defective
;
2), judge the form repairing object, if image repair then proceeds to step 5) reparation, otherwise enter step 3);
3), judge input be digital video or analog video, if analog video is then first converted into digital video by video frequency collection card, then proceed to step 4), otherwise, directly enter step 4);
4), video is decomposed framing;
5), the Lattice Boltzmann Method proposed is used to repair to the every two field picture decomposed;
6), required output format is judged, if required output is video, then synthetic video, otherwise direct output image
.
2. according to right 1 based on the video image flaw restorative procedure of Lattice Boltzmann method model, it is characterized in that:
Described video packets contains numeral or the analog video of MPEG-1, MPEG-4, AVI, RM, ASF, WMV, MOV or MKV form; Described image comprises the analog or digital image that all kinds of remotely sensed image or medical instrument imaging obtain, and its form comprises bmp, jpg, gif or png; Described flaw source comprises the color distortion that imaging device fault causes, the fading of older picture, folding line, the aging broadcasting flicker caused of film, the spot of old television data, cut.
3., according to claim 1 based on the video image flaw restorative procedure of Lattice Boltzmann method model, it is characterized in that:
Described Lattice Boltzmann method model: solve diffusion equation in the relaxation factor of Lattice Boltzmann method model be achieved by the edge of image cut-off function is embedded into, conventional two-dimensional lattice Boltzmann model comprises D2Q5---two dimension 5 direction diffusions, D2Q9, three-dimensional Lattice Boltzmann method model comprises D3Q7, D3Q15.
4., according to claim 1 based on the video image flaw restorative procedure of Lattice Boltzmann method model, it is characterized in that:
The described video image flaw reparation based on Lattice Boltzmann method model can at the hardware computing platform comprising PC or server, and runs in the operation system of software comprising Android, iOS, Linux or Windows.
5. the video based on Lattice Boltzmann method model according to claim 1 and image repair method, is characterized in that the concrete operation step of described step 5) is as follows:
1), judge to detect damaged area
, and expansion is carried out to the region detected obtain
, pre-service obtains image
2), set up LB model, initial equilibrium state distribution function is set
;
3) iterations of LB EVOLUTION EQUATION, is established
, and stopping criterion for iteration and iteration ends threshold value
;
4) coefficient of diffusion of LB EVOLUTION EQUATION, is designed
;
5), according to the D2Q9 model in LB model, the equilibrium distribution function in LB EVOLUTION EQUATION is upgraded
;
6) transition process in LB model, is calculated:
;
7) mechanism in LB model, is calculated:
；
8), each Nodes particle density of Renewal model
, and the peripheral information of area to be repaired;
9), judge whether to meet stopping criterion for iteration, if do not meet iteration stopping condition, enter step 10), otherwise repair end;
10), judge whether iterations reaches N, if do not reach iterations N, go to step 4), and repeat step 4)-9), until terminate after reaching iterations N to repair.
6., according to claim 3 based on the video image restorative procedure of Lattice Boltzmann method model, it is characterized in that:
Described edge cut-off function is non-negative monotonic decreasing function, image gradient information and curvature information can be utilized, slow down in boundary rate of propagation, at level and smooth place, rate of propagation is accelerated, and makes all have good performance to rate of propagation and edge hold facility in diffuse images process.
7., according to claim 3 based on the video image restorative procedure of Lattice Boltzmann method model, it is characterized in that:
Described relaxation factor controls rate of diffusion, realizes nonlinear diffusion, to reach protection image border characteristic, the object of restraint speckle; The design of equilibrium distribution function realizes anisotropy parameter or isotropic diffusion.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410733849.XA CN104867110A (en) | 2014-12-08 | 2014-12-08 | Lattice Boltzmann model-based video image defect repairing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410733849.XA CN104867110A (en) | 2014-12-08 | 2014-12-08 | Lattice Boltzmann model-based video image defect repairing method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104867110A true CN104867110A (en) | 2015-08-26 |
Family
ID=53912927
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410733849.XA Pending CN104867110A (en) | 2014-12-08 | 2014-12-08 | Lattice Boltzmann model-based video image defect repairing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104867110A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106251312A (en) * | 2016-08-09 | 2016-12-21 | 央视国际网络无锡有限公司 | Incomplete automatically benefit of a kind of picture paints method |
| CN106469441A (en) * | 2016-09-11 | 2017-03-01 | 江苏师范大学 | A kind of image synchronization noise reduction Enhancement Method based on LBM |
| CN111524077A (en) * | 2020-04-17 | 2020-08-11 | 三维六度(北京)文化有限公司 | Method and system for repairing image data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101673393A (en) * | 2009-09-25 | 2010-03-17 | 上海大学 | Image de-noising method based on lattice Boltzmann model |
| KR20110046904A (en) * | 2009-10-29 | 2011-05-06 | 삼성전자주식회사 | Apparatus and method for inpainting image by restricting reference image region |
| CN102163321A (en) * | 2011-06-15 | 2011-08-24 | 上海大学 | Image segmentation method based on lattice Boltzman model |
| US20120224781A1 (en) * | 2011-03-02 | 2012-09-06 | Xue-Cheng Tai | Methods and systems for generating enhanced images using euler's elastica model |
-
2014
- 2014-12-08 CN CN201410733849.XA patent/CN104867110A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101673393A (en) * | 2009-09-25 | 2010-03-17 | 上海大学 | Image de-noising method based on lattice Boltzmann model |
| KR20110046904A (en) * | 2009-10-29 | 2011-05-06 | 삼성전자주식회사 | Apparatus and method for inpainting image by restricting reference image region |
| US20120224781A1 (en) * | 2011-03-02 | 2012-09-06 | Xue-Cheng Tai | Methods and systems for generating enhanced images using euler's elastica model |
| CN102163321A (en) * | 2011-06-15 | 2011-08-24 | 上海大学 | Image segmentation method based on lattice Boltzman model |
Non-Patent Citations (1)
| Title |
|---|
| 张蕊 等: "图像修复的格子波尔兹曼方法", 《信息技术及图像处理》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106251312A (en) * | 2016-08-09 | 2016-12-21 | 央视国际网络无锡有限公司 | Incomplete automatically benefit of a kind of picture paints method |
| CN106251312B (en) * | 2016-08-09 | 2019-04-23 | 央视国际网络无锡有限公司 | A kind of incomplete mend automatically of picture draws method |
| CN106469441A (en) * | 2016-09-11 | 2017-03-01 | 江苏师范大学 | A kind of image synchronization noise reduction Enhancement Method based on LBM |
| CN111524077A (en) * | 2020-04-17 | 2020-08-11 | 三维六度(北京)文化有限公司 | Method and system for repairing image data |
| CN111524077B (en) * | 2020-04-17 | 2023-11-17 | 三维六度(北京)文化有限公司 | Method and system for repairing image data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109919008A (en) | Moving target detecting method, device, computer equipment and storage medium | |
| CN101980285B (en) | Method for restoring non-local images by combining GMRF priori | |
| CN103208123B (en) | Image partition method and system | |
| WO2017201751A1 (en) | Hole filling method and device for virtual viewpoint video or image, and terminal | |
| CN102999887A (en) | Sample based image repairing method | |
| CN103578085B (en) | Image cavity region based on variable-block method for repairing and mending | |
| CN103679749A (en) | Moving target tracking based image processing method and device | |
| CN115393727B (en) | Pavement linear crack identification method, electronic equipment and storage medium | |
| CN103914561B (en) | A kind of image search method and device | |
| CN102196292A (en) | Human-computer-interaction-based video depth map sequence generation method and system | |
| EP3182369A1 (en) | Stereo matching method, controller and system | |
| Li et al. | Image inpainting with salient structure completion and texture propagation | |
| CN103024421A (en) | Method for synthesizing virtual viewpoints in free viewpoint television | |
| CN104867110A (en) | Lattice Boltzmann model-based video image defect repairing method | |
| CN101674397A (en) | Repairing method of scratch in video sequence | |
| CN111192241B (en) | Quality evaluation method and device for face image and computer storage medium | |
| CN102722875A (en) | Visual-attention-based variable quality ultra-resolution image reconstruction method | |
| CN102194222A (en) | Image reconstruction method based on combination of motion estimation and super-resolution reconstruction | |
| CN107948586B (en) | Trans-regional moving target detecting method and device based on video-splicing | |
| WO2023221608A1 (en) | Mask recognition model training method and apparatus, device, and storage medium | |
| CN105224914B (en) | It is a kind of based on figure without constraint video in obvious object detection method | |
| CN108537868A (en) | Information processing equipment and information processing method | |
| CN102800077B (en) | Bayes non-local mean image restoration method | |
| CN113159046B (en) | Ballastless track bed foreign matter detection method and device | |
| CN115270184A (en) | Video desensitization method, vehicle video desensitization method and vehicle-mounted processing system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| EXSB | Decision made by sipo to initiate substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150826 |