CN107316283A - Algorithm is repaired based on the digital picture that multiwindow is merged - Google Patents
Algorithm is repaired based on the digital picture that multiwindow is merged Download PDFInfo
- Publication number
- CN107316283A CN107316283A CN201710575157.0A CN201710575157A CN107316283A CN 107316283 A CN107316283 A CN 107316283A CN 201710575157 A CN201710575157 A CN 201710575157A CN 107316283 A CN107316283 A CN 107316283A
- Authority
- CN
- China
- Prior art keywords
- repaired
- reparation
- algorithm
- window
- target pixel
- 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
- 238000004422 calculation algorithm Methods 0.000 title claims abstract description 32
- 230000008439 repair process Effects 0.000 claims abstract description 20
- 238000004364 calculation method Methods 0.000 claims abstract description 9
- 230000000694 effects Effects 0.000 description 7
- 230000006870 function Effects 0.000 description 6
- 238000000034 method Methods 0.000 description 6
- 230000015572 biosynthetic process Effects 0.000 description 4
- 238000003786 synthesis reaction Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 208000024754 bloodshot eye Diseases 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 201000005111 ocular hyperemia Diseases 0.000 description 1
- 238000012913 prioritisation Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The present invention relates to a kind of digital picture reparation algorithm merged based on multiwindow, including:For complex pattern I to be repaired, the reparation window ω under different sizes is setn;) repairing window ωnUnder, the reparation that algorithm completes complex pattern I to be repaired is repaired using Criminisi, the image after reparation is designated as In;Calculate the complex pattern U to be repaired after updatingI, calculate and repair variogram VI;Calculate target pixel points puPriority, and Criminisi repaired into the priority calculation of target pixel points in algorithm be changed to Pu(pu) calculation, then according to Criminisi Principles to UIRepaired, obtain final reparation result IP。
Description
Technical field
Field is restored the invention belongs to computer picture, is related to a kind of digital picture reparation calculation merged based on multiwindow
Method, available for historical relic's protection, video display special effect making etc..
Background technology
Digital Image Inpainting is the process for entering row information filling to loss of learning region in image to be repaired, its purpose
It is in order to realize the recovery of breakage image, while so that observer can not perceive the vestige that image is repaired.Digital picture is repaiied
Recovering technology is applied to various fields, to repair film, the rotten film deteriorated of reduction in photography and moviemaking, simultaneously
It can also be used to eliminate blood-shot eye illness, the date on photo, digital watermarking etc..
Digital Image Inpainting at this stage is broadly divided into two major classes, and a class is the digital picture based on partial differential equation
Algorithm is repaired, another kind of is that the digital picture based on textures synthesis repairs algorithm.Calculated compared to the reparation based on partial differential equation
Method, the digital picture based on textures synthesis, which repairs algorithm, has many significant advantages, and it has taken into account the texture and structure of image
It is easy to the realization that big region breakage image is repaired.Criminisi algorithms are the current the most frequently used digitized maps based on textures synthesis
As repairing algorithm, but there is certain defect in the algorithm, and it needs the filling that texture is carried out by the size of specified restoration window
With synthesis, when reparation window changes, easily cause and repair the larger deviation of result appearance.Therefore single reparation window is used
The Criminisi algorithms repaired need the size of rational selection reparation window.
The content of the invention
The problem of present invention exists for Criminisi algorithms proposes a kind of digital picture reparation merged based on multiwindow
Algorithm, by being merged to the reparation result under different repairing sizes windows, updates the texture and knot of broken nomogram picture to be repaired
Structure information, then completes the digital picture reparation under multiwindow fusion.Technical scheme is as follows:
It is a kind of that algorithm is repaired based on the digital picture that multiwindow is merged, comprise the following steps:
(1) for complex pattern I to be repaired, the reparation window ω under different sizes is setn, wherein n is the sequence number for repairing window
And n ∈ { 1 ..., N }, N are the number of window;
(2) window ω is being repairednUnder, the reparation that algorithm completes complex pattern I to be repaired is repaired using Criminisi, after reparation
Image be designated as In;
(3) according to formulaCalculate after updating
Complex pattern U to be repairedI, wherein d is compares order, and f is comparison function, while according to formulaCalculating is repaiied
Multiple variogram VI;
(4) according to formula Pu(pu)=Cu(pu)·Du(pu)·Eu(pu) calculate target pixel points puPriority Pu(pu),
Wherein Cu(pu) it is target pixel points puConfidence level, Du(pu) it is target pixel points puData item,
ε is constant, and the Criminisi priority calculations for repairing target pixel points in algorithm are changed into Pu(pu) calculation,
Then according to Criminisi Principles to UIRepaired, obtain final reparation result IP。
In a word, a kind of deficiency that the present invention exists for Criminisi algorithms, it is proposed that numeral merged based on multiwindow
Image repair algorithm, the reparation result according to multiwindow carries out the renewal of breakage image to be repaired, while believing according to variance is repaired
Cease to calculate the priority in traditional Criminisi algorithms and be improved.The present invention results in more accurate digitized map
As repairing effect, have a wide range of applications.
Brief description of the drawings
Fig. 1 present invention's repairs algorithm flow chart based on the digital picture that multiwindow is merged.
Fig. 2 repairs algorithm with tradition Criminisi digital pictures for the present invention and the repairing effect of image is contrasted, wherein:
Scheme (a) breakage image to be repaired (red area represents damaged area);
Scheme the repairing effect that (b) is tradition Criminisi algorithms;
It is the inventive method repairing effect to scheme (c).
Embodiment
The present invention repairs algorithm based on the digital picture that multiwindow is merged, and is mainly made up of three parts:Under many size windows
Digital picture reparation, the renewal of complex pattern to be repaired and repair the calculating of covariance information, priority and calculate the improvement of function.Specifically
Step and principle are as follows:
101:Various sizes of reparation window is set;
The size for repairing window is ωn, n=1 ..., N, n is repair the sequence number of window, and N is the number of window.Using
Criminsi repairs algorithm and image I to be repaired is repaired, and the image after reparation is designated as In:
102:Reparation window size is ωnUnder complex pattern I damaged area to be repaired and source region mark;
Complex pattern I to be repaired damaged area is designated as Ωn, edge is designated asTarget pixel points pnForOn it is any one
Point, source region is Φn, and meet relation Φn+Ωn=I.
103:Renewal to relevant parameter in I repair processes;
Pn(pn)=Cn(pn)·Dn(pn)
Wherein Pn(pn) it is target pixel points pnPriority, Cn(pn) it is target pixel points pnConfidence level, ωn(pn)
For with pnCentered on pixel, window size is ωnAny pixel composition object block;vnFor ωn(pn) and source region Φn
Pixel in common factor;Dn(pn) it is target pixel points pnData item;Represent pnPoint isophote direction vector,Respectively I level, the gradient of vertical direction;t(pn) it is damaged boundary in pnThe normal vector of the tangent line at place;α is
Constant, is worth for 255.
The maximum point of target pixel points priority is calculated according to equation below
ObtainAfterwards, its corresponding object blockMultiblock as to be repaired, calculates to be repaired according to similarity measurements flow function
The corresponding Optimum Matching point of multiblockAnd blocks and optimal matching blocks
Wherein, qnIt is Φ for source regionnInterior any pixel.
Then reparations is filled to damaged multiblock to be repaired using blocks and optimal matching blocks, and multiblock to be repaired is labeled as to have repaiied
Multiple sample block, updates damaged area and source region:
Obtain blocks and optimal matching blocksAfterwards, to damaged multiblock to be repairedReparation is filled, i.e.,:
Then damaged area and source region are updated, new damaged area Ω ' is respectively obtainednAnd new source region Φ 'nI.e.:
Then new damaged area is again marked as damaged area, new source region is again marked as source region, i.e.,:
Ωn=Ω 'n
Φn=Φ 'n
Update the confidence value for having repaired respective pixel point in sample block;
According to below equation to having repaired the confidence value of respective pixel point in sample block,
104:102-103 steps are repeated, until damaged area is empty set, the reparation of image are completed.
It is ω to obtain repairing window sizenUnder reparation result In。
105:The renewal of complex pattern to be repaired and the calculating for repairing variogram;
Part similar in result, which will be repaired, to be used to update complex pattern I to be repaired, and the complex pattern to be repaired after renewal is designated as UI, together
Shi Tongji UIIn each pixel in I1,...,INBrightness value variance in correspondence position, obtains repairing variogram VI:
Wherein z is pixel point coordinates, and d is compares order, and the complex pattern to be repaired after renewal is designated as VIN number of reparation is merged
As a result, and count with I1On the basis of, repair the covariance information of image.VI(z) it is smaller, illustrate that z points are inclined in multiple reparation results
It is poor little, row major reparation can be entered in follow-up reparation.Therefore it is rational in the calculating of priority to add covariance information
It can ensure that z points are preferentially repaired.
106:To UIRepaired using based on the Criminisi algorithms for improving prioritization functions, mark repairs window chi
Very little is ω1Under complex pattern U to be repairedIDamaged area and source region;
Complex pattern U to be repairedIDamaged area be designated as Ωu, edge is designated asTarget pixel points puForOn it is any one
Point, source region is Φu, and meet relation Φu+Ωu=UI。
107:To UIThe renewal of relevant parameter in repair process;
Pu(pu)=Cu(pu)·Du(pu)·Eu(pu)
Wherein Pu(pu) it is target pixel points puPriority, Cu(pu) it is target pixel points puConfidence level, ωu(pu)
For with puCentered on pixel, window size is ωuAny pixel composition object block;vuFor ωu(pu) and source region Φu
Pixel in common factor;Du(pu) it is target pixel points puData item;Eu(pu) it is target pixel points puReparation covariance information
VI(pu) merged with what constant ε (value is taken as 1) was constituted, it has reacted puThe deviation journey between result is repaired in previous multiwindow
Degree, is worth bigger explanation and deviates smaller, preferentially repair;Represent puThe illumination line direction of point,Point
Wei not UILevel, the gradient of vertical direction;t(pu) it is damaged boundary in puThe normal vector of the tangent line at place;α is constant, is worth and is
255。
The maximum point of target pixel points priority is calculated according to equation below
ObtainAfterwards, its corresponding object blockMultiblock as to be repaired, calculates to be repaired according to similarity measurements flow function
The corresponding Optimum Matching point of multiblockAnd blocks and optimal matching blocks
Wherein, quIt is Φ for source regionuInterior any pixel.
Then reparations is filled to damaged multiblock to be repaired using blocks and optimal matching blocks, and multiblock to be repaired is labeled as to have repaiied
Multiple sample block, updates damaged area and source region:
Obtain blocks and optimal matching blocksAfterwards, to damaged multiblock to be repairedReparation is filled, i.e.,:
Then damaged area and source region are updated, new damaged area Ω ' is respectively obtaineduAnd new source region Φ 'uI.e.:
Then new damaged area is again marked as damaged area, new source region is again marked as source region, i.e.,:
Ωu=Ω 'u
Φu=Φ 'u
Update the confidence value for having repaired respective pixel point in sample block;
According to below equation to having repaired the confidence value of respective pixel point in sample block,
108:106-107 steps are repeated, until damaged area is empty set, the reparation of image are completed.
Obtain final reparation result IP。
The feasibility of this method is verified with specific experiment below, it is described below:
Result of the test is that this method is Intel i7-3610QM, 2.3GHz in CPU, inside saves as 16G notebook computer
Obtained by upper operation, operating system is Windows 7, and simulation software is 64 Matlab R2012b.
From figure 2 it can be seen that the repairing effect obtained using traditional Criminisi reparation algorithms is not good, in chair
Edge exist it is obvious repair by mistake, and the repairing effect that the present invention is obtained after multiwindow reparation fusion is closed more naturally
Reason.
It will be appreciated by those skilled in the art that accompanying drawing is the schematic diagram of a preferred embodiment, the embodiments of the present invention
Sequence number is for illustration only, and the quality of embodiment is not represented.
The foregoing is only presently preferred embodiments of the present invention, be not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent substitution and improvements made etc. should be included in the scope of the protection.
The idiographic flow of the present invention is as follows:
(1) for complex pattern I to be repaired, the reparation window ω under different sizes is setn, wherein n is the sequence number for repairing window
And n ∈ { 1 ..., N }, N are the number of window;
(2) window ω is being repairednUnder, the reparation that algorithm completes complex pattern I to be repaired is repaired using Criminisi, after reparation
Image be designated as In;
(3) according to formulaCalculate after updating
Complex pattern U to be repairedI, wherein d is compares order, and f is comparison function, while according to formulaCalculating is repaiied
Multiple variogram VI;
(4) according to formula Pu(pu)=Cu(pu)·Du(pu)·Eu(pu) calculate target pixel points puPriority Pu(pu),
Wherein Cu(pu) it is target pixel points puConfidence level, Du(pu) it is target pixel points puData item,
ε is constant, and the Criminisi priority calculations for repairing target pixel points in algorithm are changed into Pu(pu) calculation,
Then according to Criminisi Principles to UIRepaired, obtain final reparation result IP。
Claims (1)
1. a kind of repair algorithm based on the digital picture that multiwindow is merged, comprise the following steps:
(1) for complex pattern I to be repaired, the reparation window ω under different sizes is setn, wherein n is the sequence number and n ∈ for repairing window
{ 1 ..., N }, N is the number of window;
(2) window ω is being repairednUnder, the reparation that algorithm completes complex pattern I to be repaired, the image after reparation are repaired using Criminisi
It is designated as In;
(3) according to formulaCalculate waiting after updating
Repair image UI, wherein d is compares order, and f is comparison function, while according to formulaCalculate reparation side
Difference figure VI;
(4) according to formula Pu(pu)=Cu(pu)·Du(pu)·Eu(pu) calculate target pixel points puPriority Pu(pu), wherein
Cu(pu) it is target pixel points puConfidence level, Du(pu) it is target pixel points puData item,ε
For constant, and the Criminisi priority calculations for repairing target pixel points in algorithm are changed to Pu(pu) calculation,
Then according to Criminisi Principles to UIRepaired, obtain final reparation result IP。
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710575157.0A CN107316283B (en) | 2017-07-14 | 2017-07-14 | Digital image restoration method based on multi-window fusion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710575157.0A CN107316283B (en) | 2017-07-14 | 2017-07-14 | Digital image restoration method based on multi-window fusion |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107316283A true CN107316283A (en) | 2017-11-03 |
CN107316283B CN107316283B (en) | 2020-08-14 |
Family
ID=60178786
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710575157.0A Expired - Fee Related CN107316283B (en) | 2017-07-14 | 2017-07-14 | Digital image restoration method based on multi-window fusion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107316283B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110223252A (en) * | 2019-06-10 | 2019-09-10 | 成都理工大学 | A kind of depth image reparation algorithm based on compound adaptive region growth criterion |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5268771A (en) * | 1989-02-15 | 1993-12-07 | Hitachi, Ltd. | Image restoration processing system and method |
US20040022449A1 (en) * | 2002-08-02 | 2004-02-05 | Frank Olschewski | Method, arrangement, and software for optimizing the image quality of movable subjects imaged with a microscope |
CN101661613A (en) * | 2009-08-27 | 2010-03-03 | 北京交通大学 | Image restoration method based on image segmentation, and system therefor |
CN102760285A (en) * | 2012-05-31 | 2012-10-31 | 河海大学 | Image restoration method |
US8447135B2 (en) * | 2011-03-02 | 2013-05-21 | Nanyang Technological University | Methods and systems for generating enhanced images using Euler's Elastica model |
CN103700093A (en) * | 2013-12-03 | 2014-04-02 | 杭州电子科技大学 | Criminisi image restoration method based on textures and edge features |
CN104680493A (en) * | 2015-03-12 | 2015-06-03 | 华东理工大学 | Digital image repairing method based on scale optimization |
CN104809698A (en) * | 2015-03-18 | 2015-07-29 | 哈尔滨工程大学 | Kinect depth image inpainting method based on improved trilateral filtering |
CN105389781A (en) * | 2015-10-30 | 2016-03-09 | 哈尔滨工程大学 | Vehicle light repairing method based on saliency detection and Criminisi algorithm |
CN105678712A (en) * | 2016-01-29 | 2016-06-15 | 陕西师范大学 | Improved Criminisi image restoration method with self-adaptive gradient block division and equilong transformation |
-
2017
- 2017-07-14 CN CN201710575157.0A patent/CN107316283B/en not_active Expired - Fee Related
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5268771A (en) * | 1989-02-15 | 1993-12-07 | Hitachi, Ltd. | Image restoration processing system and method |
US20040022449A1 (en) * | 2002-08-02 | 2004-02-05 | Frank Olschewski | Method, arrangement, and software for optimizing the image quality of movable subjects imaged with a microscope |
CN101661613A (en) * | 2009-08-27 | 2010-03-03 | 北京交通大学 | Image restoration method based on image segmentation, and system therefor |
US8447135B2 (en) * | 2011-03-02 | 2013-05-21 | Nanyang Technological University | Methods and systems for generating enhanced images using Euler's Elastica model |
CN102760285A (en) * | 2012-05-31 | 2012-10-31 | 河海大学 | Image restoration method |
CN103700093A (en) * | 2013-12-03 | 2014-04-02 | 杭州电子科技大学 | Criminisi image restoration method based on textures and edge features |
CN104680493A (en) * | 2015-03-12 | 2015-06-03 | 华东理工大学 | Digital image repairing method based on scale optimization |
CN104809698A (en) * | 2015-03-18 | 2015-07-29 | 哈尔滨工程大学 | Kinect depth image inpainting method based on improved trilateral filtering |
CN105389781A (en) * | 2015-10-30 | 2016-03-09 | 哈尔滨工程大学 | Vehicle light repairing method based on saliency detection and Criminisi algorithm |
CN105678712A (en) * | 2016-01-29 | 2016-06-15 | 陕西师范大学 | Improved Criminisi image restoration method with self-adaptive gradient block division and equilong transformation |
Non-Patent Citations (2)
Title |
---|
ANUPAM等: "Fast and Enhanced Algorithm for Exemplar Based Image Inpainting", 《2010 FOURTH PACIFIC-RIM SYMPOSIUM ON IMAGE AND VIDEO TECHNOLOGY》 * |
张晴: "基于样本的数字图像修复技术研究", 《中国博士学位论文全文数据库(信息科技辑)》 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110223252A (en) * | 2019-06-10 | 2019-09-10 | 成都理工大学 | A kind of depth image reparation algorithm based on compound adaptive region growth criterion |
CN110223252B (en) * | 2019-06-10 | 2022-12-06 | 成都理工大学 | Depth image restoration algorithm based on composite self-adaptive region growing criterion |
Also Published As
Publication number | Publication date |
---|---|
CN107316283B (en) | 2020-08-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104463786B (en) | A kind of mobile robot image split-joint method and device | |
CN103700101B (en) | Non-rigid brain image registration method | |
CN110111248B (en) | Image splicing method based on feature points, virtual reality system and camera | |
CN101571950B (en) | Image restoring method based on isotropic diffusion and sparse representation | |
CN109974743B (en) | Visual odometer based on GMS feature matching and sliding window pose graph optimization | |
WO2017177514A1 (en) | Mura phenomenon compensation method for display panel | |
CN110246100B (en) | Image restoration method and system based on angle sensing block matching | |
CN104200444B (en) | Image repair method based on balanced sample block | |
US20170309075A1 (en) | Image to item mapping | |
CN112991176B (en) | Panoramic image splicing method based on optimal suture line | |
CN103886561B (en) | Criminisi image inpainting method based on mathematical morphology | |
CN102760285A (en) | Image restoration method | |
CN104376535A (en) | Rapid image repairing method based on sample | |
CN107507232B (en) | Stereo matching method based on multi-scale iteration | |
CN106934780A (en) | A kind of automatic watermark minimizing technology based on image repair | |
CN108197567B (en) | Method, apparatus and computer readable medium for image processing | |
CN103413339B (en) | 1000000000 pixel high dynamic range images are rebuild and the method for display | |
CN115456990B (en) | CT image-based rib counting method, device, equipment and storage medium | |
CN103578085A (en) | Image empty region inpainting method based on variable blocks | |
CN110021000A (en) | Hair line restorative procedure and device based on figure layer deformation | |
CN106407978A (en) | Unconstrained in-video salient object detection method combined with objectness degree | |
CN107316283A (en) | Algorithm is repaired based on the digital picture that multiwindow is merged | |
CN107169498A (en) | It is a kind of to merge local and global sparse image significance detection method | |
CN112686872B (en) | Wood counting method based on deep learning | |
CN106504203A (en) | A kind of Block- matching image repair method of application screening strategy |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200814 |