CN107507137A - A kind of image repair method and system - Google Patents
A kind of image repair method and system Download PDFInfo
- Publication number
- CN107507137A CN107507137A CN201710599116.5A CN201710599116A CN107507137A CN 107507137 A CN107507137 A CN 107507137A CN 201710599116 A CN201710599116 A CN 201710599116A CN 107507137 A CN107507137 A CN 107507137A
- Authority
- CN
- China
- Prior art keywords
- image
- defect
- repaired
- repair
- algorithm
- 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
- 230000008439 repair process Effects 0.000 title claims abstract description 194
- 238000000034 method Methods 0.000 title claims abstract description 63
- 230000007547 defect Effects 0.000 claims abstract description 344
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 162
- 238000002156 mixing Methods 0.000 claims abstract description 33
- 238000012545 processing Methods 0.000 claims abstract description 6
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 claims description 22
- 238000006243 chemical reaction Methods 0.000 claims description 16
- 239000000203 mixture Substances 0.000 claims description 10
- 239000012141 concentrate Substances 0.000 claims description 7
- 238000001914 filtration Methods 0.000 claims description 7
- 230000007704 transition Effects 0.000 claims description 7
- 238000005266 casting Methods 0.000 claims description 3
- 230000002950 deficient Effects 0.000 claims description 3
- 239000002131 composite material Substances 0.000 abstract 1
- 238000004364 calculation method Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 239000011159 matrix material Substances 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000000342 Monte Carlo simulation Methods 0.000 description 2
- 230000001174 ascending effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005315 distribution function Methods 0.000 description 2
- 206010003671 Atrioventricular Block Diseases 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 239000000428 dust Substances 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 239000003755 preservative agent Substances 0.000 description 1
- 230000002335 preservative effect Effects 0.000 description 1
- 230000036632 reaction speed Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000000779 smoke Substances 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012360 testing method Methods 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/77—Retouching; Inpainting; Scratch removal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of image repair method and system, method includes：Defect area lookup is carried out to image to be repaired, obtains the defect area of complex pattern to be repaired；A kind of reparation algorithm as image to be repaired is selected from scatterplot defect repair algorithm, line defect repair algorithm, block defect repair algorithm and mixing defect repair algorithm according to the defect area of complex pattern to be repaired；Image repair is carried out to image to be repaired according to the algorithm of repairing of selection；System includes defect area searching modul, selecting module and repair module.The present invention is additionally arranged defect area and searches and choose the different processes for repairing algorithm according to defect area, it can be repaired for different types of Incomplete image using different reparation algorithms, it is more efficient and more flexible, meet people's high request growing to image repair.It the composite can be widely applied to image processing field.
Description
Technical field
The present invention relates to image processing field, especially a kind of image repair method and system.
Background technology
In today that information is flourishing, people will check the picture of magnanimity daily, and be found certainly in the image information of magnanimity
The information that oneself needs.However, the image information of magnanimity, when network transmission, artificial preservation, image understands generating unit unavoidably
Information is divided to lose.In addition, in current shooting application, camera is due to being blocked (such as dust, cut, mould by trickle foreign object
Point, water smoke etc.), the picture parts content distortion that actually photographed can be made, influence photographic effects.Higher is being required to axis information
Field of face identification, solved frequently with artificial cleaning camera lens or the mode for changing camera, cause that rehabilitation cost is high, reaction speed
Degree is slow, poor user experience.As can be seen here, the defect of image all can to the vision of people and daily life, study and work
Bring inconvenience.
To solve the deletion problem of image, image repair technology is arisen at the historic moment.Image repair refers to those in partial zones
Numeric field data partial loss or the image lost completely, using image information intact in image, calculated by various means
Or the data of loss are approached, the original information of lost regions is estimated, to fill up area to be repaired so that entire image, which reaches, to be regarded
It is continuous and complete in feel.
Existing image repair method can only use single reparation side for the certain types of Incomplete image of certain class mostly
Method is repaired, it is impossible to repaired for different types of Incomplete image using different restorative procedures, it is less efficient and not
It is enough flexible, people's high request growing to image repair can not be met.
The content of the invention
In order to solve the above technical problems, it is an object of the invention to：A kind of efficiency high and flexible, image repair side are provided
Method.
Another object of the present invention is to：A kind of efficiency high and flexible, image repair system are provided.
The technical solution used in the present invention is：
A kind of image repair method, comprises the following steps：
Defect area lookup is carried out to image to be repaired, obtains the defect area of complex pattern to be repaired；
According to the defect area of complex pattern to be repaired from scatterplot defect repair algorithm, line defect repair algorithm, block defect repair
A kind of reparation algorithm as image to be repaired is selected in algorithm and mixing defect repair algorithm；
Image repair is carried out to image to be repaired according to the algorithm of repairing of selection.
Further, lacked according to the defect area of complex pattern to be repaired from scatterplot defect repair algorithm, line defect repair algorithm, block
Damage repairs algorithm and mixes the step for a kind of reparation algorithm as image to be repaired is selected in defect repair algorithm, and it has
Body is：
If the defect area of complex pattern to be repaired is scatterplot shape defect area, scatterplot defect repair algorithm is selected as to be repaired
The reparation algorithm of multiple image；If the defect area of complex pattern to be repaired is wire defect area, selection line defect repair algorithm
Reparation algorithm as image to be repaired；If the defect area of complex pattern to be repaired is block defect area, selection block defect
Repair reparation algorithm of the algorithm as image to be repaired；Lacked if the defect area of complex pattern to be repaired is point, line and the mixing in face
Region is damaged, then reparation algorithm of the selection mixing defect repair algorithm as image to be repaired.
Further, described to repair the step for algorithm carries out image repair to image to be repaired according to selection, it is wrapped
Include：
Image to be repaired is pre-processed, obtains pretreated image；
Scatterplot scanning is carried out to pretreated image, obtains defect scatterplot；
Defect scatterplot is repaired using median filtering method, the image after being repaired.
Further, described to repair the step for algorithm carries out image repair to image to be repaired according to selection, it is wrapped
Include：
Image to be repaired is pre-processed, obtains pretreated image；
Wire defect area scanning is carried out to pretreated image, obtains the pixel of defect；
The pixel of defect is repaired using Lagrange's interpolation, the image after being repaired.
Further, described to repair the step for algorithm carries out image repair to image to be repaired according to selection, it is wrapped
Include：
Image to be repaired is pre-processed, obtains pretreated image；
Two-dimensional pixel one-dimensional is carried out to pretreated image, obtains one-dimensional pixel sequence；
Subsequence division, feature space structure and pixel division are carried out to one-dimensional pixel sequence, obtains defect pixel data
Collection and non-defect pixel data set；
According to non-defect pixel data set using the arest neighbors restorative procedure based on Euclidean distance to defect pixel data set
Repaired, the image after being repaired.
Further, described to repair the step for algorithm carries out image repair to image to be repaired according to selection, it is wrapped
Include：
Image to be repaired is carried out to mix defect area scanning, obtains defect pixel；
Coordinate of the defect pixel in image to be repaired is recorded, and using the bivariate normal integral set to lacking
The coordinate of damage pixel is handled, and generates the two-dimensional coordinate of random pixel point；
Be color value by the grayvalue transition of random pixel point according to the gray scale of setting and color conversion formula, obtain by
The color set of the color value composition of random pixel point, and statistical color concentrates the probability of each color value；
The maximum color value of probable value is taken out from color set as current color value；
The pixel of several arest neighbors around current color value and defect pixel is formed into the first stochastic variable, and obtained
The second stochastic variable that the pixel of several arest neighbors forms to around by defect pixel；
The mathematic expectaion and the mathematic expectaion and variance of variance and the second stochastic variable of the first stochastic variable are calculated respectively；
Calculated according to the mathematic expectaion and variance of the mathematic expectaion of the first stochastic variable and variance and the second stochastic variable
Confidential interval；
Judge whether the first stochastic variable meets confidential interval, if so, next step is then performed, conversely, then by current face
Colour is cast out from color set, and returns to the step that the maximum color value of probable value is taken out from color set as current color value
Suddenly；
According to current color value and the gray scale of setting and the gray value of color conversion formula calculating defect pixel, and according to
The gray value of defect pixel is repaired, the image after being repaired.
Another technical scheme for being taken of the present invention is：
A kind of image repair system, including with lower module：
Defect area searching modul, for carrying out defect area lookup to image to be repaired, obtain complex pattern to be repaired
Defect area；
Selecting module, calculated for the defect area according to complex pattern to be repaired from scatterplot defect repair algorithm, line defect repair
A kind of reparation algorithm as image to be repaired is selected in method, block defect repair algorithm and mixing defect repair algorithm；
Repair module, image repair is carried out to image to be repaired for the algorithm of repairing according to selection.
Further, the selecting module is specifically used for performing following operate：
If the defect area of complex pattern to be repaired is scatterplot shape defect area, scatterplot defect repair algorithm is selected as to be repaired
The reparation algorithm of multiple image；If the defect area of complex pattern to be repaired is wire defect area, selection line defect repair algorithm
Reparation algorithm as image to be repaired；If the defect area of complex pattern to be repaired is block defect area, selection block defect
Repair reparation algorithm of the algorithm as image to be repaired；Lacked if the defect area of complex pattern to be repaired is point, line and the mixing in face
Region is damaged, then reparation algorithm of the selection mixing defect repair algorithm as image to be repaired.
Further, the repair module includes：
Pretreatment unit, for being pre-processed to image to be repaired, obtain pretreated image；
One-dimensional processing unit, for carrying out two-dimensional pixel one-dimensional to pretreated image, obtain one-dimensional pixel sequence
Row；
Division unit, divide, obtain for carrying out subsequence division, feature space structure and pixel to one-dimensional pixel sequence
Defect pixel data set and non-defect pixel data set；
First repairs unit, for using the arest neighbors restorative procedure based on Euclidean distance according to non-defect pixel data set
Defect pixel data set is repaired, the image after being repaired.
Further, the repair module includes：
Scanning element, for carrying out mixing defect area scanning to image to be repaired, obtain defect pixel；
Random pixel point two-dimensional coordinate generation unit, for recording coordinate of the defect pixel in image to be repaired,
And the coordinate of defect pixel is handled using the bivariate normal integral of setting, the two dimension seat of generation random pixel point
Mark；
Conversion and probability statistics unit, for the gray scale according to setting and color conversion formula, by the ash of random pixel point
Angle value is converted to color value, obtains the color set being made up of the color value of random pixel point, and statistical color concentrates each color value
Probability；
First chooses unit, for taking out the maximum color value of probable value from color set as current color value；
Stochastic variable generation unit, for by the pixel of several arest neighbors around current color value and defect pixel
The first stochastic variable is formed, and obtains the second random change being made up of the pixel of several arest neighbors around defect pixel
Amount；
It is expected with variance computing unit, for calculate respectively the first stochastic variable mathematic expectaion and variance and second with
The mathematic expectaion and variance of machine variable；
Confidential interval computing unit, for the mathematic expectaion according to the first stochastic variable and variance and the second stochastic variable
Mathematic expectaion and variance calculate confidential interval；
Confidence level judging unit, for judging whether the first stochastic variable meets confidential interval, repaiied if so, then performing second
The operation of multiple unit, conversely, then casting out current color value from color set, and return to first and choose unit；
Second repairs unit, for calculating defect pixel with color conversion formula according to current color value and the gray scale of setting
The gray value of point, and repaired according to the gray value of defect pixel, the image after being repaired.
The beneficial effects of the method for the present invention is：Including carrying out defect area lookup to image to be repaired, according to be repaired
The defect area of complex pattern is repaiied from scatterplot defect repair algorithm, line defect repair algorithm, block defect repair algorithm and mixing defect
A kind of reparation algorithm as image to be repaired and the reparation algorithm according to selection are selected in double calculation method to figure to be repaired
The step of as carrying out image repair, the step of defect area is searched and chooses different reparation algorithms according to defect area is additionally arranged,
It can be repaired for different types of Incomplete image using different reparation algorithms, it is more efficient and more flexible, meet
People's high request growing to image repair.
The beneficial effect of system of the present invention is：Including defect area searching modul, selecting module and repair module, set up
Defect area lookup is carried out in defect area searching modul and different reparations are chosen according to defect area in selecting module
The process of algorithm, it can be repaired for different types of Incomplete image using different reparation algorithms, it is more efficient and more
Flexibly, people's high request growing to image repair is met.
Brief description of the drawings
Fig. 1 is a kind of overall flow chart of steps of image repair method of the present invention；
Fig. 2 is the flow chart of scatterplot defect repair algorithm of the present invention；
Fig. 3 is the flow chart of line defect repair algorithm of the present invention；
Fig. 4 is the flow chart of block defect repair algorithm of the present invention；
Fig. 5 is the flow chart of present invention mixing defect repair algorithm；
Fig. 6 is a kind of overall structure block diagram of image repair system of the present invention；
Fig. 7 is the overall flow figure of the image repair method of the embodiment of the present invention one.
Embodiment
A kind of reference picture 1, image repair method, comprises the following steps：
Defect area lookup is carried out to image to be repaired, obtains the defect area of complex pattern to be repaired；
According to the defect area of complex pattern to be repaired from scatterplot defect repair algorithm, line defect repair algorithm, block defect repair
A kind of reparation algorithm as image to be repaired is selected in algorithm and mixing defect repair algorithm；
Image repair is carried out to image to be repaired according to the algorithm of repairing of selection.
Wherein, defect area search procedure, can be come by existing region segmentation, feature extraction with the methods of positioning true
The Defect types of fixed image to be repaired.
After the completion of image repair, the present invention can also show the image after reparation, in order to which user directly sees
Reparation result is examined, it is more convenient.
Be further used as preferred embodiment, according to the defect area of complex pattern to be repaired from scatterplot defect repair algorithm,
A kind of repairing as image to be repaired is selected in line defect repair algorithm, block defect repair algorithm and mixing defect repair algorithm
The step for double calculation method, it is specially：
If the defect area of complex pattern to be repaired is scatterplot shape defect area, scatterplot defect repair algorithm is selected as to be repaired
The reparation algorithm of multiple image；If the defect area of complex pattern to be repaired is wire defect area, selection line defect repair algorithm
Reparation algorithm as image to be repaired；If the defect area of complex pattern to be repaired is block defect area, selection block defect
Repair reparation algorithm of the algorithm as image to be repaired；Lacked if the defect area of complex pattern to be repaired is point, line and the mixing in face
Region is damaged, then reparation algorithm of the selection mixing defect repair algorithm as image to be repaired.
Reference picture 2, is further used as preferred embodiment, and the reparation algorithm according to selection is to image to be repaired
The step for carrying out image repair, it includes：
Image to be repaired is pre-processed, obtains pretreated image；
Scatterplot scanning is carried out to pretreated image, obtains defect scatterplot；
Defect scatterplot is repaired using median filtering method, the image after being repaired.
The present invention selects scatterplot defect repair algorithm to carry out image repair when defect area is scatterplot.Scatterplot defect
The process chart of algorithm is repaired as shown in Fig. 2 specifically including：
After the data of image to be repaired are read, the view data of reading need to be pre-processed, i.e., the figure of reading
As data are converted to two-dimensional matrix in order to subsequent operation.
Defect area scanning is proceeded by after the completion of pretreatment., can be white finding defect area for the ease of operation
Feature of the color as defect area.
After finding defect area, using the method for medium filtering, each defect area is handled, and intermediate value is filtered
The data that ripple obtains fill up the scatterplot of defect, so as to completing the reparation of image.
Wherein, median filter method is a kind of Nonlinear harmonic oscillator method, by from some sampling window in image
Taking-up odd number data, which are ranked up, obtains a result.Intermediate value refers to be in after odd number data in window are sequentially arranged by size
That number of center, and the median filter method then result using the intermediate value of window as processing, and being carried out to current pixel
Filling.
The present invention median filter method belong to two dimension median filter method, its window shape have it is a variety of, such as wire, side
Shape, cross, circle and rhombus etc., window of different shapes can produce different filter effects.
The median filter method of the present invention, the odd number data of taking-up can be carried out ascending sort, after then taking out sequence
Element value in ordered sequence centre position, and the point that the element value of taking-up is used to fill up defect, to repair scatterplot defect
Image and the image after reparation is shown in display interface, can also prompt the image after user's preservative restoration.
Reference picture 3, is further used as preferred embodiment, and the reparation algorithm according to selection is to image to be repaired
The step for carrying out image repair, it includes：
Image to be repaired is pre-processed, obtains pretreated image；
Wire defect area scanning is carried out to pretreated image, obtains the pixel of defect；
The pixel of defect is repaired using Lagrange's interpolation, the image after being repaired.
The present invention is when defect area is wire defect area (view data loss i.e. to be repaired is line), selection line
Defect repair algorithm carries out image repair.
What if view data to be repaired was lost is horizontal straight line, then the row where losing data is just defective region
Domain.Although the data of defect area full line all lost, for the data of a certain row, the data of loss only account for very little
Ratio, influence less.Therefore, when extracting the data of a certain row, the trend of permutation data there is no change, only exist
There is loss of data in indivedual points.Then, the lagrange-interpolation commonly used in interpolation algorithm can be just utilized, is estimated each
The data of that point of defect in row, so as to reach the purpose for repairing entire image.
With the data instance of a certain row, the picture of several each intact pixels before and after a row pixel defect point
Element value is used as a data acquisition system { y1,y2,y3,…,yn, and it is x to arrange line number corresponding to y, then according to lagrange-interpolation
The basic function L of its n times interpolation can be calculatedn(x)：
Wherein, lk(x), k=0,1 ..., n are polynomials of degree n, y0For set-point, and meet：
Obtain Lagrange interpolation function Ln(x) after, it is only necessary to which the row x where defect point is substituted into Ln(x), just can estimate
Go out the pixel value of defect pixel, to reach the effect of image repair.Specific process chart such as Fig. 3 institutes of line defect repair algorithm
Show.
Reference picture 4, is further used as preferred embodiment, and the reparation algorithm according to selection is to image to be repaired
The step for carrying out image repair, it includes：
Image to be repaired is pre-processed, obtains pretreated image；
Two-dimensional pixel one-dimensional is carried out to pretreated image, obtains one-dimensional pixel sequence；
Subsequence division, feature space structure and pixel division are carried out to one-dimensional pixel sequence, obtains defect pixel data
Collection and non-defect pixel data set；
According to non-defect pixel data set using the arest neighbors restorative procedure based on Euclidean distance to defect pixel data set
Repaired, the image after being repaired.
For the present invention when defect area is image block, selection block defect repair algorithm carries out image repair.
The principle of block defect repair algorithm is：There is correlation in piece image between adjacent pixel, and have between image block
There is similitude, so on the premise of the pixel value of the interregional one part of pixel of two in piece image different images is similar,
The possibility that the pixel value of other parts is similar is very big, thus can around according to defect area the distribution of non-defect pixel it is special
Sign after determining the image-region highly similar to defect area, is repaiied based on the similarity of pixel value to block defect area
It is multiple.
From the perspective of signal transacting, digital picture is all two-dimensional discrete signal.For the ease of carrying out block defect repair
Operation, it can be scanned first with Hilbert and two-dimensional digital image information is changed into one-dimensional discrete signal.Hilbert scans
Image-region is described in detail, it all preserves very well to the neighbor information in horizontal and vertical directions.
One-dimensional pixel sequence is divided into the pixel subsequence that length is l after Hilbert scanning, and each pixel is sub
Sequence is converted to vector xi, then utilize vector xiConstruction feature space X, that is, have：
X=[x1 x2 … xn]
Wherein, xi=[p1 p2 … pi … pl], piFor pixel value.
Then, the x containing defect pixel in XiUse yiRepresent, so that it may obtain a series of x without defect pixeliIt is (i.e. non-
Defect set of pixels) and a small amount of y containing defect pixeli(i.e. defect set of pixels).
Finally, y is calculated by Euclidean distance calculation formulaiWith xiSimilarity (Euclidean distance is smaller, and similarity is bigger),
Find and yiThe x of highly similar (i.e. similarity is more than given threshold)i, can be to utilize xiIn pixel data go to fill up yiIn
Defect pixel data.
The block Incomplete image of the present invention repairs the flow chart of algorithm as indicated at 4.
Reference picture 5, is further used as preferred embodiment, and the reparation algorithm according to selection is to image to be repaired
The step for carrying out image repair, it includes：
Image to be repaired is carried out to mix defect area scanning, obtains defect pixel；
Coordinate of the defect pixel in image to be repaired is recorded, and using the bivariate normal integral set to lacking
The coordinate of damage pixel is handled, and generates the two-dimensional coordinate of random pixel point；
Be color value by the grayvalue transition of random pixel point according to the gray scale of setting and color conversion formula, obtain by
The color set of the color value composition of random pixel point, and statistical color concentrates the probability of each color value；
The maximum color value of probable value is taken out from color set as current color value；
The pixel of several arest neighbors around current color value and defect pixel is formed into the first stochastic variable, and obtained
The second stochastic variable that the pixel of several arest neighbors forms to around by defect pixel；
The mathematic expectaion and the mathematic expectaion and variance of variance and the second stochastic variable of the first stochastic variable are calculated respectively；
Calculated according to the mathematic expectaion and variance of the mathematic expectaion of the first stochastic variable and variance and the second stochastic variable
Confidential interval；
Judge whether the first stochastic variable meets confidential interval, if so, next step is then performed, conversely, then by current face
Colour is cast out from color set, and returns to the step that the maximum color value of probable value is taken out from color set as current color value
Suddenly；
According to current color value and the gray scale of setting and the gray value of color conversion formula calculating defect pixel, and according to
The gray value of defect pixel is repaired, the image after being repaired.
For the present invention when defect area is the region of point, line and the mixing in face, selection mixes defect repair algorithm to carry out
Image repair.
In daily life, the defect part of picture is often point, line and the mixing in face, if still adopted in this case
With Fig. 2, Fig. 3 or Fig. 4 restorative procedure, then seem cumbersome, it is special based on DSMC for such case, the present invention
Door establishes a kind of mixing defect repair algorithm that can be quickly repaired to the picture of this mixing defect.
DSMC (Monte-Carlo methods), also referred to as statistical simulation methods, be it is a kind of using Probability Statistics Theory as
A kind of very important numerical computation method of guidance, it is that one kind carrys out solution definitely using random number (or more conventional pseudo random number)
The method of calculation problem.The basic thought of DSMC is：When institute's Solve problems be certain chance event occur probability, or
When person is some expectation of a random variable, by the method for certain " experiment ", the Frequency Estimation that occurs with this event this
The probability of chance event, or some numerical characteristics of this stochastic variable are obtained, and as the solution of problem.Solving in fact
Mainly there is two parts work using DSMC when the problem of border：1st, during process a certain with Monte-carlo Simulation,
Need to produce the stochastic variable of various probability distribution；2nd, the numerical characteristic of model is estimated with statistical method, so as to obtain
The numerical solution of practical problem.
The broad flow diagram of mixing defect repair algorithm of the invention based on DSMC is as shown in figure 5, main bag
Include：
After defect image data is received, each pixel of picture is scanned, when scanning to a defect picture
After element, the coordinate of defect pixel in the picture is recorded, and a series of clothes are generated using the distribution function of Two dimension normal distribution
From the two-dimensional coordinate of the random pixel point of normal distribution, the distribution function of Two dimension normal distribution is shown below：
Wherein, x and y is respectively coordinate of the defect pixel in X-axis and Y-axis, fXAnd f (x)Y(y) it is respectively defect pixel
Random pixel point in the coordinate of X-axis and Y-axis, σxAnd σyRespectively mark of the random pixel point of defect pixel in X-axis and Y-axis
Poor, the μ of standardxAnd μyRespectively the random pixel point of defect pixel is in X-axis and the average of Y-axis.
, can be according to the color value of setting and grayvalue transition formula after the random pixel point for generating defect pixel, will be with
The gray value (i.e. RGB) of machine pixel is converted into color value, and can calculate random pixel point according to the probability statistics formula of setting
The probability of a variety of colors value.The color value that the present invention is set can be with grayvalue transition formula：
Colori=65536 × Ri+256×Gi+Bi
Then, calculate respectively the first stochastic variable mathematic expectaion and the mathematic expectaion of variance and the second stochastic variable and
Variance.First stochastic variable (typically first chooses the color value for maximum probability occur as full by the larger color value of probability of occurrence
The color value assumed enough, when the maximum color value of probability of detection meets that region of rejection requires after hypothesis testing, then choose appearance
The maximum color value of probability time is as the color value ... for meeting to assume untill being met the color value of hypothesis) and defect
The color value composition of several pixels of arest neighbors near pixel, and the second stochastic variable is by arest neighbors near defect pixel
Several pixels color value composition.The mathematic expectaion E (x) and variance D (x) of the two stochastic variables calculation formula is：
D (x)=E { [x-E (x)]2}
Wherein, x1,x2,x3,…,xnRepresent each color value, p1,p2,p3,…,pnFor probability corresponding to each color value.
After the mathematic expectaion and variance of two stochastic variables is calculated, verified according to the hypothesis of normal population variance
It is theoretical, it can be determined that whether the first stochastic variable meets the confidential interval of population variance.If use χ2Method of calibration, then totality side
Difference confidential interval be：
Wherein, XiFor i-th of sampled pixel color value by Hilbert scanning, μ0For population mean,WithRespectively χ2Two critical points of method of calibration region of rejection,WithValue can be by looking into χ2Table is drawn.
If meeting the confidential interval of population variance, illustrate that the larger color value of this in the first stochastic variable has met and repair
Multiple color value requirement, now calculates defect picture according to the color value by the color value and grayvalue transition formula of foregoing setting
The gray value (RGB) of element.If the confidential interval requirement of population variance can not be met, give up this larger color value, change to take
The larger color value of probability of occurrence probability less than that, recalculate its mathematic expectaion, variance and the confidence for comparing stochastic variable
Section, untill meeting the confidential interval of population variance.
Reference picture 6, a kind of image repair system, including with lower module：
Defect area searching modul, for carrying out defect area lookup to image to be repaired, obtain complex pattern to be repaired
Defect area；
Selecting module, calculated for the defect area according to complex pattern to be repaired from scatterplot defect repair algorithm, line defect repair
A kind of reparation algorithm as image to be repaired is selected in method, block defect repair algorithm and mixing defect repair algorithm；
Repair module, image repair is carried out to image to be repaired for the algorithm of repairing according to selection.
Preferred embodiment is further used as, the selecting module is specifically used for performing following operate：
If the defect area of complex pattern to be repaired is scatterplot shape defect area, scatterplot defect repair algorithm is selected as to be repaired
The reparation algorithm of multiple image；If the defect area of complex pattern to be repaired is wire defect area, selection line defect repair algorithm
Reparation algorithm as image to be repaired；If the defect area of complex pattern to be repaired is block defect area, selection block defect
Repair reparation algorithm of the algorithm as image to be repaired；Lacked if the defect area of complex pattern to be repaired is point, line and the mixing in face
Region is damaged, then reparation algorithm of the selection mixing defect repair algorithm as image to be repaired.
Preferred embodiment is further used as, the repair module includes：
Pretreatment unit, for being pre-processed to image to be repaired, obtain pretreated image；
One-dimensional processing unit, for carrying out two-dimensional pixel one-dimensional to pretreated image, obtain one-dimensional pixel sequence
Row；
Division unit, divide, obtain for carrying out subsequence division, feature space structure and pixel to one-dimensional pixel sequence
Defect pixel data set and non-defect pixel data set；
First repairs unit, for using the arest neighbors restorative procedure based on Euclidean distance according to non-defect pixel data set
Defect pixel data set is repaired, the image after being repaired.
Preferred embodiment is further used as, the repair module includes：
Scanning element, for carrying out mixing defect area scanning to image to be repaired, obtain defect pixel；
Random pixel point two-dimensional coordinate generation unit, for recording coordinate of the defect pixel in image to be repaired,
And the coordinate of defect pixel is handled using the bivariate normal integral of setting, the two dimension seat of generation random pixel point
Mark；
Conversion and probability statistics unit, for the gray scale according to setting and color conversion formula, by the ash of random pixel point
Angle value is converted to color value, obtains the color set being made up of the color value of random pixel point, and statistical color concentrates each color value
Probability；
First chooses unit, for taking out the maximum color value of probable value from color set as current color value；
Stochastic variable generation unit, for by the pixel of several arest neighbors around current color value and defect pixel
The first stochastic variable is formed, and obtains the second random change being made up of the pixel of several arest neighbors around defect pixel
Amount；
It is expected with variance computing unit, for calculate respectively the first stochastic variable mathematic expectaion and variance and second with
The mathematic expectaion and variance of machine variable；
Confidential interval computing unit, for the mathematic expectaion according to the first stochastic variable and variance and the second stochastic variable
Mathematic expectaion and variance calculate confidential interval；
Confidence level judging unit, for judging whether the first stochastic variable meets confidential interval, repaiied if so, then performing second
The operation of multiple unit, conversely, then casting out current color value from color set, and return to first and choose unit；
Second repairs unit, for calculating defect pixel with color conversion formula according to current color value and the gray scale of setting
The gray value of point, and repaired according to the gray value of defect pixel, the image after being repaired.
The present invention is further explained and illustrated with reference to specific embodiment.
Embodiment one
It can only be repaired for prior art for the certain types of Incomplete image of certain class using single restorative procedure
The problem of, the present invention proposes a kind of new image repair method and system.As shown in fig. 7, the image repair technology is main
Handling process is：Input Incomplete image to be repaired, the methods of then passing through feature location fast search go out defect area, and unite
The Pixel Information on defect area periphery is counted, is repaired further according to defect area from scatterplot reparation, wire reparation, bulk and various scarce
Damage type hybrid, which is repaired in these four reparation types, selects a kind of be used as to repair type, finally according to reparation type to be repaired
Incomplete image is repaired.Enter row information estimation to defect area present invention utilizes image information intact in picture,
To realize that image repair is handled so that entire image keeps visual continuous and complete.In addition, the present invention is repairing completion
Afterwards, the image repaired can also be shown in display interface, in order to which user carries out effect preview and repairs front and rear contrast.
As shown in fig. 7, the image repair method of the present invention specifically includes following steps：
Step 1：Image to be repaired is inputted, and the image to be repaired is shown to user.
Step 2：Defect area lookup is carried out to image to be repaired, obtains the defect area of complex pattern to be repaired.
The present invention carries out the methods of full figure travels through lookup, passes through feature location to image to be repaired and finds out defect area,
Subsequently to carry out repair process.
Step 3：According to the defect area defect situation of complex pattern to be repaired, from scatterplot defect repair algorithm, line defect repair
A kind of reparation algorithm as image to be repaired is selected in algorithm, block defect repair algorithm and mixing defect repair algorithm.
Specifically, if the defect area of complex pattern to be repaired is scatterplot shape defect area, scatterplot defect repair is selected to calculate
Reparation algorithm of the method as image to be repaired；If the defect area of complex pattern to be repaired is wire defect area, selection line lacks
Damage repairs reparation algorithm of the algorithm as image to be repaired；If the defect area of complex pattern to be repaired is block defect area,
Reparation algorithm of the selection block defect repair algorithm as image to be repaired；If the defect area of complex pattern to be repaired be point, line and
The mixing defect area in face, then selection mix reparation algorithm of the defect repair algorithm as image to be repaired.
Step 4：Image repair is carried out to image to be repaired according to the algorithm of repairing of selection.
It can be seen from step 3, image repair algorithm of the invention includes scatterplot defect repair algorithm, line defect repair is calculated
Method, block defect repair algorithm and mixing defect repair algorithm.
Following steps are specifically included as shown in Fig. 2 repairing algorithm according to scatterplot and carrying out image repair：
S11, image to be repaired is pre-processed, view data is converted to two-dimensional matrix corresponding to.
S12, scatterplot scanning is carried out to pretreated image, obtain defect scatterplot.
S13, the method using medium filtering, are handled each defect area, and the data obtained with medium filtering
To fill up the scatterplot of defect, so as to complete the reparation of image.By taking square medium filtering as an example, the odd number data of taking-up are carried out
Ascending sort, the element value in ordered sequence centre position after sequence is taken out, then the element value of taking-up is used for filling up scarce
The scatterplot of damage, you can repair the image of scatterplot defect.
As shown in figure 3, carrying out image repair according to line defect repair algorithm specifically includes following steps：
S21, image to be repaired is pre-processed, view data is converted to two-dimensional matrix corresponding to.
S22, wire defect area scanning is carried out to pretreated image, obtain the pixel of defect.
S23, using Lagrange's interpolation the pixel of defect is repaired, the image after being repaired
The line defect repair algorithm of the present invention can obtain Lagrange interpolation function after Lagrange's interpolation is carried out, then
Only need the row or column where defect pixel to substitute into Lagrange interpolation function, just can estimate the pixel of defect pixel
The pixel value estimated, is finally filled into the effect that reparation is can reach in the pixel of defect by value.
As shown in figure 4, carrying out image repair according to block defect repair algorithm specifically includes following steps：
S31, image to be repaired is pre-processed, view data is converted to two-dimensional matrix corresponding to.
S32, two-dimensional pixel one-dimensional is carried out to pretreated image, obtain one-dimensional pixel sequence.Two-dimensional pixel is one-dimensional
Changing can be scanned using Hilbert to realize.
S33, subsequence division, feature space structure and pixel division are carried out to one-dimensional pixel sequence, obtain defect pixel
Data set and non-defect pixel data set.
S33, according to non-defect pixel data set using the N neighbours restorative procedure based on Euclidean distance to defect pixel data
Calculated, concentrated from non-defect pixel data find with the N number of data of defect pixel data similarity highest, it is and N number of with this
The average of data goes to fill up defect pixel data, completes image repair.
As shown in figure 5, carrying out image repair according to mixing defect repair algorithm specifically includes following steps：
After S41, input Incomplete image data, each pixel of image is scanned, when scanning to a defect picture
After element, the coordinate of defect point in the picture is recorded, and a series of obediences are being generated just using the bivariate normal integral of setting
The two-dimensional coordinate of the random pixel point of state distribution.
S42, gray scale and color conversion formula according to setting, are color value by the grayvalue transition of random pixel point, obtain
The color set formed to the color value by random pixel point, and statistical color concentrates the probability of each color value, the setting turns
Changing formula is：
Colori=65536 × Ri+256×Gi+Bi。
S43, the maximum color value of probable value is taken out as current color value from color set.
S44, pixel the first stochastic variable of composition by several arest neighbors around current color value and defect pixel,
And obtain the second stochastic variable being made up of the pixel of several arest neighbors around defect pixel, then calculate the two with
The average and variance of machine variable.
S45, calculated according to the mathematic expectaion and variance of the two stochastic variables and selected hypothesis method of calibration it is overall
The confidential interval of variance.
S46, judge whether the first stochastic variable meets the confidential interval of population variance, if meeting the confidential interval,
Then illustrate that current color value meets the color value requirement of reparation, it is possible to the gray value of defect pixel is calculated according to the color value
(RGB) and according to the gray value of defect pixel repaired；If can not meet confidential interval, current color value is from color set
In cast out, be then back to step S43, until complete image repair.
Above is the preferable implementation to the present invention is illustrated, but the present invention is not limited to the embodiment, ripe
A variety of equivalent variations or replacement can also be made on the premise of without prejudice to spirit of the invention by knowing those skilled in the art, this
Equivalent deformation or replacement are all contained in the application claim limited range a bit.
Claims (10)
- A kind of 1. image repair method, it is characterised in that：Comprise the following steps：Defect area lookup is carried out to image to be repaired, obtains the defect area of complex pattern to be repaired；According to the defect area of complex pattern to be repaired from scatterplot defect repair algorithm, line defect repair algorithm, block defect repair algorithm A kind of reparation algorithm as image to be repaired is selected with mixing defect repair algorithm；Image repair is carried out to image to be repaired according to the algorithm of repairing of selection.
- A kind of 2. image repair method according to claim 1, it is characterised in that：According to the defect area of complex pattern to be repaired Selected from scatterplot defect repair algorithm, line defect repair algorithm, block defect repair algorithm and mixing defect repair algorithm a kind of The step for reparation algorithm as image to be repaired, it is specially：If the defect area of complex pattern to be repaired is scatterplot shape defect area, scatterplot defect repair algorithm is selected as to be repaired The reparation algorithm of image；If the defect area of complex pattern to be repaired is wire defect area, selection line defect repair algorithm conduct The reparation algorithm of image to be repaired；If the defect area of complex pattern to be repaired is block defect area, selection block defect repair Reparation algorithm of the algorithm as image to be repaired；If the defect area of complex pattern to be repaired is point, line and the mixing in face defective region Domain, then selection mix reparation algorithm of the defect repair algorithm as image to be repaired.
- A kind of 3. image repair method according to claim 2, it is characterised in that：The reparation algorithm pair according to selection Image to be repaired carries out the step for image repair, and it includes：Image to be repaired is pre-processed, obtains pretreated image；Scatterplot scanning is carried out to pretreated image, obtains defect scatterplot；Defect scatterplot is repaired using median filtering method, the image after being repaired.
- A kind of 4. image repair method according to claim 2, it is characterised in that：The reparation algorithm pair according to selection Image to be repaired carries out the step for image repair, and it includes：Image to be repaired is pre-processed, obtains pretreated image；Wire defect area scanning is carried out to pretreated image, obtains the pixel of defect；The pixel of defect is repaired using Lagrange's interpolation, the image after being repaired.
- A kind of 5. image repair method according to claim 2, it is characterised in that：The reparation algorithm pair according to selection Image to be repaired carries out the step for image repair, and it includes：Image to be repaired is pre-processed, obtains pretreated image；Two-dimensional pixel one-dimensional is carried out to pretreated image, obtains one-dimensional pixel sequence；Subsequence division, feature space structure and pixel division are carried out to one-dimensional pixel sequence, obtain defect pixel data set and Non- defect pixel data set；Defect pixel data set is carried out using the arest neighbors restorative procedure based on Euclidean distance according to non-defect pixel data set Repair, the image after being repaired.
- A kind of 6. image repair method according to claim 2, it is characterised in that：The reparation algorithm pair according to selection Image to be repaired carries out the step for image repair, and it includes：Image to be repaired is carried out to mix defect area scanning, obtains defect pixel；Coordinate of the defect pixel in image to be repaired is recorded, and using the bivariate normal integral set to defect picture The coordinate of vegetarian refreshments is handled, and generates the two-dimensional coordinate of random pixel point；According to the gray scale of setting and color conversion formula, it is color value by the grayvalue transition of random pixel point, obtains by random The color set of the color value composition of pixel, and statistical color concentrates the probability of each color value；The maximum color value of probable value is taken out from color set as current color value；By around current color value and defect pixel several arest neighbors pixel form the first stochastic variable, and obtain by Second stochastic variable of the pixel composition of several arest neighbors around defect pixel；The mathematic expectaion and the mathematic expectaion and variance of variance and the second stochastic variable of the first stochastic variable are calculated respectively；Confidence is calculated according to the mathematic expectaion and variance of the mathematic expectaion of the first stochastic variable and variance and the second stochastic variable Section；Judge whether the first stochastic variable meets confidential interval, if so, next step is then performed, conversely, then by current color value Cast out from color set, and return and the step of maximum color value of probable value is as current color value is taken out from color set；The gray value of defect pixel is calculated with color conversion formula according to current color value and the gray scale of setting, and according to defect The gray value of pixel is repaired, the image after being repaired.
- A kind of 7. image repair system, it is characterised in that：Including with lower module：Defect area searching modul, for carrying out defect area lookup to image to be repaired, obtain the defect of complex pattern to be repaired Region；Selecting module, for the defect area according to complex pattern to be repaired from scatterplot defect repair algorithm, line defect repair algorithm, block A kind of reparation algorithm as image to be repaired is selected in defect repair algorithm and mixing defect repair algorithm；Repair module, image repair is carried out to image to be repaired for the algorithm of repairing according to selection.
- A kind of 8. image repair system according to claim 7, it is characterised in that：The selecting module is specifically used for performing Operate below：If the defect area of complex pattern to be repaired is scatterplot shape defect area, scatterplot defect repair algorithm is selected as to be repaired The reparation algorithm of image；If the defect area of complex pattern to be repaired is wire defect area, selection line defect repair algorithm conduct The reparation algorithm of image to be repaired；If the defect area of complex pattern to be repaired is block defect area, selection block defect repair Reparation algorithm of the algorithm as image to be repaired；If the defect area of complex pattern to be repaired is point, line and the mixing in face defective region Domain, then selection mix reparation algorithm of the defect repair algorithm as image to be repaired.
- A kind of 9. image repair system according to claim 8, it is characterised in that：The repair module includes：Pretreatment unit, for being pre-processed to image to be repaired, obtain pretreated image；One-dimensional processing unit, for carrying out two-dimensional pixel one-dimensional to pretreated image, obtain one-dimensional pixel sequence；Division unit, divided for carrying out subsequence division, feature space structure and pixel to one-dimensional pixel sequence, obtain defect Pixel data set and non-defect pixel data set；First repairs unit, for using the arest neighbors restorative procedure based on Euclidean distance to lacking according to non-defect pixel data set Damage pixel data set is repaired, the image after being repaired.
- A kind of 10. image repair system according to claim 8, it is characterised in that：The repair module includes：Scanning element, for carrying out mixing defect area scanning to image to be repaired, obtain defect pixel；Random pixel point two-dimensional coordinate generation unit, for recording coordinate of the defect pixel in image to be repaired, and is adopted The coordinate of defect pixel is handled with the bivariate normal integral of setting, generates the two-dimensional coordinate of random pixel point；Conversion and probability statistics unit, for the gray scale according to setting and color conversion formula, by the gray value of random pixel point Color value is converted to, obtains the color set being made up of the color value of random pixel point, and statistical color concentrates the general of each color value Rate；First chooses unit, for taking out the maximum color value of probable value from color set as current color value；Stochastic variable generation unit, for the pixel of several arest neighbors around current color value and defect pixel to be formed First stochastic variable, and obtain the second stochastic variable being made up of the pixel of several arest neighbors around defect pixel；It is expected with variance computing unit, for calculating the mathematic expectaion of the first stochastic variable respectively and variance and second random becoming The mathematic expectaion and variance of amount；Confidential interval computing unit, the number for the mathematic expectaion according to the first stochastic variable and variance and the second stochastic variable Hope in term and variance calculates confidential interval；Confidence level judging unit, for judging whether the first stochastic variable meets confidential interval, if so, then performing second repairs list The operation of member, conversely, then casting out current color value from color set, and return to first and choose unit；Second repairs unit, for calculating defect pixel with color conversion formula according to current color value and the gray scale of setting Gray value, and repaired according to the gray value of defect pixel, the image after being repaired.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710599116.5A CN107507137A (en) | 2017-07-14 | 2017-07-14 | A kind of image repair method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710599116.5A CN107507137A (en) | 2017-07-14 | 2017-07-14 | A kind of image repair method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107507137A true CN107507137A (en) | 2017-12-22 |
Family
ID=60688777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710599116.5A Pending CN107507137A (en) | 2017-07-14 | 2017-07-14 | A kind of image repair method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107507137A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109976615A (en) * | 2019-03-28 | 2019-07-05 | Oppo广东移动通信有限公司 | Fingerprint image processing method and relevant apparatus |
| CN110189276A (en) * | 2019-05-31 | 2019-08-30 | 华东理工大学 | A kind of facial image restorative procedure based on very big radius circle domain |
| CN113421229A (en) * | 2021-06-04 | 2021-09-21 | 桂林电子科技大学 | Method for detecting and repairing defect of small area of image |
| CN114418899A (en) * | 2022-03-28 | 2022-04-29 | 深圳市嘉年印务有限公司 | Self-adaptive repairing method and system for self-color printing and readable storage medium |
| CN116542880A (en) * | 2023-07-05 | 2023-08-04 | 上海蜜度信息技术有限公司 | Automatic image restoration method, system, storage medium and electronic equipment |
| CN116823678A (en) * | 2023-08-29 | 2023-09-29 | 国网江西省电力有限公司超高压分公司 | Intelligent repairing system for image defect points |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7502521B2 (en) * | 2005-06-21 | 2009-03-10 | Microsoft Corporation | Image completion with structure propagation |
| CN101782526A (en) * | 2010-01-19 | 2010-07-21 | 江苏大学 | Method and device for automatically restoring, measuring and classifying steel dimple images |
| CN102509319A (en) * | 2011-10-20 | 2012-06-20 | 西北民族大学 | Method for restoring Thangka image by combining shapes and neighborhood classification of damaged piece |
| CN105627992A (en) * | 2015-12-28 | 2016-06-01 | 桂林电子科技大学 | Method for quickly surveying and mapping historic building in high-accuracy non-contact mode |
-
2017
- 2017-07-14 CN CN201710599116.5A patent/CN107507137A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7502521B2 (en) * | 2005-06-21 | 2009-03-10 | Microsoft Corporation | Image completion with structure propagation |
| CN101782526A (en) * | 2010-01-19 | 2010-07-21 | 江苏大学 | Method and device for automatically restoring, measuring and classifying steel dimple images |
| CN102509319A (en) * | 2011-10-20 | 2012-06-20 | 西北民族大学 | Method for restoring Thangka image by combining shapes and neighborhood classification of damaged piece |
| CN105627992A (en) * | 2015-12-28 | 2016-06-01 | 桂林电子科技大学 | Method for quickly surveying and mapping historic building in high-accuracy non-contact mode |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109976615A (en) * | 2019-03-28 | 2019-07-05 | Oppo广东移动通信有限公司 | Fingerprint image processing method and relevant apparatus |
| CN109976615B (en) * | 2019-03-28 | 2021-03-12 | Oppo广东移动通信有限公司 | Fingerprint image processing method and related device |
| CN110189276A (en) * | 2019-05-31 | 2019-08-30 | 华东理工大学 | A kind of facial image restorative procedure based on very big radius circle domain |
| CN113421229A (en) * | 2021-06-04 | 2021-09-21 | 桂林电子科技大学 | Method for detecting and repairing defect of small area of image |
| CN113421229B (en) * | 2021-06-04 | 2023-05-30 | 桂林电子科技大学 | Method for detecting and repairing defects of small areas of images |
| CN114418899A (en) * | 2022-03-28 | 2022-04-29 | 深圳市嘉年印务有限公司 | Self-adaptive repairing method and system for self-color printing and readable storage medium |
| CN114418899B (en) * | 2022-03-28 | 2022-08-16 | 深圳市嘉年印务有限公司 | Self-adaptive repairing method and system for self-color printing and readable storage medium |
| CN116542880A (en) * | 2023-07-05 | 2023-08-04 | 上海蜜度信息技术有限公司 | Automatic image restoration method, system, storage medium and electronic equipment |
| CN116542880B (en) * | 2023-07-05 | 2023-10-20 | 上海蜜度信息技术有限公司 | Automatic image restoration method, system, storage medium and electronic equipment |
| CN116823678A (en) * | 2023-08-29 | 2023-09-29 | 国网江西省电力有限公司超高压分公司 | Intelligent repairing system for image defect points |
| CN116823678B (en) * | 2023-08-29 | 2023-11-17 | 国网江西省电力有限公司超高压分公司 | Intelligent repairing system for image defect points |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107507137A (en) | A kind of image repair method and system | |
| Shi et al. | A framework for learning depth from a flexible subset of dense and sparse light field views | |
| Wu | Adaptive split-and-merge segmentation based on piecewise least-square approximation | |
| Saxena et al. | Make3d: Learning 3d scene structure from a single still image | |
| CN109377445B (en) | Model training method, method and device for replacing image background and electronic system | |
| CN111583097A (en) | Image processing method, image processing device, electronic equipment and computer readable storage medium | |
| CN110738207A (en) | character detection method for fusing character area edge information in character image | |
| CN106683048A (en) | Image super-resolution method and image super-resolution equipment | |
| CN104574331B (en) | A kind of data processing method, device, computer storage medium and user terminal | |
| CN109598223A (en) | Method and apparatus based on video acquisition target person | |
| US9406140B2 (en) | Method and apparatus for generating depth information | |
| CN111062331B (en) | Image mosaic detection method and device, electronic equipment and storage medium | |
| CN104270624B (en) | A kind of subregional 3D video mapping method | |
| CN114004754A (en) | Scene depth completion system and method based on deep learning | |
| CN111753782A (en) | False face detection method and device based on double-current network and electronic equipment | |
| CN103743750B (en) | A kind of generation method of distribution diagram of surface damage of heavy calibre optical element | |
| CN113205502A (en) | Insulator defect detection method and system based on deep learning | |
| CN116030498A (en) | Virtual garment running and showing oriented three-dimensional human body posture estimation method | |
| Zhou et al. | Attention transfer network for nature image matting | |
| CN110163055A (en) | Gesture identification method, device and computer equipment | |
| CN112884657B (en) | Face super-resolution reconstruction method and system | |
| CN110223356A (en) | A kind of monocular camera full automatic calibration method based on energy growth | |
| CN113298707B (en) | Image frame splicing method, video inspection method, device, equipment and storage medium | |
| CN113888509A (en) | Method, device and equipment for evaluating image definition and storage medium | |
| KR20080079443A (en) | Method and apparatus for extracting object from image |
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: 20171222 |
|
| RJ01 | Rejection of invention patent application after publication |