CN109544465A - Image damage block restorative procedure based on change of scale - Google Patents
Image damage block restorative procedure based on change of scale Download PDFInfo
- Publication number
- CN109544465A CN109544465A CN201811234096.2A CN201811234096A CN109544465A CN 109544465 A CN109544465 A CN 109544465A CN 201811234096 A CN201811234096 A CN 201811234096A CN 109544465 A CN109544465 A CN 109544465A
- Authority
- CN
- China
- Prior art keywords
- image
- damage block
- edge
- repaired
- single 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000008859 change Effects 0.000 title claims abstract description 14
- 230000008439 repair process Effects 0.000 claims abstract description 16
- 238000005070 sampling Methods 0.000 claims abstract description 15
- 230000002708 enhancing effect Effects 0.000 claims abstract description 5
- 238000003708 edge detection Methods 0.000 claims abstract description 4
- 230000009466 transformation Effects 0.000 claims description 3
- 230000000694 effects Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000011156 evaluation Methods 0.000 description 3
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003447 ipsilateral effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000011084 recovery Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000003786 synthesis reaction 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
- G06T5/77—Retouching; Inpainting; Scratch removal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
- G06T5/94—Dynamic range modification of images or parts thereof based on local image properties, e.g. for local contrast enhancement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20052—Discrete cosine transform [DCT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Processing (AREA)
Abstract
The present invention discloses the image damage block restorative procedure based on change of scale, and step is repeatedly to be decomposed to form single pixel point to be repaired using DCT method to the damage block in image；Interpolation reparation is carried out using the local edge information from adjacent pixel to single pixel point to be repaired, the edge skeleton of image where generating the single pixel point by edge detection by canny edge detector, image enhancement is carried out using the global marginal information of acquisition, picture size expands after enhancing after image enhancement by up-sampling；Image after expanding size merges with the sample that same layer up-samples, and is then interpolated into image enchancing method using aforementioned, repairs to the damage block of the sample of same layer up-sampling；It repeats the above method successively to repair the damage block of the sample of the up-sampling for the image successively repaired, until the damage block reparation of image is completed.The present invention, which can be repaired effectively, there is the case where lost blocks in digital picture.
Description
Technical field
The present invention relates to technical field of image processing, and in particular to the image damage block restorative procedure based on change of scale.
Background technique
With the arriving of Digital Media, the application of media has become part indispensable in our lives, in media
In application process, many image processing techniques have played important function, and image repair is matchmaker as a wherein important technology
Body application provides important technical support.It may be because that network congestion or mobile device signal are lost when image and transmission of video
The reasons such as mistake, cause data packetloss, so that image quality decrease.With the need of the image application program to relatively high bandwidth
It asks and increases rapidly, excite the demand to data packet-loss recovery, to improve more reliable network service and more acceptable
User experience.
Summary of the invention
In view of the technical drawbacks of the prior art, it is an object of the present invention to provide a kind of figures based on change of scale
As damage bad block repair method.
The technical solution adopted to achieve the purpose of the present invention is:
Image damage block restorative procedure based on change of scale, comprising steps of
S1 is repeatedly decomposed to form single pixel point to be repaired using DCT method to the damage block in image；
S2 carries out interpolation reparation using the local edge information from adjacent pixel to single pixel point to be repaired, borrows
The edge skeleton of image, uses the complete of acquisition where helping canny edge detector to generate the single pixel point by edge detection
Office's marginal information carries out image enhancement, is expanded picture size after enhancing by up-sampling after image enhancement；
S3, the image after expanding size merge with the sample that same layer up-samples, and are then interpolated into figure using step S2
The method of image intensifying repairs the damage block of the sample of same layer up-sampling；
S4 repeats step S2-S3, successively repairs to the damage block of the sample of the up-sampling to the image repaired, until
The damage block reparation of image is completed.
The bulk damage block in image is repeatedly decomposed to form single pixel point to be repaired using DCT method
Step is, using discrete cosine transform, the damage block in image is resolved into four frequency quadrants, by the portion that frequency is minimum
Divide and continues to decompose until damage block is compressed into single pixel.
It is described by up-sampling will after enhancing picture size expand the step of be will complete repair image use zero padding
The size that inverse transformation is up-sampled with enlarged image.
The present invention is based on the margin guide interpolation methods of multi-scale transform, can effectively repair in digital picture and lose
The case where losing block (or damage block), the reparation of the damage block especially including edge including, while be suitable for rule be distributed with
And the reparation of the damage block of random distribution.
Detailed description of the invention
Fig. 1 is the reparation flow diagram of the image damage block restorative procedure of the invention based on change of scale；
Fig. 2A, 2B, 2C are the schematic diagram of the pixel of damage and three kinds of relationships at edge respectively.
Specific embodiment
The present invention is described in further detail below in conjunction with the drawings and specific embodiments.It should be appreciated that described herein
Specific embodiment be only used to explain the present invention, be not intended to limit the present invention.
The present invention is applied to the case where data-bag lost when reparation and the image transmitting in image damage region, is become using scale
The method changed decomposes image, is repaired on the basis of decomposition, and final synthesis obtains repairing result.
As shown in Figure 1, the present invention is that the bulk damage block in image is resolved into single pixel point using DCT, it is then right
It carries out interpolation to it using the local edge information from adjacent pixel, generates compression image by canny edge detector
Edge skeleton, carry out image enhancement using global marginal information, expanded picture size by up-sampling after enhancing, and it is same
The sample of layer up-sampling merges；Sample after merging to the up-sampling for repairing the image completed, repeats above-mentioned be interpolated into
This serial procedures of image enhancement, until repairing image completely.
Firstly, the damage block in image is resolved into four frequency quadrants, L and H divide using discrete cosine transform (DCT)
Not Biao Shi low frequency and high frequency, by the minimum part of frequency (LL) continue decompose until damage block be compressed into single pixel.
Such as the damage block for being 8*8 for size, it needs to be decomposed to single pixel three times.It is 512* for full size
512 image, by the image for decomposing boil down to 64*64 three times.Dct transform can be indicated by following formula:
Wherein, BxyIt is by dct transform as a result, subscript " s " refers to the level of dct transform, Ms, Ns are s grades of damages
The length and width of bad block.For example, block, the first order s=1, M are damaged for 8*81=N1=8；S=2, M2=N2=4；M3=N3
=2；M4=N4=1.While x and y value is respectively 0 between 0 to (Ns-1), i.e., as s=1, x and y are taken between (Ms-1)
Value is between 0 to 7；As s=2, x and y value is between 0 to 3；As s=3, x and y value is between 0 to 1；Work as s=4
When, x and y value is 0.I (m, n) is input picture, and m, n are picture size, bxAnd bvIt is the constant coefficient of DCT, by following formula
It indicates:
It, can by comparing the relationship of single pixel and edge that compression obtains after being decomposed to obtained damage block
To be divided into Fig. 2A, 2B, it is horizontal edge that three kinds of situations shown in 2C, the first edge, which is located at bad point or more,；Second of edge is located at
Bad point or so is vertical edge；The third edge is the edge of cross shaped head.In Fig. 2A, 2B, 2C, black square indicates damage
Pixel, the arrow of the same direction indicate that, positioned at the ipsilateral of edge, the gray value of arrow meaning block is used to calculate estimating for the direction
Evaluation, estimated value are indicated with the dot with the arrow same direction.Relatively small, the transverse edges along the pixel value difference of edge direction
The pixel value difference in direction is relatively large, and the difference of pixel value is approximate consistent on the two sides at edge, edge.
To compressed independent pixel is decomposed, according to the estimation of the relationship of the single pixel of above-mentioned acquisition and edge
Value is repaired using orientation edge interpolation, i.e., is operated to the 64*64 image resolved into, and the edge for losing pixel increases
Strong estimated value is handled according to formula below:
Wherein, fM, nEnhance estimated value for the edge of spot failure, E is edge, and (m, n) is the starting pixels point at edge, (m+
K, n+l) be edge termination pixel, EM+k, n+lIt is Fig. 2A, 2B, the edge of the regional area of three kinds of borderline cases shown in 2C is estimated
Evaluation, RI are the regions of selected local interpolation.Estimate at edge along the direction of pixel (m, n) → pixel (m+k, n+l)
Evaluation is obtained by the average value of all usable levels of the neighbouring same direction.H, y, C respectively indicate horizontal sides mentioned above
Edge, vertical edge, overlapping edges, ωM+k, n+lFor the weighting function at edge, for being consistent with most apparent edge.Citing
For, for Fig. 2A, horizontal edge in 2B, 2C, (m, n) is spot failure, margin estimation value EM-1, nIt can be expressed as
As shown in formula (5), if EM-1, n-1, EM-1, n+1Different directions at edge, i.e. EM-1, n-1*EM-1, n+1< 0,
Then by EM-1, nBe set as 0, edge it is ipsilateral when EM-1, nFor both direction discreet value EM-1, n-1, EM-1, n+1Average value, due to close
It spends that highest edge is the most obvious in the region of local interpolation, therefore calculates margin estimation value and need different weights, weight letter
Number ωM+k, n+lIt can be indicated by marginal density, shown in following formula:
Wherein, ∑K=-1: 1, l=-1: 1, k, l ≠ 0ωM+k, n+l=1,
Using the above-mentioned margin estimation value being calculated to the interpolation of single pixel.
After the interpolation for completing single pixel point is repaired, similar side is used to the damage block having a size of 2*2,4*4,8*8
Method is modified；When interpolation is repaired, it is the interpolation for starting from damaging the exterior contour of block, is gradually repaired to interpolated value.
It after the local interpolation of pixel, needs to carry out global edge interpolation, to reduce lost blocks to edge detection
It influences.In order to estimate the major side in image, the global edge skeleton of entire image is generated by canny edge detector,
Change the variance of Gaussian filter and the minimum and maximum threshold value of efficient frontier, to change the sensitivity of Canny detector.
Judge with the presence or absence of missing gap in edge that edge detector detects, if there are missing gaps in edge
Situation, the value of missing pixel are the average value of pixel in the same direction, then can be obtained more preferably at edge and other regions
Image.
It completes after repairing, uses zero padding inverse transformation to up-sample the ruler with enlarged image the image for completing to repair
It is very little, it repeats to repair, the processes such as up-sampling, until image is restored to original size, and by entire process iteration four to five times, into one
Step repairs the spot failure omitted in previous reparation.
Present invention incorporates change of scale, local edge interpolation and global Edge Enhancements, improve normal interpolation reparation
Effect enables the damage block with edge preferably to be repaired, and improves each layer in reconstruction process by alternative manner,
Improve the effect of recovery.
By the way of the present invention using picture breakdown and successively restores, repairs structure and be similar to pyramid, to lowest frequency pixel
Point is preferentially repaired, and low frequency pixel is the place that energy is concentrated in image, and preferential repair can effectively ensure that repairing effect.
The present invention use based on local edge interpolation, the edge skeleton for generating different accuracy is filled up in edge lack between
Gap enhances the repairing effect of image by way of global edge interpolation and using iteration to handle.The present invention is according to image
The process change Gaussian filter variance of decomposition completes global edge reparation to generate the edge skeleton of different accuracy.
The present invention is suitable for black white image and color image, has preferably to the damage block of rule distribution and random distribution
Repairing effect, while present invention may also apply to the situation of noise disturbance is met in video.
The above is only a preferred embodiment of the present invention, it is noted that for the common skill of the art
For art personnel, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications
Also it should be regarded as protection scope of the present invention.
Claims (3)
1. the image damage block restorative procedure based on change of scale, which is characterized in that comprising steps of
S1 is repeatedly decomposed to form single pixel point to be repaired using DCT method to the damage block in image；
S2 carries out interpolation reparation using the local edge information from adjacent pixel to single pixel point to be repaired, by
The edge skeleton of image, uses the overall situation of acquisition where canny edge detector generates the single pixel point by edge detection
Marginal information carries out image enhancement, and picture size expands after being enhanced after image enhancement by up-sampling；
S3, the image after expanding size merge with the sample that same layer up-samples, and are then increased using the image that is interpolated into of step S2
Strong method repairs the damage block of the sample of same layer up-sampling；
S4 repeats step S2-S3, successively repairs to the damage block of the sample of the up-sampling for the image repaired, until image
Block reparation is damaged to complete.
2. the image damage block restorative procedure based on change of scale according to claim 1, which is characterized in that described to image
In bulk damage block the step of single pixel point to be repaired is repeatedly decomposed to form using DCT method be, using discrete remaining
String converts DCT, and the damage block in image is resolved into four frequency quadrants, and the minimum part of frequency is continued to decompose until damage
Block is compressed into single pixel.
3. the image damage block restorative procedure based on change of scale according to claim 2, which is characterized in that described by upper
Sampling will after enhancing picture size expand the step of be by complete repair image use zero padding inverse transformation to up-sample with
The size of enlarged image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811234096.2A CN109544465A (en) | 2018-10-23 | 2018-10-23 | Image damage block restorative procedure based on change of scale |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811234096.2A CN109544465A (en) | 2018-10-23 | 2018-10-23 | Image damage block restorative procedure based on change of scale |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109544465A true CN109544465A (en) | 2019-03-29 |
Family
ID=65844809
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811234096.2A Pending CN109544465A (en) | 2018-10-23 | 2018-10-23 | Image damage block restorative procedure based on change of scale |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109544465A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062924A (en) * | 2019-12-17 | 2020-04-24 | 腾讯科技（深圳）有限公司 | Image processing method, device, terminal and storage medium |
| CN111968059A (en) * | 2020-08-27 | 2020-11-20 | 李松涛 | Multi-patch matching golden photograph restoration method and device |
| CN113205152A (en) * | 2021-05-24 | 2021-08-03 | 西安邮电大学 | Feature fusion method for panoramic fusion |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105392015A (en) * | 2015-11-06 | 2016-03-09 | 厦门大学 | Cartoon image compression method based on explicit hybrid harmonic diffusion |
| CN106056553A (en) * | 2016-05-31 | 2016-10-26 | 哈尔滨工业大学深圳研究生院 | Image inpainting method based on tight frame feature dictionary |
| CN106875359A (en) * | 2017-02-16 | 2017-06-20 | 阜阳师范学院 | A kind of sample block image repair method based on layering boot policy |
-
2018
- 2018-10-23 CN CN201811234096.2A patent/CN109544465A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105392015A (en) * | 2015-11-06 | 2016-03-09 | 厦门大学 | Cartoon image compression method based on explicit hybrid harmonic diffusion |
| CN106056553A (en) * | 2016-05-31 | 2016-10-26 | 哈尔滨工业大学深圳研究生院 | Image inpainting method based on tight frame feature dictionary |
| CN106875359A (en) * | 2017-02-16 | 2017-06-20 | 阜阳师范学院 | A kind of sample block image repair method based on layering boot policy |
Non-Patent Citations (1)
| Title |
|---|
| 龙杰: "基于区域分类的图像修复算法研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062924A (en) * | 2019-12-17 | 2020-04-24 | 腾讯科技（深圳）有限公司 | Image processing method, device, terminal and storage medium |
| CN111968059A (en) * | 2020-08-27 | 2020-11-20 | 李松涛 | Multi-patch matching golden photograph restoration method and device |
| CN111968059B (en) * | 2020-08-27 | 2024-03-08 | 李松涛 | Multi-patch matching golden phase diagram restoration method and device |
| CN113205152A (en) * | 2021-05-24 | 2021-08-03 | 西安邮电大学 | Feature fusion method for panoramic fusion |
| CN113205152B (en) * | 2021-05-24 | 2023-12-22 | 西安邮电大学 | Feature fusion method for look-around fusion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US6421468B1 (en) | Method and apparatus for sharpening an image by scaling elements of a frequency-domain representation | |
| US20070286515A1 (en) | Method and apparatus for removing false contours | |
| KR20050031210A (en) | Method and apparatus of image denoising | |
| CN102663695A (en) | DR image denoising method based on wavelet transformation and system thereof | |
| CN104680485A (en) | Method and device for denoising image based on multiple resolutions | |
| CN104077746A (en) | Gray level image processing method and device | |
| CN104050646B (en) | Method based on iteration non-local mean suppression image impulse noise | |
| CN109544465A (en) | Image damage block restorative procedure based on change of scale | |
| EP1286306A2 (en) | Data processing apparatus and method of processing data | |
| CN103561194A (en) | Scanned image descreening method based on adaptive filtering | |
| CN106651792A (en) | Method and device for removing strip noise of satellite image | |
| CN104021523A (en) | Novel method for image super-resolution amplification based on edge classification | |
| CN101655977B (en) | Method for eliminating image impulse noise based on differential image detection and filtration by multiple windows | |
| CN105913383A (en) | Image noise reduction method based on image block prior estimation mixed framework | |
| Tauber et al. | A robust speckle reducing anisotropic diffusion | |
| CN103748613A (en) | Anisotropic gradient regularization for image denoising, compression, and interpolation | |
| CN105335930A (en) | Edge data driven robustness-based face super-resolution processing method and system | |
| CN105279742B (en) | A kind of image de-noising method quickly based on piecemeal estimation of noise energy | |
| JPH1063839A (en) | Image processing method and device therefor | |
| Irum et al. | Salt and pepper noise removal filter for 8-bit images based on local and global occurrences of grey levels as selection indicator | |
| CN102156973A (en) | Image de-noising method and equipment | |
| CN106846262A (en) | Remove the method and system of mosquito noise | |
| KR101707337B1 (en) | Multiresolution non-local means filtering method for image denoising | |
| KR20080048904A (en) | Apparatus and method for restoring loss pixel using directional interpolation | |
| Shrivastava et al. | Noise removal using first order neighborhood mean filter |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190329 |