CN106504207B - A kind of image processing method - Google Patents
A kind of image processing method Download PDFInfo
- Publication number
- CN106504207B CN106504207B CN201610948033.8A CN201610948033A CN106504207B CN 106504207 B CN106504207 B CN 106504207B CN 201610948033 A CN201610948033 A CN 201610948033A CN 106504207 B CN106504207 B CN 106504207B
- Authority
- CN
- China
- Prior art keywords
- matrix
- image
- denoising
- noise
- array
- 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.)
- Active
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 15
- 239000011159 matrix material Substances 0.000 claims abstract description 134
- 238000000034 method Methods 0.000 claims abstract description 48
- 239000013598 vector Substances 0.000 claims abstract description 15
- 230000003044 adaptive effect Effects 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims description 23
- 238000000354 decomposition reaction Methods 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 230000000007 visual effect Effects 0.000 abstract description 9
- 230000008901 benefit Effects 0.000 abstract description 8
- 230000000694 effects Effects 0.000 description 20
- 230000015654 memory Effects 0.000 description 12
- 238000010586 diagram Methods 0.000 description 11
- 230000006870 function Effects 0.000 description 11
- 238000012360 testing method Methods 0.000 description 9
- 238000011156 evaluation Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000012544 monitoring process Methods 0.000 description 5
- 238000011160 research Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 108090000623 proteins and genes Proteins 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 235000002566 Capsicum Nutrition 0.000 description 1
- 240000004760 Pimpinella anisum Species 0.000 description 1
- 241000758706 Piperaceae Species 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000011109 contamination Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000000151 deposition Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000009792 diffusion process Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 208000013021 vision distortion Diseases 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
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The embodiment of the present invention provides a kind of image processing method, which comprises carries out total variation to noise image matrix, obtains the first denoising image array；According to the first denoising image array and the noise image matrix, residual image matrix is obtained；The residual image matrix is subjected to adaptive wiener filter, obtains filtered residual image matrix；The first denoising image array, the filtered residual image matrix and weight vectors are subjected to second of denoising according to the first preset rules, obtain the second denoising image array.The method takes full advantage of the prior information of image, to preferably remain edge, the details of image, while obtaining high s/n ratio, structural similarity also keeps higher level, sufficiently meets people's visual effect while favorably removal noise.
Description
Technical field
The present invention relates to technical field of image processing, in particular to a kind of image processing method.
Background technique
Due to inevitably will receive the interference of noise, causing during acquisition, transimission and storage picture signal
The information of image is submerged, and causes serious influence to the visual quality of image.A large amount of image border and minutia are flooded
Not yet, very big difficulty is brought to the analysis of image and subsequent processing.Elimination to picture noise is one in image preprocessing
A important research content.The processing such as edge detection, image segmentation, feature extraction and pattern-recognition for subsequent progress provide well
Basis.Therefore in the case where keeping the contrast of clarity and image of image detail, how effectively to remove noise becomes people
Research hotspot.
With the proposition of non local thought, people start to be converted into the pass to picture structure from the self-similarity of image
Note.Wherein Alessandro Foi, Giacomo Boracchi propose a kind of non local self-similarity algorithm based on funnel
(Anis.Fov.NL-Means).Basic thought is: being accorded with based on the space HVS using spill to substitute phase in the non local denoising of tradition
Like the calculating of distance, increase the fidelity at structure and edge with the concave operator of radial anisotropic.Airspace and transform domain
In conjunction with, image it is structural be converted into picture element matrix it is structural when, rank of matrix as a kind of measurement rectangular array or
The index of capable correlation can describe the structural information of similar matrix well.The video clipping of still camera capture has one
A specific low-rank attribute, can carry out background modeling and foreground extraction on this basis.This also shows by the similar benefit in part
The matrix that fourth is formed under natural image is low-rank, can be used for the recovery of high-performance image.Therefore by from image array
Restoring potential image information in degradation model becomes research hotspot, the i.e. approximation of low-rank matrix.Due to owing convex and non-convex
The fast development of technology has some low-rank matrixes approximately to study in recent years, and many important models and algorithm.
Low-rank matrix approximation generally can be divided into two classes: low-rank matrix decomposition method (LRMF) and nuclear norm minimizing method (NNM).
When LRMF is directed to set matrix Y, a matrix X is found, makes it on certain data fidelity functions as close possible to Y.Together
When can also be decomposed into the products of two low-rank matrixes.Many algorithms based on LRMF are suggested, including dividing from classical singular value
Solve the stable algorithm of many L1-norm.NNM is another approximate form of low-rank matrix, is different from LRMF and is
Core standard is minimised as while finding approximate matrix X.And the advantage of NNM is it in fidelity term specific data problem
On be convex, and LRMF is non-convex, therefore has attracted the great research interest of scholar in recent years.Candes and Recht
Prove most of low-rank matrixes can by a NNM problem thus to restore；Caietal [is demonstrated along with to F-
Research of the norm data fidelity to vibration operation singular value problem, can easily be solved with NNM.Although NNM is wide
It is general approximate for low-rank matrix, but it still has some problems, in order to guarantee convexity, has ignored the prior information of image, nothing
The edge details of method reservation image.
Summary of the invention
In view of this, the embodiment of the present invention is designed to provide a kind of image processing method, to solve above-mentioned ask
Topic.
To achieve the goals above, technical solution used in the embodiment of the present invention is as follows:
The embodiment of the invention provides a kind of image processing methods, which comprises to noise image matrix into
Row total variation obtains the first denoising image array；According to the first denoising image array and the noise image matrix, obtain
Take residual image matrix；The residual image matrix is subjected to adaptive wiener filter, obtains filtered residual image matrix；
By the first denoising image array, the filtered residual image matrix and weight vectors, according to the first preset rules, into
Second of denoising of row obtains the second denoising image array.
Compared with prior art, a kind of image processing method provided in an embodiment of the present invention, by noise image
Matrix carries out total variation first, and carries out adaptive wiener filter to residual image matrix, and denoise image for described first
Matrix, the filtered residual image matrix and weight vectors carry out second denoising processing, obtain according to the first preset rules
The second denoising image array is taken, the prior information of image is taken full advantage of, thus while favorably removal noise, preferably
Edge, the details for remaining image, while obtaining high s/n ratio, structural similarity also keeps higher level, sufficiently full
Sufficient people's visual effect.
To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, preferred embodiment is cited below particularly, and cooperate
Appended attached drawing, is described in detail below.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attached
Figure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pair
The restriction of range for those of ordinary skill in the art without creative efforts, can also be according to this
A little attached drawings obtain other relevant attached drawings.
Fig. 1 is the structural block diagram of server provided in an embodiment of the present invention.
Fig. 2 is a kind of flow chart for image processing method that first embodiment of the invention provides.
Fig. 3 is a kind of flow chart for image processing method that second embodiment of the invention provides.
Fig. 4 is a kind of structural block diagram for image data processing system that third embodiment of the invention provides.
Fig. 5 is third embodiment of the invention and existing method (AFNM, BM3D, WNNM) to lena256X256 test chart
Denoise effect contrast figure.
Fig. 6 is third embodiment of the invention and existing method (AFNM, BM3D, WNNM) to Monarch256X256 test chart
Denoising effect contrast figure.
Fig. 7 is third embodiment of the invention and existing method (AFNM, BM3D, WNNM) to peppers256X256 test chart
Denoising effect contrast figure.
Fig. 8 is that third embodiment of the invention and existing method (BM3D, WNNM) imitate the denoising of certain workshop monitoring test chart
Fruit comparison diagram.
Fig. 9 is that third embodiment of the invention and existing method (BM3D, WNNM) monitor noise level effect pair to certain workshop
Than figure.
Figure 10 is that third embodiment of the invention and existing method (BM3D, WNNM) monitor survey IQA evaluation comparison to certain workshop
Figure.
Specific embodiment
Below in conjunction with attached drawing in the embodiment of the present invention, technical solution in the embodiment of the present invention carries out clear, complete
Ground description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Usually exist
The component of the embodiment of the present invention described and illustrated in attached drawing can be arranged and be designed with a variety of different configurations herein.Cause
This, is not intended to limit claimed invention to the detailed description of the embodiment of the present invention provided in the accompanying drawings below
Range, but it is merely representative of selected embodiment of the invention.Based on the embodiment of the present invention, those skilled in the art are not doing
Every other embodiment obtained under the premise of creative work out, shall fall within the protection scope of the present invention.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need that it is further defined and explained in subsequent attached drawing.Meanwhile of the invention
In description, term " first ", " second " etc. are only used for distinguishing description, are not understood to indicate or imply relative importance.
As shown in Figure 1, being the block diagram of server 200.The server 200 includes memory 201, processor
202 and network module 203.
Memory 201 can be used for storing software program and module, such as the image real time transfer side in the embodiment of the present invention
Method and the corresponding program instruction/module of device, processor 202 by the software program that is stored in memory 201 of operation and
Module, thereby executing various function application and data processing, i.e. application issue recommended method in the realization embodiment of the present invention.
Memory 201 may include high speed random access memory, may also include nonvolatile memory, as one or more magnetic storage fills
It sets, flash memory or other non-volatile solid state memories.Further, the software program and module in above-mentioned memory 201
It may also include that operating system 221 and service module 222.Wherein operating system 221, may be, for example, LINUX, UNIX,
WINDOWS may include various for management system task (such as memory management, storage equipment control, power management etc.)
Component software and/or driving, and can mutually be communicated with various hardware or component software, to provide the operation of other software component
Environment.Service module 222 operates on the basis of operating system 221, and is come from by the network service of operating system 221 monitoring
The request of network completes corresponding data processing according to request, and returns to processing result to client.That is, service mould
Block 222 is used to provide network service to client.
Network module 203 is for receiving and transmitting network signal.Above-mentioned network signal may include wireless signal or have
Line signal.
It is appreciated that structure shown in FIG. 1 is only to illustrate, the server 200 may also include it is more than shown in Fig. 1 or
The less component of person, or with the configuration different from shown in Fig. 1.Each component shown in Fig. 1 can using hardware, software or
A combination thereof is realized.In addition, the server in the embodiment of the present invention can also include the server of multiple specific different function.
Fig. 2 shows a kind of flow charts for image processing method that first embodiment of the invention provides, and please refer to figure
2, the present embodiment describes the process flow of server, which comprises
Step S310 carries out total variation to noise image matrix, obtains the first denoising image array.
Total variation refers to the minimization problem that image denoising is modeled as to an energy function, so that image reaches flat
Sliding state while smooth noise, can make by the anisotropic diffusion equation of partial differential equation to noise image processing
Edge is maintained, and preferably solves restoring image detail and inhibits the contradiction between noise.
Step S320 obtains residual image matrix according to the first denoising image array and the noise image matrix.
As an implementation, the noise image matrix can be subtracted to the square of the first denoising image array
Battle array, as the residual image matrix.
The residual image matrix is carried out adaptive wiener filter, obtains filtered residual image square by step S330
Battle array.
Adaptive wiener filter has preferable filter effect compared with other filters, and selectivity is good, can be better
Retain the edge and high frequency detail of image.It therefore, can be in the structure for retaining residual image missing image using Wiener filtering
Inhibit noise while information well.
Step S340, by the first denoising image array, the filtered residual image matrix and weight vectors, root
According to the first preset rules, second of denoising is carried out, obtains the second denoising image array.
As an implementation, first preset rules refer to according to by the first denoising image array and described
Filtered residual image matrix be added the superposition matrix to be formed, low-rank matrix corresponding with the superposition matrix, weight vectors,
First parameter and the Weighted Kernel Norm Model of the second parameter building.
Wherein, the corresponding low-rank matrix of the described and superposition matrix refers to image information corresponding with the superposition matrix
Low-rank matrix, image information refer to the free of contamination image being made of pure image texture, structure etc..
Further, the Weighted Kernel Norm Model are as follows:
Wherein, YiFor be added by the first denoising image array and the filtered residual image matrix formed it is folded
Adding matrix, Xi is low-rank matrix corresponding with the superposition matrix, and λ is the first parameter, and σ n is the second parameter,W=[w1,w2.....wi] it is weight vectors,σi
(Yi) it is YiI-th of singular value, i=1,2......n.
As a kind of mode, the second parameter σ n is noise variance value.First parameter lambda is soft-threshold adjustment
The factor.
Weight vectors w=[the w1,w2.....wi] in each element calculation formula it is as follows:
Wherein, c is a constant greater than zero, and n is third parameter, ε=10-16For fixed value,σi(Yi) it is YiI-th of singular value, i=1,2......n, YiFor by described first
Denoising image array and the filtered residual image matrix are added the superposition matrix to be formed.
A kind of image processing method provided in an embodiment of the present invention, by carrying out totality first to noise image matrix
Variation, and adaptive wiener filter carried out to residual image matrix, and described first will denoise image array, described filtered
Residual image matrix and weight vectors carry out second denoising processing according to the first preset rules, obtain the second denoising image moment
Battle array, takes full advantage of the prior information of image, to preferably remain the side of image while favorably removal noise
Edge, details, while obtaining high s/n ratio, structural similarity also keeps higher level, sufficiently meets people's visual effect.
Fig. 3 shows a kind of flow chart of image processing method of second embodiment of the invention offer, please refers to figure
3, the present embodiment describes the process flow of server, which comprises
Step S410 carries out total variation to noise image matrix, obtains the first denoising image array.
Step S420 obtains residual image matrix according to the first denoising image array and the noise image matrix.
The residual image matrix is carried out adaptive wiener filter, obtains filtered residual image square by step S430
Battle array.
Step S440 is added the first denoising image array and the filtered residual image matrix to form superposition
Matrix.
The superposition matrix is divided into multiple submatrixs according to noise variance value by step S450.
According to the difference of noise variance value, the superposition matrix can be divided according to different sizes, for example, working as
Noise variance value is respectively σn≤ 25,25 < σn≤ 40,40 < σn≤ 60,60 < σnWhen, it can be corresponding by the size of submatrix respectively
It is set as 90 × 90,90 × 90,130 × 130,140 × 140.
Step S460 obtains the corresponding similar matrix of each submatrix according to the second preset rules respectively.
Wherein, the mode for obtaining the corresponding similar matrix of each submatrix, can be by the submatrix carry out into
The submatrix can be divided into different by the division of one step according to the difference of noise variance value as an implementation
Size, for example, when noise variance value is respectively σn≤ 25,25 < σnWhen, can respectively by submatrix it is corresponding according to 35 × 35,40 ×
40 are divided, and small dimension matrix is obtained.After submatrix is divided further division, carried out according to the small dimension matrix after division
Search matching obtains the corresponding similar matrix of each submatrix.
Wherein, second preset rules are regularization iterative processing.
As an implementation, the regularization iterative formula of the regularization iterative processing isWherein,y(0)=y, y are the first denoising image array and the filtering
Residual image matrix afterwards is superimposed the superposition matrix to be formed, and δ is iteration step length parameter.
Wherein, the number of iterations of k setting is followed successively by 6,8,9,11 according to the difference of noise equation value.
When iteration step length parameter δ=0.1, when noise variance value is less than and waits with 40, soft-threshold Dynamic gene λ is 0.56,
When noise variance value is greater than 40, soft-threshold Dynamic gene λ is 0.58.
Step S470 carries out singular value decomposition to each similar matrix respectively, obtains each similar matrix pair
The first matrix, singular value diagonal matrix and the second matrix answered.
Assuming that the similar matrix yi is m × n rank matrix, so then exist one decompose so that
[U, Σ, V]=SVD (yi)=U Σ V
Wherein the first matrix U is m × m rank unitary matrice；Σ is m × n rank singular value diagonal matrix；Second matrix V is n × n
Rank unitary matrice.It is such to decompose the singular value decomposition for being referred to as yi.Element Σ i, i on Σ diagonal line are the singular value of yi.
Step S480, respectively by the corresponding singular value diagonal matrix of each similar matrix and the weight vectors phase
Multiply, obtains the corresponding third matrix of each similar matrix.
As an implementation, the weight vectors w=[w1,w2.....wi] in each elementWherein, c is a constant greater than zero, and n is third parameter, ε=10-16For fixed value,σi(Yi) it is YiI-th of singular value, i=1,2......n, Yi is by described the
One denoising image array and the filtered residual image matrix are superimposed the superposition matrix to be formed.
Further, each element can be calculated according to following formula in the third matrix:
Sw(Σ)ii=max (Σii-wi, 0), i=1,2......n
Wherein, the ΣiiFor the value of element in the singular value diagonal matrix, wiFor element in the weight vectors
Value.
Step S490, respectively by corresponding first matrix of each similar matrix, the third matrix and described second
The transposed matrix of matrix is multiplied, and obtains the corresponding second denoising submatrix of each similar matrix.
It can be calculated according to following formula:
Wherein, U is the first matrix, Sw(Σ) is third matrix, and V is the second matrix.
Step S500 will all the second denoising submatrix sum, and obtain the second denoising image array.
Further, in order to illustrate the beneficial effect of the embodiment of the present invention, using standard common in image denoising
Lenas256X256, Manchar256X256, peppers256X256 image, and standard deviation sigma is separately added into this three width image
=10,25,50,70,100 zero-mean additive white Gaussian noise, the raw 21 noisy figures of common property are as test data.Make respectively
With (1) three-dimensional Block- matching denoising, this method abbreviation BM3D；(2) the non local denoising of funnel self similarity, this method abbreviation AFNM；(3)
Weight nuclear norm denoising, this method abbreviation WNNM；(4) image processing method of the embodiment of the present invention, wherein adaptive wiener
The threshold value of filtering is set as 3X3, and parameter c uses 2.8284 in second of denoising, according to different noise variance value σn≤25、
25<σn≤40、40<σn≤60、60<σn, search window is successively set as 7 × 7,7 × 7,8 × 8,9 × 9, and the number of iterations K is successively
It is set as 6,8,9,11.σnNon local search window is sized to 35 when≤25, remaining is all 40.(1), (2) are used simultaneously
And methods herein processes comparison to a frame of the monitoring video image of a certain workshop.
The evaluation of image denoising effect is divided into subjective assessment standard and objectivity evaluation criterion two major classes.Subjective criterion master
If the vision by human eye directly observes image, to make evaluation to picture quality, picture quality is good, feels clear
It is clear, it is good to denoise effect, otherwise denoising effect is then poor.Subjective, the embodiment of the present invention is weighed using structural similarity (SSIM)
The denoising effect of spirogram picture, objectively, the present invention uses the denoising effect of Y-PSNR (PSNR) Lai Hengliang image.Reality
There is no muting image in life, will generally fail to the evaluation criterion such as PSNR and SSIM etc. of denoising image.The present invention
Using the evaluation method for combining fuzziness and noise level, fuzziness is measured with average edge width, with the noise of smooth region
Information representation noise level, referred to as IQA, the smaller picture quality of IQA are better.
By four kinds of denoising methods to tri- width image of Lena, Manchar, peppers respectively noise criteria difference σ=10,
25, the results are shown in Table 1 for the Y-PSNR in the case where 50,70,100.The different denoising methods of table 1 are to different images in difference
Y-PSNR PSNR and SSIM under noise intensity.
Table 1
As can be seen from Table 1 the method for the present embodiment no matter on signal-to-noise ratio or structural similarity all than BM3D and AFLM
Will be high, especially under strong noise, it can more show superiority.Compared with original weights nuclear norm Denoising Algorithm, although right
The processing result of lena is slightly not good enough on signal-to-noise ratio, but is preferable on SSIM.Processing to other two images
Also being satisfied with as a result, illustrate that the present invention can effectively improve the denoising effect of image, meet human visual system's
Impression.
Fig. 5 shows third embodiment of the invention and existing method (AFNM, BM3D, WNNM) and tests lena256X256
The denoising effect contrast figure of figure.Fig. 6 shows third embodiment of the invention and existing method (AFNM, BM3D, WNNM) is right
The denoising effect contrast figure of Monarch256X256 test chart.Fig. 7 be with third embodiment of the invention and existing method (AFNM,
BM3D, WNNM) to the denoising effect contrast figure of peppers256X256 test chart.
Fig. 5,6,7 show that in noise criteria difference be σ=25, when 50,70,100, the denoising effect picture of above-mentioned four kinds of methods
Comparison.AFNM denoising and BM3D denoising can eliminate noise to a certain extent it can be seen from Fig. 5,6,7, but BM3D is denoised
Although being effectively maintained the detail edges of image under high noise levels, but go out to introduce a large amount of vacation in detail textures
Striped, and the image after AFNM denoising has seriously colored piebald effect, making that treated, image fault is larger.It is gone using WNNM
It makes an uproar and can effectively solve the problem that vision distortion caused by leave request line and piebald phenomenon, also can preferably remove noise while retaining edge
It is complete with texture, but more or less also there is deceptive information.Using the secondary image denoising based on WNNM, sufficiently use
The advantages of WNNM, and while reducing the number of iterations good visual experience is brought, although the processing to lena exists
On signal-to-noise ratio almost than WNNM, but SSIM is integrally improved.It can be seen that being gone using the secondary image based on WNNM
Obtained visual quality of images of making an uproar to go than using other converter techniques denoise come it is good.
Fig. 8 is that third embodiment of the invention and existing method (BM3D, WNNM) imitate the denoising of certain workshop monitoring test chart
Fruit comparison diagram.Fig. 9 is that third embodiment of the invention and existing method (BM3D, WNNM) monitor noise level effect pair to certain workshop
Than figure, abscissa is noise variance value in Fig. 9, and ordinate is noise level.Figure 10 is third embodiment of the invention and existing side
Method (BM3D, WNNM), which monitors certain workshop, surveys IQA evaluation comparison diagram, and abscissa is noise variance value in Figure 10, and ordinate is
IQA。
Compare Fig. 8,9,10, which give BM3D, AFNM and the embodiment of the present invention, imitates the denoising of certain workshop monitoring image
Fruit.As can be seen from Figure 8 three kinds of algorithms can effectively remove noise, but the visual effect of BM3D is apparently without other two kinds
It is good.It can be seen that the present invention from Fig. 9, Figure 10 and denoise that effect is best, and noise level only has 0.0034, the IQA to be in σ=36
1.3607, it is more excellent than other both of which, and as the IQA for increasing by three kinds of noise denoisings and noise level of σ is increasing, it can
To draw a conclusion, using one kind based on weighting the improved secondary image denoising method of nuclear norm either to natural image emulation at
Reason, or the processing to strong noise image in real life, can obtain good effect.
Image processing method provided in an embodiment of the present invention, by carrying out overall change first to noise image matrix
Point, and adaptive wiener filter carried out to residual image matrix, and described first will denoise image array, described filtered residual
Difference image matrix and weight vectors carry out second denoising processing according to the first preset rules, obtain the second denoising image array,
Its prior information for taking full advantage of image preferably remains the edge, thin of image thus while favorably removal noise
Section, while obtaining high s/n ratio, structural similarity also keeps higher level, sufficiently meets people's visual effect.
Please refer to the functional block diagram that Fig. 4 is image data processing system 600 provided in an embodiment of the present invention.It is described
Image real time transfer 600 includes the first denoising module 610, processing module 620, filter module 630, the second denoising module 640.
The first denoising module 610 obtains the first denoising image moment for carrying out total variation to noise image matrix
Battle array.
The processing module 620, for obtaining residual according to the first denoising image array and the noise image matrix
Difference image matrix.
The filter module 630 obtains filtered for the residual image matrix to be carried out adaptive wiener filter
Residual image matrix.
The second denoising module 640, for denoising image array, the filtered residual image square for described first
Battle array and weight vectors carry out second of denoising according to the first preset rules, obtain the second denoising image array.
Above each module can be by software code realization, at this point, above-mentioned each module can be stored in depositing for server 200
In reservoir 201.Above each module can equally be realized by hardware such as IC chip.
It should be noted that all the embodiments in this specification are described in a progressive manner, each embodiment weight
Point explanation is the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.
The technical effect of image data processing system provided by the embodiment of the present invention, realization principle and generation and aforementioned
Embodiment of the method is identical, and to briefly describe, Installation practice part does not refer to place, can refer to corresponding in preceding method embodiment
Content.
In several embodiments provided herein, it should be understood that disclosed device and method can also pass through
Other modes are realized.The apparatus embodiments described above are merely exemplary, for example, flow chart and block diagram in attached drawing
Show the device of multiple embodiments according to the present invention, the architectural framework in the cards of method and computer program product,
Function and operation.In this regard, each box in flowchart or block diagram can represent the one of a module, section or code
Part, a part of the module, section or code, which includes that one or more is for implementing the specified logical function, to be held
Row instruction.It should also be noted that function marked in the box can also be to be different from some implementations as replacement
The sequence marked in attached drawing occurs.For example, two continuous boxes can actually be basically executed in parallel, they are sometimes
It can execute in the opposite order, this depends on the function involved.It is also noted that every in block diagram and or flow chart
The combination of box in a box and block diagram and or flow chart can use the dedicated base for executing defined function or movement
It realizes, or can realize using a combination of dedicated hardware and computer instructions in the system of hardware.
In addition, each functional module in each embodiment of the present invention can integrate one independent portion of formation together
Point, it is also possible to modules individualism, an independent part can also be integrated to form with two or more modules.
It, can be with if the function is realized and when sold or used as an independent product in the form of software function module
It is stored in a computer readable storage medium.Based on this understanding, technical solution of the present invention is substantially in other words
The part of the part that contributes to existing technology or the technical solution can be embodied in the form of software products, the meter
Calculation machine software product is stored in a storage medium, including some instructions are used so that a computer equipment (can be a
People's computer, server or network equipment etc.) it performs all or part of the steps of the method described in the various embodiments of the present invention.
And storage medium above-mentioned includes: USB flash disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), arbitrary access
The various media that can store program code such as memory (RAM, Random Access Memory), magnetic or disk.It needs
It is noted that herein, relational terms such as first and second and the like are used merely to an entity or operation
It is distinguished with another entity or operation, without necessarily requiring or implying between these entities or operation, there are any this
Actual relationship or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to nonexcludability
It include so that the process, method, article or equipment for including a series of elements not only includes those elements, but also to wrap
Include other elements that are not explicitly listed, or further include for this process, method, article or equipment intrinsic want
Element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including described want
There is also other identical elements in the process, method, article or equipment of element.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.It should also be noted that similar label and letter exist
Similar terms are indicated in following attached drawing, therefore, once being defined in a certain Xiang Yi attached drawing, are then not required in subsequent attached drawing
It is further defined and explained.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (5)
1. a kind of image processing method, it is characterised in that, the described method includes:
Total variation is carried out to noise image matrix, obtains the first denoising image array；
According to the first denoising image array and the noise image matrix, residual image matrix is obtained；
The residual image matrix is subjected to adaptive wiener filter, obtains filtered residual image matrix；
It is added the first denoising image array and the filtered residual image matrix to form superposition matrix；
According to noise variance value, the superposition matrix is divided into multiple submatrixs；
According to the second preset rules, the corresponding similar matrix of each submatrix is obtained respectively；
Singular value decomposition is carried out to each similar matrix respectively, obtain corresponding first matrix of each similar matrix,
Singular value diagonal matrix and the second matrix；
The corresponding singular value diagonal matrix of each similar matrix and weight vectors are multiplied respectively, obtain each similar matrix
Corresponding third matrix；
Respectively by corresponding first matrix of each similar matrix, the transposed matrix of the third matrix and second matrix
It is multiplied, obtains the corresponding second denoising submatrix of each similar matrix；
It will all the second denoising submatrix sum, and obtain the second denoising image array.
2. according to the first denoising image array and described making an uproar the method according to claim 1, wherein described
Acoustic image matrix obtains residual image matrix, comprising:
The noise image matrix is subtracted into the matrix that the first denoising image array obtains, as the residual image square
Battle array.
3. the method according to claim 1, wherein second preset rules are regularization iterative processing.
4. according to the method described in claim 3, it is characterized in that, the regularization iterative formula of the regularization iterative processing isWherein,y(0)=y, y are the first denoising image array and the filtering
Residual image matrix afterwards is superimposed the superposition matrix to be formed, and δ is iteration step length parameter, k 6,8,9 or 11.
5. the method according to claim 1, wherein the weight vectors w=[w1,w2,.....,wi] in it is each
ElementWherein, c is a constant greater than zero, and n is third parameter, ε=10-16For fixed value,σi(Yi) it is YiI-th of singular value, i=1,2 ..., n, Yi be by described
First denoising image array and the filtered residual image matrix are superimposed the superposition matrix to be formed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610948033.8A CN106504207B (en) | 2016-10-24 | 2016-10-24 | A kind of image processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610948033.8A CN106504207B (en) | 2016-10-24 | 2016-10-24 | A kind of image processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106504207A CN106504207A (en) | 2017-03-15 |
| CN106504207B true CN106504207B (en) | 2019-08-02 |
Family
ID=58322902
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610948033.8A Active CN106504207B (en) | 2016-10-24 | 2016-10-24 | A kind of image processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106504207B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108765330B (en) * | 2018-05-21 | 2022-05-27 | 西北大学 | Image denoising method and device based on global and local prior joint constraint |
| CN110675344B (en) * | 2019-09-24 | 2022-07-05 | 福州大学 | Low-rank denoising method and device based on real color image self-similarity |
| CN111383196B (en) * | 2020-03-13 | 2023-07-28 | 浙江大华技术股份有限公司 | Infrared image stripe eliminating method, infrared detector and storage device |
| CN117581556A (en) * | 2021-07-07 | 2024-02-20 | 高通股份有限公司 | Cascaded image processing for noise reduction |
| CN113378415B (en) * | 2021-08-12 | 2021-11-02 | 西南科技大学 | Multimedia data self-adaptive recovery method and device based on local and global constraints |
| CN113850632B (en) * | 2021-11-29 | 2022-03-01 | 平安科技（深圳）有限公司 | User category determination method, device, equipment and storage medium |
| CN115601232B (en) * | 2022-12-14 | 2023-03-10 | 华东交通大学 | Color image decoloring method and system based on singular value decomposition |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103208104A (en) * | 2013-04-16 | 2013-07-17 | 浙江工业大学 | Non-local theory-based image denoising method |
| US9015294B2 (en) * | 2006-10-30 | 2015-04-21 | Dell Products L.P. | System and method for assigning addresses to information handling systems |
-
2016
- 2016-10-24 CN CN201610948033.8A patent/CN106504207B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9015294B2 (en) * | 2006-10-30 | 2015-04-21 | Dell Products L.P. | System and method for assigning addresses to information handling systems |
| CN103208104A (en) * | 2013-04-16 | 2013-07-17 | 浙江工业大学 | Non-local theory-based image denoising method |
Non-Patent Citations (1)
| Title |
|---|
| Weighted Nuclear Norm Minimization with Application to Image Denoising;Shuhang Gu 等;《IEEE Conference on Computer Vision and Pattern Recognition（CVPR）》;20141231;第1-8页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106504207A (en) | 2017-03-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106504207B (en) | A kind of image processing method | |
| Gu et al. | No-reference quality assessment of screen content pictures | |
| Zhang et al. | Convolutional sparse and low-rank coding-based rain streak removal | |
| CN105913393B (en) | A kind of adaptive wavelet threshold image de-noising method and device | |
| CN104023230B (en) | A kind of non-reference picture quality appraisement method based on gradient relevance | |
| CN113450288B (en) | Single image rain removing method and system based on deep convolutional neural network and storage medium | |
| CN108134937B (en) | Compressed domain significance detection method based on HEVC | |
| CN110070539A (en) | Image quality evaluating method based on comentropy | |
| Wen et al. | VIDOSAT: High-dimensional sparsifying transform learning for online video denoising | |
| CN106133788A (en) | Process the image processing apparatus of digital picture | |
| CN103208097A (en) | Principal component analysis collaborative filtering method for image multi-direction morphological structure grouping | |
| Ono et al. | Color-line regularization for color artifact removal | |
| CN106204502B (en) | Based on mixing rank L0Regularization fuzzy core estimation method | |
| Patel et al. | Separated component-based restoration of speckled SAR images | |
| CN103116873A (en) | Image noise reducing method | |
| CN116309178A (en) | Visible light image denoising method based on self-adaptive attention mechanism network | |
| Zeng et al. | Screen content video quality assessment model using hybrid spatiotemporal features | |
| Kumar et al. | Image denoising based on fractional gradient vector flow and overlapping group sparsity as priors | |
| CN107578406A (en) | Based on grid with Wei pool statistical property without with reference to stereo image quality evaluation method | |
| CN110930314A (en) | Image banding noise suppression method and device, electronic device and storage medium | |
| Ouyang et al. | Research on DENOISINg of cryo-em images based on deep learning | |
| Wang et al. | Edge structure preserving image denoising using OAGSM/NC statistical model | |
| CN113487491A (en) | Image restoration method based on sparsity and non-local mean self-similarity | |
| Zeng et al. | Perceptual quality assessment of denoised images | |
| AKINTOYE et al. | COMPOSITE MEDIAN WIENER FILTER BASED TECHNIQUE FOR IMAGE ENHANCEMENT. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |