CN108765322A - Image de-noising method and device - Google Patents
Image de-noising method and device Download PDFInfo
- Publication number
- CN108765322A CN108765322A CN201810468116.6A CN201810468116A CN108765322A CN 108765322 A CN108765322 A CN 108765322A CN 201810468116 A CN201810468116 A CN 201810468116A CN 108765322 A CN108765322 A CN 108765322A
- Authority
- CN
- China
- Prior art keywords
- image
- structure group
- denoising
- dictionary
- mixed noise
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000010276 construction Methods 0.000 claims abstract description 13
- 238000004422 calculation algorithm Methods 0.000 claims description 19
- 239000011159 matrix material Substances 0.000 claims description 15
- 239000000203 mixture Substances 0.000 claims description 7
- 238000000926 separation method Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 238000000354 decomposition reaction Methods 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000006116 polymerization reaction Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 9
- 238000010586 diagram Methods 0.000 description 10
- 230000006870 function Effects 0.000 description 10
- 230000000694 effects Effects 0.000 description 6
- 239000006002 Pepper Substances 0.000 description 5
- 238000001914 filtration Methods 0.000 description 4
- 230000002093 peripheral effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 150000003839 salts Chemical class 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000004590 computer program Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 235000008534 Capsicum annuum var annuum Nutrition 0.000 description 1
- 240000008384 Capsicum annuum var. annuum Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000002620 method output Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000000007 visual effect Effects 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
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
An embodiment of the present invention provides a kind of image de-noising method and devices, are related to technical field of image processing.Method is applied to an electronic equipment, the method includes：It obtains containing mixed noise after denoising image, waits for that denoising image is filtered to described, obtain initial pictures；It is based on the initial pictures again, construct multiple structure groups and dictionary construction is carried out to each structure group in the multiple structure group, obtains the corresponding study dictionary of each structure group；It is then based on the corresponding study dictionary of each structure group, establishes mixed noise removal model；And loop iteration calculating is carried out to mixed noise removal model, iterations are preset until meeting, export the image after denoising.This method is more acurrate and improves computational efficiency.
Description
Technical field
The present invention relates to technical field of image processing, in particular to a kind of image de-noising method and device.
Background technology
Image is easy to be influenced by factors such as sensor, camera shake and transmission channels in transmission process, leads to figure
As being polluted while being easy by a variety of noises, such as mixed Gaussian noise and impulsive noise, so as to cause the visual quality of image
Decline.In order to remove the mixed noise in image, priori and background knowledge by means of image are needed, and the acquisition of priori
Modeling dependent on image.Image de-noising method accuracy is low at present.
Invention content
The purpose of the present invention is to provide a kind of image de-noising method and devices, to improve the above problem.On realizing
Purpose is stated, the technical solution adopted by the present invention is as follows：
In a first aspect, an embodiment of the present invention provides a kind of image de-noising method, it is applied to an electronic equipment, the method
Including：It obtains and waits for denoising image containing mixed noise；It waits for that denoising image is filtered to described, obtains initial pictures；
Based on the initial pictures, multiple structure groups are constructed；Dictionary construction is carried out to each structure group in the multiple structure group, is obtained
Obtain the corresponding study dictionary of each structure group；Based on the corresponding study dictionary of each structure group, mixed noise is established
Remove model；Loop iteration calculating is carried out to mixed noise removal model, iterations is preset until meeting, exports denoising
Image afterwards.
Second aspect, an embodiment of the present invention provides a kind of image denoising devices, run on an electronic equipment, described device
Including acquiring unit, filter unit, structure group structural unit, dictionary structural unit, establish unit and iterative calculation unit.It obtains
Unit waits for denoising image for obtaining containing mixed noise.Filter unit, for waiting for that denoising image is filtered place to described
Reason obtains initial pictures.Structure group structural unit constructs multiple structure groups for being based on the initial pictures.Dictionary construction is single
Member obtains corresponding of each structure group for carrying out dictionary construction to each structure group in the multiple structure group
Handwriting practicing allusion quotation.Unit is established, for being based on the corresponding study dictionary of each structure group, establishes mixed noise removal model.Repeatedly
It is defeated until meeting default iterations for carrying out loop iteration calculating to mixed noise removal model for computing unit
Go out the image after denoising.
An embodiment of the present invention provides a kind of image de-noising method and devices, are applied to an electronic equipment, the method packet
It includes：It obtains containing mixed noise after denoising image, waits for that denoising image is filtered to described, obtain initial pictures；
It is based on the initial pictures again, construct multiple structure groups and dictionary structure is carried out to each structure group in the multiple structure group
It makes, obtains the corresponding study dictionary of each structure group；It is then based on the corresponding study dictionary of each structure group, is established
Mixed noise removes model；And loop iteration calculating is carried out to mixed noise removal model, preset iteration until meeting
Number exports the image after denoising.This method is more acurrate and improves computational efficiency.
Other features and advantages of the present invention will be illustrated in subsequent specification, also, partly be become from specification
It is clear that by implementing understanding of the embodiment of the present invention.The purpose of the present invention and other advantages can be by saying what is write
Specifically noted structure is realized and is obtained in bright book, claims and attached drawing.
Description of the drawings
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 structure diagram that can be applied to electronic equipment provided in an embodiment of the present invention；
Fig. 2 is the flow chart of image de-noising method provided in an embodiment of the present invention；
Fig. 3 is the denoising result comparison schematic diagram of Lena images in image de-noising method provided in an embodiment of the present invention；
Fig. 4 is the denoising result comparison schematic diagram of boat images in image de-noising method provided in an embodiment of the present invention；
Fig. 5 is the structure diagram for the image denoising device that the embodiment of the present invention improves.
Specific implementation mode
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.The present invention being usually described and illustrated herein in the accompanying drawings is implemented
The component of example can be arranged and be designed with a variety of different configurations.Therefore, below to the reality of the present invention provided in the accompanying drawings
The detailed description for applying example is not intended to limit the range of claimed invention, but is merely representative of the selected implementation of the present invention
Example.Based on the embodiments of the present invention, those of ordinary skill in the art are obtained without creative efforts
Every other embodiment, shall fall within the protection scope of the present invention.
It should be noted that：Similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined, then it further need not be defined and explained in subsequent attached drawing in a attached drawing.Meanwhile the present invention's
In description, term " first ", " second " etc. are only used for distinguishing description, are not understood to indicate or imply relative importance.
Present inventor has found that the prior art is in invention the embodiment of the present application：Traditional traditional rarefaction representation and plus
The image blend denoising scheme summary of power coding has：The degraded image y of the mixed noise of the spiced salt containing gaussian sum pollution is inputted first, if
Dictionary is φ, and carrying out adaptive median filter to degraded image y obtains a width initial pictures x(0), then initialize coded residual e
=y-x(0), and μ=0 is initialized, μ indicates non local similar image block coding coefficient vector.Secondly, to mixed Gaussian and the spiced salt
Noise carries out joint modeling, and the final model for removing mixed noise is represented by：
Wherein, w indicates weighting matrix, diagonal entry wii=wi。wiiIt is expressed asIn the range of wii
∈[0,1].A is a constant more than 0, for inhibiting weight wiiAttenuation degree.λ indicates regularization factors, | | | |1Table
Show l1Norm.Indicate the code coefficient vector of estimation.It is finally based on model (1) and carries out K loop iteration operation, it is final to obtain
Image x=φ α after denoising(K)。
Further, present inventor has found the image blend denoising side based on traditional rarefaction representation and weighted coding
Method analyzes the spike heavytailed distribution characteristic of mixed noise, and has shown that this distribution is difficult to build using conventional parameter model
The conclusion of mould.In conjunction with traditional rarefaction representation and non local self-similarity prior information, by being weighted modeling to coded residual
To inhibit mixed noise, the effect of preferable removal mixed noise is achieved.But conventional method coded residual merely by
The difference of raw noise image and all pixels point average value obtains, when calculating non local self-similar image block priori, still
Traditional way is so utilized, i.e., without excluding impulsive noise, causes the priori obtained not accurate enough.On the other hand based on tradition
The image blend denoising method of rarefaction representation and weighted coding is required for each image block in dictionary learning and cataloged procedure
Thus individually processing needs the Large-scale Optimization Problems for solving a high complexity, is easy to cause sparse coding coefficient in this way
Solve not accurate enough, the computational efficiency for eventually leading to the program is relatively low, it is difficult to meet the needs handled in real time.
In consideration of it, the embodiment of the present application provides a kind of image de-noising method and device, imitated with improving accuracy and calculating
Rate.
Fig. 1 shows a kind of structure diagram for the electronic equipment 100 that can be applied in the embodiment of the present invention.As shown in Figure 1,
Electronic equipment 100 may include memory 102, storage control 104, one or more (one is only shown in Fig. 1) processors
106, Peripheral Interface 108, input/output module 110, audio-frequency module 112, display module 114, radio-frequency module 116 and image denoising
Device.
Memory 102, storage control 104, processor 106, Peripheral Interface 108, input/output module 110, audio mould
Directly or indirectly be electrically connected between block 112, display module 114,116 each element of radio-frequency module, with realize data transmission or
Interaction.For example, can realize electrical connection by one or more communication bus or signal bus between these elements.Image denoising
Method respectively includes at least one software work(that can be stored in the form of software or firmware (firmware) in memory 102
Can module, such as software function module or computer program that described image denoising device includes.
Memory 102 can store various software programs and module, such as image denoising side provided by the embodiments of the present application
Method and the corresponding program instruction/module of device.Processor 106 by run storage software program in the memory 102 and
Module realizes the image de-noising method in the embodiment of the present application to perform various functions application and data processing.
Memory 102 can include but is not limited to random access memory (Random Access Memory, RAM), only
Read memory (Read Only Memory, ROM), programmable read only memory (Programmable Read-Only
Memory, PROM), erasable read-only memory (Erasable Programmable Read-Only Memory, EPROM),
Electricallyerasable ROM (EEROM) (Electric Erasable Programmable Read-Only Memory, EEPROM) etc..
Processor 106 can be a kind of IC chip, have signal handling capacity.Above-mentioned processor can be general
Processor, including central processing unit (Central Processing Unit, abbreviation CPU), network processing unit (Network
Processor, abbreviation NP) etc.；It can also be digital signal processor (DSP), application-specific integrated circuit (ASIC), field-programmable
Gate array (FPGA) either other programmable logic device, discrete gate or transistor logic, discrete hardware components.It can
To realize or execute disclosed each method, step and the logic diagram in the embodiment of the present application.General processor can be micro-
Processor or the processor can also be any conventional processor etc..
The Peripheral Interface 108 couples various input/output devices to processor 106 and memory 102.At some
In embodiment, Peripheral Interface 108, processor 106 and storage control 104 can be realized in one single chip.Other one
In a little examples, they can be realized by independent chip respectively.
The interaction that input/output module 110 is used to that user input data to be supplied to realize user and electronic equipment 100.It is described
Input/output module 110 may be, but not limited to, mouse and keyboard etc..
Audio-frequency module 112 provides a user audio interface, may include that one or more microphones, one or more raises
Sound device and voicefrequency circuit.
Display module 114 provides an interactive interface (such as user interface) between electronic equipment 100 and user
Or it is referred to user for display image data.In the present embodiment, the display module 114 can be liquid crystal display or touch
Control display.Can be that the capacitance type touch control screen or resistance-type of single-point and multi-point touch operation is supported to touch if touch control display
Control screen etc..Single-point and multi-point touch operation is supported to refer to touch control display and can sense on the touch control display one or more
The touch control operation generated simultaneously at a position, and transfer to processor 106 to be calculated and handled the touch control operation that this is sensed.
Radio-frequency module 116 is used to receive and transmit electromagnetic wave, realizes the mutual conversion of electromagnetic wave and electric signal, thus with
Communication network or other equipment are communicated.
It is appreciated that structure shown in FIG. 1 is only to illustrate, electronic equipment 100 may also include it is more than shown in Fig. 1 or
Less component, or with the configuration different from shown in Fig. 1.Each component shown in Fig. 1 may be used hardware, software or its
Combination is realized.
In the embodiment of the present invention, electronic equipment 100 can be used as user terminal, or as server.User terminal
Can be PC (personal computer) computer, tablet computer, mobile phone, laptop, smart television, set-top box, vehicle-mounted
The terminal devices such as terminal.
In the present embodiment, it is 64 Windows7 operating systems and Matlab2010 that environment is equipped in electronic equipment.
The hardware configuration of electronic equipment includes CPU 2.6G and memory 8G.
Referring to Fig. 2, an embodiment of the present invention provides a kind of image de-noising method, it is applied to an electronic equipment, the side
Method includes：Step S200, step S210, step S220, step S230, step S240 and step S250.
Step S200：It obtains and waits for denoising image containing mixed noise.
In the present embodiment, mixed noise may include Gaussian noise and salt-pepper noise.For example, waiting for that denoising image y can be with
To be 20 by Gaussian noise variance, the degraded image for the mixed noise pollution simultaneously that salt-pepper noise ratio is 30%.
Step S210：It waits for that denoising image is filtered to described, obtains initial pictures.
Optionally, step S210 may include：
It waits for that denoising image is filtered to described based on adaptive median filter algorithm, obtains initial pictures.
It is filtered for example, treating denoising image y and carrying out adaptive median filter algorithm, obtains initial pictures x(0), then
Initialize coded residual e=y-x(0)。
After step S200, before step S210, the method further includes：
Initialize relevant parameter.
For example, the relevant parameter may include the row value B of image block matrixs, the number c of similar block, regularization factors
μ, λ.
Step S220：Based on the initial pictures, multiple structure groups are constructed.
Optionally, step S220 may include：
The initial pictures are divided into multiple images block；Based on block matching algorithm, respectively in described multiple images block
Each image block carry out adjacent block search, the corresponding multiple similar images of each image block are obtained according to search result
Block；By one structure group of each corresponding multiple similar image blocks compositions of image block, to obtain multiple structure groups.
For example, initial pictures x(0)Size be N × N, by initial pictures be divided into multiple sizes be Bs×BsImage block
xk, such as BsValue 16, wherein k=1,2 ..., n.Then to each image block, using block matching algorithm search and this image block
Similar c similar image block is most matched, and c similar image block is constituted into setIt then will setIn all phases
A matrix is lined up like image block and constitutes a structure group, is denoted asSimilarly, it obtains and corresponds to each image block
Multiple similar image blocks constitute a structure group, to obtain multiple structure groups.
In the embodiment of the present application, in conjunction with the non local similitude and sparse prior of image, finding has similar image
Block simultaneously constitutes a structure group, using single structure group as the basic unit of sparse representation model, rather than single image block
As basic unit, it is contemplated that the correlation between image block improves accuracy.
Step S230：Dictionary construction is carried out to each structure group in the multiple structure group, obtains each structure
The corresponding study dictionary of group.
Optionally, step S230 may include：
Dictionary construction is carried out to each structure group in the multiple structure group based on singular value decomposition method, described in acquisition
Each corresponding study dictionary of structure group.
For example, to each structure groupCorresponding study dictionary is constructed using singular value decomposition method
In the present embodiment, there are one corresponding dictionaries for each structure group, have preferable picture material adaptive
Property, accuracy is improved, computational efficiency is improved.
Step S240：Based on the corresponding study dictionary of each structure group, mixed noise removal model is established.
Step S240 may include：
Establishing mixed noise removal model is：
Wherein,For the estimated value of code coefficient vector, wGFor structure group weighting matrix,It is weighted for i-th of structure group
Matrix,It is expressed as A is the constant more than 0, and y is described
Wait for denoising image,For the corresponding study dictionary of i-th of structure group, φ in the multiple structure groupGIt is allCombination, αG
For the corresponding sparse coding coefficient matrix of i-th of structure group, α in the multiple structure groupGIt is allCombination, λ indicate just
Then change the factor, | | αG||0Indicate αGL0Norm.
In the embodiment of the present application, mixed noise removal model is group sparse representation model, sparse based on tradition to overcome
Indicate the computational efficiency problem of the image blend denoising method of model and weighted coding.
Step S250：Loop iteration calculating is carried out to mixed noise removal model, iterations are preset until meeting,
Export the image after denoising.
As an implementation, step S250 may include：
It calls separation Bregman iterative algorithms to solve and calculates the mixed noise removal modelIterations are preset until meeting, export the image after denoising.
Further, the default iterations are K, and the calling separation Bregman iterative algorithms solve described in calculating
Mixed noise removes modelIterations are preset until meeting, output is gone
Image after making an uproar, including：
Definitiong(αG)=λ | | αG||0And u=φGαG；
Separation Bregman iterative algorithms are called, it willIt is converted into following three formulas
It is solved respectively：Andμ, λ indicate the regularization factors for 0；
Bregman iterative algorithms (Split Bregman Iteration, SBI) are detached to solveOptimization problem process it is as follows：
Given αG, forPartial derivative is asked to u, is calculated
To the estimated value of u I indicates identity matrix；
Given u, forTo αGPartial derivative is sought, is calculated
To αGEstimated value Τ=(λ × Bs×c)/(μN)
Indicate that a threshold value, N indicate the line number or columns of image array, Indicate study dictionaryInverse square
Battle array, hard () indicate that hard -threshold operator, ⊙ indicate that two vectorial Hadamard accumulate operation, and 1 indicates that all elements are 1
Vector；
It is based onCalculate u(k+1)；
Calculate r(k+1)=u(k+1)-b(k+1), Τ=(λ × Bs×c)/(μN)；
It calculatesUpdate residual error e(k)=y-x(k), according toCalculate structure group weight wG；
By all study dictionariesPolymerization is to update dictionary φG；
All sparse coding coefficients are polymerize to update
It is based onUpdate b(k+1)；
Based on the above, after carrying out K loop iteration operation, the image x=φ after denoising are exportedGαG (K).Repeat step
S230- step S240 after carrying out K loop iteration operation, export the image after denoising.
In order to which the denoising performance of image de-noising method provided by the embodiments of the present application is further illustrated, two width have been selected
It waits for denoising image, is Lena and Boat respectively, pixel size is 512 × 512.Meanwhile image provided by the embodiments of the present application
Compared with denoising method has carried out performance with 2 kinds of mixed noise filtering methods in the prior art, 2 kinds of mixing in the prior art are made an uproar
Sound filtering method includes document 1 i.e. J.Jiang, L.Zhang, J.Yang.Mixed noise removal by weighted
encoding with sparse nonlocal regularization.IEEE Transactions on Image
Processing,2014,23(6):2651-2662 and document 2 are K.Dabov, A.Foi, V.Katkovnik, and
K.Egiazarian.Image Denoising by Sparse 3-D Transform-Domain Collaborative
Filtering.IEEE Transactions Image Processing,2007,16(8):2080-2095, document 1 and document
The parameter of 2 methods is configured according to the value suggested in original text.
In image de-noising method provided by the embodiments of the present application, parameter is set as：BsValue be 8, i.e. tile size Bs×
BsIt is 88；The similar image block number c that each structure group includes is set as 60；The size of structure group is 64 × 60；Similar image
Interval between block is set as 4, and the size of structure group searching window is 40 × 40；The value of regularization factors λ is 0.5；The value of μ is set
For 2.5e-3。
Fig. 3 and Fig. 4 are please referred to, the denoising effect comparing result that two width wait for denoising image is set forth in Fig. 3 to Fig. 4.Its
In, (a) indicates the Lena images polluted by mixed noise in Fig. 3, and Gaussian noise variance is 20, and salt-pepper noise density is
30%, Fig. 3 (b) indicates that the Lena images after 2 denoising method denoising of document, Fig. 3 (c) indicate to pass through 1 denoising method of document
Lena images after denoising, Fig. 3 (d) indicate the figures of the Lena after image de-noising method denoising provided by the embodiments of the present application
Picture；(a) indicates the Boat images polluted by mixed noise in Fig. 4, and Gaussian noise variance is 20, and salt-pepper noise density is
30%, Fig. 4 (b) indicates that the Boat images after 2 denoising method denoising of document, Fig. 4 (c) indicate to pass through 1 denoising method of document
Boat images after denoising, Fig. 4 (d) indicate the figures of the Boat after image de-noising method denoising provided by the embodiments of the present application
Picture.
The embodiment of the present application is can be seen that from Fig. 3 and Fig. 4 to the denoising result of Lena images and Boat images respectively to provide
Image de-noising method output denoising after Lena images and Boat images all have preferable destination edge and keep effect, and texture
Structure is more clear, excess smoothness phenomenon does not occur.Image is utilized in image de-noising method provided by the embodiments of the present application
Structure group rarefaction representation and non local similitude reach preferable mixed noise filtering effect in combination with weighted coding method
Fruit.
Further, objective evaluation is carried out to mixing Denoising Algorithm, the present embodiment uses (peak signal-to-
Noise ratio, PSNR) and feature structure similarity (structural similarity index, FSIM) to filtered
Picture quality is evaluated.Table 1 give Lena and Boat images Gaussian noise variance and salt-pepper noise density be respectively 10,
20%；20,30%；The performance comparison result of 3 kinds of denoising methods.From table 1 it follows that image provided by the embodiments of the present application
Denoising method is all different degrees of in terms of Y-PSNR and structural similarity index to be improved, and removes mixed Gaussian green pepper
The effect of salt noise is preferable.
Results of property after 1 Lena and Boat image denoisings of table compares
An embodiment of the present invention provides a kind of image de-noising methods, are applied to an electronic equipment, the method includes：It obtains
It containing mixed noise after denoising image, waits for that denoising image is filtered to described, obtains initial pictures；It is based on institute again
Initial pictures are stated, multiple structure groups are constructed and dictionary construction is carried out to each structure group in the multiple structure group, obtain institute
State the corresponding study dictionary of each structure group；It is then based on the corresponding study dictionary of each structure group, establishes mixed noise
Remove model；And loop iteration calculating is carried out to mixed noise removal model, iterations are preset until meeting, output
Image after denoising.This method is more acurrate and improves computational efficiency.
Referring to Fig. 5, an embodiment of the present invention provides a kind of image denoising device 400, an electronic equipment is run on, it is described
Device 400 includes acquiring unit 410, filter unit 420, structure group structural unit 430, dictionary structural unit 440, establishes unit
450 and iterative calculation unit 460.
Acquiring unit 410 waits for denoising image for obtaining containing mixed noise.
Filter unit 420 obtains initial pictures for waiting for that denoising image is filtered to described.
Filter unit 420 waits for that denoising image is filtered to described for being based on adaptive median filter algorithm, obtains
Obtain initial pictures.
Structure group structural unit 430 constructs multiple structure groups for being based on the initial pictures.
The structure group structural unit 430, is used for：The initial pictures are divided into multiple images block；Based on Block- matching
Algorithm, the adjacent block search of each image block progress in the block to described multiple images, obtains described every according to search result respectively
The corresponding multiple similar image blocks of a image block；By one structure of each corresponding multiple similar image blocks compositions of image block
Group, to obtain multiple structure groups.
Dictionary structural unit 440 obtains institute for carrying out dictionary construction to each structure group in the multiple structure group
State the corresponding study dictionary of each structure group.
The dictionary structural unit 440 is used for based on singular value decomposition method to each knot in the multiple structure group
Structure group carries out dictionary construction, obtains the corresponding study dictionary of each structure group.
Unit 450 is established, for being based on the corresponding study dictionary of each structure group, establishes mixed noise removal mould
Type.
Unit 450 is established, is for establishing mixed noise removal model：
Wherein,For the estimated value of code coefficient vector, wGFor structure group weighting matrix,For i-th of structure group weighting matrix,
It is expressed asA is the constant more than 0, and y waits for denoising figure to be described
Picture,For the corresponding study dictionary of i-th of structure group, φ in the multiple structure groupGIt is allCombination, αGIt is described more
The corresponding sparse coding coefficient matrix of i-th of structure group, α in a structure groupGIt is allCombination, λ indicate regularization because
Son, | | αG||0Indicate αGL0Norm.
Unit 460 is iterated to calculate, it is pre- until meeting for carrying out loop iteration calculating to mixed noise removal model
If iterations, the image after denoising is exported.
Unit 460 is iterated to calculate, mould is removed for calling separation Bregman iterative algorithms solution to calculate the mixed noise
TypeIterations are preset until meeting, export the image after denoising.
The default iterations are K, iterate to calculate unit 460, are used for：
Definitiong(αG)=λ | | αG||0And u=φGαG；
Separation Bregman iterative algorithms are called, it willIt is converted into following three
Formula is solved respectively：
Andμ, λ indicate the regularization factors for 0；
Given αG, forPartial derivative is asked to u, is calculated
To the estimated value of u I indicates identity matrix；
Given u, forTo αGPartial derivative is sought, is calculated
To αGEstimated value Τ=(λ × Bs×c)/(μN)
Indicate that a threshold value, N indicate the line number or columns of image array, Indicate study dictionaryInverse square
Battle array, hard () indicate that hard -threshold operator, ⊙ indicate that two vectorial Hadamard accumulate operation, and 1 indicates that all elements are 1
Vector；
It is based onCalculate u(k+1)；
Calculate r(k+1)=u(k+1)-b(k+1), Τ=(λ × Bs×c)/(μN)；
It calculatesUpdate residual error e(k)=y-x(k), according toCalculate structure group weight wG；
By all study dictionariesPolymerization is to update dictionary φG；
All sparse coding coefficients are polymerize to update
It is based onUpdate b(k+1)；
Based on the above, after carrying out K loop iteration operation, the image x=φ after denoising are exportedGαG (K)。
The above each unit can be by software code realization, at this point, above-mentioned each unit can be stored in memory 102.
The above each unit can equally be realized by hardware such as IC chip.
The technique effect of image denoising device 400 provided in an embodiment of the present invention, realization principle and generation and aforementioned side
Method embodiment is identical, and to briefly describe, device embodiment part does not refer to place, can refer in corresponding in preceding method embodiment
Hold.
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, the flow chart in attached drawing and block diagram
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 for 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 at some as in the realization method replaced, the function of being marked in box can also be to be different from
The sequence marked in attached drawing occurs.For example, two continuous boxes can essentially be basically executed in parallel, they are sometimes
It can execute in the opposite order, this is depended on the functions 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 function or the dedicated base of action as defined in executing
It realizes, or can be realized using a combination of dedicated hardware and computer instructions in the system of hardware.
In addition, each function module in each embodiment of the present invention can integrate to form an independent portion
Point, can also be modules individualism, can also two or more modules be integrated to form an independent part.
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 read/write memory medium.Based on this understanding, technical scheme 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 expressed 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
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 are deposited
The various media that can store program code such as reservoir (RAM, Random Access Memory), magnetic disc or CD.It needs
Illustrate, herein, relational terms such as first and second and the like be used merely to by an entity or operation with
Another entity or operation distinguish, and without necessarily requiring or implying between these entities or operation, there are any this realities
The relationship or sequence on border.Moreover, the terms "include", "comprise" or its any other variant are intended to the packet of nonexcludability
Contain, so that the process, method, article or equipment including a series of elements includes not only those elements, but also includes
Other elements that are not explicitly listed, or further include for elements inherent to such a process, method, article, or device.
In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including the element
Process, method, article or equipment in there is also other identical elements.
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, any made by repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
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 (10)
1. a kind of image de-noising method, which is characterized in that it is applied to an electronic equipment, the method includes：
It obtains and waits for denoising image containing mixed noise；
It waits for that denoising image is filtered to described, obtains initial pictures；
Based on the initial pictures, multiple structure groups are constructed；
Dictionary construction is carried out to each structure group in the multiple structure group, obtains the corresponding study word of each structure group
Allusion quotation；
Based on the corresponding study dictionary of each structure group, mixed noise removal model is established；
Loop iteration calculating is carried out to mixed noise removal model, iterations are preset until meeting, after exporting denoising
Image.
2. according to the method described in claim 1, it is characterized in that, it is described be based on the initial pictures, construct multiple structure groups,
Including：
The initial pictures are divided into multiple images block；
Based on block matching algorithm, each image block in the block to described multiple images carries out adjacent block search respectively, according to search
As a result the corresponding multiple similar image blocks of each image block are obtained；
By one structure group of each corresponding multiple similar image blocks compositions of image block, to obtain multiple structure groups.
3. according to the method described in claim 1, it is characterized in that, each structure group in the multiple structure group into
Row dictionary constructs, and obtains the corresponding study dictionary of each structure group, including：
Dictionary construction is carried out to each structure group in the multiple structure group based on singular value decomposition method, is obtained described each
The corresponding study dictionary of structure group.
4. according to the method described in claim 1, it is characterized in that, described be based on the corresponding study word of each structure group
Allusion quotation establishes mixed noise removal model, including：
Establishing mixed noise removal model is：
Wherein,For the estimated value of code coefficient vector, wGFor structure group weighting matrix,Square is weighted for i-th of structure group
Battle array,It is expressed as A is the constant more than 0, and y is described waits for
Denoising image,For the corresponding study dictionary of i-th of structure group, φ in the multiple structure groupGIt is allCombination, αGFor
The corresponding sparse coding coefficient matrix of i-th of structure group, α in the multiple structure groupGIt is allCombination, λ indicate canonical
Change the factor, | | αG||0Indicate αGL0Norm.
5. according to the method described in claim 4, it is characterized in that, described recycle repeatedly mixed noise removal model
In generation, calculates, and iterations are preset until meeting, and exports the image after denoising, including：
It calls separation Bregman iterative algorithms to solve and calculates the mixed noise removal modelIterations are preset until meeting, export the image after denoising.
6. according to the method described in claim 5, it is characterized in that, the default iterations are K, the calling detaches
Bregman iterative algorithms, which solve, calculates the mixed noise removal model
Iterations are preset until meeting, export the image after denoising, including：
Definitiong(αG)=λ | | αG||0And u=φGαG；
Separation Bregman iterative algorithms are called, it willIt is converted into following three
Formula is solved respectively：
Andμ, λ indicate the regularization factors for 0；
Given αG, forPartial derivative is asked to u, u is calculated
Estimated value I indicates identity matrix；
Given u, forTo αGPartial derivative is sought, α is calculatedG
Estimated value Τ=(λ × Bs× c)/one threshold of (μ N) expression
Value, N indicate the line number or columns of image array, Indicate study dictionaryInverse matrix, hard ()
Indicate that hard -threshold operator, ⊙ indicate that two vectorial Hadamard accumulate operation, 1 expression all elements are 1 vector；
It is based onCalculate u(k+1)；
Calculate r(k+1)=u(k+1)-b(k+1), Τ=(λ × Bs×c)/(μN)；
It calculatesUpdate residual error e(k)=y-x(k), according toCalculate structure group weight wG；
By all study dictionariesPolymerization is to update dictionary φG；
All sparse coding coefficients are polymerize to update
It is based onUpdate b(k+1)；
Based on the above, after carrying out K loop iteration operation, the image x=φ after denoising are exportedGαG (K)。
7. according to the method described in claim 1, it is characterized in that, described wait for that denoising image is filtered to described, obtain
Initial pictures are obtained, including：
It waits for that denoising image is filtered to described based on adaptive median filter algorithm, obtains initial pictures.
8. a kind of image denoising device, which is characterized in that run on an electronic equipment, described device includes：
Acquiring unit waits for denoising image for obtaining containing mixed noise；
Filter unit obtains initial pictures for waiting for that denoising image is filtered to described；
Structure group structural unit constructs multiple structure groups for being based on the initial pictures；
Dictionary structural unit obtains described each for carrying out dictionary construction to each structure group in the multiple structure group
The corresponding study dictionary of structure group；
Unit is established, for being based on the corresponding study dictionary of each structure group, establishes mixed noise removal model；
Unit is iterated to calculate, for carrying out loop iteration calculating to mixed noise removal model, iteration is preset until meeting
Number exports the image after denoising.
9. device according to claim 8, which is characterized in that the structure group structural unit is used for：By the initial graph
As being divided into multiple images block；Based on block matching algorithm, each image block in the block to described multiple images carries out adjacent respectively
Block search obtains the corresponding multiple similar image blocks of each image block according to search result；By each image block pair
The multiple similar image blocks answered constitute a structure group, to obtain multiple structure groups.
10. device according to claim 8, which is characterized in that the dictionary structural unit, for being based on singular value decomposition
Method carries out dictionary construction to each structure group in the multiple structure group, obtains the corresponding study word of each structure group
Allusion quotation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810468116.6A CN108765322B (en) | 2018-05-16 | 2018-05-16 | Image denoising method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810468116.6A CN108765322B (en) | 2018-05-16 | 2018-05-16 | Image denoising method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108765322A true CN108765322A (en) | 2018-11-06 |
| CN108765322B CN108765322B (en) | 2021-04-27 |
Family
ID=64008201
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810468116.6A Active CN108765322B (en) | 2018-05-16 | 2018-05-16 | Image denoising method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108765322B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109712205A (en) * | 2018-12-10 | 2019-05-03 | 重庆邮电大学 | A kind of compression of images perception method for reconstructing based on non local self similarity model |
| CN110322407A (en) * | 2019-06-03 | 2019-10-11 | 辽宁师范大学 | Image salt-pepper noise minimizing technology based on depth residual error network |
| CN110661549A (en) * | 2019-11-11 | 2020-01-07 | 广东石油化工学院 | PLC signal reconstruction method and system by utilizing dictionary atoms |
| CN110728641A (en) * | 2019-10-12 | 2020-01-24 | 浙江工业大学 | Remote sensing image impulse noise removing method and device |
| CN111242137A (en) * | 2020-01-13 | 2020-06-05 | 江西理工大学 | Salt and pepper noise filtering method and device based on morphological component analysis |
| WO2021012965A1 (en) * | 2019-07-23 | 2021-01-28 | 广州市百果园信息技术有限公司 | Image processing method and apparatus, mobile terminal video processing method and apparatus, device and medium |
| CN113837958A (en) * | 2021-09-09 | 2021-12-24 | 南方医科大学 | Diffusion weighted image denoising algorithm, medium, and apparatus |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050128355A1 (en) * | 2003-12-11 | 2005-06-16 | Samsung Electronics Co., Ltd. | Method of removing noise from digital moving picture data |
| CN105761234A (en) * | 2016-01-28 | 2016-07-13 | 华南农业大学 | Structure sparse representation-based remote sensing image fusion method |
| CN106157254A (en) * | 2015-04-21 | 2016-11-23 | 南京理工大学 | Rarefaction representation remote sensing images denoising method based on non local self-similarity |
| CN106204482A (en) * | 2016-07-08 | 2016-12-07 | 桂林电子科技大学 | Based on the mixed noise minimizing technology that weighting is sparse |
| CN106254720A (en) * | 2016-07-19 | 2016-12-21 | 四川大学 | A kind of video super-resolution method for reconstructing based on associating regularization |
| WO2017143334A1 (en) * | 2016-02-19 | 2017-08-24 | New York University | Method and system for multi-talker babble noise reduction using q-factor based signal decomposition |
| CN107301629A (en) * | 2017-06-28 | 2017-10-27 | 重庆大学 | A kind of image reconstructing method represented based on transform domain joint sparse |
-
2018
- 2018-05-16 CN CN201810468116.6A patent/CN108765322B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050128355A1 (en) * | 2003-12-11 | 2005-06-16 | Samsung Electronics Co., Ltd. | Method of removing noise from digital moving picture data |
| CN106157254A (en) * | 2015-04-21 | 2016-11-23 | 南京理工大学 | Rarefaction representation remote sensing images denoising method based on non local self-similarity |
| CN105761234A (en) * | 2016-01-28 | 2016-07-13 | 华南农业大学 | Structure sparse representation-based remote sensing image fusion method |
| WO2017143334A1 (en) * | 2016-02-19 | 2017-08-24 | New York University | Method and system for multi-talker babble noise reduction using q-factor based signal decomposition |
| CN106204482A (en) * | 2016-07-08 | 2016-12-07 | 桂林电子科技大学 | Based on the mixed noise minimizing technology that weighting is sparse |
| CN106254720A (en) * | 2016-07-19 | 2016-12-21 | 四川大学 | A kind of video super-resolution method for reconstructing based on associating regularization |
| CN107301629A (en) * | 2017-06-28 | 2017-10-27 | 重庆大学 | A kind of image reconstructing method represented based on transform domain joint sparse |
Non-Patent Citations (6)
| Title |
|---|
| JIELIN JIANG ET AL: ""Mixed Noise Removal by Weighted Encoding With Sparse Nonlocal Regularization"", 《IEEE TRANSACTIONS ON IMAGE PROCESSING》 * |
| LICHENG LIU ET AL: ""Weighted Joint Sparse Representation for Removing Mixed Noise in Image"", 《IEEE TRANSACTIONS ON CYBERNETICS》 * |
| SHUTAO LI ET AL: ""Group-Sparse Representation With Dictionary Learning for Medical Image Denoising and Fusion"", 《 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING》 * |
| 张健: ""基于稀疏表示模型的图像复原技术研究"", 《中国博士学位论文全文数据库·信息科技辑》 * |
| 赵井坤: ""基于非局部相似性和稀疏表示的图像去噪技术研究"", 《中国优秀硕士学位论文全文数据库·信息科技辑》 * |
| 韩玉兰: ""基于非局部均值和稀疏表示理论的图像去噪研究"", 《中国优秀硕士学位论文全文数据库·信息科技辑》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109712205A (en) * | 2018-12-10 | 2019-05-03 | 重庆邮电大学 | A kind of compression of images perception method for reconstructing based on non local self similarity model |
| CN110322407A (en) * | 2019-06-03 | 2019-10-11 | 辽宁师范大学 | Image salt-pepper noise minimizing technology based on depth residual error network |
| WO2021012965A1 (en) * | 2019-07-23 | 2021-01-28 | 广州市百果园信息技术有限公司 | Image processing method and apparatus, mobile terminal video processing method and apparatus, device and medium |
| CN110728641A (en) * | 2019-10-12 | 2020-01-24 | 浙江工业大学 | Remote sensing image impulse noise removing method and device |
| CN110661549A (en) * | 2019-11-11 | 2020-01-07 | 广东石油化工学院 | PLC signal reconstruction method and system by utilizing dictionary atoms |
| CN110661549B (en) * | 2019-11-11 | 2021-05-04 | 广东石油化工学院 | PLC signal reconstruction method and system by utilizing dictionary atoms |
| CN111242137A (en) * | 2020-01-13 | 2020-06-05 | 江西理工大学 | Salt and pepper noise filtering method and device based on morphological component analysis |
| CN111242137B (en) * | 2020-01-13 | 2023-05-26 | 江西理工大学 | Spiced salt noise filtering method and device based on morphological component analysis |
| CN113837958A (en) * | 2021-09-09 | 2021-12-24 | 南方医科大学 | Diffusion weighted image denoising algorithm, medium, and apparatus |
| CN113837958B (en) * | 2021-09-09 | 2023-08-04 | 南方医科大学 | Diffusion weighted image denoising algorithm, medium and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108765322B (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108765322A (en) | Image de-noising method and device | |
| Yang et al. | BM3D-Net: A convolutional neural network for transform-domain collaborative filtering | |
| Zhang et al. | Image denoising method based on a deep convolution neural network | |
| CN105335947B (en) | Image de-noising method and image denoising device | |
| Deng et al. | Hessian matrix-based fourth-order anisotropic diffusion filter for image denoising | |
| Minaee et al. | An ADMM approach to masked signal decomposition using subspace representation | |
| Zhang et al. | Removal of impulse noise in color images based on convolutional neural network | |
| Cui et al. | Image compressed sensing using non-local neural network | |
| Bouboulis et al. | Adaptive kernel-based image denoising employing semi-parametric regularization | |
| CN108805840A (en) | Method, apparatus, terminal and the computer readable storage medium of image denoising | |
| CN113362250B (en) | Image denoising method and system based on dual-tree quaternary wavelet and deep learning | |
| Ma et al. | Decomposition approach for low-rank matrix completion and its applications | |
| Li et al. | Improved BM3D denoising method | |
| Ono et al. | Color-line regularization for color artifact removal | |
| Singh et al. | ResDNN: deep residual learning for natural image denoising | |
| Chen et al. | Removing Gaussian noise for colour images by quaternion representation and optimisation of weights in non‐local means filter | |
| CN114444565A (en) | Image tampering detection method, terminal device and storage medium | |
| Wen et al. | Learning flipping and rotation invariant sparsifying transforms | |
| CN110163095B (en) | Loop detection method, loop detection device and terminal equipment | |
| Xie et al. | Color and direction-invariant nonlocal self-similarity prior and its application to color image denoising | |
| Zhang et al. | An image denoising method based on BM4D and GAN in 3D shearlet domain | |
| Zhu et al. | Generative high-capacity image hiding based on residual CNN in wavelet domain | |
| Yeganli et al. | Super-resolution using multiple structured dictionaries based on the gradient operator and bicubic interpolation | |
| Shi et al. | A competent image denoising method based on structural information extraction | |
| CN108596959A (en) | A kind of extracting method of video image space-time characteristic point |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20231214 Address after: 230000 floor 1, building 2, phase I, e-commerce Park, Jinggang Road, Shushan Economic Development Zone, Hefei City, Anhui Province Patentee after: Dragon totem Technology (Hefei) Co.,Ltd. Address before: No. 401, Zhimin Avenue, Xinzhou District, Shangrao City, Jiangxi Province Patentee before: SHANGRAO NORMAL University |
|
| TR01 | Transfer of patent right |