CN106228528B - A kind of multi-focus image fusing method based on decision diagram and rarefaction representation - Google Patents
A kind of multi-focus image fusing method based on decision diagram and rarefaction representation Download PDFInfo
- Publication number
- CN106228528B CN106228528B CN201610605703.6A CN201610605703A CN106228528B CN 106228528 B CN106228528 B CN 106228528B CN 201610605703 A CN201610605703 A CN 201610605703A CN 106228528 B CN106228528 B CN 106228528B
- Authority
- CN
- China
- Prior art keywords
- image
- fusion
- decision
- region
- point
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000010586 diagram Methods 0.000 title claims abstract description 38
- 230000004927 fusion Effects 0.000 claims abstract description 65
- 238000002156 mixing Methods 0.000 claims abstract description 14
- 238000000354 decomposition reaction Methods 0.000 claims description 13
- 238000005070 sampling Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 235000013399 edible fruits Nutrition 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000014509 gene expression Effects 0.000 claims description 3
- 238000012549 training Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 abstract description 10
- 230000008569 process Effects 0.000 abstract description 10
- 230000001976 improved effect Effects 0.000 abstract description 4
- 230000000007 visual effect Effects 0.000 abstract description 4
- 230000000694 effects Effects 0.000 description 13
- 238000005516 engineering process Methods 0.000 description 9
- 238000012545 processing Methods 0.000 description 8
- 238000003384 imaging method Methods 0.000 description 6
- 230000009466 transformation Effects 0.000 description 6
- 239000013598 vector Substances 0.000 description 5
- 238000013459 approach Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 230000006835 compression Effects 0.000 description 3
- 238000007906 compression Methods 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 239000000284 extract Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000000877 morphologic effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000007704 transition 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
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/77—Retouching; Inpainting; Scratch removal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20048—Transform domain processing
- G06T2207/20064—Wavelet transform [DWT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20092—Interactive image processing based on input by user
- G06T2207/20104—Interactive definition of region of interest [ROI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/513—Sparse representations
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of multi-focus image fusing method based on decision diagram and rarefaction representation, this method proposes the multi-focus image fusion frame different from previous Multi-focus image fusion based on the characteristic of human visual system, and it is analyzed and is studied for the transitional region of multiple focussing image, to avoid its influence to fusion results, the quality of blending image is improved.Realization process: on the basis of the low scale image to multiple focussing image carries out clarity analysis, decision diagram is generated, and fusion results are obtained according to the decision diagram；In view of the clarity judgement of transitional region is there are deviation, will lead to the decision diagram of generation, there are errors, it is necessary to determine transitional region and be handled using the Multi-focus image fusion based on rarefaction representation, obtain the fusion results of transitional region；Finally, by fusion results and transitional region fusion results progress mean operation based on decision diagram, final blending image is obtained.
Description
Technical field
The invention belongs to technical field of image processing, are related to compression of images, image sparse expression, image space frequency ratio
Compared with, Mathematical Morphology technology and image fusion technology, can be used for machine vision, target identification, digital camera blur-free imaging
Equal fields.
Background technique
In shooting process, after the focal length setting of imaging system, due to the limitation of the camera lens depth of field, only camera lens conjugate planes
The a certain range of object imaging in front and back be not then fuzzy in the object imaging of this range clearly.It is practical
When in some scene imaging, since subject is different at a distance from camera lens, imaging is frequently not all clear
Clear.To obtain the full clear image of scene, a feasible method is exactly to focus respectively to object each in scene, obtains it
Clear image, and its region is extracted and is fused together.To how accurately to extract the clear of image
Region, i.e., how to obtain the decision diagram for accurately dividing clear area and fuzzy region becomes the pass for measuring fused image quality
Key.In general, the transitional region between clear area and fuzzy region is difficult clearly to be divided, this is the work that clear area is extracted
Difficulty is brought, leading to obtained decision diagram, there are errors, reduce the quality of blending image.In view of human visual system has
There is following characteristic: when quickly distinguishing clear area and fuzzy region, being more likely to directly find clear area and fuzzy region
Boundary and be not primarily upon the detailed information of image.If human eye vision feature quick obtaining decision diagram can be utilized, and root
The transitional region between clear area and fuzzy region is determined according to boundary line contained by the decision diagram, further to the region
It is handled to improve the accuracy rate of clear area extraction, this will make it possible the raising of fusion mass.In view of such it is assumed that
And, as foundation, to be proposed based on decision diagram and sparse table the characteristics of the image-forming principle of multiple focussing image and its low scale image
The multi-focus image fusing method shown.This method compares by compression of images, the rarefaction representation of image, image space frequency, mathematics
Morphology technology and image fusion technology are constituted, and main thought can be divided into five steps and be described；(1) poly is considered
The low scale image of burnt image has the characteristics that mainly to reflect picture structure and do not include excessive image detail information, using more rulers
Degree wavelet transformation decomposes multiple focussing image, and obtaining size is the big half low scale image of former multiple focussing image；(2)
It analyzes the clarity of low scale image each point and each point is divided into focusing, defocus, uncertain three classes, to generate initial decision map,
And up-sampling operation is carried out to initial decision map, make the in the same size of the multiple focussing image of its size and input；(3) space is used
Each point in the uncertain region of initial decision map is further divided into focusing and two class of defocus by frequency approach, obtains final decision
Figure is determined to define the boundary line for focusing domain, defocus domain；(4) influence of the transitional region to decision plot quality is considered, according to most
Whole decision diagram merges the multiple focussing image of input, and detects transitional region according to boundary line obtained, and utilize
The region is merged based on the Multi-focus image fusion of rarefaction representation；(5) fusion results for combining the two, obtain most
Whole blending image.This method is compared to traditional multi-focus image fusion algorithm, and there are following two features: one, traditional poly
Burnt Image Fusion is directly focused feature extraction to multiple focussing image, and our rule is according to image low frequency subband image
Possessed characteristic, the low scale image of multiple focussing image is decomposited using wavelet transformation, and then utilizes complete rarefaction representation
Model analyzes clarity, that is, focus characteristics of the low scale image.In this way, the computation complexity of method is not only reduced, and
And not only guarantee that the focus characteristics of former multiple focussing image are constant, but also reduce the difficulty of multiple focussing image clarity analysis.Two, traditional
Multi-focus image fusion have ignored influence of the transitional region to fusion mass of multiple focussing image, and this method is to poly
The transitional region of burnt image is analyzed and is merged using the multi-focus image fusion based on rarefaction representation to it, is reduced
The region adverse effect, to improve the effect of blending image.Therefore, using this method, blending image can be improved
Effect, also, this method is compared with the multiple focussing image algorithm based on rarefaction representation, has higher timeliness.
Wavelet transformation is function, operator or data to be divided into various frequency contents, then carry out to it multiple dimensioned thin
Change.As a kind of mathematical tool, wavelet transformation has played effect in terms of image co-registration, is with various directions picture breakdown
The subband signal of feature, go forward side by side line frequency domain analysis and time-domain analysis.Under normal circumstances, it is carried out at image co-registration using wavelet algorithm
When reason, image is considered as 2D signal, and multi-level decomposition is carried out to it.Image after wavelet decomposition produce one it is low
Frequency component and three high fdrequency components, they are the expressions to the different content in original image, in order to simpler to obtain
Take the feature in picture.The low frequency subband image of wavelet decomposition still maintains the general picture and spatial character of original image, only size
Become smaller.Therefore, present invention characteristic according to possessed by the low frequency subband image of wavelet decomposition, using wavelet transformation to multi-focus
Source images carry out level of decomposition and will acquire low frequency sub-band as low scale image, further to use complete rarefaction representation side
Method carries out rarefaction representation to it and lays the foundation.
In recent years, excessively complete rarefaction representation effectively indicates that model is widely used in image denoising, image restores as a kind of
Etc. in the tasks such as field of image processings and image recognition.The excessively complete rarefaction representation of signal is to seek in excessively complete dictionary
The linear combination of minimum atom indicates signal.The data information of natural image also has redundancy under normal circumstances, therefore can
To carry out rarefaction representation in redundant dictionary.Cross in complete rarefaction representation problem, the excessively complete dictionary of selection whether appropriate relation
To the sparsity for indicating that signal is expressed.For the image of different structure and feature, the dictionary or base that should take various forms
Function.The construction method for crossing complete dictionary can be divided into two classes: (1) constructing dictionary according to mathematical model；(2) according to sample
To design dictionary.Second category dictionary construction method can learn the dictionary for being suitble to certain a kind of signal characteristic by training sample,
To ensure sparsity that signal indicates.Common method that can be compatible with sparse decomposition algorithm now is that ELAD was proposed in 2006
K-SVD learning algorithm.This method is alternately performed signal in the rarefaction representation of current dictionary and the renewal process of atom, to reach
Learn the purpose of dictionary.In addition to dictionary construction step, quickly and effectively sparse decomposition algorithm is equally to be related to signal for design
Indicate whether optimal committed step.Existing sparse decomposition algorithm is summarized as three classes: greedy tracing algorithm (OMP) is based on model
Algorithm, the iterative shrinkage algorithm of number canonical.Wherein, greedy algorithm be using Greedy idea as core, in each step iterative process,
Selection participates in sparse bayesian learning with the most matched atom of residual error in dictionary.Such algorithm mainly includes matching pursuit algorithm
(and improved orthogonal matching pursuit algorithm on its basis).The present invention is sparse using complete dictionary K-SVD and OMP was based on
The sparse representation model of encryption algorithm carries out rarefaction representation to low scale image, and then analyzes each pixel in low scale image
Clarity obtain clarity score value figure, lay the foundation for the generation of decision diagram.
Spatial frequency reflects the overall active degree of piece image, and the bigger image of spatial frequency is more active, more clear.It is empty
Between frequency got by line frequency and the calculating of column frequency, i.e. the change frequency of horizontal and vertical directions.This concept it is extensive
With, for the researchs such as visual characteristic, the transmission of figural perception and vision system signal, the processing of information provide one it is new
Approach.Therefore, the present invention makees further clarity analysis, the bigger explanation of spatial frequency to uncertain region using spatial frequency
The clarity of the point is higher.
Mathematical morphology (Mathematical Morphology) is built upon the Men Xueke on the basis of set theory, non-
The often geometrical morphological analysis of suitable signal and description.Its basic thought is to carry out " detection " to signal using structural element, is retained
Principal shape deletes irrelevant shape (such as noise, burr).Form and difference operation, i.e. expansion and corrosion are mathematical morphologies
Basis.Mathematical morphology first processing is bianry image, referred to as binary mathematical morphology (Binary Morphology).Two
Value mathematical morphology is a kind for the treatment of process for set.Morphological images processing is mobile structural elements in the picture
Then structural element and following bianry image are carried out the set operations such as intersecting and merging by element.First corrode the process expanded afterwards to be known as
Opening operation.It, which has, eliminates small objects, in the effect of very thin place's separating objects and smooth larger object boundary.The present invention is just
It is using the small area region in opening operation removal bianry image.
As sparse representation theory is widely used in image co-registration processing, the blending algorithm based on rarefaction representation
Research is also by the concern of domestic and foreign scholars.Sparse representation theory is applied in image co-registration by Yang B. etc., and this method improves
The effect of fusion, but training dictionary process is than relatively time-consuming.2013, Chen L. et al. and Yin H. et al. were by rarefaction representation
It is combined with other blending algorithms and is used for image co-registration, improve fused image quality, but the orthogonal matching in rarefaction representation is calculated
Method calculation amount is very big, and the image co-registration of two width, time-consuming.Liu Y. et al. is by by multi-scale geometric analysis method and sparse table
Representation model combines, and proposes a kind of general image co-registration framework, overcome it is simple based on multi-scale geometric analysis method or
Inherent defect present in the fusion method of person's rarefaction representation.They also by sparse representation model simultaneously be applied to image co-registration with
Image denoising.Relative to traditional multi-scale geometric analysis method, the Image Fusion based on rarefaction representation is merged
The effect of image is preferably and the strong robustness of algorithm.Therefore, the present invention is calculated using the multi-focus image fusion based on rarefaction representation
Method merges transitional region to improve the quality of final fusion results.
Summary of the invention
It is an object of the present invention to propose that a kind of new multi-focus image fusion frame is based on decision diagram and rarefaction representation
Multi-focus image fusing method, the fusion of several multiple focussing images not only may be implemented, existing multi-focus figure can also be improved
As the low disadvantage of blending algorithm applicability, and improve the quality of multi-focus image fusion image.
The technical scheme is that inputting the two width polies about a certain scene obtained from a certain digital camera first
Burnt image, decomposing input picture to obtain size by multi-scale wavelet transformation is the big half low scale image of input picture,
Then complete sparse representation model was used to analyze it with intelligibility measure method image block based, by low scalogram
As each pixel is divided into three classes: focusing, is not known defocus, is generated initial decision map and is simultaneously carried out up-sampling treatment to it, so that
The size of initial decision map and input picture it is in the same size.In order to obtain the only final decision figure comprising focusing domain, defocus domain,
Using spatial frequency method the uncertain region in initial decision map is handled, determines to divide and focus domain, defocus domain
Line of demarcation.According to final decision figure, extracts the clear area of each input picture and they combine acquisition blending image.So
And due to being difficult that the pixel of transitional region is accurately divided into two classes: focusing, defocus, lead to the final decision figure generated
There are errors.Therefore, it is necessary to determine transitional region and analyze it and handle, to reduce shadow brought by error
It rings, improves fusion mass.For the present invention centered on detected boundary line, the rectangular area of appropriate radius is considered as transition region
Domain carries out the fusion results that fusion obtains the region using the Multi-focus image fusion based on rarefaction representation to it.Finally,
In conjunction with decision diagram blending image obtained is based on, i.e., knot is merged with the decision diagram of corresponding region to the fusion results of transitional region
Fruit carries out mean operation, generates final blending image.
Specific step is as follows:
One, to be fused multiple focussing image of two width of input about a certain scene, to the multiple focussing image of this two width input
A wavelet decomposition is carried out, is obtained,,,, four width sub-band images, wherein willLow frequency subband image is as this hair
Bright low scale image.
Multiple focussing image to be fused can be gray level image and be also possible to color image but must be about Same Scene
Multiple focussing image；Wherein, by a wavelet decomposition, the size of low scale image obtained is that input multiple focussing image is big
Small half.
Two, rarefaction representation is carried out to low scale image using excessively complete sparse representation model, that is, extracts sparse spy
Sign constitutes the corresponding degree of rarefication figure of two width.
Three, compare and measure the clarity of each point in degree of rarefication figure by intelligibility measure method image block based, obtain
Obtain the corresponding clarity score value figure of two width.
Four, according to above-mentioned clarity score value figure obtained, binarization operation is carried out to it and obtains corresponding bianry image.
Wherein, disconnected small area region can occur due to erroneous judgement in bianry image, need to open fortune by Mathematical Morphology technology
It calculates and eliminates the region.Then, according to the decision rule of setting to two width treated bianry image makes decisions obtain a width at the beginning of
Beginning decision diagram.Since the size of initial decision map is only the half of input picture, up-sampling treatment is carried out to it, keeps it big
It is small in the same size with input picture.
Wherein, binary conversion treatment can be described as: threshold value appropriate is arranged, the point that definition values are greater than threshold value is assigned a value of 1；
The point for being unsatisfactory for above-mentioned condition will be assigned 0.
Three processing, initial decision map contain through the above steps: focus domain, defocus domain, uncertain region these three
Each point in the multiple focussing image of input is divided into above-mentioned three classes by region, in brief, decision diagram.
Five, the uncertain region in each multiple focussing image of input is divided into muti-piece sub-block using smooth window setting technique, counted
Each point in uncertain region is further divided into two classes according to comparing result: focused by the spatial frequency for calculating and comparing relevant block
With defocus, to obtain final decision diagram.
Wherein, decision diagram becomes only comprising two regions: focusing domain, defocus domain, while by the boundary of this two region division
Line is also determined therewith.
Uncertain region not necessarily belongs to the rectangular area of rule, therefore in order to be divided into son of the same size
Block, positioned at the pixel of edges of regions, corresponding to the content of sub-block not come under uncertain region.
Six, it is based on final decision figure, input multiple focussing image is extracted in the clear area of its divided regional location
Come, then combines extracted region according to the fusion rule of setting, obtain the fusion results based on decision diagram.
Seven, according to the boundary line in the obtained final decision figure of above-mentioned steps four, transitional region is determined.Using based on dilute
The Multi-focus image fusion that dredging indicates merges transitional region, obtains the fusion results in the region.
The point centered on the boundary line in final decision figure, it is radius that parameter appropriate, which is arranged, is formed around central point
Rectangular area be transitional region.
Eight, it finally, the fusion results of fusion results and transitional region based on decision diagram are carried out mean operation, obtains most
Whole blending image.
The mode of mean operation: the pixel value for taking mean value final as the point corresponding position pixel.
In order to improve the effect of multi-focus image fusion, the characteristic according to possessed by human visual system, the present invention is proposed
A kind of new multi-focus image fusion frame is the multi-focus image fusing method based on decision diagram and rarefaction representation, this method
By compression of images, image sparse indicates that image space frequency comparison, Mathematical Morphology technology rise in conjunction with image fusion technology
Come, analyze the characteristic of the low scale image of multiple focussing image, generate decision diagram, fusion results are obtained according to decision diagram；And consider
The influence to the syncretizing effect based on decision diagram is judged by accident to transitional region, and transitional region and progress are further determined according to decision diagram
Processing reduces transitional region erroneous judgement bring error, improves the quality of fusion picture.
Detailed description of the invention
The present invention is based on the multi-focus image fusing method flow charts of decision diagram and rarefaction representation by Fig. 1.
Fig. 2 final decision figure product process figure of the present invention.
The operational flowchart that Fig. 3 degree of rarefication figure of the present invention and clarity score value figure generate.
Two multiple focussing images to be fused of Fig. 4, 。
Fig. 5 present invention obtains two low scale images after carrying out wavelet decomposition to two images to be fused, 's
Effect picture.
Fig. 6 present invention used complete sparse representation model and block-based intelligibility measure method respectively to low scalogram
Picture, It carries out clarity analysis and obtains its corresponding clarity score value figure, Effect picture.
Fig. 7 present invention is respectively to clarity score value figure, It carries out binary conversion treatment and obtains its corresponding bianry image, Effect picture.
Fig. 8 present invention is using mathematical morphology open operator respectively to bianry image, It is handled after being handled
Bianry image afterwards, Effect picture.
Fig. 9 present invention is respectively to two width of place treated bianry image, Make decisions acquisition initial decision map's
Effect picture.
Figure 10 present invention carries out uncertain region contained by initial decision map using block-based spatial frequency measurement method
Classification obtains final decision figureEffect picture.
Figure 11 present invention is according to final decision figureFusion is carried out to input picture and obtains fusion resultsEffect picture.
Figure 12 present invention is according to final decision figureDetermine transitional region in contained boundary lineEffect picture.
Figure 13 present invention is using the multi-focus image fusing method based on rarefaction representation to transitional regionCarry out fusion acquisition
Fusion resultsEffect picture.
Figure 14 present invention is by fusion resultsWith fusion resultsIt carries out mean operation and obtains final fusion resultsEffect
Fruit figure.
Specific embodiment
A specific embodiment of the invention is described in further detail below.
As shown in Figure 1, a kind of multi-focus image fusing method flow chart based on decision diagram and rarefaction representation of the present invention, first
First input two multiple focussing images to be fused, , this two images is once decomposed using daub1 small echo, by institute
The low frequency subband image of acquisitionLow scale image as corresponding every width multiple focussing image,。
It is the process that final decision figure generates in next step.As shown in Fig. 2 final decision figure product process figure of the present invention:
Input parameter: low scale image, ；
Export result: final decision figure；
(1) it used complete sparse representation model to carry out rarefaction representation to two low scale images respectively first and obtains it accordingly
Degree of rarefication figure.Concrete mode is as follows: if low scale image is,It is big for it
Small, the size of excessively complete dictionary K-SVD is, using smooth window setting technique, from the low scale image upper left corner to the lower right corner, with one
Pixel is step-length, is successively divided into low scale imageThe image block of size,Indicate the quantity of image block.Often
A image block all corresponds to a column vector, and these column vectors are combined sequentially into matrix.Utilize OMP sparse coding algorithm
According to formulaCalculate the corresponding sparse coefficient of each column vector,
Wherein,It was the atom in complete dictionary K-SVD and had；And according to ruleTo each sparse vectorIt is handled.Then, according to it is each it is sparse to
AmountIn the position of low scale image, each sparse vector is polymerize and is reconstructed equal-sized sparse with low scale image
Degree figure.Concrete operations show as shown in Figure 3.
(2) secondly, using intelligibility measure method image block based to degree of rarefication figureIntelligibility measure is carried out to obtain
Obtain corresponding clarity score value figure.Concrete operations mode are as follows: smooth window setting technique is used, from degree of rarefication figureThe upper left corner to the right side
The figure is successively divided by inferior horn using a pixel as step-lengthThe image subblock of size,Indicate image
The quantity of block.The degree of rarefication of pixel contained in every piece of image block is summed, the clarity of every piece of image block is obtained.Then,
The equal-sized score value figure of three width sizes Yu degree of rarefication figure is set, wherein the value of score value figure each point is all 0.According to
The comparison rule of setting, compares the clarity of corresponding image block in two width degree of rarefication figures, and comparison result is recorded in score value figure
In, whereinIt is mainly responsible for and records the number that each pixel compares.Comparison rule are as follows: for every two pieces of corresponding degree of rarefications
The sub-block of figure, it is assumed that the sub-block of one group of degree of rarefication figure, If sub-blockClarity be greater than sub-block, then by score value figure
Each point value of the position of corresponding sub block increases by 1；Otherwise in score value figureEach point value of the position of corresponding sub block increases by 1；Meanwhile often
One group of sparse sub-block comparison terminates, by score value figureEach point value of the position of corresponding sub block increases by 1.Successively compare, until completing institute
There is the clarity of sub-block to compare.Finally, respectively by score value figure,The value and score value figure of middle each pointThe value of middle each point is removed
Method operation obtains required degree of rarefication score value figure.Specific operation process is as shown in Figure 3.
(3) threshold value appropriate is set, by each score value figureIn be greater than the threshold value point value be assigned a value of 1, be otherwise 0,
Thus to obtain bianry image.Small area region discrete in binary map is eliminated using the opening operation of Mathematical Morphology technology, and
And according to treated bianry imageAnd judgement formula, generate
Initial decision diagram.Judgement formula may be interpreted as: if a certain pixelIn score value figureIn value be 1 and
Score value figureIn value be 0, then the point is referred to as focus point, and be assigned a value of 1；If point is in score value figureIn value be 0
And in score value figureIn value be 1, then the point is referred to as focal point, and be assigned a value of 0；If the point is all unsatisfactory for above-mentioned item
Part, then referred to as uncertainty node, and it is assigned a value of 0.5.
(4) finally, further being divided to the point of uncertain region using spatial frequency comparative approach.Using smoothing windows
The region is successively divided by technology from uncertain region upper left corner the to the lower right cornerThe image block of size, according to
FormulaCalculate every piece of sub-block of uncertain region in each input multiple focussing image
Spatial frequency, whereinIt indicates with certain pointCentered on the sub-block region put；Respectively indicate a horizontal component with
The difference of vertical component.According to judgement formula, point corresponding to uncertain region
It makes decisions, generates final decision figure.Decision rule may be interpreted as: if pointCorresponding sub-block is in multiple focussing imageIn spatial frequency values be greater than its in multiple focussing imageThe value of middle corresponding sub block, which just belongs to focus point, and is assigned a value of
1, it is otherwise focal point, and be assigned a value of 0.
It is based on final decision figure in next step again, according to fusion rule
The multiple focussing image of input is merged.Fusion rule may be interpreted as: extracting each region in decision diagram and is inputting multi-focus
Clear area in image, and these clear areas are constituted into fusion results according to corresponding position grouping。
It is according to final decision figure in next step since there are errors for the judgement to decision diagramThe boundary line for being included
Determine transitional region.The point centered on each pixel of boundary line is arranged value appropriate as above and below, the radius of left and right, draws
Surely around the matrix area of central point as transitional region.Using the Multi-focus image fusion based on rarefaction representation to the area
Domain is merged, and the fusion results of transitional region are obtained。
Finally, being by fusion results in next stepWith transitional region fusion resultsThe point of corresponding region carries out mean value fortune
It calculates, obtains final fusion results。
The above content is the present invention is based on the detailed step of decision diagram and the multi-focus image fusing method of rarefaction representation and
Implementation method.Any change made under the premise of not departing from design of the invention for those skilled in the art belongs to this
Within the protection scope of invention.
Claims (8)
1. the multi-focus image fusing method based on decision diagram and rarefaction representation, method includes the following steps:
Step 1: inputting two width multiple focussing images to be fused first, a wavelet decomposition is carried out to image respectively, by respective institute
Low scale image of the low-frequency image of acquisition as multiple focussing image；
Step 2: secondly, carrying out sparse table using excessively complete sparse representation model low scale image obtained to above-mentioned steps
Show, i.e. extraction sparse features, constitutes corresponding degree of rarefication figure；
Step 3: using smooth window setting technique and intelligibility measure method image block based to above-mentioned degree of rarefication figure obtained
It carries out intelligibility measure and is compared, obtain corresponding clarity score value figure；
Step 4: binaryzation, mathematical morphology open operator operation being carried out to above-mentioned clarity score value figure obtained, handled
Binary map afterwards；According to treated binary map, image each point is divided into three classes according to set decision rule: focus, from
It is burnt and uncertain, and generate initial decision map；Initial decision map is subjected to up-sampling treatment, up to its size and is inputted more
Focusedimage is in the same size；
Step 5: smooth window setting technique is used, using a pixel as step-length, from the upper left corner in region to the lower right corner successively by every width
The uncertain region of multiple focussing image is divided into several sub-blocks；The spatial frequency of sub-block corresponding to each pixel is calculated, and is compared
Each pixel is determined as focus point or focal point according to comparison result by the result of every group of sub-block of respective pixel point；Finally
Obtain the only final decision figure comprising focusing domain and defocus domain；
Step 6: according to the final decision figure of above-mentioned acquisition, obtaining the fusion knot based on decision diagram according to the fusion rule of setting
Fruit；
Step 7: transitional region being determined according to the boundary line in above-mentioned final decision figure obtained, and using based on sparse table
The multi-focus image fusion algorithm shown carries out the fusion results that fusion obtains transitional region to it；
Step 8: to the fusion results and the progress mean operation acquisition of transitional region fusion results based on decision diagram of above-mentioned acquisition
Final blending image.
2. according to the method described in claim 1, wherein in step 1 respectively to the two width multiple focussing images to be fused of input into
Wavelet decomposition of row, using the low-frequency image respectively obtained as its low scale image, wherein the size of low scale image is defeated
Enter the half of image size.
According to the method for claim 1,3. wherein being calculated in step 2 using based on training dictionary K-SVD and OMP sparse coding
The excessively complete sparse representation model of method carries out degree of rarefication expression to low scale image, and constitutes the corresponding degree of rarefication figure of two width.
According to the method for claim 1,4. wherein every low scale image is drawn first with smooth window setting technique in step 3
It is divided into the image subblock that several sizes are a certain value；Secondly, every group of phase is measured using intelligibility measure method image block based
It answers the clarity of sub-block and is compared to each other；Finally according to every group of comparison result, assignment is carried out to each point value in sub-block, is obtained
Corresponding two width clarity score value figure.
According to the method for claim 1,5. wherein the binary conversion treatment in step 4 is explained are as follows: threshold value appropriate is set, it will
Point in above-mentioned clarity score value figure obtained greater than the threshold value is assigned a value of 1, is otherwise 0, thus to obtain corresponding binary map
Picture；Then further judgement is made to bianry image according to set decision rule, generates initial decision map.
According to the method for claim 1,6. the position for the uncertain region for wherein being included according to initial decision map in step 5
It sets, detects corresponding uncertain region of the uncertain region in two width multiple focussing images of input, it will be in initial decision map
Uncertain region in each point be further divided into focusing and two class of defocus, generate final decision figure.
According to the method for claim 1,7. wherein being extracted most in step 6 according to above-mentioned final decision figure obtained
The clear area of each region in the input image in whole decision diagram, and these clear areas are constituted according to corresponding position grouping
Fusion results.
According to the method for claim 1,8. value appropriate is wherein arranged in step 7 for above and below, the radius of left and right, finally to determine
Point centered on each point for the boundary line that plan figure is included, delimiting out rectangular area is transitional region；And using based on rarefaction representation
Multi-focus method to transitional region carry out fusion obtain transitional region fusion results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610605703.6A CN106228528B (en) | 2016-07-29 | 2016-07-29 | A kind of multi-focus image fusing method based on decision diagram and rarefaction representation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610605703.6A CN106228528B (en) | 2016-07-29 | 2016-07-29 | A kind of multi-focus image fusing method based on decision diagram and rarefaction representation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106228528A CN106228528A (en) | 2016-12-14 |
| CN106228528B true CN106228528B (en) | 2019-09-27 |
Family
ID=57534529
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610605703.6A Expired - Fee Related CN106228528B (en) | 2016-07-29 | 2016-07-29 | A kind of multi-focus image fusing method based on decision diagram and rarefaction representation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106228528B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107680072A (en) * | 2017-11-01 | 2018-02-09 | 淮海工学院 | It is a kind of based on the positron emission fault image of depth rarefaction representation and the fusion method of MRI |
| CN108364273B (en) * | 2018-01-30 | 2022-02-25 | 中南大学 | Method for multi-focus image fusion in spatial domain |
| CN108510465B (en) * | 2018-01-30 | 2019-12-24 | 西安电子科技大学 | Multi-focus image fusion method based on consistency constraint non-negative sparse representation |
| CN109785282B (en) * | 2019-01-22 | 2021-03-26 | 厦门大学 | Multi-focus image fusion method |
| CN112508828A (en) * | 2019-09-16 | 2021-03-16 | 四川大学 | Multi-focus image fusion method based on sparse representation and guided filtering |
| CN111127375B (en) * | 2019-12-03 | 2023-04-07 | 重庆邮电大学 | Multi-focus image fusion method combining DSIFT and self-adaptive image blocking |
| CN111127380B (en) * | 2019-12-26 | 2023-05-23 | 云南大学 | Multi-focus image fusion method based on intuitionistic fuzzy similarity measurement technology |
| CN111582283A (en) * | 2020-04-27 | 2020-08-25 | 安创生态科技(深圳)有限公司 | Double-view image fusion method and device |
| CN112184646B (en) * | 2020-09-22 | 2022-07-29 | 西北工业大学 | Image fusion method based on gradient domain oriented filtering and improved PCNN |
| CN112233049B (en) * | 2020-12-14 | 2021-03-02 | 成都中轨轨道设备有限公司 | Image fusion method for improving image definition |
| CN115393243B (en) * | 2022-10-28 | 2023-02-24 | 北京开运联合信息技术集团股份有限公司 | Multi-focus image fusion method and device and electronic equipment |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678723A (en) * | 2015-12-29 | 2016-06-15 | 内蒙古科技大学 | Multi-focus image fusion method based on sparse decomposition and differential image |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102053051A (en) * | 2009-10-30 | 2011-05-11 | 西门子公司 | Body fluid analysis system as well as image processing device and method for body fluid analysis |
-
2016
- 2016-07-29 CN CN201610605703.6A patent/CN106228528B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678723A (en) * | 2015-12-29 | 2016-06-15 | 内蒙古科技大学 | Multi-focus image fusion method based on sparse decomposition and differential image |
Non-Patent Citations (1)
| Title |
|---|
| 多聚焦图像像素级融合方法研究;李华锋;《中国博士学位论文全文数据库》;20130515(第05期);第19-102页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106228528A (en) | 2016-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106228528B (en) | A kind of multi-focus image fusing method based on decision diagram and rarefaction representation | |
| CN106339998B (en) | Multi-focus image fusing method based on contrast pyramid transformation | |
| CN109377448B (en) | Face image restoration method based on generation countermeasure network | |
| CN105956582B (en) | A kind of face identification system based on three-dimensional data | |
| CN107480772A (en) | A kind of car plate super-resolution processing method and system based on deep learning | |
| CN103455991B (en) | A kind of multi-focus image fusing method | |
| CN112733950A (en) | Power equipment fault diagnosis method based on combination of image fusion and target detection | |
| CN106296638A (en) | Significance information acquisition device and significance information acquisition method | |
| CN105354558B (en) | Humanface image matching method | |
| CN110110646A (en) | A kind of images of gestures extraction method of key frame based on deep learning | |
| CN107909560A (en) | A kind of multi-focus image fusing method and system based on SiR | |
| CN105917353A (en) | Feature extraction and matching and template update for biometric authentication | |
| CN108573222A (en) | The pedestrian image occlusion detection method for generating network is fought based on cycle | |
| CN108550166B (en) | Spatial target image matching method | |
| CN106447640B (en) | Multi-focus image fusing method and device based on dictionary learning, rotation guiding filtering | |
| CN110348263A (en) | A kind of two-dimensional random code image recognition and extracting method based on image recognition | |
| CN107944437B (en) | A kind of Face detection method based on neural network and integral image | |
| CN114758288A (en) | Power distribution network engineering safety control detection method and device | |
| CN105869166A (en) | Human body action identification method and system based on binocular vision | |
| CN113312973B (en) | Gesture recognition key point feature extraction method and system | |
| CN111507426A (en) | No-reference image quality grading evaluation method and device based on visual fusion characteristics | |
| Wang et al. | Multi-focus image fusion based on quad-tree decomposition and edge-weighted focus measure | |
| Zhu et al. | Towards automatic wild animal detection in low quality camera-trap images using two-channeled perceiving residual pyramid networks | |
| CN116128820A (en) | Pin state identification method based on improved YOLO model | |
| Luo et al. | Bi-GANs-ST for perceptual image super-resolution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190927 Termination date: 20200729 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |