CN110223264A - Image difference characteristic attribute fusion availability distributed structure and synthetic method based on intuition possibility collection - Google Patents
Image difference characteristic attribute fusion availability distributed structure and synthetic method based on intuition possibility collection Download PDFInfo
- Publication number
- CN110223264A CN110223264A CN201910343217.5A CN201910343217A CN110223264A CN 110223264 A CN110223264 A CN 110223264A CN 201910343217 A CN201910343217 A CN 201910343217A CN 110223264 A CN110223264 A CN 110223264A
- Authority
- CN
- China
- Prior art keywords
- difference characteristic
- amplitude
- algorithm
- image
- difference
- 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
- 230000004927 fusion Effects 0.000 title claims abstract description 67
- 238000010189 synthetic method Methods 0.000 title claims abstract description 9
- 238000002156 mixing Methods 0.000 claims abstract description 69
- 230000000694 effects Effects 0.000 claims abstract description 36
- 238000000034 method Methods 0.000 claims abstract description 22
- 238000009826 distribution Methods 0.000 claims abstract description 16
- 230000010287 polarization Effects 0.000 claims description 17
- 230000002902 bimodal effect Effects 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000003384 imaging method Methods 0.000 claims description 5
- 230000015572 biosynthetic process Effects 0.000 claims description 4
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000011160 research Methods 0.000 claims description 3
- 238000003786 synthesis reaction Methods 0.000 claims description 3
- 235000013399 edible fruits Nutrition 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 abstract description 7
- 238000001514 detection method Methods 0.000 description 7
- 230000008859 change Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 230000009466 transformation Effects 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 206010049155 Visual brightness Diseases 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000155 melt Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000004451 qualitative analysis Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000010008 shearing Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- 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
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The present invention proposes difference characteristic attribute fusion availability distributed structure and synthetic method based on intuition possibility collection.First, determine the type of difference characteristic, calculating blending image is between source images at a distance from difference characteristic amplitude, according to algorithm to the syncretizing effect of each difference characteristic amplitude, its distance is divided into, is bad, the non-intuition possibility collection for getting well 3 sections of non-evil idea and merge availability to each difference characteristic as algorithm, proposes the fusion availability distributed structure method based on intuition possibility collection；Secondly, for the multiple amplitudes section of each difference characteristic, it is proposed a kind of distribution synthetic method based on the sequence of intuition possibility collection, calculate the comprehensive evaluation value that fusion availability scoring event and algorithm of each algorithm on difference characteristic difference amplitude section merge availability to each difference characteristic amplitude, as according to for different difference characteristics, the preferable blending algorithm of syncretizing effect is chosen.
Description
Technical field
The invention belongs to image co-registration fields, and the image difference characteristic attribute fusion specially based on intuition possibility collection has
Validity distributed structure and synthetic method.
Background technique
Infrared polarization respectively detects objective attribute target attribute by the polarization of infrared ray and strength information with light intensity imaging, is
Bimodal detection system in the fields such as the monitoring of high-performance aviation integral, the driving of unmanned aircraft remote sensing observations, intelligent automobile
Important component, both modalities which image co-registration are conducive to effective storage, characteristic synthetic and the understanding of detection image, significantly improve
The image quality of detection system, detection accuracy.However, the diversity of the otherness of imaging mechanism, detection environment and target causes
Difference characteristic between bimodal image with scene is complicated and changeable, and (fusion is effective for fusion performance of each blending algorithm to difference characteristic
Degree) it is different, it is difficult to meet different demands using state algorithm；Only melted accordingly according to different difference characteristic driving selections
Hop algorithm is just able to achieve blending algorithm adaptive change with the variation of image difference feature.Currently, not according to difference characteristic
It is the key technology and research for improving bimodal infrared image fusion specific aim and validity with attribute dynamic adjustment blending algorithm
Hot spot.
In recent years, bimodal infrared image Fusion Model by the types of some known image difference characteristics of qualitative analysis with
Connection between multiple blending algorithms determines the preferable algorithm of syncretizing effect.Such as: by establishing luminance difference feature and being based on top cap
The connection between (Top-Hat) and support transformation (SVT) blending algorithm is converted, the contrast of blending image is improved；There is document to grind
Study carefully the infrared polarization based on weighted average, pyramid, wavelet transformation, wavelet packet, NSCT and NSST and merges calculation with intensity image
The performance of method establishes the corresponding relationship of itself and image difference feature, achieves preferable syncretizing effect；Establish statistical discrepancy spy
Contacting between linking the blending algorithm in conjunction with PCNN and NSCT with adaptive binary channels unit is levied, there is blending image abundant
Detailed information.These algorithms usually have special scenes good syncretizing effect.
However, in true target acquisition, between two class images difference characteristic other than this attribute of type, amplitude this
Attribute and its variation also have having a significant impact to fusion results；And type, amplitude of difference characteristic etc. belong in actual detection image
Property all changes at random, especially dynamic instrumentation scene, and each attribute change of difference characteristic is increasingly complex between image, existing method
Algorithm is only only accounted for difference characteristic single attribute --- the syncretizing effect of type can not reflect difference characteristic different attribute
The influence of (such as type, amplitude) to algorithm picks, and can not quantitative description algorithm to difference characteristic attribute merge availability
Variable condition, cause syncretizing effect difference even fail.Therefore, specific aim is only had according to the selection of difference characteristic different attribute
Algorithm, the fusion mass of bimodal infrared image could be improved.
In infrared polarization is merged with intensity image, pre-selected blending algorithm can not to difference characteristic difference amplitude
Always be able to maintain preferable syncretizing effect, thus blending algorithm to the syncretizing effect and on-fixed of difference characteristic difference amplitude not
Become, but the dynamic change with the difference of difference characteristic amplitude, there is uncertainty.For actual detection image, for retouching
The fusion availability for stating the syncretizing effect superiority and inferiority degree of difference characteristic difference value is mostly according to existing, limited and close scene
The fusion results of image are predicted and are estimated that having to the measurement of fusion validity may be predictive.It can therefore, it is necessary to use
Fusion availability change procedure of the energy property distribution description algorithm to difference characteristic attribute.However, it is possible to which though property distribution can reflect this
The dynamic change of kind fusion availability, but " being this or that " property can only be described, algorithm can not be described, difference characteristic attribute is melted
In non-good non-bad intermediate state in conjunction availability, and the fusion availability in intermediate state has very greatly algorithms selection
It influences.Therefore, present applicant proposes the difference characteristic attributes based on intuition possibility collection to merge availability building method；For more
A difference characteristic, propose it is a kind of based on intuition possibility collection sequence distribution synthetic method, calculate each algorithm difference characteristic not
With on amplitude section fusion availability scoring event and algorithm to each difference characteristic amplitude fusion availability overall merit
Value chooses the preferable blending algorithm of syncretizing effect as according to for different difference characteristics.
Summary of the invention
The present invention is in view of the above-mentioned problems, provide the image difference characteristic attribute fusion availability based on intuition possibility collection
Distributed structure and synthetic method.
The present invention adopts the following technical scheme that realization: the image difference characteristic attribute based on intuition possibility collection melts
Close availability distributed structure and synthetic method, comprising the following steps:
S1: source images determine: choosing infrared light intensity and polarization scene figure as research object；
S2: it the determination of difference characteristic type: compares the feature of image of scene figure in S1 and combines the infrared light intensity of bimodal
With the imaging mechanism of polarization image, the difference characteristic type of bimodal infrared image is determined, choose gray average (Mean, M), mark
Quasi- poor (Standard Deviation, SD), edge strength (EdgeIntensity, EI) and spatial frequency
(SpatialFrequency, SF) is used as difference characteristic T；
S3: it establishes difference characteristic intuition possibility and integrates: the different values for setting domain X as difference characteristic amplitude, by algorithm pair
The syncretizing effect of difference characteristic attribute is used as well 1 possibility collectionThenIt indicates when difference feature amplitude value is x,
A possibility that algorithm is high to amplitude x fusion availability,Indicate that, when amplitude value is x, algorithm merges amplitude x effective
Spend high non-possibility, i.e. algorithm merges a possibility that availability is low to amplitude x,Reflecting algorithm has amplitude x fusion
A possibility that validity Fei Gaofei is low；
S4: blending image between source images at a distance from difference characteristic amplitude calculate: choose a variety of blending algorithms and with selection
Blending algorithm the source images in S1 are merged, by after the source images and various algorithm fusions in S1 blending image carry out
Piecemeal processing, obtains n image block, utilizes difference characteristic between blending image and source images in each image block of distance measure calculating
Distance between amplitude,Wherein, DXFor blending image and all figures of source images
As the n-dimensional vector of distance between block difference characteristic amplitude,Indicate the equal of i-th image block difference characteristic amplitude in blending image
Value,WithThe mean value of i-th of image block difference characteristic amplitude in infrared polarization image and infrared intensity image is respectively indicated,
X is this 4 class difference characteristic of M, SD, EI and SF, and D is all obtained by calculation in the blending image after each algorithm fusionX；
S5: fusion availability distributed structure: (1) according to algorithm to the syncretizing effect of difference characteristic amplitude, by blending image
It is divided into 3 sections at a distance between source images difference characteristic amplitude, respectively
With
(2) difference characteristic amplitude is divided into K section, counts each amplitude section X respectivelykPacket in (k=1,2 ..., K)
Total number containing image blockAnd 3 image block numbers for being included apart from section in each amplitude sectionWithIn conjunction with the calculating of distance between blending image and source images difference characteristic amplitude, propose that the fusion based on intuition possibility collection has
Validity distributed structure method, formula are as follows:
And have
X is this 4 class difference characteristic of M, SD, EI and SF；
(3) each blending algorithm is calculated to the fusion availability of difference characteristic amplitude, to obtain intuition using above-mentioned formula
The fusion availability of possibility collection is distributed；
S6: intuition may collect ranking method: intuition possibility collectionScore value and exact value be respectively:Wherein,For a certain algorithm
To fusion availability of the difference characteristic on different amplitude sections, if score value is bigger, illustrate algorithm to difference characteristic in the width
The sequence being worth on section is higher, and syncretizing effect is better；And in the identical situation of score value, exact value is bigger, then algorithm is to difference
Sequence of the different feature on respective magnitudes section is also higher, and the synthesis knot of availability distribution is merged according to multiple difference characteristic attributes
Fruit obtains the algorithm good to difference characteristic syncretizing effect, and then can targetedly be selected according to the difference of difference characteristic attribute
Select the blending algorithm best to this feature syncretizing effect.
It is distributed using the difference characteristic attribute fusion availability based on intuition possibility collection of the application construction, and uses base
Multiple distributions are synthesized in intuition possibility collection sort method, it can be according to difference characteristic attribute (type, amplitude) no
Together, the blending algorithm best to this feature syncretizing effect is targetedly selected.
Detailed description of the invention
Fig. 1 is the application method flow diagram.
Fig. 2 is 5 groups of infrared polarizations and light intensity experimental image.
Fig. 3 be different difference characteristic amplitudes under source images between blending image feature at a distance from scatter plot.
Fig. 4 is fusion availability distribution map of the NSST algorithm to each difference characteristic amplitude.
Fig. 5 is that algorithms of different merges availability distribution to the intuition possibility collection of difference characteristic mean value amplitude.
Fig. 6 is fusion availability comprehensive evaluation value schematic diagram of the algorithm to each difference characteristic amplitude.
Fig. 7 is three groups of authentication images.
Fig. 8 is the blending image of first group of authentication image in Fig. 7.
Fig. 9 is the blending image of second group of authentication image in Fig. 7.
Figure 10 is the blending image of third group authentication image in Fig. 7.
Specific embodiment
The determination of difference characteristic type
The main reason for difference characteristic formation is that the two has differences on imaging characteristic between infrared polarization and intensity image,
Radiation characteristic difference, characteristics of atmospheric transmission difference and imager response characteristic difference including target and background etc..The application will
5 groups of Fig. 2 by rigid registrations, the significant infrared polarization of vision difference and intensity image as experimental image (having a size of 256 ×
256), it can be seen that the features such as edge, details are more apparent in infrared polarization image but brightness is darker；And infrared intensity image tool
There is significant brightness, but edge, minutia are unobvious, therefore two class images are in the features such as brightness, edge and details
Difference it is obvious.Since luminance information, standard deviation and edge strength that the gray average of image reflects image can reflect image
Marginal information, spatial frequency the detailed information of image can be described, therefore the application determines that the type of difference characteristic T is that gray scale is equal
It is worth (Mean, M), standard deviation (Standard Deviation, SD), edge strength (Edge Intensity, EI) and space frequency
Rate (Spatial Frequency, SF).
Difference characteristic amplitude based on intuition possibility collection merges availability distributed structure
Intuition possibility collection
If X is domain, if there are 2 mappings on XX → [0,1] andX → [0,1], so that
With
And meet condition
Then claimWith1 intuition possibility collection on domain X has been determinedIt can be abbreviated as
Claim respectivelyWithForA possibility that distribution and non-possibility distrabtion,WithFor element x about
Possibility degree and non-possibility degree.It is x pairsHesitation degree.Element xj(j=1,2 ..., n) it is rightPossibility degreeWith non-possibility degreeIt is abbreviated as respectivelyWith For xjBelong to1 possibility
Degree and non-possibility degree ordered pair.
In infrared polarization is merged with intensity image, domain X is the different values of difference characteristic amplitude, by algorithm to difference
The syncretizing effect of characteristic attribute is used as well 1 possibility collectionSoIt indicates to calculate when difference feature amplitude value is x
A possibility that method is high to amplitude x fusion availability,Indicate that algorithm merges availability to amplitude x when amplitude value is x
A possibility that high non-possibility, i.e. algorithm are low to amplitude x fusion availability,Algorithm is reflected to merge effectively amplitude x
A possibility that Du Feigaofei is low.
Blending image calculates at a distance from difference characteristic amplitude between source images
The application chooses 12 kinds from the blending algorithms class such as multi-scale transform, rarefaction representation, variation, morphology and pixel domain
Typical blending algorithm merges 5 groups of experimental images in Fig. 2.These algorithms be respectively as follows: wavelet transform (DWT),
Quaternion wavelet converts (QWT), dual-tree complex wavelet transform (DTCWT), wavelet package transforms (WPT), non-lower sampling shearing wave conversion
(NSST), Laplacian Pyramid Transform (LP), grad pyramid transformation (GP), KSVD, weighted average fusion (WA), principal component
Analyze (PCA), top cap converts (TH), guiding filtering (GF).
By in Fig. 2 experimental image and blending image by 16 × 16 window carry out piecemeal processing, obtain n image block,
Using distance measure calculate in each image block blending image between source images between difference characteristic amplitude at a distance from, see formula (5) institute
Show.
Wherein, DXThe n-dimensional vector of distance between blending image and all image block difference characteristic amplitudes of source images,Table
Show the mean value of i-th of image block difference characteristic amplitude in blending image,WithRespectively indicate infrared polarization image and infrared light
The mean value of i-th of image block difference characteristic amplitude in strong image, X can be this 4 class difference characteristic of M, SD, EI and SF.
Using each amplitude of difference characteristic as horizontal axis, by difference characteristic width between blending image (by taking NSST algorithm as an example) and source images
Normalized cumulant between value is the longitudinal axis, obtains the discrete point diagram of different difference characteristics, as shown in Figure 3.
Distributed structure
The purpose of two class image co-registrations is the reservation as much as possible under the premise of not losing source images important information as far as possible
The different information of source images.Therefore, if blending image is closer with infrared intensity image on brightness, while blending image
It is closer with infrared polarization image in edge contour, grain details, illustrate that syncretizing effect is better, i.e. fusion availability is higher；
Otherwise it is lower.It is as follows to merge availability distributed structure process:
It 1), will be between blending image and source images difference characteristic amplitude according to algorithm to the syncretizing effect of difference characteristic amplitude
Distance is divided into 3 sections, respectivelyWithWhereinIndicate the difference characteristic amplitude obtained using weighted mean method
Mean value.When in image block blending image between source images difference characteristic amplitude at a distance from be inWhen, it says
The syncretizing effect of the bright blending image in these image blocks is better than the blending image effect obtained based on weighted mean method, therefore
The algorithm set high to difference characteristic amplitude fusion availabilityOn should have a possibility that certain；When blending image is poor with source images
Distance between different feature amplitude is inWhen, illustrate blending image not by difference characteristic with strong complementarity
Merged, thus in these image blocks algorithm the low collection of difference characteristic amplitude fusion availability is closed should have it is certain can
It can property；When being at a distance from blending image is between source images difference characteristic amplitude
When, can not define in these image blocks algorithm to the fusion availability of difference characteristic amplitude be height be it is low, i.e., it is low in Fei Gaofei
Intermediate state.
2) difference characteristic amplitude is divided into K section, counts each amplitude section X respectivelykInclude in (k=1,2 ..., K)
The total number of image blockAnd 3 image block numbers for being included apart from section in each amplitude sectionWithIn conjunction with the calculating of distance between blending image and source images difference characteristic amplitude, propose that the fusion based on intuition possibility collection has
Validity distributed structure method, formula are as follows:
And haveX can be this 4 class difference characteristic of M, SD, EI and SF.
3) formula (6)-(8) are utilized, calculate blending algorithm to the fusion availability of difference characteristic amplitude, to obtain intuition
The fusion availability of possibility collection is distributed, as shown in figure 4, wherein horizontal axis indicates each difference characteristic amplitude section, longitudinal axis expression is melted
Fusion availability of the hop algorithm (by taking NSST algorithm as an example) to difference characteristic amplitude.
Similarly, can construct respectively DWT, QWT, DTCWT, WPT, LP, GP, KSVD, WA, PCA, TH and GF scheduling algorithm to M,
Fusion availability distribution of this 4 class difference characteristic of SD, EI and SF on its amplitude section.Algorithms of different is to difference characteristic mean value width
The intuition possibility collection fusion availability distribution of value is as shown in Figure 5.
The fusion availability distribution synthesis of sequence may be collected based on intuition
Intuition may collect ranking method
Algorithms of different A can be obtained according to the above methodi(i=1,2 ... m) to the differences feature amplitude such as M, SD, EI and SF
Intuition possibility collection merges availability matrix.By taking difference gray average M as an example, each algorithm is to the matrix F on its amplitude section:
So as to obtain algorithm AiTo the comprehensive evaluation value V of difference gray average amplitudek(k=1,2 ... K) are as follows:
Obviously, Vi(i=1,2 ..., m) it is also intuition possibility collection.If intuition possibility collectionScore
Value and exact value are respectively:
Wherein,For a certain algorithm to difference characteristic on different amplitude sections
It merges availability and illustrates that algorithm is higher to sequence of the difference characteristic on the amplitude section, syncretizing effect is got over if score value is bigger
It is good；And in the identical situation of score value, exact value is bigger, then sequence of the algorithm to difference characteristic on respective magnitudes section
It is higher.That is:
(1) ifThen
(2) ifThen:
1) whenWhen, then
2) whenWhen, then
3) whenWhen, then
According to formula (10), scores of the algorithms of different to difference gray average on the 1st to the 10th amplitude section can be obtained,
As shown in table 1, table 2 is scores of the algorithms of different to difference gray average on the 11st to the 20th amplitude section.
Scores of 1 algorithms of different of table to difference gray average on the 1st to the 10th amplitude section
| M1 | M2 | M3 | M4 | M5 | M6 | M7 | M8 | M9 | M10 | |
| A1 | 0.2143 | 0.4742 | 0.6535 | 0.6526 | 0.8095 | 0.9182 | 0.8829 | 0.8962 | 0.9394 | 0.9167 |
| A2 | 0.3571 | 0.3918 | 0.2970 | 0.2211 | 0.2143 | 0.1818 | 0.0450 | 0.0189 | 0.0101 | 0.0167 |
| A3 | 0.0429 | 0.0206 | -0.0198 | 0.1474 | 0.1270 | 0.1091 | -0.0270 | -0.0094 | -0.0101 | -0.0500 |
| A4 | -0.6286 | -0.4845 | -0.4356 | -0.2737 | -0.0635 | 0.0727 | 0 | 0 | -0.0101 | -0.0333 |
| A5 | 0.3571 | 0.5464 | 0.6139 | 0.7053 | 0.7540 | 0.8273 | 0.8378 | 0.7264 | 0.6465 | 0.6167 |
| A6 | 0.3857 | 0.5670 | 0.6733 | 0.6632 | 0.6190 | 0.7818 | 0.8108 | 0.6887 | 0.7374 | 0.6167 |
| A7 | 0.2429 | 0.4124 | 0.5050 | 0.4947 | 0.5159 | 0.5818 | 0.6306 | 0.4906 | 0.5253 | 0.3833 |
| A8 | 0.7000 | 0.7010 | 0.7228 | 0.5368 | 0.5794 | 0.5091 | 0.4865 | 0.5000 | 0.5859 | 0.5333 |
| A9 | 1.0000 | 0.0515 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| A10 | 0.7571 | 0.6392 | 0.3762 | 0.2842 | 0.2778 | 0.1455 | 0.0721 | 0.0943 | 0.0606 | 0.0500 |
| A11 | 0.9143 | 0.1856 | 0.0198 | 0.0105 | 0 | 0 | 0 | 0 | 0 | 0 |
| A12 | 0.1286 | 0.5670 | 0.7129 | 0.6737 | 0.8254 | 0.9182 | 0.8739 | 0.7642 | 0.8081 | 0.8167 |
Scores of 2 algorithms of different of table to difference gray average on the 11st to the 20th amplitude section
According to the scores of Tables 1 and 2, it can be seen that amplitude section each for difference gray average, algorithm score
Value illustrates that its syncretizing effect is better closer to 1；Score value closer -1, illustrates that syncretizing effect is poorer.It similarly, can using intuition
Algorithms of different can be obtained to the scoring event in other each amplitude sections of 3 class difference characteristic in energy property collection ranking method.By by different calculations
Method synthesizes each difference characteristic amplitude section fusion availability, and algorithm A can be obtainediHave to the fusion of each difference characteristic amplitude
Validity comprehensive evaluation value, as shown in table 3.Fig. 6 is comprehensive evaluation value schematic diagram.
3 algorithm A of tableiTo the fusion availability comprehensive evaluation value of each difference characteristic amplitude
| M | SD | EI | SF | |
| A1 | 0.7467 | 0.4579 | 0.3354 | 0.3260 |
| A2 | 0.1738 | 0.1923 | 0.1901 | 0.2168 |
| A3 | 0.1447 | 0.2418 | 0.1036 | 0.2181 |
| A4 | 0.1819 | 0.1115 | 0.0817 | 0.1921 |
| A5 | 0.6794 | 0.4127 | 0.2383 | 0.1781 |
| A6 | 0.7301 | 0.5755 | 0.4967 | 0.4929 |
| A7 | 0.4823 | 0.1355 | 0.1871 | 0.2020 |
| A8 | 0.4635 | 0.3241 | 0.0651 | 0.0303 |
| A9 | 0.4007 | 0.0912 | 0.2193 | 0.3080 |
| A10 | 0.2447 | 0.2817 | 0.1657 | 0.1079 |
| A11 | 0.3973 | 0.1899 | 0.1811 | 0.2653 |
| A12 | 0.7867 | 0.5007 | 0.2744 | 0.3739 |
It is right as can be seen that merging the composite result of availability distribution according to multiple difference characteristic attributes from table 3 and Fig. 6
Availability comprehensive evaluation value highest is merged in the fusion of difference gray average, GF algorithm, i.e. syncretizing effect is best；For difference mark
For quasi- poor, difference edge intensity and difference spatial frequency, fusion availability overall merit of the LP algorithm to this 3 class difference characteristic
Value is highest.
Experimental analysis
In order to verify the validity that the difference characteristic attribute fusion availability based on intuition possibility collection of proposition is distributed, this
Application optionally takes the progress experimental verification (as shown in Figure 7) of three groups of bimodal infrared images under different scenes.
Subjective assessment: for first group of authentication image, DWT, LP and GF algorithm fusion image is more complete remain it is red
The brightness of outer intensity image, and other blending images are whole partially dark；Keeping blending image that there is higher brightness and comparison
Under the premise of degree, LP algorithm can be more by the texture detail information of infrared polarization image and edge contour information (see Fig. 8's
Boxed area) it is dissolved into blending image, and visual effect and clarity are also relatively good.In second group of authentication image, for
3 regions (see the boxed area of Fig. 9) that source images differ greatly, from visual brightness, contrast, grain details and edge contour
Etc., the preferable algorithm of syncretizing effect is LP and GF algorithm, and especially the texture detail information of LP algorithm fusion image is more
It is abundant, and other blending images are whole partially dark, cause visual effect poor.For third group authentication image, blending image is being kept
Under the premise of with higher brightness and contrast, especially in the region of front cover (see the white edge position of Figure 10), syncretizing effect phase
It is GF, LP, GP and DWT to best algorithm.
Objectively evaluate: the application is similar using gray average (mean), standard deviation (std), edge strength (es), edge
Spend (QAB/f) and spatial frequency (sf) different blending algorithms are objectively evaluated.Wherein, gray average reflects the bright of image
Feature is spent, standard deviation reflects the contrast metric of image, and edge strength and edge similar degree reflect the edge wheel of image
Wide feature, spatial frequency reflects the clarity of image, and These parameters are the bigger the better.
The gray average of 4 three groups of image algorithms of different of table
Other index values of 5 three groups of image algorithms of different of table
For three groups of authentication images, mean index of the GF algorithm in above-mentioned algorithm be it is highest, illustrate the algorithm
It can luminance difference feature preferably between blending image；Std, es, Q of LP algorithmAB/fWith the indexs such as sf in above-mentioned algorithm
It is substantially highest.(when LP algorithm index appearance be not highest the case where when, value in 12 kinds of algorithms be also it is especially forward
).Illustrate edge contour, contrast and the detail textures feature that the algorithm can preferably between blending image.
Show the best calculation of obtained syncretizing effect for different difference characteristic attributes by subjective and objective analysis
Method is substantially consistent, and the algorithm for showing that the fusion availability according to the application construction is distributed to select syncretizing effect best is feasible
With it is effective.Therefore, it is distributed using the difference characteristic attribute fusion availability based on intuition possibility collection of the application construction, and
Multiple distributions are synthesized using based on intuition possibility collection sort method, it can be according to difference characteristic attribute (type, width
Value) difference, targetedly select the blending algorithm best to this feature syncretizing effect.
Claims (1)
1. image difference characteristic attribute fusion availability distributed structure and synthetic method, feature based on intuition possibility collection exist
In the following steps are included:
S1: source images determine: choosing infrared light intensity and polarization scene figure as research object；
S2: the determination of difference characteristic type: comparison S1 in scene figure feature of image and combine the infrared light intensity of bimodal and partially
The imaging mechanism of vibration image, determines the difference characteristic type of bimodal infrared image, and it is strong to choose gray average, standard deviation, edge
Degree and spatial frequency are as difference characteristic T；
S3: it establishes difference characteristic intuition possibility and integrates: the different values for setting domain X as difference characteristic amplitude, by algorithm to difference
The syncretizing effect of characteristic attribute is used as well 1 possibility collectionThenIt indicates when difference feature amplitude value is x, algorithm
A possibility that high to amplitude x fusion availability,It indicates when amplitude value is x, algorithm is high to amplitude x fusion availability
Non- possibility, i.e., algorithm to amplitude x fusion availability it is low a possibility that,It reflects algorithm and availability is merged to amplitude x
A possibility that Fei Gaofei is low；
S4: blending image calculates at a distance from difference characteristic amplitude between source images: choosing a variety of blending algorithms and is melted with what is chosen
Hop algorithm merges the source images in S1, and the blending image after the source images and various algorithm fusions in S1 is carried out piecemeal
Processing, obtains n image block, utilizes difference characteristic amplitude between blending image and source images in each image block of distance measure calculating
Between distance,Wherein, DXFor blending image and all image blocks of source images
The n-dimensional vector of distance between difference characteristic amplitude,Indicate the mean value of i-th of image block difference characteristic amplitude in blending image,WithThe mean value of i-th of image block difference characteristic amplitude in infrared polarization image and infrared intensity image is respectively indicated, X is
M, D is all obtained by calculation in this 4 class difference characteristic of SD, EI and SF, the blending image after each algorithm fusionX；
S5: fusion availability distributed structure: (1) according to algorithm to the syncretizing effect of difference characteristic amplitude, by blending image and source
Distance between image difference feature amplitude is divided into 3 sections, respectively
With
(2) difference characteristic amplitude is divided into K section, counts each amplitude section X respectivelykThe interior total number comprising image blockAnd 3 image block numbers for being included apart from section in each amplitude sectionWithIn conjunction with blending image
The calculating of distance between source images difference characteristic amplitude proposes the fusion availability distributed structure side based on intuition possibility collection
Method, formula are as follows:
And haveX is this 4 class difference characteristic of M, SD, EI and SF；
(3) each blending algorithm is calculated to the fusion availability of difference characteristic amplitude using above-mentioned formula, so that obtaining intuition may
Property collection fusion availability distribution；
S6: intuition may collect ranking method: intuition possibility collectionScore value and exact value be respectively:Wherein,For a certain algorithm pair
Fusion availability of the difference characteristic on different amplitude sections illustrates algorithm to difference characteristic in the amplitude if score value is bigger
Sequence on section is higher, and syncretizing effect is better；And in the identical situation of score value, exact value is bigger, then algorithm is to difference
Sequence of the feature on respective magnitudes section is also higher, and the synthesis knot of availability distribution is merged according to multiple difference characteristic attributes
Fruit obtains the algorithm good to difference characteristic syncretizing effect, and then can targetedly be selected according to the difference of difference characteristic attribute
Select the blending algorithm best to this feature syncretizing effect.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910343217.5A CN110223264B (en) | 2019-04-26 | 2019-04-26 | Image difference characteristic attribute fusion validity distribution structure based on intuition possibility set and synthesis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910343217.5A CN110223264B (en) | 2019-04-26 | 2019-04-26 | Image difference characteristic attribute fusion validity distribution structure based on intuition possibility set and synthesis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110223264A true CN110223264A (en) | 2019-09-10 |
| CN110223264B CN110223264B (en) | 2022-03-25 |
Family
ID=67820011
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910343217.5A Expired - Fee Related CN110223264B (en) | 2019-04-26 | 2019-04-26 | Image difference characteristic attribute fusion validity distribution structure based on intuition possibility set and synthesis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110223264B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111445430A (en) * | 2020-03-31 | 2020-07-24 | 中北大学 | Bimodal infrared image fusion algorithm selection method based on difference characteristic amplitude interval fusion validity distribution |
| CN111445464A (en) * | 2020-03-31 | 2020-07-24 | 中北大学 | Difference characteristic frequency distribution construction method based on nonparametric estimation |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1326169A (en) * | 2000-05-31 | 2001-12-12 | 三星电子株式会社 | Method and device for measuring similarity between images |
| US20110052029A1 (en) * | 2009-08-27 | 2011-03-03 | David Connah | Method and system for generating accented image data |
| CA2822150A1 (en) * | 2013-07-26 | 2015-01-26 | Rui Shen | Method and system for fusing multiple images |
| CN106846288A (en) * | 2017-01-17 | 2017-06-13 | 中北大学 | A kind of many algorithm fusion methods of bimodal infrared image difference characteristic Index |
| CN106846289A (en) * | 2017-01-17 | 2017-06-13 | 中北大学 | A kind of infrared light intensity and polarization image fusion method based on conspicuousness migration with details classification |
| CN108038852A (en) * | 2017-12-14 | 2018-05-15 | 天津师范大学 | A kind of Quality Measures for Image Fusion represented based on joint sparse |
| CN109035188A (en) * | 2018-07-16 | 2018-12-18 | 西北工业大学 | A kind of intelligent image fusion method based on target signature driving |
| CN109255368A (en) * | 2018-08-07 | 2019-01-22 | 平安科技（深圳）有限公司 | Randomly select method, apparatus, electronic equipment and the storage medium of feature |
| CN109636766A (en) * | 2018-11-28 | 2019-04-16 | 南京理工大学 | Polarization differential and intensity image Multiscale Fusion method based on marginal information enhancing |
-
2019
- 2019-04-26 CN CN201910343217.5A patent/CN110223264B/en not_active Expired - Fee Related
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1326169A (en) * | 2000-05-31 | 2001-12-12 | 三星电子株式会社 | Method and device for measuring similarity between images |
| US20110052029A1 (en) * | 2009-08-27 | 2011-03-03 | David Connah | Method and system for generating accented image data |
| CA2822150A1 (en) * | 2013-07-26 | 2015-01-26 | Rui Shen | Method and system for fusing multiple images |
| CN106846288A (en) * | 2017-01-17 | 2017-06-13 | 中北大学 | A kind of many algorithm fusion methods of bimodal infrared image difference characteristic Index |
| CN106846289A (en) * | 2017-01-17 | 2017-06-13 | 中北大学 | A kind of infrared light intensity and polarization image fusion method based on conspicuousness migration with details classification |
| CN108038852A (en) * | 2017-12-14 | 2018-05-15 | 天津师范大学 | A kind of Quality Measures for Image Fusion represented based on joint sparse |
| CN109035188A (en) * | 2018-07-16 | 2018-12-18 | 西北工业大学 | A kind of intelligent image fusion method based on target signature driving |
| CN109255368A (en) * | 2018-08-07 | 2019-01-22 | 平安科技（深圳）有限公司 | Randomly select method, apparatus, electronic equipment and the storage medium of feature |
| CN109636766A (en) * | 2018-11-28 | 2019-04-16 | 南京理工大学 | Polarization differential and intensity image Multiscale Fusion method based on marginal information enhancing |
Non-Patent Citations (5)
| Title |
|---|
| LINNA JI 等: "Image Fusion Algorithm Selection Based on Fusion Validity Distribution Combination of Difference Features", 《ELECTRONICS》 * |
| SICHENG ZHAO 等: "Approximating Discrete Probability Distribution of Image Emotions by Multi-Modal Features Fusion", 《PROCEEDINGS OF THE TWENTY-SIXTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE 》 * |
| 张雷 等: "差异特征指数测度的红外偏振与光强图像多算法融合", 《火力指挥与控制》 * |
| 杨风暴: "红外偏振与光强图像的拟态融合原理和模型研究", 《中北大学学报（自然科学版）》 * |
| 袭肖明 等: "基于手指静脉和手指轮廓的个性化融合方法", 《计算机研究与发展》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111445430A (en) * | 2020-03-31 | 2020-07-24 | 中北大学 | Bimodal infrared image fusion algorithm selection method based on difference characteristic amplitude interval fusion validity distribution |
| CN111445464A (en) * | 2020-03-31 | 2020-07-24 | 中北大学 | Difference characteristic frequency distribution construction method based on nonparametric estimation |
| CN111445430B (en) * | 2020-03-31 | 2022-03-29 | 中北大学 | Bimodal infrared image fusion algorithm selection method based on difference characteristic amplitude interval fusion validity distribution |
| CN111445464B (en) * | 2020-03-31 | 2022-04-19 | 中北大学 | Difference characteristic frequency distribution construction method based on nonparametric estimation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110223264B (en) | 2022-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107341786B (en) | The infrared and visible light image fusion method that wavelet transformation and joint sparse indicate | |
| CN101478693B (en) | Method for evaluating star-loaded optical remote sensing image compression quality | |
| CN102932605B (en) | Method for selecting camera combination in visual perception network | |
| CN104299232B (en) | SAR image segmentation method based on self-adaptive window directionlet domain and improved FCM | |
| CN106803242A (en) | Multi-focus image fusing method based on quaternion wavelet conversion | |
| CN107704807A (en) | A kind of dynamic monitoring method based on multi-source remote sensing sequential images | |
| CN110232389A (en) | A kind of stereoscopic vision air navigation aid based on green crop feature extraction invariance | |
| CN109583435A (en) | A kind of adaptive ballot method for detecting lane lines | |
| CN105651263A (en) | Shallow sea water depth multi-source remote sensing fusion inversion method | |
| CN110223264A (en) | Image difference characteristic attribute fusion availability distributed structure and synthetic method based on intuition possibility collection | |
| CN108765414A (en) | Based on wavelet decomposition and natural scene count without referring to stereo image quality evaluation method | |
| CN102509104A (en) | Confidence map-based method for distinguishing and detecting virtual object of augmented reality scene | |
| CN108154094A (en) | The non-supervisory band selection method of high spectrum image divided based on subinterval | |
| CN114782298B (en) | Infrared and visible light image fusion method with regional attention | |
| CN102982517A (en) | Remote-sensing image fusion method based on local correlation of light spectrum and space | |
| CN109523583A (en) | A kind of power equipment based on feedback mechanism is infrared and visible light image registration method | |
| CN102096913B (en) | Multi-strategy image fusion method under compressed sensing framework | |
| Gu et al. | Nonlinear additive model based saliency map weighting strategy for image quality assessment | |
| CN106846288A (en) | A kind of many algorithm fusion methods of bimodal infrared image difference characteristic Index | |
| CN112669249A (en) | Infrared and visible light image fusion method combining improved NSCT (non-subsampled Contourlet transform) transformation and deep learning | |
| CN109816618A (en) | A kind of region energy photon counting Image Fusion based on adaptive threshold | |
| CN108846329A (en) | A kind of EO-1 hyperion face identification method based on waveband selection and Fusion Features | |
| CN113298147B (en) | Image fusion method and device based on regional energy and intuitionistic fuzzy set | |
| CN109215006B (en) | Image fusion method based on total significant variation model and L2-norm optimization | |
| CN104951800A (en) | Resource exploitation-type area-oriented remote sensing image fusion method |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220325 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |