CN107220974A - A kind of full reference image quality appraisement method and device - Google Patents
A kind of full reference image quality appraisement method and device Download PDFInfo
- Publication number
- CN107220974A CN107220974A CN201710598627.5A CN201710598627A CN107220974A CN 107220974 A CN107220974 A CN 107220974A CN 201710598627 A CN201710598627 A CN 201710598627A CN 107220974 A CN107220974 A CN 107220974A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- mnk
- structural similarity
- shearing
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Abstract
The invention discloses a kind of full reference image quality appraisement method and device, discrete inseparable shearing wave conversion is carried out to reference picture and distorted image respectively, the shearing wave sub-band coefficients of different directions under different scale are obtained；Calculate the structural similarity of each discrete inseparable each block of shearing wave varitron band in direction in each yardstick；Calculate the structural similarity that each discrete inseparable shearing wave in direction in each yardstick converts sub-band coefficients；By calculating the structural similarity of the discrete inseparable shearing each yardstick of wave conversion, the structural similarity for calculating general image is used as the quality evaluation value of distorted image.Therefore, described a kind of full reference image quality appraisement method and device can solve the problem that in the prior art to the problem of image quality evaluation is not comprehensive, accuracy is not high.
Description
Technical field
The present invention relates to technical field of image processing, a kind of full reference image quality appraisement method and device are particularly related to.
Background technology
At present, image quality evaluation has obtained substantial amounts of application in many images and field of video processing, such as compression,
Storage, fusion, enhancing, recovery etc..The target of Objective image quality evaluation is that the evaluation method for developing quantization goes predicted distortion
The quality of image, enables predicted value and subjective assessment to be consistent.Traditional Objective image quality evaluation method includes peak value
Signal to noise ratio (peak signal to noise, PSNR) and mean square error (mean squared error, MSE), both sides
Method is widely used.Although both approaches have a clearly physical significance, and calculate simple, its result of calculation often and
Subjective assessment is inconsistent, because they only calculate the difference between reference picture and distorted image pixel, does not account for
The characteristic of human visual system.
Sheikh et al. proposes fidelity of information criterion method (the Visual Information of natural scene analysis
Fidelity, VIF) (Hamid Rahim Sheikh, Alan C.Bovik, " Image information and visual
Quality Sheikh ", IEEE Trans.Image Process., 2006,15 (2), pp.430-444), the method is information
Theory applied in eyefidelity measurement, by evaluate and test reference picture and distorted image common information number evaluate distortion
The quality of image.WangZhou proposes structural similarity method (structural similarity, SSIM) (Zhou
Wang, Alan Conrad Bovik, Hamid Rahim Sheikh, Eero P.Simoncelli, " Image quality
assessment:From error visibility to structural similarity ", IEEE Trans.Image
Process.,2004,13(4),pp.600-612).SSIM methods extract scene with assuming human visual system's height adaptive
Structural information, the method is of considerable interest, but the method can not correctly evaluate the quality of blurred picture, and is not had
There is the visual characteristic with reference to human eye.Based on SSIM methods, Chun-Ling Yang propose the structural similarity of wavelet transformed domain
Method (discrete wavelet transform-based structural similarity, DWT-SSIM) (Chun-
Ling Yang, Wen-RuiGao, Lai-Man Po, " discrete wavelet transform-based structural
Similarity for image quality assessment ", Proc.Int.Conf.Image Process.ICIP, San
Diego,United states,October 2008).But wavelet transformation is merely capable of optimally representing one-dimensional singular point number
According to, it is impossible to optimally represent the view data of two dimension.Wang-Q Lim propose discrete inseparable shearing wave conversion
(discrete nonseparableshearlet transform, DNST) (Wang-Q Lim,
“Nonseparableshearlet transform”,IEEE Trans.Image Process.,2013,22(5),
pp.2056-2065).Discrete inseparable shearing wave conversion is upwardly formed the tight frame of localized waveform in each yardstick and side,
Image most can be sparsely represented, the defect of wavelet transformation is overcome.
The content of the invention
In view of this, it is an object of the invention to propose a kind of full reference image quality appraisement method and device, it can solve
Certainly in the prior art to the problem of image quality evaluation is not comprehensive, accuracy is not high.
Full reference image quality appraisement method is provided based on the above-mentioned purpose present invention, including：
Discrete inseparable shearing wave conversion is carried out to reference picture and distorted image respectively, obtains different under different scale
The shearing wave sub-band coefficients in direction；
Calculate the structural similarity of each discrete inseparable each block of shearing wave varitron band in direction in each yardstick；
According to the structural similarity of each block, calculate each discrete inseparable shearing wave in direction in each yardstick and become
Change the structural similarity of sub-band coefficients；
By calculating the structural similarity of the discrete inseparable shearing each yardstick of wave conversion, the structure of general image is calculated
Similarity, is used as the quality evaluation value of distorted image.
In the present invention, the shearing wave sub-band coefficients for obtaining different directions under different scale, including：XmnAnd YmnPoint
Biao Shi not reference picture and the discrete inseparable sub-band coefficients for shearing n-th of direction under m grades of yardsticks of wave conversion of distorted image；
It is theoretical using Affine Systems in shearing wave conversion；For two-dimentional system, the formula of composite expanded Affine Systems：
Wherein, ψ ∈ L2(R2), A and B are 2 × 2 invertible matrix, and | detB |=1；Expandable matrix AjHave with change of scale
Close, matrix BlIt is relevant with rotating, shearing；
Wherein, A=A0It is anisotropic expandable matrix, B=B0It is shearing matrix, their form：
Structural similarity is calculated to shearing wave conversion sub-band coefficients；
The form of contrast sensitivity characteristic curvilinear function：
H (f)=2.6 × (0.0192+0.114f) × exp [- (0.114f)1.1]
Wherein, f is spatial frequency.
In the present invention, SSIMmnkRepresent reference picture and the discrete inseparable shearing m grades of chis of wave conversion of distorted image
The structural similarity of lower k-th piece of n-th of the directional subband of degree：
SSIMmnk(xmnk,ymnk)=[l (xmnk,ymnk)]α×[c(xmnk,ymnk)]β×[s(xmnk,ymnk)]γ
Wherein, xmnkAnd ymnkReference picture and the discrete inseparable shearing m grades of chis of wave conversion of distorted image are represented respectively
The coefficient of lower k-th piece of n-th of the directional subband of degree；L (x, y) represents xmnkAnd ymnkBrightness ratio represent x compared with, c (x, y)mnkWith
ymnkContrast compare, s (x, y) represent xmnkAnd ymnkStructure compare；α, β, γ represent coefficient, are all set to 1.
In some embodiments of the invention, l (x, y), c (x, y) and s (x, y) form：
Wherein, μxAnd μyX is represented respectivelymnkAnd ymnkAverage value, σxAnd σyX is represented respectivelymnkAnd ymnkStandard deviation, σxy
Represent xmnkAnd ymnkCovariance, C1、C2And C3Represent small constant.
In the present invention, the discrete inseparable shearing wave conversion in each direction in each yardstick is calculated using following formula
The structural similarity of sub-band coefficients, wherein SSIMmnRepresent the discrete inseparable shearing wave varitron in n-th of direction of m grades of yardsticks
Structural similarity with coefficient：
In some embodiments of the invention, the structure phase for calculating the discrete inseparable shearing each yardstick of wave conversion
Include like degree：
Wherein, SSIMmRepresent the structural similarity of m grades of yardsticks, ωmnRepresent under m grades of yardsticks n-th directional subband
Weights.
In the present invention, using PLCCmnIt is used as the weights ω of m n-th of directional subband of levelmn, ωmn=PLCCmn；
Wherein,aiIt is each m grades of yardsticks n-th of distorted image
The structural similarity SSIM of individual directional subbandmn, biIt is the subjective assessment value of each distorted image.
In the present invention, the structural similarity NTST-SSIM (X, Y) for calculating general image, including：
Wherein, ωmRepresent the weights of m grades of yardsticks.
In some embodiments of the invention, ωm=H (f) |F=m, and
H (f)=2.6 × (0.0192+0.114f) × exp [- (0.114f)1.1]。
In addition, the present invention is always according to full reference image quality appraisement method recited above, there is provided a kind of complete with reference to figure
As quality evaluation device, including：
Sub-band coefficients acquiring unit, becomes for carrying out discrete inseparable shearing wave to reference picture and distorted image respectively
Change, obtain the shearing wave sub-band coefficients of different directions under different scale；
Block structure similarity calculated, for calculating each discrete inseparable shearing wave conversion in direction in each yardstick
The structural similarity of each block of subband；
Sub-band coefficients structural similarity computing unit, for the structural similarity according to each block, calculates each chi
Each discrete inseparable shearing wave in direction converts the structural similarity of sub-band coefficients in degree；
Image structure similarity computing unit, for the knot by calculating the discrete inseparable shearing each yardstick of wave conversion
Structure similarity, calculates the structural similarity of general image, is used as the evaluation of estimate of distorted image.
From the above it can be seen that the present invention a kind of full reference image quality appraisement method and device that provide, from
Dissipate in inseparable shearing wavelet domain, extract the structural similarity of different scale and different directions subband, and using different
Weights summation, be used as evaluation score.Method proposed by the present invention and subjective assessment have good uniformity, and can be fine
Prediction monotonicity and accuracy.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of full reference image quality appraisement method in the embodiment of the present invention；
Fig. 2 is the structural representation of full reference image quality appraisement device in the embodiment of the present invention.
Embodiment
For the object, technical solutions and advantages of the present invention are more clearly understood, below in conjunction with specific embodiment, and reference
Accompanying drawing, the present invention is described in more detail.
It should be noted that all statements for using " first " and " second " are for differentiation two in the embodiment of the present invention
The entity of individual same names non-equal or the parameter of non-equal, it is seen that " first " " second " should not only for the convenience of statement
The restriction to the embodiment of the present invention is interpreted as, subsequent embodiment no longer illustrates one by one to this.
As shown in fig.1, be the structural representation of full reference image quality appraisement method and device in the embodiment of the present invention,
Including step：
Step 101, discrete inseparable shearing wave conversion is carried out to reference picture and distorted image respectively, obtains different chis
The shearing wave sub-band coefficients of the lower different directions of degree.
In embodiment, XmnAnd YmnReference picture and the discrete inseparable shearing wave conversion m of distorted image are represented respectively
The sub-band coefficients in n-th of direction under level yardstick.
In addition, in shearing wave conversion, having used the Affine Systems of classics theoretical.It is composite expanded imitative for two-dimentional system
Shown in the formula such as formula (1) for penetrating system.Wherein, ψ ∈ L2(R2), A and B are 2 × 2 invertible matrix, and | detB |=1.Expand square
Battle array AjIt is relevant with change of scale, matrix BlIt is relevant with rotating, shearing.
Wherein, A=A0It is anisotropic expandable matrix, B=B0It is shearing matrix, shown in their form such as formula (2)：
Shearing wave has the directional sensitivity and spatial domain localization property of height.Discrete inseparable shearing wave conversion is used
The compact schemes shearing wave generated by the inseparable factor.Using discrete inseparable shearing wave conversion, an image can be represented
Multiple subbands of different directions under into different scale, each subband still maintains the elementary contour of original image.One image
After discrete inseparable shearing wave conversion is carried out, the structural information between pixel is not lost.So, can directly it calculate
The structural similarity of shearing wave sub-band coefficients.
Human eye has different sensitiveness to different spatial frequencys.Contrast sensitivity characteristic curve (contrast
Sensitivity function, CSF) this characteristic of human visual system, contrast sensitivity characteristic curve letter can be described
Shown in several forms such as formula (3).Wherein, f is spatial frequency.Contrast sensitivity characteristic curvilinear function shows that human eye is to middle frequency
Rate has higher sensitiveness than high and low frequency.After discrete inseparable shearing wave conversion is carried out to image, image quilt
Resolve into multiple subbands with different space frequency.
H (f)=2.6 × (0.0192+0.114f) × exp [- (0.114f)1.1] (3)
Step 102, the structure of the discrete inseparable each block of shearing wave varitron band in each direction in each yardstick is calculated
Similarity.Its specific implementation process includes：
SSIMmnkRepresent n-th of side under reference picture and the discrete inseparable shearing m grades of yardsticks of wave conversion of distorted image
To k-th piece of structural similarity of subband, as shown in formula (4)：
SSIMmnk(xmnk,ymnk)=[l (xmnk,ymnk)]α×[c(xmnk,ymnk)]β×[s(xmnk,ymnk)]γ(4)
Wherein, xmnkAnd ymnkReference picture and the discrete inseparable shearing m grades of chis of wave conversion of distorted image are represented respectively
The coefficient of lower k-th piece of n-th of the directional subband of degree；L (x, y) represents xmnkAnd ymnkBrightness ratio represent x compared with, c (x, y)mnkWith
ymnkContrast compare, s (x, y) represent xmnkAnd ymnkStructure compare.α, β, γ represent coefficient, are all set to 1.
Further, shown in l (x, y), c (x, y) and s (x, y) concrete form such as formula (5)~(7)：
Wherein, μxAnd μyX is represented respectivelymnkAnd ymnkAverage value, σxAnd σyX is represented respectivelymnkAnd ymnkStandard deviation, σxy
Represent xmnkAnd ymnkCovariance, C1、C2And C3Represent small constant.
Step 103, the structure phase that each discrete inseparable shearing wave in direction in each yardstick converts sub-band coefficients is calculated
Like degree.
As embodiment, the discrete inseparable shearing wave in each direction in each yardstick is calculated using following formula (8)
Convert the structural similarity of sub-band coefficients, wherein SSIMmnRepresent that the discrete inseparable shearing wave in n-th of direction of m grades of yardsticks becomes
Change the structural similarity of sub-band coefficients.
Step 104, the structural similarity of the discrete inseparable shearing each yardstick of wave conversion is calculated.Specific implementation process
Including：
As shown in formula (9)：
Wherein, SSIMmRepresent the structural similarity of m grades of yardsticks, ωmnRepresent under m grades of yardsticks n-th directional subband
Weights.Calculate n-th of direction structure similar value SSIM under subjective assessment value and m grades of yardsticksmn(Xmn,Ymn) Spearman etc.
Level coefficient correlation PLCCmn。PLCCmnThe degree of correlation of n-th of directional subband predicted value of m levels and subjective assessment value is represented, so
It can be used for representing sensitivity of the human eye to this subband.In the embodiment invented at this, PLCC is usedmnIt is used as m levels n-th
The weights ω of individual directional subbandmn, as shown in formula (10)：ωmn=PLCCmn (10)
Wherein,aiIt is each m grades of chis of distorted image
Spend the structural similarity SSIM of n-th of directional subbandmn, biIt is the subjective assessment value of each distorted image.
Step 105, the structural similarity NTST-SSIM (X, Y) of general image is calculated.It is specific to implement to include：
As shown in formula (11)：
Wherein, ωmRepresent the weights of m grades of yardsticks.Contrast sensitivity characteristic curve can be used to calculate ωm。
Wherein, ωm=H (f) |F=m, and
H (f)=2.6 × (0.0192+0.114f) × exp [- (0.114f)1.1]。
As the embodiment of another aspect of the present invention, as shown in fig.2, being full reference picture quality in the embodiment of the present invention
The structural representation of evaluating apparatus.In embodiment, described full reference image quality appraisement device includes the son being sequentially connected
Band coefficient acquiring unit 201, block structure similarity calculated 202, sub-band coefficients structural similarity computing unit 203 and figure
As structural similarity computing unit 204.Wherein, sub-band coefficients acquiring unit 201 is carried out to reference picture and distorted image respectively
Discrete inseparable shearing wave conversion, obtains the shearing wave sub-band coefficients of different directions under different scale.Block structure similarity meter
Calculate the structural similarity that unit 202 calculates each discrete inseparable each block of shearing wave varitron band in direction in each yardstick.
Sub-band coefficients structural similarity computing unit 203 calculates each side in each yardstick according to the structural similarity of each block
The structural similarity of sub-band coefficients is converted to discrete inseparable shearing wave.Image structure similarity computing unit 204 passes through meter
The structural similarity of the discrete inseparable shearing wave conversion of each yardstick is calculated, the structural similarity of general image is calculated.
It should be noted that when being operated between the complete each unit of reference image quality appraisement device and unit
The step of, be described in detail in process and the superincumbent full reference image quality appraisement method of various embodiments, herein not
Remake repetition discussion.
A kind of full reference image quality appraisement method and device that the present invention is provided, in discrete inseparable shearing wave conversion
In domain, the structural similarity of different scale and different directions subband is extracted, and is summed using different weights, is divided as evaluating
Number.
LIVE image quality evaluations storehouse can be used to assess NTST-SSIM methods proposed by the present invention.LIVE database bags
Include 29 reference pictures, 779 distorted images, type of distortion in having 5：JPEG2000 compressions (can be represented with JP2000),
JPEG compression (can be represented with JPEG), white noise (can be represented with WN), Gaussian Blur (can be represented with GB), using quick
The Rayleigh channel that fails transmits the error of transmission (can be represented with FF) of JPEG2000 images.
It is used for measuring the performance of method for objectively evaluating using 3 evaluation indexes.Spearman coefficient of rank correlations
(Spearman rank-order correlation coefficient, SROCC) is used for measuring the monotonicity of prediction.Use
Nonlinear regression is that subjective evaluation of estimate and predicted value provide Nonlinear Mapping.After Nonlinear Mapping is carried out, use respectively
Pearson linearly dependent coefficients (linear correlation coefficient, PLCC) and root-mean-square error (root
Mean square error, RMSE) it is used for correlation and the degree of accuracy of evaluation and foreca.SROCC values and PLCC values are bigger, RMSE
It is smaller, represent that estimated performance is better.
The comparative result of three evaluation indexes of method proposed by the present invention and classical way is as shown in table 1~3, best property
Energy parameter is represented using runic.The classical way compared includes PSNR, SSIM, DWT-SSIM, VIF.Proposed by the present invention
In method DNST-SSIM, 5 grades of discrete inseparable shearing wave conversions, from lowest scale to highest yardstick, each yardstick are used
Direction quantity is 8,8,8,16,16 successively.
Table 1：The comparison of SROCC indexs
Table 2：The comparison of PLCC indexs
Table 3：The comparison of RMSE indexs
From table 1~3, can clearly it find out, method DNST-SSIM proposed by the present invention performance is in all distortion classes
All other methods are exceeded in type.Moreover, the overall performance of DNST-SSIM methods exceeded in these three indexs it is other
Method.
In the present invention, it is proposed that a kind of new full ginseng based on discrete inseparable shearing wavelet domain structural similarity
Examine image quality evaluating method DSNT-SSIM.In this method, discrete inseparable shearing wave conversion has been used, and has been considered
The characteristic of human visual system is arrived.Experimental result surface, method DNST-SSIM proposed by the present invention has exceeded it in performance
Its method, and subjective assessment have more preferable uniformity, and can be good at prediction monotonicity and accuracy.
In summary, a kind of full reference image quality appraisement method and device that provide of the present invention, creatively using cutting
Cut the feature of wave conversion sub-band coefficients, it is proposed that image quality evaluating method and dress that a kind of and human visual system is consistent
Put；Structural similarity is calculated to the subband of discrete inseparable shearing wave conversion, the structural similarity of each subband can be obtained
Value, then sets different weights to different subbands.According to contrast sensitivity characteristic curve, human eye has to different frequencies
Different sensitiveness, so as to obtain different weights；The present invention has extensive, great dissemination；Finally, it is whole described
Full reference image quality appraisement method and device are compact, it is easy to control.
Those of ordinary skills in the art should understand that：The discussion of any of the above embodiment is exemplary only, not
It is intended to imply that the scope of the present disclosure (including claim) is limited to these examples；Under the thinking of the present invention, above example
Or can also not be combined between the technical characteristic in be the same as Example, step can be realized with random order, and be existed such as
Many other changes of upper described different aspect of the invention, for simplicity, they are provided not in details.
Embodiments of the invention be intended to fall within the broad range of appended claims it is all it is such replace,
Modifications and variations.Therefore, within the spirit and principles of the invention, any omission, modification, equivalent substitution, the improvement made
Deng should be included in the scope of the protection.
Claims (10)
1. a kind of full reference image quality appraisement method, it is characterised in that including：
Discrete inseparable shearing wave conversion is carried out to reference picture and distorted image respectively, different directions under different scale are obtained
Shearing wave sub-band coefficients；
Calculate the structural similarity of each discrete inseparable each block of shearing wave varitron band in direction in each yardstick；
According to the structural similarity of each block, the discrete inseparable shearing wave varitron in each direction in each yardstick is calculated
Structural similarity with coefficient；
By calculating the structural similarity of the discrete inseparable shearing each yardstick of wave conversion, the structure for calculating general image is similar
Degree, is used as the quality evaluation value of distorted image.
2. according to the method described in claim 1, it is characterised in that the shearing marble for obtaining different directions under different scale
Band coefficient, including：XmnAnd YmnReference picture and the discrete inseparable shearing m grades of yardsticks of wave conversion of distorted image are represented respectively
The sub-band coefficients in lower n-th of direction；
It is theoretical using Affine Systems in shearing wave conversion；For two-dimentional system, the formula of composite expanded Affine Systems：
Wherein, ψ ∈ L2(R2), A and B are 2 × 2 invertible matrix, and | detB |=1；Expandable matrix AjIt is relevant with change of scale,
Matrix BlIt is relevant with rotating, shearing；
Wherein, A=A0It is anisotropic expandable matrix, B=B0It is shearing matrix, their form：
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>A</mi>
<mn>0</mn>
</msub>
<mo>=</mo>
<mfenced open = "(" close = ")">
<mtable>
<mtr>
<mtd>
<mn>4</mn>
</mtd>
<mtd>
<mn>0</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>2</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>B</mi>
<mn>0</mn>
</msub>
<mo>=</mo>
<mfenced open = "(" close = ")">
<mtable>
<mtr>
<mtd>
<mn>1</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
</mtd>
<mtd>
<mn>1</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Structural similarity is calculated to shearing wave conversion sub-band coefficients；
The form of contrast sensitivity characteristic curvilinear function：
H (f)=2.6 × (0.0192+0.114f) × exp [- (0.114f)1.1]
Wherein, f is spatial frequency.
3. according to the method described in claim 1, it is characterised in that SSIMmnkRepresent reference picture and distorted image is discrete can not
The structural similarity of n-th of k-th piece of directional subband under separation shearing m grades of yardsticks of wave conversion：
SSIMmnk(xmnk,ymnk)=[l (xmnk,ymnk)]α×[c(xmnk,ymnk)]β×[s(xmnk,ymnk)]γ
Wherein, xmnkAnd ymnkRepresent respectively under reference picture and the discrete inseparable shearing m grades of yardsticks of wave conversion of distorted image
The coefficient of n-th of k-th piece of directional subband；L (x, y) represents xmnkAnd ymnkBrightness ratio represent x compared with, c (x, y)mnkAnd ymnk's
Contrast compares, and s (x, y) represents xmnkAnd ymnkStructure compare；α, β, γ represent coefficient, are all set to 1.
4. method according to claim 3, it is characterised in that l (x, y), c (x, y) and s (x, y) form：
<mrow>
<mi>c</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<mn>2</mn>
<msub>
<mi>&sigma;</mi>
<mi>x</mi>
</msub>
<msub>
<mi>&sigma;</mi>
<mi>y</mi>
</msub>
<mo>+</mo>
<msub>
<mi>c</mi>
<mn>2</mn>
</msub>
</mrow>
<mrow>
<msubsup>
<mi>&sigma;</mi>
<mi>x</mi>
<mn>2</mn>
</msubsup>
<mo>+</mo>
<msubsup>
<mi>&sigma;</mi>
<mi>y</mi>
<mn>2</mn>
</msubsup>
<mo>+</mo>
<msub>
<mi>c</mi>
<mn>2</mn>
</msub>
</mrow>
</mfrac>
</mrow>
<mrow>
<mi>S</mi>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>&sigma;</mi>
<mrow>
<mi>x</mi>
<mi>y</mi>
</mrow>
</msub>
<mo>+</mo>
<msub>
<mi>c</mi>
<mn>3</mn>
</msub>
</mrow>
<mrow>
<msub>
<mi>&sigma;</mi>
<mi>x</mi>
</msub>
<mo>+</mo>
<msub>
<mi>&sigma;</mi>
<mi>y</mi>
</msub>
<mo>+</mo>
<msub>
<mi>c</mi>
<mn>3</mn>
</msub>
</mrow>
</mfrac>
</mrow>
Wherein, μxAnd μyX is represented respectivelymnkAnd ymnkAverage value, σxAnd σyX is represented respectivelymnkAnd ymnkStandard deviation, σxyRepresent
xmnkAnd ymnkCovariance, C1、C2And C3Represent small constant.
5. method according to claim 3, it is characterised in that calculate each direction in each yardstick using following formula
Discrete inseparable shearing wave converts the structural similarity of sub-band coefficients, wherein SSIMmnRepresent n-th of direction of m grades of yardsticks from
Dissipate the structural similarity that inseparable shearing wave converts sub-band coefficients：
<mrow>
<msub>
<mi>SSIM</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>X</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>Y</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>K</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>K</mi>
</munderover>
<msub>
<mi>SSIM</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
<mi>k</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mo>.</mo>
</mrow>
6. method according to claim 5, it is characterised in that the discrete inseparable shearing each chi of wave conversion of calculating
The structural similarity of degree, including：
<mrow>
<msub>
<mi>SSIM</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>X</mi>
<mi>m</mi>
</msub>
<mo>,</mo>
<msub>
<mi>Y</mi>
<mi>m</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>&Sigma;</mi>
<mi>n</mi>
</msub>
<msub>
<mi>&omega;</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
<msub>
<mi>SSIM</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>X</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>Y</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>&Sigma;</mi>
<mi>n</mi>
</msub>
<msub>
<mi>&omega;</mi>
<mrow>
<mi>m</mi>
<mi>n</mi>
</mrow>
</msub>
</mrow>
</mfrac>
</mrow>
Wherein, SSIMmRepresent the structural similarity of m grades of yardsticks, ωmnRepresent the power of n-th of directional subband under m grades of yardsticks
Value.
7. method according to claim 6, it is characterised in that use PLCCmnIt is used as the power of m n-th of directional subband of level
Value ωmn, ωmn=PLCCmn；
Wherein,aiIt is each n-th of yardstick of m grades of distorted image
The structural similarity SSIM of directional subbandmn, biIt is the subjective assessment value of each distorted image.
8. method according to claim 6, it is characterised in that the structural similarity NTST- of the calculating general image
SSIM (X, Y), including：
<mrow>
<mi>N</mi>
<mi>T</mi>
<mi>S</mi>
<mi>T</mi>
<mo>-</mo>
<mi>S</mi>
<mi>S</mi>
<mi>I</mi>
<mi>M</mi>
<mrow>
<mo>(</mo>
<mi>X</mi>
<mo>,</mo>
<mi>Y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mrow>
<msub>
<mi>&Sigma;</mi>
<mi>m</mi>
</msub>
<msub>
<mi>&omega;</mi>
<mi>m</mi>
</msub>
<msub>
<mi>SSIM</mi>
<mi>m</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>X</mi>
<mi>m</mi>
</msub>
<mo>,</mo>
<msub>
<mi>Y</mi>
<mi>m</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<msub>
<mi>&Sigma;</mi>
<mi>m</mi>
</msub>
<msub>
<mi>&omega;</mi>
<mi>m</mi>
</msub>
</mrow>
</mfrac>
</mrow>
Wherein, ωmRepresent the weights of m grades of yardsticks.
9. method according to claim 8, it is characterised in that ωm=H (f) |F=m, and H (f)=2.6 × (0.0192+
0.114f)×exp[-(0.114f)1.1]。
10. a kind of full reference image quality appraisement device, it is characterised in that the side according to any one of claim 1 to 9
Method, including：
Sub-band coefficients acquiring unit, for carrying out discrete inseparable shearing wave conversion to reference picture and distorted image respectively,
Obtain the shearing wave sub-band coefficients of different directions under different scale；
Block structure similarity calculated, for calculating each discrete inseparable shearing wave varitron band in direction in each yardstick
The structural similarity of each block；
Sub-band coefficients structural similarity computing unit, for the structural similarity according to each block, is calculated in each yardstick
Each discrete inseparable shearing wave in direction converts the structural similarity of sub-band coefficients；
Image structure similarity computing unit, for the structure phase by calculating the discrete inseparable shearing each yardstick of wave conversion
Like spending, the structural similarity of general image is calculated, the evaluation of estimate of distorted image is used as.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710598627.5A CN107220974A (en) | 2017-07-21 | 2017-07-21 | A kind of full reference image quality appraisement method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710598627.5A CN107220974A (en) | 2017-07-21 | 2017-07-21 | A kind of full reference image quality appraisement method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107220974A true CN107220974A (en) | 2017-09-29 |
Family
ID=59953405
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710598627.5A Pending CN107220974A (en) | 2017-07-21 | 2017-07-21 | A kind of full reference image quality appraisement method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107220974A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108600745A (en) * | 2018-08-06 | 2018-09-28 | 北京理工大学 | A kind of method for evaluating video quality based on time-space domain slice multichannel chromatogram configuration |
| CN109360194A (en) * | 2018-09-27 | 2019-02-19 | 北京印刷学院 | A kind of image quality evaluating method based on discrete inseparable shearing wave conversion and human-eye visual characteristic |
| CN110827241A (en) * | 2019-10-21 | 2020-02-21 | 国家广播电视总局广播电视规划院 | Low-brightness enhanced picture full-reference method based on color distortion and contrast enhancement |
| CN112918116A (en) * | 2021-01-27 | 2021-06-08 | 北京印刷学院 | Method for improving printing precision of color printing system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101621709A (en) * | 2009-08-10 | 2010-01-06 | 浙江大学 | Method for evaluating objective quality of full-reference image |
| CN102421007A (en) * | 2011-11-28 | 2012-04-18 | 浙江大学 | Image quality evaluating method based on multi-scale structure similarity weighted aggregate |
| CN102821301A (en) * | 2012-08-31 | 2012-12-12 | 天津大学 | HVS (Human Visual System)-based objective quality evaluation method for three-dimensional (3D) information |
| CN103475898A (en) * | 2013-09-16 | 2013-12-25 | 北京理工大学 | Non-reference image quality assessment method based on information entropy characters |
| CN103996188A (en) * | 2014-04-27 | 2014-08-20 | 嘉兴学院 | Full-reference-type image quality evaluation method based on Gabor weighted characteristics |
| CN104282012A (en) * | 2013-07-05 | 2015-01-14 | 江南大学 | Wavelet domain based semi-reference image quality evaluating algorithm |
| CN104318577A (en) * | 2014-11-06 | 2015-01-28 | 天津大学 | Multi-scale and full-reference image quality evaluation method |
-
2017
- 2017-07-21 CN CN201710598627.5A patent/CN107220974A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101621709A (en) * | 2009-08-10 | 2010-01-06 | 浙江大学 | Method for evaluating objective quality of full-reference image |
| CN102421007A (en) * | 2011-11-28 | 2012-04-18 | 浙江大学 | Image quality evaluating method based on multi-scale structure similarity weighted aggregate |
| CN102821301A (en) * | 2012-08-31 | 2012-12-12 | 天津大学 | HVS (Human Visual System)-based objective quality evaluation method for three-dimensional (3D) information |
| CN104282012A (en) * | 2013-07-05 | 2015-01-14 | 江南大学 | Wavelet domain based semi-reference image quality evaluating algorithm |
| CN103475898A (en) * | 2013-09-16 | 2013-12-25 | 北京理工大学 | Non-reference image quality assessment method based on information entropy characters |
| CN103996188A (en) * | 2014-04-27 | 2014-08-20 | 嘉兴学院 | Full-reference-type image quality evaluation method based on Gabor weighted characteristics |
| CN104318577A (en) * | 2014-11-06 | 2015-01-28 | 天津大学 | Multi-scale and full-reference image quality evaluation method |
Non-Patent Citations (3)
| Title |
|---|
| CHUN-LING YANG等: "DISCRETE WAVELET TRANSFORM-BASED STRUCTURAL SIMILARITY FOR IMAGE QUALITY ASSESSMENT", 《2008 15TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING》 * |
| WU DONG等: "Fast Digital Shearlet Transform Based Image Quality Assessment", 《2016 JOINT INTERNATIONAL CONFERENCE ON SERVICE SCIENCE, MANAGEMENT AND ENGINEERING (SSME 2016) AND INTERNATIONAL CONFERENCE ON INFORMATION SCIENCE AND TECHNOLOGY (IST 2016)》 * |
| 任玉玲: "基于自然场景统计特性的图像质量评价", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108600745A (en) * | 2018-08-06 | 2018-09-28 | 北京理工大学 | A kind of method for evaluating video quality based on time-space domain slice multichannel chromatogram configuration |
| CN108600745B (en) * | 2018-08-06 | 2020-02-18 | 北京理工大学 | Video quality evaluation method based on time-space domain slice multi-map configuration |
| CN109360194A (en) * | 2018-09-27 | 2019-02-19 | 北京印刷学院 | A kind of image quality evaluating method based on discrete inseparable shearing wave conversion and human-eye visual characteristic |
| CN109360194B (en) * | 2018-09-27 | 2022-02-22 | 北京印刷学院 | Image quality evaluation method based on discrete inseparable shear wave transformation and human eye visual characteristics |
| CN110827241A (en) * | 2019-10-21 | 2020-02-21 | 国家广播电视总局广播电视规划院 | Low-brightness enhanced picture full-reference method based on color distortion and contrast enhancement |
| CN112918116A (en) * | 2021-01-27 | 2021-06-08 | 北京印刷学院 | Method for improving printing precision of color printing system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Reduced-reference image quality assessment using a wavelet-domain natural image statistic model | |
| KR101664913B1 (en) | Method and system for determining a quality measure for an image using multi-level decomposition of images | |
| CN107220974A (en) | A kind of full reference image quality appraisement method and device | |
| CN101562675B (en) | No-reference image quality evaluation method based on Contourlet transform | |
| Eskicioglu et al. | A survey of quality measures for gray scale image compression | |
| CN101562758B (en) | Method for objectively evaluating image quality based on region weight and visual characteristics of human eyes | |
| CN103200421B (en) | No-reference image quality evaluation method based on Curvelet transformation and phase coincidence | |
| CN102547368B (en) | Objective evaluation method for quality of stereo images | |
| CN101976444B (en) | Pixel type based objective assessment method of image quality by utilizing structural similarity | |
| KR20140019358A (en) | Method and system for increasing robustness of visual quality metrics using spatial shifting | |
| Okarma | Combined image similarity index | |
| CN104202594B (en) | A kind of method for evaluating video quality based on 3 D wavelet transformation | |
| CN101127926A (en) | Image quality evaluation method based on multi-scale geometric analysis | |
| CN104376565A (en) | Non-reference image quality evaluation method based on discrete cosine transform and sparse representation | |
| CN103354617B (en) | Boundary strength compressing image quality objective evaluation method based on DCT domain | |
| CN107948635A (en) | It is a kind of based on degenerate measurement without refer to sonar image quality evaluation method | |
| CN106023214A (en) | Image quality evaluation method and system based on foveal vision gradient structural similarity | |
| CN102547363B (en) | No-reference image quality evaluating method on basis of contourlet transform domain image energy features | |
| CN102930545A (en) | Statistical measure method for image quality blind estimation | |
| CN103258326A (en) | Information fidelity method for image quality blind evaluation | |
| CN103996188A (en) | Full-reference-type image quality evaluation method based on Gabor weighted characteristics | |
| CN111968073B (en) | No-reference image quality evaluation method based on texture information statistics | |
| CN103955956B (en) | A kind of image combined reconstruction method towards compressed sensing | |
| Vora et al. | Analysis of compressed image quality assessments, m | |
| Fu et al. | Screen content image quality assessment using Euclidean distance |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170929 |