CN107105223B - A kind of tone mapping method for objectively evaluating image quality based on global characteristics - Google Patents
A kind of tone mapping method for objectively evaluating image quality based on global characteristics Download PDFInfo
- Publication number
- CN107105223B CN107105223B CN201710164242.8A CN201710164242A CN107105223B CN 107105223 B CN107105223 B CN 107105223B CN 201710164242 A CN201710164242 A CN 201710164242A CN 107105223 B CN107105223 B CN 107105223B
- Authority
- CN
- China
- Prior art keywords
- denoted
- pixel value
- indicate
- fitting
- value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/02—Diagnosis, testing or measuring for television systems or their details for colour television signals
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of tone mapping method for objectively evaluating image quality based on global characteristics comprising two processes of training stage and test phase；Influence that takes into account natural scene statistical nature and Color Statistical feature to tone mapping, extract the global characteristics vector of tone mapping image, then it is trained using global characteristics vector of the support vector regression to all tone mapping images that training image is concentrated, constructs quality prediction model；In test phase, it is used as the global characteristics vector of the tone mapping image of test by calculating, and the quality prediction model constructed according to the training stage, prediction obtains the Objective Quality Assessment predicted value of the tone mapping image, since the global characteristics Vector Message of acquisition has stronger stability, and it can preferably reflect the quality change situation of tone mapping image, therefore effectively improve the correlation objectively evaluated between result and subjective perception.
Description
Technical field
The present invention relates to a kind of image quality evaluating methods, more particularly, to a kind of tone mapping figure based on global characteristics
As assessment method for encoding quality.
Background technique
With the fast development of display technology, high dynamic range images (HDR) more and more attention has been paid to.High dynamic range
The level of image is abundant, can achieve the effect of shadow that reality is more approached more than normal image.However, traditional display equipment is only
The display of low-dynamic range can be supported to export.In order to which the dynamic range for solving real scene and traditional display equipment is unmatched
Contradiction is currently suggested tone mapping (Tone Mapping) algorithm of many high dynamic range images.High dynamic range images
The target of tone-mapping algorithm be by the luminance compression of high dynamic range images to traditional display equipment acceptable model
It encloses, while retaining the detailed information of original image as far as possible, and avoid image flaw.Therefore, it how accurately, objectively to evaluate not
With the performance of tone mapping method, have a very important role to guidance content production and post-processing.
And for tone mapping image quality evaluation, if directly existing image quality evaluating method is applied to
Tone mapping image, then image only has high dynamic range images as reference due to tone mapping, therefore will lead to can not be accurate
Prediction obtains objectively evaluating value.Therefore, how visual signature is efficiently extracted out in evaluation procedure, so that objectively evaluating result
More feel to meet human visual system, is to need to research and solve during carrying out evaluating objective quality to tone mapping image
The problem of.
Summary of the invention
Technical problem to be solved by the invention is to provide a kind of, and the tone mapping picture quality based on global characteristics is objective
Evaluation method can effectively improve the correlation objectively evaluated between result and subjective perception.
The technical scheme of the invention to solve the technical problem is: a kind of tone mapping figure based on global characteristics
As assessment method for encoding quality, it is characterised in that including two processes of training stage and test phase；
The specific steps of the training stage process are as follows:
1. _ 1, choosing N width tone mapping image construction training image collection, it is denoted asWherein, N > 1,1≤
K≤N,It indicatesIn kth width tone mapping image,In every width tone mapping
The width of image is W, and height is H；
1. _ 2, calculatingIn every width tone mapping image natural scene statistical nature vector, willNatural scene statistical nature vector be denoted asWherein,Dimension be 5 × 1；
1. _ 3, calculatingIn every width tone mapping image Color Statistical characteristic vector, will's
Color Statistical characteristic vector is denoted asWherein,Dimension be 18 × 1；
1. _ 4, willIn every width tone mapping image natural scene statistical nature vector sum color
Statistical nature vector is constitutedIn every width tone mapping image global characteristics vector, willThe overall situation
Characteristic vector is denoted as Fk,Wherein, FkDimension be 23 × 1, symbol " [] " be vector representation symbol,Indicating willWithIt connects to form a global characteristics vector；
1. _ 5, willIn the respective global characteristics vector sum mean subjective of all tone mapping images
Score difference composing training sample data sets, includes the N number of average master of N number of global characteristics vector sum in training sample data set
See scoring difference；Then method of the support vector regression as machine learning is used, to all in training sample data set
Global characteristics vector is trained, so that by the error between the obtained regression function value of training and mean subjective scoring difference
Minimum, fitting obtain optimal weight vector woptWith optimal bias term bopt；Followed by optimal weight vector woptMost
Excellent bias term bopt, quality prediction model is constructed, f (F) is denoted as,Wherein, f () is function table
Show form, F is used to indicate the global characteristics vector of tone mapping image, and the input vector as quality prediction model, (wopt
)TFor woptTransposition,For the linear function of F；
The specific steps of the test phase process are as follows:
2. being used as the tone mapping image I of test for any one widthtest, 1. _ 2 1. _ 4 identical to step according to step
Operation obtains ItestGlobal characteristics vector, be denoted as Ftest；Then according to the quality prediction model of training stage construction to Ftest
It is tested, prediction obtains FtestCorresponding predicted value, using the predicted value as ItestObjective Quality Assessment predicted value, be denoted as
Qtest,Wherein, ItestWidth be W', and height be H', FtestDimension be 23 × 1,Indicate FtestLinear function.
The step 1. _ 2 inAcquisition process are as follows:
1. _ 2a, calculatingIn every width tone mapping image in all pixels point pixel value it is equal
Value, willIn the mean value of pixel value of all pixels point be denoted as ρ,Then it calculatesIn every width tone mapping image in all pixels point pixel value standard deviation, willIn institute
There is the standard deviation of the pixel value of pixel to be denoted as δ,Then it calculates
In every width tone mapping image in all pixels point pixel value the degree of bias, willIn all pixels point pixel value
The degree of bias be denoted as θ,And it calculatesIn every width tone mapping image in
All pixels point pixel value kurtosis, willIn the kurtosis of pixel value of all pixels point be denoted as κ,It calculates againIn every width tone mapping image in all pixels points
The entropy of pixel value, willIn the entropy of pixel value of all pixels point be denoted as η,Wherein, 1≤x≤W, 1
≤ y≤H,It indicatesMiddle coordinate position is the pixel value of the pixel of (x, y), 0≤g≤255, pgIt indicatesIn
All pixels point pixel value in belong to the probability density function values of g-th of density value,
1. _ 2b, by pairIn all tone mapping images respectively in all pixels point pixel
The mean value of value carries out Gauss Distribution Fitting, and fitting obtainsIn all tone mapping images mean value Gauss
Then fitting of distribution curve obtains the match value of ρ according to the mean value Gauss Distribution Fitting curve that fitting obtains, is denoted as fρ,Equally, by rightIn all tone mapping images respectively in
The standard deviation of pixel value of all pixels point carry out Gauss Distribution Fitting, fitting obtainsIn institute it is coloured
The standard deviation Gauss Distribution Fitting curve of mapping image is adjusted, is then obtained according to the standard deviation Gauss Distribution Fitting curve that fitting obtains
The match value for obtaining δ, is denoted as fδ,By rightIn all tones
Map image respectively in the degree of bias of pixel value of all pixels point carry out Gauss Distribution Fitting, fitting obtainsIn all tone mapping images degree of bias Gauss Distribution Fitting curve, then according to fitting obtain it is inclined
The match value that Gauss Distribution Fitting curve obtains θ is spent, f is denoted asθ,By rightIn all tone mapping images respectively in all pixels point pixel value kurtosis carry out Gaussian Profile
Fitting, fitting obtainIn all tone mapping images kurtosis Gauss Distribution Fitting curve, then basis
It is fitted the match value that obtained kurtosis Gauss Distribution Fitting curve obtains κ, is denoted as fκ,
By rightIn all tone mapping images respectively in all pixels point pixel value entropy carry out Gauss
Fitting of distribution, fitting obtainIn all tone mapping images entropy Gauss Distribution Fitting curve, then basis
It is fitted the match value that obtained entropy Gauss Distribution Fitting curve obtains η, is denoted as fη,
Wherein, μρAnd σρIndicate that the parameter value of mean value Gauss Distribution Fitting curve, exp () are indicated using natural radix e as the index letter at bottom
Number, μδAnd σδIndicate the parameter value of standard deviation Gauss Distribution Fitting curve, μθAnd σθIndicate the ginseng of degree of bias Gauss Distribution Fitting curve
Numerical value, μκAnd λκIndicate the parameter value of kurtosis Gauss Distribution Fitting curve, μηAnd σηIndicate the parameter of entropy Gauss Distribution Fitting curve
Value；
1. _ 2c, by fρ、fδ、fθ、fκAnd fηArranged in sequence obtains Wherein, symbol
" [] " is vector representation symbol.
The step 1. _ 3 inAcquisition process are as follows:
1. _ 3a, generalIt is denoted as { R respectively in three components of RGB colork(x,y)}、{Gk(x, y) } and { Bk(x,
Y) }, wherein 1≤x≤W, 1≤y≤H, Rk(x, y) indicates { Rk(x, y) } in coordinate position be (x, y) pixel pixel
Value, Gk(x, y) indicates { Gk(x, y) } in coordinate position be (x, y) pixel pixel value, Bk(x, y) indicates { Bk(x,y)}
Middle coordinate position is the pixel value of the pixel of (x, y)；
1. _ 3b, to { Rk(x, y) } operation is normalized, by { Rk(x, y) } obtained after normalization operation image note
ForIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { Gk(x, y) } operation is normalized, by { Gk(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { Bk(x, y) } operation is normalized, by { Bk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Rk(x, y) } in all pixels point pixel value mean value, Indicate { Rk(x, y) } in all pixels point pixel value standard deviation, Indicate { Gk(x, y) } in all pixels point pixel value mean value, Indicate { Gk(x, y) } in all pixels point pixel value standard deviation, Indicate { Bk(x, y) } in all pixels point pixel value mean value, Indicate { Bk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3c, using Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gR(h),And use Generalized Gaussian
Distributed model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gG(h),Using Generalized Gaussian Distribution Model pairDistribution of color into
Row fitting, fitting obtainMatched curve, be denoted as gB(h),
Wherein, 0≤h≤255, αRIndicate matched curve gR(h) scale parameter, βRIndicate matched curve gR(h) form parameter, exp
() indicates using natural radix e as the exponential function at bottom, symbol " | | " it is the symbol that takes absolute value,t
For integration variable, αGIndicate matched curve gG(h) scale parameter, βGIndicate matched curve gG(h) form parameter,αBIndicate matched curve gB(h) scale parameter, βBIndicate matched curve gB(h) shape ginseng
Number,
1. _ 3d, generalIt is denoted as { L respectively in three components of CIELAB color spacek(x,y)}、{ak(x, y) } and { bk
(x, y) }, wherein Lk(x, y) indicates { Lk(x, y) } in coordinate position be (x, y) pixel pixel value, ak(x, y) is indicated
{ak(x, y) } in coordinate position be (x, y) pixel pixel value, bk(x, y) indicates { bk(x, y) } in coordinate position be (x,
Y) pixel value of pixel；
1. _ 3e, to { Lk(x, y) } operation is normalized, by { Lk(x, y) } obtained after normalization operation image note
ForIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { ak(x, y) } operation is normalized, by { ak(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { bk(x, y) } operation is normalized, by { bk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Lk(x, y) } in all pixels point pixel value mean value, Indicate { Lk(x, y) } in all pixels point pixel value standard deviation, Indicate { ak(x, y) } in all pixels point pixel value mean value, Indicate { ak(x, y) } in all pixels point pixel value standard deviation, Indicate { bk(x, y) } in all pixels point pixel value mean value, Indicate { bk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3f, using Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gL(h),And using Generalized Gaussian point
Cloth model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as ga(h),Using Generalized Gaussian Distribution Model pairDistribution of color carry out
Fitting, fitting obtainMatched curve, be denoted as gb(h),Its
In, αLIndicate matched curve gL(h) scale parameter, βLIndicate matched curve gL(h) form parameter,αaIndicate matched curve ga(h) scale parameter, βaIndicate matched curve ga(h) shape ginseng
Number,αbIndicate matched curve gb(h) scale parameter, βbIndicate matched curve gb(h) shape
Parameter,
1. _ 3g, generalIt is denoted as { Y respectively in three components of YCbCr color spacek(x,y)}、{Uk(x, y) } and { Vk
(x, y) }, wherein Yk(x, y) indicates { Yk(x, y) } in coordinate position be (x, y) pixel pixel value, Uk(x, y) is indicated
{Uk(x, y) } in coordinate position be (x, y) pixel pixel value, Vk(x, y) indicates { Vk(x, y) } in coordinate position be (x,
Y) pixel value of pixel；
1. _ 3h, to { Yk(x, y) } operation is normalized, by { Yk(x, y) } obtained after normalization operation image note
ForIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { Uk(x, y) } operation is normalized, by { Uk(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { Vk(x, y) } operation is normalized, by { Vk(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Yk(x, y) } in all pixels point pixel value mean value, Indicate { Yk(x, y) } in all pixels point pixel value standard deviation, Indicate { Uk(x, y) } in all pixels point pixel value mean value, Indicate { Uk(x, y) } in all pixels point pixel value standard deviation, Indicate { Vk(x, y) } in all pixels point pixel value mean value, Indicate { Vk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3i, using Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gY(h),And use Generalized Gaussian
Distributed model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gU(h),Using Generalized Gaussian Distribution Model pairDistribution of color carry out
Fitting, fitting obtainMatched curve, be denoted as gV(h),Its
In, αYIndicate matched curve gY(h) scale parameter, βYIndicate matched curve gY(h) form parameter,αUIndicate matched curve gU(h) scale parameter, βUIndicate matched curve gU(h) shape ginseng
Number,αVIndicate matched curve gV(h) scale parameter, βVIndicate matched curve gV(h) shape
Parameter,
1. _ 3j, by αR、βR、αG、βG、αB、βB、αL、βL、αa、βa、αb、βb、αY、βY、αU、βU、αVAnd βVArranged in sequence obtains Wherein, symbol " [] "
For vector representation symbol.
Compared with the prior art, the advantages of the present invention are as follows:
The method of the present invention considers the influence of natural scene statistical nature and Color Statistical feature to tone mapping, extracts
The global characteristics vector of tone mapping image, all tone mapping figures that then training image is concentrated using support vector regression
The global characteristics vector of picture is trained, and constructs quality prediction model；In test phase, reflected by calculating the tone for being used as and testing
The global characteristics vector of image is penetrated, and according to the quality prediction model that the training stage constructs, prediction obtains the tone mapping image
Objective Quality Assessment predicted value and can be preferably since the global characteristics Vector Message of acquisition has stronger stability
Reflect the quality change situation of tone mapping image, therefore effectively improves the phase objectively evaluated between result and subjective perception
Guan Xing.
Detailed description of the invention
Fig. 1 is that the overall of the method for the present invention realizes block diagram.
Specific embodiment
The present invention will be described in further detail below with reference to the embodiments of the drawings.
A kind of tone mapping method for objectively evaluating image quality based on global characteristics proposed by the present invention, it is overall to realize
Block diagram is as shown in Figure 1 comprising two processes of training stage and test phase.
The specific steps of the training stage process are as follows:
1. _ 1, choosing N width tone mapping image construction training image collection, it is denoted asWherein, N > 1,
120 width tone mapping images in TMID database are chosen in the present embodiment, choose 1811 width colors in ESPL-LIVE database
It adjusts and maps image, 1≤k≤N,It indicatesIn kth width tone mapping image,
In every width tone mapping image width be W, and height be H.
1. _ 2, calculatingIn every width tone mapping image natural scene statistics
(SceneNaturalness) characteristic vector, willNatural scene statistical nature vector be denoted asWherein,Dimension
It is 5 × 1.
In the present embodiment, step 1. _ 2 inAcquisition process are as follows:
1. _ 2a, calculatingIn every width tone mapping image in all pixels point pixel value it is equal
Value, willIn the mean value of pixel value of all pixels point be denoted as ρ,Then it calculatesIn every width tone mapping image in all pixels point pixel value standard deviation, willIn institute
There is the standard deviation of the pixel value of pixel to be denoted as δ,Then it calculates
In every width tone mapping image in all pixels point pixel value the degree of bias, willIn all pixels point pixel value
The degree of bias be denoted as θ,And it calculatesIn every width tone mapping image in
All pixels point pixel value kurtosis, willIn the kurtosis of pixel value of all pixels point be denoted as κ,It calculates againIn every width tone mapping image in all pixels
The entropy of the pixel value of point, willIn the entropy of pixel value of all pixels point be denoted as η,Wherein, 1≤
X≤W, 1≤y≤H,It indicatesMiddle coordinate position is the pixel value of the pixel of (x, y), 0≤g≤255, pgTable
ShowIn all pixels point pixel value in belong to the probability density function values of g-th of density value,
1. _ 2b, by pairIn all tone mapping images respectively in all pixels point picture
The mean value of element value carries out Gauss Distribution Fitting, and fitting obtainsIn all tone mapping images mean value it is high
Then this fitting of distribution curve obtains the match value of ρ according to the mean value Gauss Distribution Fitting curve that fitting obtains, is denoted as fρ,Equally, by rightIn all tone mapping images respectively
In the standard deviation of pixel value of all pixels point carry out Gauss Distribution Fitting, fitting obtainsIn it is all
The standard deviation Gauss Distribution Fitting curve of tone mapping image, the standard deviation Gauss Distribution Fitting curve then obtained according to fitting
The match value for obtaining δ, is denoted as fδ,By rightIn institute it is coloured
Adjust mapping image respectively in the degree of bias of pixel value of all pixels point carry out Gauss Distribution Fitting, fitting obtainsIn all tone mapping images degree of bias Gauss Distribution Fitting curve, then according to fitting obtain it is inclined
The match value that Gauss Distribution Fitting curve obtains θ is spent, f is denoted asθ,By rightIn all tone mapping images respectively in the kurtosis of pixel value of all pixels point to carry out Gaussian Profile quasi-
It closes, fitting obtainsIn all tone mapping images kurtosis Gauss Distribution Fitting curve, then according to quasi-
The match value that obtained kurtosis Gauss Distribution Fitting curve obtains κ is closed, f is denoted asκ,It is logical
It is right to crossIn all tone mapping images respectively in all pixels point pixel value entropy carry out Gauss minute
Cloth fitting, fitting obtainIn all tone mapping images entropy Gauss Distribution Fitting curve, then basis
It is fitted the match value that obtained entropy Gauss Distribution Fitting curve obtains η, is denoted as fη,
Wherein, μρAnd σρThe parameter value for indicating mean value Gauss Distribution Fitting curve, takes μ in the present embodimentρ=121.70, σρ=
36.11, exp () indicated using natural radix e as the exponential function at bottom, μδAnd σδIndicate the ginseng of standard deviation Gauss Distribution Fitting curve
Numerical value takes μ in the present embodimentδ=56.47, σδ=18.43, μθAnd σθIndicate the parameter value of degree of bias Gauss Distribution Fitting curve,
μ is taken in the present embodimentθ=0.15, σθ=0.89, μκAnd λκThe parameter value for indicating kurtosis Gauss Distribution Fitting curve, in this reality
It applies and takes μ in exampleκ=2.82, λκ=18.86, μηAnd σηThe parameter value for indicating entropy Gauss Distribution Fitting curve, takes in the present embodiment
μη=7.56, ση=0.27.
1. _ 2c, by fρ、fδ、fθ、fκAnd fηArranged in sequence obtains Wherein, symbol
" [] " is vector representation symbol.
1. _ 3, calculatingIn every width tone mapping image Color Statistical (Chromatic
Information) characteristic vector, willColor Statistical characteristic vector be denoted asWherein,Dimension be 18 × 1.
In the present embodiment, step 1. _ 3 inAcquisition process are as follows:
1. _ 3a, generalIt is denoted as { R respectively in three components of RGB colork(x,y)}、{Gk(x, y) } and { Bk(x,
Y) }, wherein 1≤x≤W, 1≤y≤H, Rk(x, y) indicates { Rk(x, y) } in coordinate position be (x, y) pixel pixel
Value, Gk(x, y) indicates { Gk(x, y) } in coordinate position be (x, y) pixel pixel value, Bk(x, y) indicates { Bk(x,y)}
Middle coordinate position is the pixel value of the pixel of (x, y).
1. _ 3b, to { Rk(x, y) } operation is normalized, by { Rk(x, y) } obtained after normalization operation image note
ForIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { Gk(x, y) } operation is normalized, by { Gk(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { Bk(x, y) } operation is normalized, by { Bk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Rk(x, y) } in all pixels point pixel value mean value, Indicate { Rk(x, y) } in all pixels point pixel value standard deviation, Indicate { Gk(x, y) } in all pixels point pixel value mean value, Indicate { Gk(x, y) } in all pixels point pixel value standard deviation, Indicate { Bk(x, y) } in all pixels point pixel value mean value, Indicate { Bk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3c, using existing generalized Gaussian distribution (GGD) model pairDistribution of color be fitted, intend
Conjunction obtainsMatched curve, be denoted as gR(h),And using existing
Some generalized Gaussian distribution (GGD) models pairDistribution of color be fitted, fitting obtainsFitting
Curve is denoted as gG(h),Using existing generalized Gaussian distribution (GGD) mould
Type pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gB(h),Wherein, 0≤h≤255, αRIndicate matched curve gR(h) scale ginseng
Number, βRIndicate matched curve gR(h) form parameter, exp () indicate using natural radix e as the exponential function at bottom, symbol " | | "
For the symbol that takes absolute value,T is integration variable, αGIndicate matched curve gG(h) scale parameter,
βGIndicate matched curve gG(h) form parameter,αBIndicate matched curve gB(h) scale ginseng
Number, βBIndicate matched curve gB(h) form parameter,
1. _ 3d, generalIt is denoted as { L respectively in three components of CIELAB color spacek(x,y)}、{ak(x, y) } and { bk
(x, y) }, wherein Lk(x, y) indicates { Lk(x, y) } in coordinate position be (x, y) pixel pixel value, ak(x, y) is indicated
{ak(x, y) } in coordinate position be (x, y) pixel pixel value, bk(x, y) indicates { bk(x, y) } in coordinate position be (x,
Y) pixel value of pixel.
1. _ 3e, to { Lk(x, y) } operation is normalized, by { Lk(x, y) } obtained after normalization operation image note
ForIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { ak(x, y) } operation is normalized, by { ak(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { bk(x, y) } operation is normalized, by { bk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Lk(x, y) } in all pixels point pixel value mean value, Indicate { Lk(x, y) } in all pixels point pixel value standard deviation, Indicate { ak(x, y) } in all pixels point pixel value mean value, Indicate { ak(x, y) } in all pixels point pixel value standard deviation, Indicate { bk(x, y) } in all pixels point pixel value mean value, Indicate { bk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3f, using existing generalized Gaussian distribution (GGD) model pairDistribution of color be fitted, intend
Conjunction obtainsMatched curve, be denoted as gL(h),And using existing
Some generalized Gaussian distribution (GGD) models pairDistribution of color be fitted, fitting obtainsFitting
Curve is denoted as ga(h),Using existing generalized Gaussian distribution (GGD) mould
Type pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gb(h),Wherein, αLIndicate matched curve gL(h) scale parameter, βLIndicate fitting
Curve gL(h) form parameter,αaIndicate matched curve ga(h) scale parameter, βaIndicate quasi-
Close curve ga(h) form parameter,αbIndicate matched curve gb(h) scale parameter, βbIt indicates
Matched curve gb(h) form parameter,
1. _ 3g, generalIt is denoted as { Y respectively in three components of YCbCr color spacek(x,y)}、{Uk(x, y) } and { Vk
(x, y) }, wherein Yk(x, y) indicates { Yk(x, y) } in coordinate position be (x, y) pixel pixel value, Uk(x, y) is indicated
{Uk(x, y) } in coordinate position be (x, y) pixel pixel value, Vk(x, y) indicates { Vk(x, y) } in coordinate position be (x,
Y) pixel value of pixel.
1. _ 3h, to { Yk(x, y) } operation is normalized, by { Yk(x, y) } obtained after normalization operation image note
ForIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { Uk(x, y) } operation is normalized, by { Uk(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { Vk(x, y) } operation is normalized, by { Vk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Yk(x, y) } in all pixels point pixel value mean value, Indicate { Yk(x, y) } in all pixels point pixel value standard deviation, Indicate { Uk(x, y) } in all pixels point pixel value mean value, Indicate { Uk(x, y) } in all pixels point pixel value standard deviation, Indicate { Vk(x, y) } in all pixels point pixel value mean value, Indicate { Vk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3i, using existing generalized Gaussian distribution (GGD) model pairDistribution of color be fitted, intend
Conjunction obtainsMatched curve, be denoted as gY(h),And using existing
Generalized Gaussian distribution (GGD) model pairDistribution of color be fitted, fitting obtainsFitting it is bent
Line is denoted as gU(h),Using existing generalized Gaussian distribution (GGD) model
It is rightDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gV(h),Wherein, αYIndicate matched curve gY(h) scale parameter, βYIndicate fitting
Curve gY(h) form parameter,αUIndicate matched curve gU(h) scale parameter, βUIt indicates
Matched curve gU(h) form parameter,αVIndicate matched curve gV(h) scale parameter, βVTable
Show matched curve gV(h) form parameter,
1. _ 3j, by αR、βR、αG、βG、αB、βB、αL、βL、αa、βa、αb、βb、αY、βY、αU、βU、αVAnd βVArranged in sequence obtains Wherein, symbol " [] "
For vector representation symbol.
1. _ 4, willIn every width tone mapping image natural scene statistical nature vector sum color
Statistical nature vector is constitutedIn every width tone mapping image global characteristics vector, willThe overall situation
Characteristic vector is denoted as Fk,Wherein, FkDimension be 23 × 1, symbol " [] " be vector representation symbol,Indicating willWithIt connects to form a global characteristics vector.
1. _ 5, willIn the respective global characteristics vector sum mean subjective of all tone mapping images
Score difference composing training sample data sets, includes the N number of average master of N number of global characteristics vector sum in training sample data set
See scoring difference；Then method of the support vector regression as machine learning is used, to all in training sample data set
Global characteristics vector is trained, so that by the error between the obtained regression function value of training and mean subjective scoring difference
Minimum, fitting obtain optimal weight vector woptWith optimal bias term bopt；Followed by optimal weight vector woptMost
Excellent bias term bopt, quality prediction model is constructed, f (F) is denoted as,Wherein, f () is function table
Show form, F is used to indicate the global characteristics vector of tone mapping image, and the input vector as quality prediction model, (wopt
)TFor woptTransposition,For the linear function of F.
The specific steps of the test phase process are as follows:
2. being used as the tone mapping image I of test for any one widthtest, 1. _ 2 1. _ 4 identical to step according to step
Operation obtains ItestGlobal characteristics vector, be denoted as Ftest；Then according to the quality prediction model of training stage construction to Ftest
It is tested, prediction obtains FtestCorresponding predicted value, using the predicted value as ItestObjective Quality Assessment predicted value, be denoted as
Qtest,Wherein, ItestWidth be W', and height be H', W' can be identical as W or not identical,
H' can be identical as H or not identical, FtestDimension be 23 × 1,Indicate FtestLinear function.
In the present embodiment, difficult to understand using the TMID database and Texas ,Usa university of Canadian University of Waterloo (CA) Waterloo, Ontario, N2L3GI Canada foundation
For the ESPL-LIVE database that sting branch school is established as tone mapping image data base, TMID database includes that 120 width tones reflect
Image is penetrated, ESPL-LIVE database includes 1811 width tone mapping images.It is normal using 2 of assessment image quality evaluating method
Use objective parameter as evaluation index, i.e., Pearson linearly dependent coefficient (the Pearson linear under the conditions of nonlinear regression
Correlation coefficient, PLCC) and Spearman order coefficient of rank correlation (Spearman rank order
Correlation coefficient, SROCC).The higher evaluation result for illustrating the method for the present invention of PLCC and SROCC and average
The correlation of subjective scoring difference is better.Table 1 gives the Objective Quality Assessment predicted value and average master that the method for the present invention obtains
The correlation between scoring difference is seen, from table 1 it follows that the quality of the tone mapping image obtained using the method for the present invention
Objectively evaluate predicted value and mean subjective scoring difference between correlation be it is very high, show to objectively evaluate result and human eye master
The result that perception is known is more consistent, it is sufficient to illustrate the validity of the method for the present invention.
It is related between the Objective Quality Assessment predicted value that table 1 uses the method for the present invention to obtain and mean subjective scoring difference
Property
| Database | PLCC | SROCC |
| TMID | 0.744 | 0.698 |
| ESPL-LIVE | 0.639 | 0.629 |
Claims (2)
1. a kind of tone mapping method for objectively evaluating image quality based on global characteristics, it is characterised in that including the training stage and
Two processes of test phase；
The specific steps of the training stage process are as follows:
1. _ 1, choosing N width tone mapping image construction training image collection, it is denoted asWherein, N > 1,1≤k≤
N,It indicatesIn kth width tone mapping image,In every width tone mapping figure
The width of picture is W, and height is H；
1. _ 2, calculatingIn every width tone mapping image natural scene statistical nature vector, will's
Natural scene statistical nature vector is denoted asWherein,Dimension be 5 × 1；
The step 1. _ 2 inAcquisition process are as follows:
1. _ 2a, calculatingIn every width tone mapping image in all pixels point pixel value mean value,
It willIn the mean value of pixel value of all pixels point be denoted as ρ,Then it calculatesIn every width tone mapping image in all pixels point pixel value standard deviation, willIn institute
There is the standard deviation of the pixel value of pixel to be denoted as δ,Then it calculates
In every width tone mapping image in all pixels point pixel value the degree of bias, willIn all pixels point pixel value
The degree of bias be denoted as θ,And it calculatesIn every width tone mapping image in
All pixels point pixel value kurtosis, willIn the kurtosis of pixel value of all pixels point be denoted as κ,It calculates againIn every width tone mapping image in all pixels
The entropy of the pixel value of point, willIn the entropy of pixel value of all pixels point be denoted as η,Wherein, 1
≤ x≤W, 1≤y≤H,It indicatesMiddle coordinate position is the pixel value of the pixel of (x, y), 0≤g≤255, pg
It indicatesIn all pixels point pixel value in belong to the probability density function values of g-th of density value,
1. _ 2b, by pairIn all tone mapping images respectively in all pixels points pixel value
Mean value carries out Gauss Distribution Fitting, and fitting obtainsIn all tone mapping images mean value Gaussian Profile
Then matched curve obtains the match value of ρ according to the mean value Gauss Distribution Fitting curve that fitting obtains, is denoted as fρ,Equally, by rightIn all tone mapping images respectively in
The standard deviation of pixel value of all pixels point carry out Gauss Distribution Fitting, fitting obtainsIn institute it is coloured
The standard deviation Gauss Distribution Fitting curve of mapping image is adjusted, is then obtained according to the standard deviation Gauss Distribution Fitting curve that fitting obtains
The match value for obtaining δ, is denoted as fδ,By rightIn all tones
Map image respectively in the degree of bias of pixel value of all pixels point carry out Gauss Distribution Fitting, fitting obtainsIn all tone mapping images degree of bias Gauss Distribution Fitting curve, then according to fitting obtain it is inclined
The match value that Gauss Distribution Fitting curve obtains θ is spent, f is denoted asθ,By rightIn all tone mapping images respectively in all pixels point pixel value kurtosis carry out Gaussian Profile
Fitting, fitting obtainIn all tone mapping images kurtosis Gauss Distribution Fitting curve, then according to quasi-
The match value that obtained kurtosis Gauss Distribution Fitting curve obtains κ is closed, f is denoted asκ,
By rightIn all tone mapping images respectively in all pixels point pixel value entropy carry out Gauss
Fitting of distribution, fitting obtainIn all tone mapping images entropy Gauss Distribution Fitting curve, then basis
It is fitted the match value that obtained entropy Gauss Distribution Fitting curve obtains η, is denoted as fη,
Wherein, μρAnd σρIndicate that the parameter value of mean value Gauss Distribution Fitting curve, exp () are indicated using natural radix e as the index letter at bottom
Number, μδAnd σδIndicate the parameter value of standard deviation Gauss Distribution Fitting curve, μθAnd σθIndicate the ginseng of degree of bias Gauss Distribution Fitting curve
Numerical value, μκAnd λκIndicate the parameter value of kurtosis Gauss Distribution Fitting curve, μηAnd σηIndicate the parameter of entropy Gauss Distribution Fitting curve
Value；
1. _ 2c, by fρ、fδ、fθ、fκAnd fηArranged in sequence obtains Wherein, symbol " [] "
For vector representation symbol；
1. _ 3, calculatingIn every width tone mapping image Color Statistical characteristic vector, willColor
Statistical nature vector is denoted asWherein,Dimension be 18 × 1；
1. _ 4, willIn every width tone mapping image natural scene statistical nature vector sum Color Statistical
Characteristic vector is constitutedIn every width tone mapping image global characteristics vector, willGlobal characteristics
Vector is denoted as Fk,Wherein, FkDimension be 23 × 1, symbol " [] " be vector representation symbol,
Indicating willWithIt connects to form a global characteristics vector；
1. _ 5, willIn the respective global characteristics vector sum mean subjective scoring of all tone mapping images
Difference composing training sample data sets are commented comprising N number of N number of mean subjective of global characteristics vector sum in training sample data set
Divide difference；Then method of the support vector regression as machine learning is used, to all overall situations in training sample data set
Characteristic vector is trained, so that most by the error between the obtained regression function value of training and mean subjective scoring difference
Small, fitting obtains optimal weight vector woptWith optimal bias term bopt；Followed by optimal weight vector woptWith it is optimal
Bias term bopt, quality prediction model is constructed, f (F) is denoted as,Wherein, f () is function representation
Form, F are used to indicate the global characteristics vector of tone mapping image, and the input vector as quality prediction model, (wopt)T
For woptTransposition,For the linear function of F；
The specific steps of the test phase process are as follows:
2. being used as the tone mapping image I of test for any one widthtest, according to step 1. _ 2 to step 1. _ 4 identical operation,
Obtain ItestGlobal characteristics vector, be denoted as Ftest；Then according to the quality prediction model of training stage construction to FtestIt is surveyed
Examination, prediction obtain FtestCorresponding predicted value, using the predicted value as ItestObjective Quality Assessment predicted value, be denoted as Qtest,Wherein, ItestWidth be W', and height be H', FtestDimension be 23 × 1,Table
Show FtestLinear function.
2. a kind of tone mapping method for objectively evaluating image quality based on global characteristics according to claim 1, special
1. _ 3 sign is the step inAcquisition process are as follows:
1. _ 3a, generalIt is denoted as { R respectively in three components of RGB colork(x,y)}、{Gk(x, y) } and { Bk(x, y) },
Wherein, 1≤x≤W, 1≤y≤H, Rk(x, y) indicates { Rk(x, y) } in coordinate position be (x, y) pixel pixel value, Gk
(x, y) indicates { Gk(x, y) } in coordinate position be (x, y) pixel pixel value, Bk(x, y) indicates { Bk(x, y) } in coordinate
Position is the pixel value of the pixel of (x, y)；
1. _ 3b, to { Rk(x, y) } operation is normalized, by { Rk(x, y) } image that obtains after normalization operation is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { Gk(x, y) } operation is normalized, by { Gk(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { Bk(x, y) } operation is normalized, by { Bk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Rk(x, y) } in all pixels point pixel value mean value, Indicate { Rk(x, y) } in all pixels point pixel value standard deviation, Indicate { Gk(x, y) } in all pixels point pixel value mean value, Indicate { Gk(x, y) } in all pixels point pixel value standard deviation, Indicate { Bk(x, y) } in all pixels point pixel value mean value, Indicate { Bk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3c, using Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtains's
Matched curve is denoted as gR(h),And use Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gG(h),Using Generalized Gaussian Distribution Model pairDistribution of color into
Row fitting, fitting obtainMatched curve, be denoted as gB(h),
Wherein, 0≤h≤255, αRIndicate matched curve gR(h) scale parameter, βRIndicate matched curve gR(h) form parameter, exp
() indicates using natural radix e as the exponential function at bottom, symbol " | | " it is the symbol that takes absolute value,
T is integration variable, αGIndicate matched curve gG(h) scale parameter, βGIndicate matched curve gG(h) form parameter,αBIndicate matched curve gB(h) scale parameter, βBIndicate matched curve gB(h) shape ginseng
Number,
1. _ 3d, generalIt is denoted as { L respectively in three components of CIELAB color spacek(x,y)}、{ak(x, y) } and { bk(x,
Y) }, wherein Lk(x, y) indicates { Lk(x, y) } in coordinate position be (x, y) pixel pixel value, ak(x, y) indicates { ak
(x, y) } in coordinate position be (x, y) pixel pixel value, bk(x, y) indicates { bk(x, y) } in coordinate position be (x, y)
Pixel pixel value；
1. _ 3e, to { Lk(x, y) } operation is normalized, by { Lk(x, y) } image that obtains after normalization operation is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { ak(x, y) } operation is normalized, by { ak(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { bk(x, y) } operation is normalized, by { bk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,It indicatesIn all pixels point pixel value mean value, Indicate { Lk(x, y) } in all pixels point pixel value standard deviation, Indicate { ak(x, y) } in all pixels point pixel value mean value, Indicate { ak(x, y) } in all pixels point pixel value standard deviation, Indicate { bk(x, y) } in all pixels point pixel value mean value, Indicate { bk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3f, using Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtains's
Matched curve is denoted as gL(h),And use Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as ga(h),Using Generalized Gaussian Distribution Model pairDistribution of color carry out
Fitting, fitting obtainMatched curve, be denoted as gb(h),Its
In, αLIndicate matched curve gL(h) scale parameter, βLIndicate matched curve gL(h) form parameter,αaIndicate matched curve ga(h) scale parameter, βaIndicate matched curve ga(h) shape ginseng
Number,αbIndicate matched curve gb(h) scale parameter, βbIndicate matched curve gb(h) shape
Parameter,
1. _ 3g, generalIt is denoted as { Y respectively in three components of YCbCr color spacek(x,y)}、{Uk(x, y) } and { Vk(x,
Y) }, wherein Yk(x, y) indicates { Yk(x, y) } in coordinate position be (x, y) pixel pixel value, Uk(x, y) indicates { Uk
(x, y) } in coordinate position be (x, y) pixel pixel value, Vk(x, y) indicates { Vk(x, y) } in coordinate position be (x, y)
Pixel pixel value；
1. _ 3h, to { Yk(x, y) } operation is normalized, by { Yk(x, y) } image that obtains after normalization operation is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as And to { Uk(x, y) } operation is normalized, by { Uk(x, y) } it is obtained after normalization operation
Image be denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as To { Vk(x, y) } operation is normalized, by { Vk(x, y) } obtained after normalization operation
Image is denoted asIt willMiddle coordinate position is that the pixel value of the pixel of (x, y) is denoted as Wherein,Indicate { Yk(x, y) } in all pixels point pixel value mean value, Indicate { Yk(x, y) } in all pixels point pixel value standard deviation, Indicate { Uk(x, y) } in all pixels point pixel value mean value, Indicate { Uk(x, y) } in all pixels point pixel value standard deviation, Indicate { Vk(x, y) } in all pixels point pixel value mean value, Indicate { Vk(x, y) } in all pixels point pixel value standard deviation,
1. _ 3i, using Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtains's
Matched curve is denoted as gY(h),And use Generalized Gaussian Distribution Model pairDistribution of color be fitted, fitting obtainsMatched curve, be denoted as gU(h),Using Generalized Gaussian Distribution Model pairDistribution of color carry out
Fitting, fitting obtainMatched curve, be denoted as gV(h),Its
In, αYIndicate matched curve gY(h) scale parameter, βYIndicate matched curve gY(h) form parameter,αUIndicate matched curve gU(h) scale parameter, βUIndicate matched curve gU(h) shape ginseng
Number,αVIndicate matched curve gV(h) scale parameter, βVIndicate matched curve gV(h) shape
Parameter,
1. _ 3j, by αR、βR、αG、βG、αB、βB、αL、βL、αa、βa、αb、βb、αY、βY、αU、βU、αVAnd βVArranged in sequence obtains Wherein, symbol " [] " is arrow
Amount indicates symbol.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710164242.8A CN107105223B (en) | 2017-03-20 | 2017-03-20 | A kind of tone mapping method for objectively evaluating image quality based on global characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710164242.8A CN107105223B (en) | 2017-03-20 | 2017-03-20 | A kind of tone mapping method for objectively evaluating image quality based on global characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107105223A CN107105223A (en) | 2017-08-29 |
| CN107105223B true CN107105223B (en) | 2018-12-07 |
Family
ID=59675339
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710164242.8A Active CN107105223B (en) | 2017-03-20 | 2017-03-20 | A kind of tone mapping method for objectively evaluating image quality based on global characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107105223B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108322733B (en) * | 2018-01-17 | 2019-09-03 | 宁波大学 | A kind of no reference high dynamic range images method for evaluating objective quality |
| CN109218716B (en) * | 2018-10-22 | 2020-11-06 | 天津大学 | No-reference tone mapping image quality evaluation method based on color statistics and information entropy |
| CN109788275A (en) * | 2018-12-28 | 2019-05-21 | 天津大学 | Naturality, structure and binocular asymmetry are without reference stereo image quality evaluation method |
| CN109919959B (en) * | 2019-01-24 | 2023-01-20 | 天津大学 | Tone mapping image quality evaluation method based on color, naturalness and structure |
| CN114223015A (en) * | 2019-08-15 | 2022-03-22 | 杜比实验室特许公司 | Efficient user-defined SDR to HDR conversion using model templates |
| CN110717892B (en) * | 2019-09-18 | 2022-06-28 | 宁波大学 | Tone mapping image quality evaluation method |
| CN110910346A (en) * | 2019-10-17 | 2020-03-24 | 浙江工商职业技术学院 | Tone mapping image quality evaluation method based on dense scale invariant feature transformation |
| CN113099215B (en) * | 2021-03-19 | 2022-06-21 | 宁波大学 | Cartoon image quality evaluation method |
| CN117041531B (en) * | 2023-09-04 | 2024-03-15 | 无锡维凯科技有限公司 | Mobile phone camera focusing detection method and system based on image quality evaluation |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103338380A (en) * | 2013-06-06 | 2013-10-02 | 宁波大学 | Adaptive image quality objective evaluation method |
| CN105741328A (en) * | 2016-01-22 | 2016-07-06 | 西安电子科技大学 | Shot image quality evaluation method based on visual perception |
| CN105825500A (en) * | 2016-03-10 | 2016-08-03 | 江苏商贸职业学院 | Camera image quality evaluation method and device |
-
2017
- 2017-03-20 CN CN201710164242.8A patent/CN107105223B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103338380A (en) * | 2013-06-06 | 2013-10-02 | 宁波大学 | Adaptive image quality objective evaluation method |
| CN105741328A (en) * | 2016-01-22 | 2016-07-06 | 西安电子科技大学 | Shot image quality evaluation method based on visual perception |
| CN105825500A (en) * | 2016-03-10 | 2016-08-03 | 江苏商贸职业学院 | Camera image quality evaluation method and device |
Non-Patent Citations (3)
| Title |
|---|
| SUPPORTING BINOCULAR VISUAL QUALITY PREDICTION USING MACHINE LEARNING;Shanshan Wang;《Multimedia and Expo Workshops (ICMEW), 2014 IEEE International Conference on》;20140718;第2.2节 * |
| 基于双目特征联合的无参考立体图像质量评价;李柯蒙;《光电子.激光》;20151130;第26卷(第11期);全文 * |
| 高动态范围图像客观质量评价方法;管非凡;《计算机应用》;20170310;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107105223A (en) | 2017-08-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107105223B (en) | A kind of tone mapping method for objectively evaluating image quality based on global characteristics | |
| CN110046673B (en) | No-reference tone mapping image quality evaluation method based on multi-feature fusion | |
| CN107172418B (en) | A kind of tone scale map image quality evaluating method based on exposure status analysis | |
| CN103996192B (en) | Non-reference image quality evaluation method based on high-quality natural image statistical magnitude model | |
| CN108010024B (en) | Blind reference tone mapping image quality evaluation method | |
| CN106709958A (en) | Gray scale gradient and color histogram-based image quality evaluation method | |
| CN108428227A (en) | Non-reference picture quality appraisement method based on full convolutional neural networks | |
| CN108090902A (en) | A kind of non-reference picture assessment method for encoding quality based on multiple dimensioned generation confrontation network | |
| CN105261013B (en) | A kind of scan image quality overall evaluation method and evaluation system | |
| CN105574901B (en) | A kind of general non-reference picture quality appraisement method based on local contrast pattern | |
| CN107040775B (en) | A kind of tone mapping method for objectively evaluating image quality based on local feature | |
| CN107396095B (en) | A kind of no reference three-dimensional image quality evaluation method | |
| CN104376565B (en) | Based on discrete cosine transform and the non-reference picture quality appraisement method of rarefaction representation | |
| CN104036493B (en) | No-reference image quality evaluation method based on multifractal spectrum | |
| CN109191460B (en) | Quality evaluation method for tone mapping image | |
| CN107146220B (en) | A kind of universal non-reference picture quality appraisement method | |
| CN110516716A (en) | Non-reference picture quality appraisement method based on multiple-limb similarity network | |
| CN102722888A (en) | Stereoscopic image objective quality evaluation method based on physiological and psychological stereoscopic vision | |
| CN109859166A (en) | It is a kind of based on multiple row convolutional neural networks without ginseng 3D rendering method for evaluating quality | |
| CN109191450A (en) | A kind of remote sensing image fusion quality evaluating method | |
| CN110717892B (en) | Tone mapping image quality evaluation method | |
| CN106709504A (en) | Detail-preserving high fidelity tone mapping method | |
| CN105976351A (en) | Central offset based three-dimensional image quality evaluation method | |
| CN112950596A (en) | Tone mapping omnidirectional image quality evaluation method based on multi-region and multi-layer | |
| CN107018410A (en) | A kind of non-reference picture quality appraisement method based on pre- attention mechanism and spatial dependence |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20201202 Address after: Room 1,020, Nanxun Science and Technology Pioneering Park, No. 666 Chaoyang Road, Nanxun District, Huzhou City, Zhejiang Province, 313000 Patentee after: Huzhou You Yan Intellectual Property Service Co.,Ltd. Address before: 315211 Zhejiang Province, Ningbo Jiangbei District Fenghua Road No. 818 Patentee before: Ningbo University |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20221221 Address after: Floor 4, Unit 1, Building J, No. 6, Zhongtai Road, Loucun Community, Xinhu Street, Guangming District, Shenzhen, Guangdong 518107 Patentee after: Shenzhen Weishi Photoelectric Technology Co.,Ltd. Address before: 313000 room 1020, science and Technology Pioneer Park, 666 Chaoyang Road, Nanxun Town, Nanxun District, Huzhou, Zhejiang. Patentee before: Huzhou You Yan Intellectual Property Service Co.,Ltd. |
|
| TR01 | Transfer of patent right |