CN111598829B - Image quality evaluation method and device based on extended Prewitt operator - Google Patents
Image quality evaluation method and device based on extended Prewitt operator Download PDFInfo
- Publication number
- CN111598829B CN111598829B CN202010090192.5A CN202010090192A CN111598829B CN 111598829 B CN111598829 B CN 111598829B CN 202010090192 A CN202010090192 A CN 202010090192A CN 111598829 B CN111598829 B CN 111598829B
- Authority
- CN
- China
- Prior art keywords
- image
- reference image
- distorted
- factor
- gradient
- 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
- 238000000034 method Methods 0.000 title claims abstract description 40
- 238000013441 quality evaluation Methods 0.000 title claims abstract description 32
- 238000011156 evaluation Methods 0.000 claims abstract description 52
- 238000007781 pre-processing Methods 0.000 claims abstract description 20
- 238000012545 processing Methods 0.000 claims description 22
- 238000001914 filtration Methods 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 8
- 108010063499 Sigma Factor Proteins 0.000 claims description 4
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 claims description 3
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 claims description 3
- 230000007547 defect Effects 0.000 abstract description 4
- 238000004422 calculation algorithm Methods 0.000 description 24
- 238000010586 diagram Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 238000001303 quality assessment method Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 238000005562 fading Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000003646 Spearman's rank correlation coefficient Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000005266 casting Methods 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
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
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Abstract
The invention provides an image quality evaluation method and device based on an extended Prewitt operator, wherein the method comprises the steps of obtaining a reference image and a distortion image to be evaluated; preprocessing the reference image and the distorted image, and determining brightness factors and contrast factors of the reference image and the distorted image; performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image; determining gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value; and determining an objective evaluation result of the image quality by combining the gradient similarity factor, the brightness factor and the contrast factor. The invention improves the defects of the traditional Prewitt operator, can evaluate the quality of the noise image, and has better consistency with the subjective evaluation result.
Description
Technical Field
The invention relates to the field of image processing and image quality evaluation, in particular to an image quality evaluation method and device based on an extended Prewitt operator.
Background
Image quality assessment (Image Quality Assessment, IQA) is a commonly used image processing technique by which image characteristics can be studied and image distortion can be determined. With the rapid popularity of digital image and communication technologies, image Quality Assessment (IQA) has become an important issue in many applications such as image acquisition, transmission, compression, recovery and enhancement. From the perspective of participants, image quality assessment methods can be categorized into subjective and objective assessment methods, and since subjective IQA methods are not readily applicable to many scenarios, such as real-time systems and automated systems, it is necessary to develop objective IQA indicators to automatically and robustly measure image quality. Meanwhile, the estimated evaluation result by using the objective IQA index should be consistent with the statistical result of human observers.
Currently, there are a variety of objective IQA methods for evaluating image quality, in which peak signal-to-noise ratio (PSNR) and Mean Square Error (MSE) are classical full-reference objective image quality evaluation methods. The two methods are easy to understand and convenient to calculate, but only consider the comparison of each pixel point of the image, and do not consider the structural relationship and the like existing among each pixel point of the image, and have deviation from being really seen by human eyes.
The existing structural similarity SSIM (Structural Similarity) algorithm comprehensively compares the difference between three different information of brightness, contrast and structural similarity of an original undistorted image and an image to be evaluated, and considers the structural relationship among pixels, but the problems of poor detail grasp, difficult index parameter determination and the like under the condition of serious blurring are solved.
Based on SSIM, a gradient information-based image quality judging method, namely gradient-based Structural Similarity-based structure similarity (GSSIM), is proposed, and the GSSIM model is more in line with the characteristics of a human eye vision system than the SSIM and PSNR (MSE) models in consideration of the fact that gradients are highly sensitive to image distortion, so that the quality of a blurred image can be well evaluated.
The GSSIM algorithm for detecting the image edge by adopting the traditional Prewitt operator has a smoothing effect on noise while detecting the edge, but only the gradient information in the horizontal direction and the vertical direction is considered, partial detail information can be lost, and the edge is blurred to a certain extent.
Disclosure of Invention
The invention aims to provide an image quality evaluation method and device based on an extended Prewitt operator, which are used for solving the problem that partial detail information is lost when the traditional Prewitt operator is adopted to detect the image edge, so that the image edge has a certain blur.
In order to solve the above problems, the present invention provides an image quality evaluation method based on an extended Prewitt operator, including:
acquiring a reference image and a distortion image to be evaluated;
preprocessing the reference image and the distorted image, and determining brightness factors and contrast factors of the reference image and the distorted image;
Performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image;
Determining gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value;
and combining the gradient similarity factor, the brightness factor and the contrast factor to determine an objective evaluation result of the image quality.
Optionally, preprocessing the reference image and the distorted image further includes:
Graying the reference image and the distorted image;
according to a filtering operator, filtering the reference image and the distorted image after the graying treatment; wherein the filter operator is built using an 11 x 11 gaussian template with a standard deviation of 1.5 pixels.
Optionally, determining the luminance factors of the reference image and the distorted image further includes: by means ofCalculating brightness factors of the reference image and the distorted image;
wherein, X represents the reference image, y represents the distorted image, l (x, y) is a brightness factor, mu x and mu y respectively represent average brightness of the reference image and the distorted image, a constant C1 is used for avoiding instability, N is the number of distorted images to be evaluated, i represents the ith image, x i represents the ith reference image, and y i represents the ith distorted image.
Optionally, determining the contrast factor of the reference image and the distorted image further comprises: by means ofCalculating contrast factors of the reference image and the distorted image;
wherein, X represents the reference image, y represents the distorted image, C (x, y) represents a contrast factor, σ x and σ y represent standard deviations of pixels of the reference image and the distorted image, σ xy represents covariance of the reference image and the distorted image, μ x and μ y represent average brightness of the reference image and the distorted image respectively, a constant C2 is used for avoiding instability, N is the number of distorted images to be evaluated, and i represents the ith image.
Optionally, performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator further includes that the extended Prewitt operator is an operator selecting eight directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 ° and 315 °, and an attenuation factor is introduced to eliminate the possibility of data overflow.
Optionally, determining the gradient similarity factor of the reference image and the distorted image according to the first gradient value and the second gradient value further includes:
by using Determining a gradient similarity factor of the reference image and the distorted image; where G x(i,j),Gy (i, j) is the gradient magnitude of the reference image x and the distorted image y at (i, j), respectively, a constant C3 is used to avoid instability.
Optionally, in combination with the gradient similarity factor, the brightness factor, and the contrast factor, determining an objective evaluation result of image quality further includes: determining a Prewitt-GSSIM index of the reference image and the distorted image by using Prewitt-GSSIM (x, y) = [ l (x, y) ] α·[c(x,y)]β·[g(x,y)]λ, wherein Prewitt-GSSIM (x, y) is a comprehensive Prewitt-GSSIM index, alpha >0, beta >0 and gamma >0, and alpha, beta and gamma are parameters for adjusting relative importance of three parts of brightness, contrast and gradient similarity respectively;
From the Prewitt-GSSIM index of each sub-image block, adopting
Determining an objective evaluation result of the quality of the whole image; and M is the number of the sliding overlapping blocks of the distortion image to be evaluated by adopting a filtering factor.
The invention also provides an image quality evaluation device based on the extended Prewitt operator, which comprises:
the image acquisition module is used for acquiring a reference image and a distortion image to be evaluated;
The image preprocessing module is used for preprocessing the reference image and the distorted image and determining brightness factors and contrast factors of the reference image and the distorted image;
The gradient value calculation module is used for carrying out convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image;
The gradient similarity factor calculation module is used for determining the gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value;
And the objective evaluation result determining module is used for combining the gradient similarity factor, the brightness factor and the contrast factor to determine an objective evaluation result of the image quality.
Optionally, the image preprocessing module further includes:
the graying processing submodule is used for carrying out graying processing on the reference image and the distorted image;
the filtering processing sub-module is used for carrying out filtering processing on the reference image and the distorted image after the graying processing according to a filtering operator; wherein the filter operator is built using an 11 x 11 gaussian template with a standard deviation of 1.5 pixels.
Optionally, the extended Prewitt operator is an operator for selecting eight directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, 315 °.
The invention provides an image quality evaluation method and device based on an extended Prewitt operator, wherein the image quality evaluation method based on the extended Prewitt operator is implemented by acquiring a reference image and a distortion image to be evaluated; preprocessing the reference image and the distorted image, and determining brightness factors and contrast factors of the reference image and the distorted image; performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image; determining gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value; and determining an objective evaluation result of the image quality by combining the gradient similarity factor, the brightness factor and the contrast factor. The invention correspondingly improves the defects of the traditional Prewitt operator, can evaluate the quality of the noise image, has better consistency with subjective evaluation results, and solves the problem that partial detail information is lost when the traditional Prewitt operator is adopted to detect the image edge, so that the image edge has a certain blur.
Drawings
FIG. 1 is a flow chart of an image quality evaluation method based on an extended Prewitt operator;
FIG. 2 is a schematic diagram of a direction template representation of a conventional Prewitt operator;
FIG. 3 is a schematic diagram of a direction template representation of a six-direction extended Prewitt operator provided by the present invention;
Fig. 4 is a scatter diagram of PSNR and subjective evaluation value DMOS;
Fig. 5 is a scatter diagram of SSIM and subjective evaluation value DMOS;
fig. 6 is a scatter diagram of GSSIM and subjective evaluation value DMOS;
FIG. 7 is a scatter plot of GSSIM-Prewitt and subjective evaluation value DMOS;
FIG. 8 is a scatter plot of Sobel-GSSIM and subjective evaluation value DMOS;
FIG. 9 is a scatter plot of the Prewitt-GSSIM and the subjective evaluation value DMOS;
Fig. 10 is a schematic structural diagram of an image quality evaluation device based on an extended Prewitt operator according to an embodiment of the present invention.
Detailed Description
In order that the invention may be readily understood, a more complete description of the invention will be rendered by reference to the appended drawings. Preferred embodiments of the present invention are shown in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and/or" as used herein includes any and all combinations of one or more of the associated listed items.
In the prior art, although the gray level picture clustering performance based on non-negative matrix factorization is good, the dimension reduction performance is not excellent, and the color picture clustering performance is not satisfactory; the dimension reduction performance based on non-negative tensor decomposition is outstanding, but the conventional non-negative tensor decomposition has strict limitation on the rank, so that the clustering effect of the color pictures with special structures is not satisfactory.
The inventor researches and discovers that the GSSIM algorithm adopts a gradient operator as a Sobel operator when the image edge is considered, and the common 3*3 gradient operator also comprises a prewitt operator, and the inventor compares the difference of the GSSIM algorithm on the Sobel operator and the prewitt operator, expands the traditional prewitt operator in 8 directions and applies the traditional prewitt operator to image quality evaluation.
Referring to fig. 1, an embodiment of the present invention provides an image quality evaluation method based on an extended Prewitt operator, which includes:
Step S1: acquiring a reference image and a distortion image to be evaluated;
Step S2: preprocessing the reference image and the distorted image, and determining brightness factors and contrast factors of the reference image and the distorted image;
specifically, the process of preprocessing the reference image and the distortion image to be evaluated in this step may specifically include:
step S201: gray scale processing is carried out on the reference image and the distorted image, so that only brightness information can be reserved for carrying out subsequent image quality analysis, and color information is removed;
Step S202: and establishing a predefined filtering operator, and carrying out filtering processing on the reference image and the distorted image after the graying processing.
The predefined filter operator may use an 11 x 11 gaussian template with a standard deviation of 1.5 pixels, the function of which is defined as follows
Wherein f (u, v) is a Gaussian function, (u, v) is a coordinate value of the image pixel point, sigma is a standard deviation, and the value is determined according to an experimental result.
By means ofCalculating brightness factors of the reference image and the distorted image; wherein,X represents the reference image, y represents the distorted image, l (x, y) is a brightness factor, mu x and mu y respectively represent average brightness of the reference image and the distorted image, a constant C1 is used for avoiding instability, N is the number of distorted images to be evaluated, i represents the ith image, x i represents the ith reference image, and y i represents the ith distorted image.
By means ofCalculating contrast factors of the reference image and the distorted image; wherein/>X represents the reference image, y represents the distorted image, C (x, y) represents a contrast factor, σ x and σ y represent standard deviations of pixels of the reference image and the distorted image, σ xy represents covariance of the reference image and the distorted image, μ x and μ y represent average brightness of the reference image and the distorted image respectively, a constant C2 is used for avoiding instability, N is the number of distorted images to be evaluated, and i represents the ith image.
Step S3: performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image;
The extension is performed for the traditional Prewitt operator, and the gradient amplitude is calculated. Referring to fig. 2 and 3, the extended Prewitt operator extends six directions of 45 °, 135 °, 180 °, 225 °, 270 °, 315 ° based on the original horizontal direction and vertical direction, and the extended Prewitt operator selects eight directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, 315 °. These template operators depend only on coefficients 0, 1 and-1, and they are all axisymmetric in form along the direction, and observing the template can also find that the addition result of weights everywhere in the neighborhood in the edge template is 0. Meanwhile, in order to overcome the overflow of the processing result under the extreme condition, an attenuation factor scale is introduced, the bad calculation result is divided by the attenuation factor scale, the possibility of data overflow is eliminated, and meanwhile, an undistorted gray level edge map is obtained. In the embodiment of the present invention, the attenuation factor may be 3.
Similar to gx and gy obtained by convolution operation of horizontal gradient and vertical gradient in classical algorithm, when the algorithm is implemented, convolution operation is performed on 8 templates for each pixel point of the image, and the maximum value is selected to be divided by attenuation factor scale, specifically, the following formula is shown. And the direction indicated by the template corresponding to the maximum value is the edge direction of the pixel point.
Gx and gy are gradient components obtained by convolution operation of a horizontal direction template and a vertical direction template of a traditional Prewitt operator, and g_45, g_135, g_180, g_225, g_270 and g_315 are gradient components obtained by convolution operation of the templates.
Step S4: determining gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value;
In particular, the embodiment of the invention adopts
Determining a gradient similarity factor of the reference image and the distorted image; where G x(i,j),Gy (i, j) is the gradient magnitude of the reference image x and the distorted image y at (i, j), respectively, and the constant C3 is a small constant for avoiding instability and for preventing denominator from being 0.
Step S5: and combining the gradient similarity factor, the brightness factor and the contrast factor to determine an objective evaluation result of the image quality.
The gradient similarity factor, the brightness factor and the contrast factor are combined to obtain the comprehensive Prewitt-GSSIM index between the reference image x and the distorted image y, wherein the comprehensive Prewitt-GSSIM index is as follows: prewitt-GSSIM (x, y) = [ l (x, y) ] α·[c(x,y)]β·[g(x,y)]λ; wherein, alpha >0, beta >0 and gamma >0, alpha, beta and gamma are parameters for adjusting the relative importance of three parts of brightness, contrast and gradient similarity respectively, and the embodiment of the invention can take alpha=beta=gamma=1;
the index value of the Prewitt-GSSIM of the whole image based on the extended Prewitt operator can be obtained by the average value of the Prewitt-GSSIM of each sub-image block; from the Prewitt-GSSIM index of each sub-image block, adopting Determining an objective evaluation result of the quality of the whole image; and M is the number of the sliding overlapping blocks of the distortion image to be evaluated by adopting a filtering factor.
According to the image quality evaluation method based on the extended Prewitt operator, a reference image and a distortion image to be evaluated are obtained; preprocessing the reference image and the distorted image, and determining brightness factors and contrast factors of the reference image and the distorted image; performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image; determining gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value; and determining an objective evaluation result of the image quality by combining the gradient similarity factor, the brightness factor and the contrast factor. The invention correspondingly improves the defects of the traditional Prewitt operator, can evaluate the quality of the noise image, has better consistency with subjective evaluation results, and solves the problem that partial detail information is lost when the traditional Prewitt operator is adopted to detect the image edge, so that the image edge has a certain blur.
On the premise of the full reference image, the method can evaluate the quality of the noise image through simple calculation and has better consistency with subjective scores. The test results of the examples of the present invention will be analyzed from two sides.
1) PSNR, SSIM, GSSIM, GSSIM-Prewitt (GSSIM adopting Prewitt operator), sobel-GSSIM (GSSIM algorithm adopting extended 8-direction Sobel operator), and nonlinear fitting curve graph of Prewitt-GSSIM algorithm and DMOS value provided by the embodiment of the invention;
2) The quality evaluation method provided by the embodiment of the invention is compared with quantitative performance evaluation of other objective evaluation algorithms.
According to the embodiment of the invention, a simulation experiment is carried out by adopting an image quality estimation database (2) provided by the university of American TEXAS image and video engineering laboratory, and the performance of the improved algorithm Prewitt-GSSIM provided by the embodiment of the invention relative to objective algorithms such as PSNR, SSIM and the like is compared. The image library includes 5 kinds of distorted images: JPEG, JPEG2000, gaussian Blur (Gaussian Blur), FASTFADING (image distorted by error in the course of fast-casting channel transmission of JPEG2000 code stream), white Noise (White Noise), 779 total distorted images. Also shown in the gallery are the "subjective difference scores" (DIFFERENCE MEAN opion score, DMOS) for all distorted images, describing the difference between subjective scores (Mean Opinion Score, MOS) and the full score of 100 minutes, so that a larger DMOS indicates a worse image quality, a smaller DMOS indicates a better image quality, and a range of values for DMOS is [0,100].
In order to better compare the performance of the image objective quality judgment model, four common objective parameters are selected as objective indexes for evaluating the methods: correlation coefficient under nonlinear regression (CC), spearman rank correlation coefficient between subjective and objective nonlinear regression Scores (SROCC), mean Absolute Error (MAE) and square Root of Mean Square Error (RMSE).
When the algorithm comparison is carried out, differences in the dimension and unit of each algorithm are generated, so that nonlinear regression is carried out on objective image quality scores obtained by the algorithm to be evaluated, and a Logistic function is used as a nonlinear mapping function to carry out nonlinear regression on objective image quality original scores obtained by the algorithm to be evaluated, which is provided by the embodiment of the invention:
Where z represents the original quality score given by the algorithm to be evaluated, β 1,β2,β3,β4,β5 is the parameter adaptively adjusted in the nonlinear regression process.
(1) The average absolute error (Mean Absolute Error, MAE) between the scores after subjective and objective nonlinear regression reflects the average error level of the objective quality evaluation result and the subjective evaluation result, and the smaller the average error level is, the higher the accuracy of the image quality evaluation result is, and the formula is described as follows:
(2) The root mean square error (Root Mean Square Error, RMSE) between the scores after subjective and objective nonlinear regression reflects the accuracy of the objective evaluation result, and the smaller the root mean square error is, the higher the accuracy of the image quality evaluation result is, and the formula is described as follows:
(3) The pearson linear correlation coefficient (Correlation Coefficient, CC) between the scores after subjective and objective nonlinear regression reflects the consistency and accuracy of objective evaluation results, the value range is [ -1,1], the absolute value of the results is closer to 1, and the correlation of the subjective and objective evaluation method is better, wherein the formula is described as follows:
(4) The spearman grade correlation coefficient (SPEARMAN RANK Order Correlation Coefficient, SROCC) between the scores after subjective and objective nonlinear regression is a widely applied non-parametric statistical analysis method, reflects the monotonicity of objective quality evaluation results and subjective evaluation results, has the value range of [ -1,1], and has the absolute value of the result being close to 1, and the consistency of the subjective and objective evaluation method is better, and the formula is described as follows:
Wherein the formula variables in (1) to (4) have the meanings: p is the size of the image library; v o (p) represents the value of the objective evaluation value of the p-th image after nonlinear fitting, and v s (p) represents the subjective evaluation value of the p-th image; and/> And gamma s(p)、γo (p) represents the ranking of the subjective and objective evaluation values of the p-th image in all evaluation values of the whole image library.
The magnitude of the CC value reflects the correlation between the objective evaluation method and DMOS, and the MAE and RMSE values reflect the accuracy of the objective evaluation model. And SROCC is mainly used for analyzing monotonicity, and the numerical value reflects the relation level between the predicted quality value of the model and the DMOS.
fig. 4 to 9 respectively show a PSNR, SSIM, GSSIM, GSSIM-Prewitt, sobel-GSSIM, a pretitt-GSSIM and other objective evaluation algorithms provided by the embodiments of the present invention, and a scatter diagram of a subjective evaluation value DMOS, where each point in the graph represents an image, an abscissa of the point is an objective quality evaluation score of the algorithm on the image, an ordinate is a subjective evaluation DMOS value of the image, and a solid line represents a fitted curve. The more closely the scattered points are distributed near the fitting curve, the better the consistency of the algorithm and subjective evaluation results is expressed, and the better the algorithm is. It can be seen that 982 scatter diagrams of the method provided by the invention are closest to the fitting curve, the Prewitt-GSSIM scatter diagrams are more densely concentrated near the fitting curve, and 4 objective parameters of the Prewitt-GSSIM algorithm are better than other objective evaluation algorithms such as SSIM and GSSIM, so that the effect of the method provided by the invention is better than that of other objective evaluation algorithms after nonlinear fitting.
Table 1 LIVE image database image quality evaluation method Performance comparison
| Evaluation method and Standard | MAE | RMSE | CC | SROCC |
| PSNR | 7.2743 | 9.0871 | 0.8256 | 0.8197 |
| SSIM | 5.1030 | 6.7240 | 0.9087 | 0.8999 |
| GSSIM | 5.0671 | 6.7082 | 0.9091 | 0.9046 |
| GSSIM-Prewitt | 4.9916 | 6.6345 | 0.9112 | 0.9064 |
| Sobel-GSSIM | 5.0420 | 6.6904 | 0.9096 | 0.9051 |
| Prewitt-GSSIM | 4.9334 | 6.5754 | 0.9128 | 0.9067 |
Experimental results show that the Prewitt-GSSIM model is more in line with the characteristics of a human eye vision system than the SSIM and PSNR (MSE) models, and can better evaluate the quality of a blurred image.
The image quality evaluation method based on the extended Prewitt operator provided by the embodiment of the invention is further elaborated by combining a specific scene, firstly, a LIVE gallery is arranged, wherein the LIVE gallery comprises 5 sub-galleries, namely JPEG, JPEG2000, gaussian blue (Gaussian Blur), FAST FADING (images distorted by errors in the process of transmitting a JPEG2000 code stream in a fast-coding channel) and White Noise (White Noise). There are 233 images in the JPEG gallery, 169 of which are distorted images, 233 of which are JPEG gallery, and 175 of which are distorted images. Three libraries of Gaussian Blur, FAST FADING, white Noise are 174 images and 145 are distorted images. And after finishing, correspondingly finding out the DMOS value of each picture according to the information of each picture library.
Another embodiment of the image quality evaluation method based on the extended Prewitt operator provided by the invention comprises the following steps:
step S100: preprocessing the reference image and the distorted image by using a Gaussian weight function, (using overlapped blocks with standard deviation of 1.5 pixels and window size of 11 x 11 by default);
step S101: respectively calculating brightness factors of the reference image and the distorted image;
Step S102: respectively calculating contrast factors of the reference image and the distorted image;
Step S103: expanding the traditional Prewitt operator;
step S104: performing convolution operation on the reference image and the distorted image by using an extended Prewitt operator respectively, and calculating a gradient value;
Step S105: calculating a gradient similarity factor by using gradient values obtained by the reference image and the distorted image;
Step S106: and combining the brightness factor, the contrast factor and the gradient similarity factor to obtain an objective evaluation value.
Step S107: and averaging the objective values of the obtained overlapping blocks of each image to obtain a final result.
Referring to fig. 10, an embodiment of an image quality evaluation apparatus based on an extended Prewitt operator according to the present invention may include:
the image acquisition module is used for acquiring a reference image and a distortion image to be evaluated;
The image preprocessing module is used for preprocessing the reference image and the distorted image and determining brightness factors and contrast factors of the reference image and the distorted image;
The gradient value calculation module is used for carrying out convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image;
The gradient similarity factor calculation module is used for determining the gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value;
And the objective evaluation result determining module is used for combining the gradient similarity factor, the brightness factor and the contrast factor to determine an objective evaluation result of the image quality.
As a specific implementation manner, in the image quality evaluation device based on the extended Prewitt operator provided by the embodiment of the present invention, the image preprocessing module further includes:
the graying processing submodule is used for carrying out graying processing on the reference image and the distorted image;
the filtering processing sub-module is used for carrying out filtering processing on the reference image and the distorted image after the graying processing according to a filtering operator; wherein the filter operator is built using an 11 x 11 gaussian template with a standard deviation of 1.5 pixels.
The extended Prewitt operator is an operator for selecting eight directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, 315 °.
The image quality evaluation device based on the extended Prewitt operator provided by the embodiment of the invention is characterized by acquiring a reference image and a distortion image to be evaluated; preprocessing the reference image and the distorted image, and determining brightness factors and contrast factors of the reference image and the distorted image; performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image; determining gradient similarity factors of the reference image and the distorted image according to the first gradient value and the second gradient value; and determining an objective evaluation result of the image quality by combining the gradient similarity factor, the brightness factor and the contrast factor. The invention correspondingly improves the defects of the traditional Prewitt operator, can evaluate the quality of the noise image, has better consistency with subjective evaluation results, and solves the problem that partial detail information is lost when the traditional Prewitt operator is adopted to detect the image edge, so that the image edge has a certain blur.
In this specification, each embodiment is described in a progressive manner, and each embodiment is mainly described in a different point from other embodiments, so that the same or similar parts between the embodiments are referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein. Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above in terms of function in a generic sense for clarity of understanding of the interchangeability of hardware and software. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
In the embodiments provided in the present application, it should be understood that the disclosed system, apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical or other form.
The units described as separate units may or may not be physically separate, and units shown as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a read-only memory (ROM), a random access memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (4)
1. An image quality evaluation method based on an extended Prewitt operator is characterized by comprising the following steps:
acquiring a reference image and a distortion image to be evaluated;
preprocessing the reference image and the distorted image, and determining brightness factors and contrast factors of the reference image and the distorted image;
Determining the luminance factors of the reference image and the distorted image further comprises: by means of Calculating brightness factors of the reference image and the distorted image;
wherein, X represents the reference image, y represents the distorted image, l (x, y) is a brightness factor, mu x and mu y respectively represent average brightness of the reference image and the distorted image, a constant C1 is used for avoiding instability, N is the number of distorted images to be evaluated, i represents the ith image, x i represents the ith reference image, and y i represents the ith distorted image;
determining the contrast factor of the reference image and the distorted image further comprises: by means of Calculating contrast factors of the reference image and the distorted image;
wherein, X represents the reference image, y represents the distorted image, C (x, y) represents a contrast factor, σ x and σ y represent standard deviations of pixels of the reference image and the distorted image, σ xy represents covariance of the reference image and the distorted image, μ x and μ y represent average brightness of the reference image and the distorted image respectively, a constant C2 is used for avoiding instability, N is the number of distorted images to be evaluated, and i represents an ith image;
Performing convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image;
the extended Prewitt operator is a template operator for selecting eight directions of 0 degree, 45 degree, 90 degree, 135 degree, 180 degree, 225 degree, 270 degree and 315 degree, and each template operator only depends on coefficients 0, 1 and-1 and is axisymmetric along the direction;
Introducing an attenuation factor, eliminating the possibility of data overflow, and simultaneously obtaining an undistorted gray level edge map, which specifically comprises the following steps: carrying out convolution operation on eight templates on each pixel point of an image to obtain corresponding gradient components, and selecting the maximum value in all gradient components to be divided by the attenuation factor, wherein the direction represented by the template corresponding to the maximum value is the edge direction of the pixel point;
Determining a gradient similarity factor for the reference image and the distorted image according to the first gradient value and the second gradient value, including:
by using Determining a gradient similarity factor of the reference image and the distorted image; wherein G x(i,j),Gy (i, j) is the gradient magnitude of the reference image x and the distorted image y at (i, j), respectively, and a constant C3 is used to avoid instability;
Combining the gradient similarity factor, the brightness factor and the contrast factor to determine an objective evaluation result of image quality;
combining the gradient similarity factor, the brightness factor and the contrast factor, determining an objective evaluation result of image quality further comprises: determining a Prewitt-GSSIM index of the reference image and the distorted image by using Prewitt-GSSIM (x, y) = [ l (x, y) ] α·[c(x,y)]β·[g(x,y)]λ, wherein Prewitt-GSSIM (x, y) is a comprehensive Prewitt-GSSIM index, alpha >0, beta >0 and gamma >0, and alpha, beta and gamma are parameters for adjusting relative importance of three parts of brightness, contrast and gradient similarity respectively;
From the Prewitt-GSSIM index of each sub-image block, adopting
Determining an objective evaluation result of the quality of the whole image; and M is the number of the sliding overlapping blocks of the distortion image to be evaluated by adopting a filtering factor.
2. The extended Prewitt operator-based image quality evaluation method of claim 1, wherein preprocessing the reference image and the distorted image further comprises:
Graying the reference image and the distorted image;
according to a filtering operator, filtering the reference image and the distorted image after the graying treatment; wherein the filter operator is built using an 11 x 11 gaussian template with a standard deviation of 1.5 pixels.
3. An image quality evaluation device based on an extended Prewitt operator, comprising:
the image acquisition module is used for acquiring a reference image and a distortion image to be evaluated;
The image preprocessing module is used for preprocessing the reference image and the distorted image and determining brightness factors and contrast factors of the reference image and the distorted image;
Determining the luminance factors of the reference image and the distorted image further comprises: by means of Calculating brightness factors of the reference image and the distorted image;
wherein, X represents the reference image, y represents the distorted image, l (x, y) is a brightness factor, mu x and mu y respectively represent average brightness of the reference image and the distorted image, a constant C1 is used for avoiding instability, N is the number of distorted images to be evaluated, i represents the ith image, x i represents the ith reference image, and y i represents the ith distorted image;
determining the contrast factor of the reference image and the distorted image further comprises: by means of Calculating contrast factors of the reference image and the distorted image;
wherein, X represents the reference image, y represents the distorted image, C (x, y) represents a contrast factor, σ x and σ y represent standard deviations of pixels of the reference image and the distorted image, σ xy represents covariance of the reference image and the distorted image, μ x and μ y represent average brightness of the reference image and the distorted image respectively, a constant C2 is used for avoiding instability, N is the number of distorted images to be evaluated, and i represents an ith image;
The gradient value calculation module is used for carrying out convolution operation on the reference image and the distorted image according to the extended Prewitt operator, and calculating a first gradient value corresponding to the reference image and a second gradient value corresponding to the distorted image;
the extended Prewitt operator is a template operator for selecting eight directions of 0 degree, 45 degree, 90 degree, 135 degree, 180 degree, 225 degree, 270 degree and 315 degree, and each template operator only depends on coefficients 0, 1 and-1 and is axisymmetric along the direction;
Introducing an attenuation factor, eliminating the possibility of data overflow, and simultaneously obtaining an undistorted gray level edge map, which specifically comprises the following steps: carrying out convolution operation on eight templates on each pixel point of an image to obtain corresponding gradient components, and selecting the maximum value in all gradient components to be divided by the attenuation factor, wherein the direction represented by the template corresponding to the maximum value is the edge direction of the pixel point;
A gradient similarity factor calculation module, configured to determine a gradient similarity factor of the reference image and the distorted image according to the first gradient value and the second gradient value, including:
by using Determining a gradient similarity factor of the reference image and the distorted image; wherein G x(i,j),Gy (i, j) is the gradient magnitude of the reference image x and the distorted image y at (i, j), respectively, and a constant C3 is used to avoid instability;
The objective evaluation result determining module is used for determining an objective evaluation result of image quality by combining the gradient similarity factor, the brightness factor and the contrast factor;
Combining the gradient similarity factor, the brightness factor and the contrast factor, determining an objective evaluation result of image quality further comprises: determining a Prewitt-GSSIM index of the reference image and the distorted image by using Prewitt-GSSIM (x, y) = [ l (x, y) ] α·[c(x,y)]β · [ g (x, y) ] lambda, wherein the Prewitt-GSSIM (x, y) is a comprehensive Prewitt-GSSIM index, alpha >0, beta >0 and gamma >0, and alpha, beta and gamma are parameters for adjusting the relative importance of three parts of brightness, contrast and gradient similarity respectively;
From the Prewitt-GSSIM index of each sub-image block, adopting
Determining an objective evaluation result of the quality of the whole image; and M is the number of the sliding overlapping blocks of the distortion image to be evaluated by adopting a filtering factor.
4. The extended Prewitt operator-based image quality evaluation apparatus of claim 3, wherein the image preprocessing module further comprises:
the graying processing submodule is used for carrying out graying processing on the reference image and the distorted image;
the filtering processing sub-module is used for carrying out filtering processing on the reference image and the distorted image after the graying processing according to a filtering operator; wherein the filter operator is built using an 11 x 11 gaussian template with a standard deviation of 1.5 pixels.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010090192.5A CN111598829B (en) | 2020-02-13 | 2020-02-13 | Image quality evaluation method and device based on extended Prewitt operator |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010090192.5A CN111598829B (en) | 2020-02-13 | 2020-02-13 | Image quality evaluation method and device based on extended Prewitt operator |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111598829A CN111598829A (en) | 2020-08-28 |
| CN111598829B true CN111598829B (en) | 2024-04-19 |
Family
ID=72184956
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010090192.5A Active CN111598829B (en) | 2020-02-13 | 2020-02-13 | Image quality evaluation method and device based on extended Prewitt operator |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111598829B (en) |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108053393A (en) * | 2017-12-08 | 2018-05-18 | 广东工业大学 | A kind of gradient similarity graph image quality evaluation method and device |
-
2020
- 2020-02-13 CN CN202010090192.5A patent/CN111598829B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108053393A (en) * | 2017-12-08 | 2018-05-18 | 广东工业大学 | A kind of gradient similarity graph image quality evaluation method and device |
Non-Patent Citations (1)
| Title |
|---|
| 邓杰航 等.基于扩展梯度算子的结构相似度图像质量评价方法.《科学技术与工程》.2018,第18卷(第27期),第42-47页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111598829A (en) | 2020-08-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Shen et al. | Hybrid no-reference natural image quality assessment of noisy, blurry, JPEG2000, and JPEG images | |
| Li et al. | Content-partitioned structural similarity index for image quality assessment | |
| Gu et al. | No-reference quality assessment of screen content pictures | |
| Yang et al. | Perceptual quality assessment of screen content images | |
| Ma et al. | Objective quality assessment for color-to-gray image conversion | |
| CN108109147B (en) | No-reference quality evaluation method for blurred image | |
| CN109978854B (en) | Screen content image quality evaluation method based on edge and structural features | |
| CN109003265B (en) | No-reference image quality objective evaluation method based on Bayesian compressed sensing | |
| Liang et al. | Comparison-based image quality assessment for selecting image restoration parameters | |
| CN108230325A (en) | The compound degraded image quality evaluating method and system decomposed based on cartoon texture | |
| Attar et al. | Image quality assessment using edge based features | |
| CN108053393A (en) | A kind of gradient similarity graph image quality evaluation method and device | |
| Kerouh et al. | A no reference quality metric for measuring image blur in wavelet domain | |
| Wang et al. | Screen content image quality assessment with edge features in gradient domain | |
| Morzelona | Human visual system quality assessment in the images using the IQA model integrated with automated machine learning model | |
| CN110910347B (en) | Tone mapping image non-reference quality evaluation method based on image segmentation | |
| CN109754390B (en) | No-reference image quality evaluation method based on mixed visual features | |
| Liu et al. | UIQI: a comprehensive quality evaluation index for underwater images | |
| Rohil et al. | An improved model for no-reference image quality assessment and a no-reference video quality assessment model based on frame analysis | |
| CN111598829B (en) | Image quality evaluation method and device based on extended Prewitt operator | |
| Gao et al. | A content-based image quality metric | |
| CN104123723A (en) | Structure compensation based image quality evaluation method | |
| Moreno et al. | Towards no-reference of peak signal to noise ratio | |
| CN111445435A (en) | No-reference image quality evaluation method based on multi-block wavelet transform | |
| Escobar et al. | Defining a no-reference image quality assessment by means of the self-affine analysis |
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 |