CN109523533A - A kind of image quality evaluating method and device - Google Patents
A kind of image quality evaluating method and device Download PDFInfo
- Publication number
- CN109523533A CN109523533A CN201811352578.8A CN201811352578A CN109523533A CN 109523533 A CN109523533 A CN 109523533A CN 201811352578 A CN201811352578 A CN 201811352578A CN 109523533 A CN109523533 A CN 109523533A
- Authority
- CN
- China
- Prior art keywords
- image
- decomposition level
- block
- similarity
- level
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20016—Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Quality & Reliability (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of image quality evaluating method and device, after obtaining the reference picture of image to be detected and image to be detected, image to be detected and reference picture are decomposed respectively, obtain corresponding at least two second images with different decomposition level of corresponding at least two first images and reference picture with different decomposition level of image to be detected；To same the first image and the second image for decomposing level: determining the first similarity and the second similarity between the first image of the decomposition level and the second image, according to the first similarity and the second similarity between each the first image and the second image for decomposing level, image quality evaluation is carried out to image to be detected, wherein the first similarity and the second similarity are obtained based on the pixel interdependence between same the first image and the second image for decomposing level, so that considering the influence of pixel interdependence when carrying out image quality evaluation to image to be detected, thus the accuracy of image quality evaluation is improved.
Description
Technical field
The invention belongs to technical field of image processing, more specifically more particularly to a kind of image quality evaluating method and
Device.
Background technique
Image inevitably will cause the damage of picture quality in the image processing process such as compression, coding and storage,
Thus need to evaluate image to be detected by image quality evaluating method (has original image certain after image procossing
The image of distortion) picture quality, such as evaluate distortion level of the image to be detected relative to original image.
Image quality evaluating method is using original image as reference picture at present, according to the information of the reference picture got
Amount includes: full reference image quality appraisement (Full Reference Assessment) method, half reference image quality appraisement
(Reduced Reference Assessment) method and non-reference picture quality appraisement (No Reference
Assessment) method.The wherein available all information to reference picture of full reference image quality appraisement method, with basis
Image to be detected and reference picture are compared to the figure of evaluation image to be detected by all information of the reference picture got
Image quality amount；The available partial information to reference picture of semi-reference image quality evaluation algorithm, according to the reference got
Image to be detected and reference picture are compared to the picture quality of evaluation image to be detected by the partial information of image；Without ginseng
Examining image quality evaluating method is only evaluated image to be detected in any information for not getting reference picture.
But full reference image quality appraisement method and semi-reference image quality evaluation algorithm are all by comparing with reference to figure
Pixel difference in picture and image to be detected between each pixel is anisotropic, so that the accuracy of image quality evaluation is by a fixed limit
System.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of image quality evaluating method and devices, for improving image
The accuracy of quality evaluation.Technical solution is as follows:
The present invention provides a kind of image quality evaluating method, which comprises
Obtain the reference picture of image to be detected and described image to be detected；
Described image to be detected and the reference picture are decomposed respectively, it is corresponding extremely to obtain described image to be detected
Few two the first images and corresponding at least two second image of the reference picture, wherein corresponding different points of each first image
Level is solved, each second image corresponds to different decomposition level, and for any first image: first image and all second figures
Second image decomposition level having the same as in；
To same the first image and the second image for decomposing level: determining the first image and the second image of the decomposition level
Between the first similarity and determine the decomposition level the first image and the second image between the second similarity, this first
Pixel interdependence between similarity and second similarity the first image and the second image based on the decomposition level obtains；
According to the first similarity and each decomposition level between each the first image and the second image for decomposing level
The first image and the second image between the second similarity, to described image to be detected carry out image quality evaluation.
Preferably, described to same the first image and the second image for decomposing level: to determine the first figure of the decomposition level
The first similarity between picture and the second image includes:
The first image and the second image of the decomposition level are split respectively, obtain the first image of the decomposition level
At least two image blocks and the decomposition level the second image at least two image blocks；
Determine i-th of image block X in the first image of the decomposition leveliWith in the second image of the decomposition level
I image block XiBetween similarity, the value of i is 1 to n, and n is the sum of the image block of the decomposition level；
According to i-th of image block X in the first image of the decomposition leveliWith in the second image of the decomposition level
I image block XiBetween similarity, obtain the first similarity between the first image of the decomposition level and the second image.
Preferably, i-th of image block X in the first image of the determination decomposition leveliWith the of the decomposition level
I-th of image block X in two imagesiBetween similarity include:
Determine the discrete grey of discrete grey's collection of the first image of the decomposition level and the second image of the decomposition level
Collection；
To i-th of image block X in the first image of the decomposition leveliWith i-th in the second image of the decomposition level
A image block Xi, it is based on formula:
Obtain the decomposition level
I-th of image block X in first imageiWith i-th of image block X in the second image of the decomposition leveliBetween mutual information, T
For the second image, F is the first image, LTCollect for the discrete grey of the second image, LFCollect for the discrete grey of the first image, r the
The pixel grey scale of two images, f are the pixel grey scale of the first image, and u is the deformation parameter in the corresponding deformation function of the first image,
p(r,f,u,Xi) be the first image and the second image joint probability density function, p (r, Xi) it is i-th of image block XiRelative to
The marginal probability density of second image, p (f, u, Xi) it is i-th of image block XiRelative to the marginal probability density of the first image, i
Value be 1 to n, n be the decomposition level image block sum；
By i-th of image block X in the first image of the decomposition leveliWith i-th in the second image of the decomposition level
A image block XiBetween mutual information determine are as follows: i-th of image block X in the first image of the decomposition leveliWith the decomposition level
The second image in i-th of image block XiBetween similarity.
Preferably, i-th of image block X in first image according to the decomposition leveliWith the of the decomposition level
I-th of image block X in two imagesiBetween similarity, obtain between the first image of the decomposition level and the second image
First similarity includes:
According to i-th of image block XiSimilarity between adjacent image block determines i-th of image block of the decomposition level
XiWeight；
According to i-th of image block X in the first image of the decomposition leveliWith in the second image of the decomposition level
I image block XiBetween similarity and i-th of image block XiWeight, obtain the first image and second of the decomposition level
The first similarity between image.
Preferably, the second similarity between the first image and the second image of the determination decomposition level includes:
The first image and the second image of the decomposition level are split respectively, obtain the first image of the decomposition level
At least two image blocks and the decomposition level the second image at least two image blocks, and determine the decomposition level first
The local entropy of each image block in image；
According to the local entropy of each image block in the first image of the decomposition level, from the first image of the decomposition level
Selected part image block in all image blocks；
Determine j-th of image block Y of the decomposition level chosenjWith in the second image of the decomposition level with j-th of figure
As block YjSimilarity between corresponding image block, the value of j are 1 to m, and m is the image block selected in the decomposition level
Sum；
According to j-th of image block Y of the decomposition level of selectionjWith in the second image of the decomposition level with j-th of figure
As block YjSimilarity and j-th of image block Y between corresponding image blockjWeight, obtain the first image of the decomposition level
And second the second similarity between image, wherein the weight of j-th of image block is according to the image block adjacent with j-th of image block
Similarity obtain.
Preferably, first similarity according between each the first image and the second image for decomposing level and every
The second similarity between a the first image and the second image for decomposing level carries out picture quality to described image to be detected and comments
Valence includes:
To each decomposition level: according between the first image and the second image of the decomposition level the first similarity and
The second similarity between the first image and the second image of the decomposition level, determines the First Eigenvalue of the decomposition level；
According to each the First Eigenvalue for decomposing level, the mark between described image to be detected and the reference picture is obtained
Standard is poor and/or obtains the variance between described image to be detected and the reference picture；
According to standard deviation and/or described image to be detected between described image to be detected and the reference picture and described
Variance between reference picture carries out image quality evaluation to described image to be detected.
The present invention also provides a kind of image quality evaluation device, described device includes:
Acquiring unit, for obtaining the reference picture of image to be detected and described image to be detected；
Decomposition unit obtains described to be checked for decomposing respectively to described image to be detected and the reference picture
Corresponding at least two first image of altimetric image and corresponding at least two second image of the reference picture, wherein each first
Image corresponds to different decomposition level, and each second image corresponds to different decomposition level, and for any first image: first figure
Picture and second image decomposition level having the same in all second images；
Similarity determining unit, for same the first image and the second image for decomposing level: determining the decomposition level
The first image and the second image between the first similarity and determine the decomposition level the first image and the second image it
Between the second similarity, first similarity and first image and second image of second similarity based on the decomposition level it
Between pixel interdependence obtain；
Evaluation unit, for according to it is each decompose level the first image and the second image between the first similarity and
The second similarity between each the first image and the second image for decomposing level carries out picture quality to described image to be detected
Evaluation.
Preferably, the similarity determining unit, comprising:
Divide subelement, for same the first image and the second image for decomposing level: to the first of the decomposition level
Image and the second image are split respectively, obtain at least two image blocks and the decomposition layer of the first image of the decomposition level
At least two image blocks of the second image of grade；
First determines subelement, i-th of image block X in the first image for determining the decomposition leveliWith the decomposition
I-th of image block X in second image of leveliBetween similarity, the value of i is 1 to n, and n is the image of the decomposition level
The sum of block；
First similarity determines subelement, for i-th of image block X in the first image according to the decomposition leveliWith
I-th of image block X in second image of the decomposition leveliBetween similarity, obtain the decomposition level the first image and
The first similarity between second image；
Image block chooses subelement, the local entropy of each image block in the first image for determining the decomposition level, and
According to the local entropy of each image block in the first image of the decomposition level, from all images of the first image of the decomposition level
Selected part image block in block；
Second determines subelement, for determining j-th of image block Y of the decomposition level chosenjWith the decomposition level
In second image with j-th of image block YjSimilarity between corresponding image block, the value of j are 1 to m, and m is the decomposition level
In the sum of image block that selects；
Second similarity determines subelement, for j-th of image block Y according to the decomposition level of selectionjWith the decomposition
In second image of level with j-th of image block YjSimilarity and j-th of image block Y between corresponding image blockjPower
Weight, obtains the second similarity between the first image of the decomposition level and the second image, wherein the weight root of j-th of image block
It is obtained according to the similarity of the image block adjacent with j-th of image block.
Preferably, it is described first determine subelement, for determine the decomposition level the first image discrete grey collection and
The discrete grey of second image of the decomposition level collects；To i-th of image block X in the first image of the decomposition leveliAnd this
Decompose i-th of image block X in the second image of leveli, it is based on formula:
Obtain the decomposition level
I-th of image block X in first imageiWith i-th of image block X in the second image of the decomposition leveliBetween mutual information,
I-th of the image block X mutual information being determined as in the first image of the decomposition leveliWith the second image of the decomposition level
In i-th of image block XiBetween similarity；
Wherein T is the second image, and F is the first image, LTCollect for the discrete grey of the second image, LFFor first image from
Gray scale collection is dissipated, r is the pixel grey scale of the second image, and f is the pixel grey scale of the first image, and u is the corresponding deformation letter of the first image
Deformation parameter in number, p (r, f, u, Xi) be the first image and the second image joint probability density function, p (r, Xi) it is i-th
A image block XiRelative to the marginal probability density of the second image, p (f, u, Xi) it is i-th of image block XiRelative to the first image
Marginal probability density, the value of i is 1 to n, and n is the sum of the image block of the decomposition level.
Preferably, first similarity determines subelement, for according to i-th of image block XiBetween adjacent image block
Similarity, determine i-th of image block X of the decomposition leveliWeight；And according in the first image of the decomposition level
I image block XiWith i-th of image block X in the second image of the decomposition leveliBetween similarity and i-th of image block
XiWeight, obtain the first similarity between the first image of the decomposition level and the second image.
Preferably, the evaluation unit, comprising:
Characteristic value determines subelement, for each decomposition level: according to the first image and the second figure of the decomposition level
The first similarity as between and the second similarity between the first image and the second image of the decomposition level determine this point
Solve the First Eigenvalue of level；
Evaluation parameter obtains subelement, for obtaining the mapping to be checked according to each the First Eigenvalue for decomposing level
Standard deviation between picture and the reference picture and/or obtain the variance between described image to be detected and the reference picture；
Subelement is evaluated, for according to standard deviation between described image to be detected and the reference picture and/or described
Variance between image to be detected and the reference picture carries out image quality evaluation to described image to be detected.
The present invention also provides a kind of storage medium, computer program code, the calculating are stored on the storage medium
Machine program code realizes above-mentioned image quality evaluating method when executing.
Compared with prior art, above-mentioned technical proposal provided by the invention has the advantages that
From above-mentioned technical proposal it is found that being treated respectively after obtaining the reference picture of image to be detected and image to be detected
Detection image and reference picture are decomposed, and corresponding at least two first with different decomposition level of image to be detected are obtained
Image and corresponding at least two second images with different decomposition level of reference picture；To same the first figure for decomposing level
Picture and the second image: determining the first similarity between the first image of the decomposition level and the second image and determines the decomposition
The second similarity between the first image and the second image of level, according to each the first image and the second image for decomposing level
Between the first similarity and it is each decompose level the first image and the second image between the second similarity, to be detected
Image carries out image quality evaluation, wherein each the first similarity for decomposing level and the second similarity are based on being in same decomposition
Pixel interdependence between the first image and the second image of level obtains, so that commenting carrying out picture quality to image to be detected
The influence of pixel interdependence is considered when valence, and influence of the pixel interdependence to image quality evaluation is greater than the pixel difference opposite sex to image
The influence of quality evaluation, therefore commented by the picture quality of the first similarity and the second similarity that are obtained based on pixel interdependence
The accuracy of image quality evaluation can be improved in valence.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is the present invention
Some embodiments for those of ordinary skill in the art without creative efforts, can also basis
These attached drawings obtain other attached drawings.
Fig. 1 is a kind of flow chart of image quality evaluating method provided in an embodiment of the present invention；
Fig. 2 is that the first similarity and second similar is determined in a kind of image quality evaluating method provided in an embodiment of the present invention
The flow chart of degree；
Fig. 3 is a kind of structural schematic diagram of image quality evaluation device provided in an embodiment of the present invention；
Fig. 4 is the structural schematic diagram of similarity determining unit in image quality evaluation device provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
Referring to Fig. 1, being used for it illustrates a kind of flow chart of image quality evaluating method provided in an embodiment of the present invention
The accuracy for improving image quality evaluation, may comprise steps of:
S101: the reference picture of image to be detected and image to be detected is obtained.It is to be understood that image to be detected is to need
The piece image of image quality evaluation is carried out, the reference picture of image to be detected is then to carry out picture quality to image to be detected
Referenced image when evaluation, the reference picture of image to be detected can be the either default ginseng of original image of image to be detected
Some reference picture in image library is examined, for how to determine the reference picture of the image to be detected in preset reference image library,
Existing reference picture acquisition methods are referred to, this present embodiment is no longer illustrated.
S102: decomposing image to be detected and reference picture respectively, obtains image to be detected corresponding at least two
First image and corresponding at least two second image of reference picture, wherein each first image corresponds to different decomposition level, often
A second image corresponds to different decomposition level, and for any first image: one in first image and all second images
A second image decomposition level having the same decomposes level for showing and carries out which layer decomposition to image.
It is so-called to decompose the image for referring to obtain the different resolution of piece image, as being 128*128 to a resolution ratio
For image, the image that the image and resolution ratio that available resolution ratio is 64*64 are 32*32, specifically can to it is above-mentioned to
Detection image and reference picture carry out low frequency and high-frequency decomposition, by taking image to be detected as an example, to the first layer of image to be detected into
Row low frequency and high-frequency decomposition, obtain the first of image to be detected two the first images for decomposing level: first decomposes the low of level
The high frequency imaging of frequency image and the first decomposition level, the height of the low-frequency image for decomposing level to first respectively and the first decomposition level
Frequency image carries out low frequency and high-frequency decomposition again, obtains the second of the low-frequency image of the first decomposition level two first for decomposing level
Image: first decomposes the low-frequency image and high frequency imaging of the second decomposition level of the low-frequency image of level, and obtains first point
Solve two the first images of the high frequency imaging of level: first decomposes the low-frequency image of the second decomposition level of the high frequency imaging of level
And high frequency imaging, and so on, until decomposed to N decomposition layer grade, wherein N decompose level show to piece image into
Row n-th is decomposed.
Such as N layers of wavelet decomposition are carried out respectively to image to be detected and reference picture, then to image to be detected and with reference to figure
As one layer of wavelet decomposition of every progress, multiple first images and multiple second images are obtained, the corresponding number of plies of this layer of wavelet decomposition is then
It is to decompose level, i-th layer of wavelet decomposition such as is carried out to image to be detected and reference picture, then decomposing level is i, and wherein N is big
It, can be depending on image to be detected, to this present embodiment to it without limiting for the value of N in 1 natural number.
S103: to same the first image and the second image for decomposing level: determining the first image and the of the decomposition level
The first similarity between two images and the second similarity between the first image and the second image of the determining decomposition level,
Pixel interdependence between first similarity and second similarity the first image and the second image based on the decomposition level
It obtains, so that considering influence of the pixel interdependence to image quality evaluation in image quality evaluation, is commented with improving picture quality
The accuracy of valence.
In the present embodiment, although the first similarity and the second similarity are all based on pixel interdependence and obtain, the
One similarity and the second similarity are to consider to obtain from the different aspect of image, such as the first similarity and the second similarity wherein it
First is that consider to obtain from the full content of image, the other is from the partial content of image (in such as image in addition to background content
Content) consider to obtain, the determination process of the first similarity and the second similarity is illustrated below in conjunction with attached drawing, such as determine appoint
One decomposes the first similarity and any the first image for decomposing level and the between the first image and the second image of level
The process of the second similarity between two images as shown in Fig. 2, it is illustrated by taking a decomposition level as an example, may include with
Lower step:
S1031: being split the first image and the second image of the decomposition level respectively, obtains the of the decomposition level
At least two image blocks of the second image of at least two image blocks and decomposition level of one image.
Wherein, the first image of the decomposition level and the second image are split using identical partitioning scheme, are such as adopted
The first image and the second image are split with conventional images partitioning algorithm, a kind of such as image segmentation algorithm for using can be with
It is: the first image of the decomposition level and the second image is respectively seen as input picture, following steps is executed to input picture:
(1), input picture is divided into n × n initial pictures block；
(2), the similar matrix of each initial pictures block is calculated；
(3), similar two initial pictures blocks are determined according to the similar matrix of each initial pictures block；
(4), to any similar two initial pictures blocks: the energy after calculating similar two initial pictures merged blocks
EijThe sum of the corresponding energy of two initial pictures blocks similar with this E=Ei+Ej, wherein EiAnd EjIt is this similar two
Image block S in initial pictures blockiAnd SjEnergy, Ei=MI (I, Si)=H (I)+H (Si)-H(I,Si), I is input picture,
EjAnd EijCalculation it is identical, this is no longer described in detail；
(5) if, E-Eij< θ (fixed value, value is without explanation), then merge image block SiAnd Sj, after merging
Image block be considered as initial pictures block and continue to calculate the similar matrix of the initial pictures block, and according to the similar of the initial pictures block
Matrix determines initial pictures block similar with its, to the similar initial pictures block of the initial pictures block and the initial pictures block
It executes (4)；
(6) if, E-Eij>=θ then forbids merging similar two initial pictures blocks；
(7), after completing above-mentioned processing to all similar two initial pictures blocks, by initial pictures remaining after processing
Block is determined as the image block of the input picture.
Herein it should be noted is that: the segmentation in the present embodiment refer to some resolution ratio the first image
Piecemeal is carried out with the second image with some resolution ratio, obtains the multiple images of each first image and each second image
Block, each image block include the partial content in image, and for same first image, multiple figures of first image
As block resolution ratio having the same, that is to say, that after decomposing to image to be detected and reference picture, obtain each decomposition
(the first image and the second image be still at this time for first image of level (one decompose level corresponding a resolution ratio) and the second image
It is so a complete image) the first image and the second image are split later.With image to be detected for a facial image
For, each the first image for decomposing level is obtained after decomposing, the first image is still a facial image at this time, only
It is that image information in the facial image reduces for image to be detected, then the first image is split, is obtained
The image block of different piece including facial image such as obtains the image block for respectively including eyes, nose, mouth.
S1032: i-th of image block X in the first image of the decomposition level is determinediWith the second image of the decomposition level
In i-th of image block XiBetween similarity, the value of i is 1 to n, and n is the sum of the image block of the decomposition level.At this
In embodiment, i-th of image block X in the first image of the decomposition level is determinediIn the second image of the decomposition level
I-th of image block XiBetween similarity a kind of feasible pattern it is as follows:
Determine the discrete grey of discrete grey's collection of the first image of the decomposition level and the second image of the decomposition level
Collection, wherein a kind of method of determination that the discrete grey of each image collects is: determine the corresponding coordinate system of the image, it will be on coordinate system
Reference axis be divided into checkerboard type grid, obtain the gray value in checkerboard type grid on each intersection position, each intersection position
On gray value composition image discrete grey collection.
To i-th of image block X in the first image of the decomposition leveliWith i-th in the second image of the decomposition level
A image block Xi, it is based on formula:
Obtain the decomposition level
I-th of image block X in first imageiWith i-th of image block X in the second image of the decomposition leveliMutual information, will be mutual
Information is determined as i-th of image block X in the first image of the decomposition leveliWith i-th in the second image of the decomposition level
A image block XiBetween similarity.
Wherein T is the second image, and F is the first image, LTCollect for the discrete grey of the second image, LFFor first image from
Gray scale collection is dissipated, r is the pixel grey scale of the second image, and f is the pixel grey scale of the first image, and u is deformation corresponding with the first image
Deformation parameter in function, p (r, f, u, Xi) be the first image and the second image joint probability density function, p (r, Xi) it is the
I image block XiRelative to the marginal probability density of the second image, p (f, u, Xi) it is i-th of image block XiRelative to the first image
Marginal probability density, and the similarity why can be determined as mutual information between image block is because when two image blocks
Intersection is bigger, and the correlation between two image blocks is also bigger, and mutual information is bigger, it is possible thereby to embody two by mutual information
Similarity between a image block.
From it was found from the calculation formula of above-mentioned mutual information: being introduced into when calculating mutual information in the corresponding deformation function of the first image
Deformation parameter, and be introduced into deformation parameter be because the first image in there may be some divergent margin points, these divergent margins
Point can not accurately indicate the edge of the first image, thus need by deformation parameter to each image block in the first image into
Row processing, to achieve the purpose that smooth edges, then again to the figure in the image block and the second image after deformation in the first image
As the influence that block carries out mutual information calculating, and thus reduction divergent margin point calculates mutual information.The wherein corresponding shape of the first image
Varying function can be any one existing function, such as can be g (x)=x+u (x), and u (x) indicates displacement, and u is deformation parameter.
The pixel grey scale of above-mentioned first image is obtained according to the gray value of all pixels point in the first image, the second image
Pixel grey scale is obtained according to the gray value of all pixels point in the second image, and the pixel grey scale of such as the first image is in the first image
The average value of the gray value of all pixels point, the pixel grey scale of the second image are the gray values of all pixels point in the second image
Weighted mean, the calculation of the pixel grey scale of the pixel grey scale of the first image and the second image in the calculation formula of mutual information
It is identical, but the present embodiment does not limit which kind of calculation it specifically uses.
S1033: according to i-th of image block X in the first image of the decomposition leveliWith the second image of the decomposition level
In i-th of image block XiBetween similarity, obtain the first phase between the first image of the decomposition level and the second image
Like degree.In the present embodiment, the feasible side of the first similarity between the first image of the decomposition level and the second image is obtained
Formula is as follows:
According to i-th of image block XiSimilarity between adjacent image block determines i-th of image block of the decomposition level
XiWeight, and according to i-th of image block X in the first image of the decomposition leveliIn the second image of the decomposition level
I-th of image block XiBetween similarity and i-th of image block XiWeight, obtain the first image and of the decomposition level
The first similarity between two images.
Such as i-th of image block XiFor: its adjacent image block include: image block above the image block,
Image block below the image block, the image block positioned at the image block of the image block left and positioned at the image block right,
Then the weight of the image block can be determined, such as the image based on the mutual information of at least two image blocks in this four image blocks
The weight w (i) of block=(LOn+LUnder+LIt is left+LIt is right)/4, w (i) is the weight of i-th of image block, LOnFor the figure above i-th of image block
As similarity (particularly i-th of figure in the image block and the second image in the first image above i-th of image block between block
As the similarity between the image block above block, the similarity between image block above referred to as i-th of image block), LUnderFor
The similarity between image block below i-th of image block, LIt is leftFor the similarity between the image block of i-th of image block left,
LIt is rightFor the similarity between the image block of i-th of image block right, this similarity according between adjacent image block is determined
Weight consider the pixel interdependence between the image block and adjacent image block, and then obtained according to weight and similarity
First similarity has also contemplated the pixel interdependence between image block, it is possible thereby to think that the first similarity can be based on pixel phase
Closing property obtains.
Feasible pattern for obtaining the first similarity between the first image of the decomposition level and the second image can be with
It is: based on formula:First similarity is obtained, w (i) is i-th of image block
XiWeight, L (T, F, u, Xi) be the decomposition level the first image in i-th of image block XiWith the second of the decomposition level
I-th of image block X in imageiBetween similarity so that the first similarity is the first image from the decomposition level
The full content of second image of full content and the decomposition level is considered to obtain, such as from the first image of the decomposition level and the
The all images block of two images is considered to obtain.
S1034: the local entropy of each image block in the first image of the decomposition level is determined.Why using local entropy be
Because some image details are divided into image background, and these contents are to picture quality when being split to the first image
The influence of evaluation is very small or even can be ignored, and needs all figures by the way of local entropy from the decomposition level thus
As selecting the image block being affected to image quality evaluation in block, the accuracy of image quality evaluation is improved, and reduce meter
Calculation amount.
In the present embodiment, the calculation of local entropy may is thatObtain any figure
As the local entropy of block Ω, x ∈ Ω indicates that x is any pixel point in image block Ω, and p (Ω) indicates the image grayscale of image block Ω
Probability density.In the present embodiment, can be at least one image in the first image and the second image of the decomposition level
Standard determines the local entropy of each image block at least one image.
S1035: according to the local entropy of each image block in the first image of the decomposition level, from the first of the decomposition level
Selected part image block in all image blocks of image.Inventor has found that in image block different gray values number
The value of local entropy depending on image block, when the value of local entropy is smaller, gray value is only several in image block, works as part
Grey value profile is more dispersed when the value of entropy is larger, in image block and the quantity of gray value has more, and the gray value the how right
The influence of image quality evaluation is bigger, therefore can be chosen according to local entropy in selected part image block, such as chooses pre-
If the value of the local entropy of quantity is greater than the image block of the value of the local entropy of other image blocks.
S1036: j-th of image block Y of the decomposition level chosen is determinedjWith in the second image of the decomposition level
J image block YjBetween similarity, the value of j is 1 to m, and m is the sum of the image block selected in the decomposition level.
S1037: according to j-th of image block Y of the decomposition level of selectionjWith in the second image of the decomposition level with
J image block YjSimilarity and j-th of image block Y between corresponding image blockjWeight, obtain the of the decomposition level
The second similarity between one image and the second image, wherein the weight of j-th of image block is according to adjacent with j-th of image block
The similarity of image block obtains.
It can be right refering to the explanation in step step S1032 and step S1033 for the mode that obtains of similarity and weight
This present embodiment is no longer described in detail, because of j-th of image block YjThe determination of weight consider the image block and adjacent image block
Between pixel interdependence, and then image is had also contemplated according to the second similarity that the similarity between weight and image block obtains
Pixel interdependence between block, it is possible thereby to think that the second similarity can be obtained based on pixel interdependence.And second is similar
Degree is that the image block based on selection obtains, therefore the second similarity is the portion of the first image and the second image from the decomposition level
Point content is considered to obtain, and such as considers to obtain from the parts of images block of the first image of the decomposition level and the second image.
In the present embodiment, the feasible of the second similarity between the first image of the decomposition level and the second image is obtained
Mode may is that based on formula:Obtain second similarity, w (i) be based on
J-th of image block Y in the image block that local entropy selectsjWeight, L (T, F, u, Yj) it is the figure selected based on local entropy
As j-th of image block Y in blockjWith in the second image with j-th of image block YjSimilarity between corresponding image block, such as may be used
It is indicated by mutual information, this present embodiment is no longer illustrated.Need exist for explanation be a bit: in the second image with j-th of figure
As block YjCorresponding image block is: j-th of the image block Y for sorting and selecting in the second imagejIt sorts in the first image identical
Image block.
S104: according to the first similarity and each decomposition between each the first image and the second image for decomposing level
The second similarity between the first image and the second image of level carries out image quality evaluation to image to be detected.It is wherein right
The process that image to be detected carries out image quality evaluation is as follows:
To each decomposition level: according between the first image and the second image of the decomposition level the first similarity and
The second similarity between the first image and the second image of the decomposition level, determines the First Eigenvalue of the decomposition level, root
According to each the First Eigenvalue for decomposing level, obtains the standard deviation between image to be detected and reference picture and/or obtain to be checked
Variance between altimetric image and reference picture, and according to standard deviation between image to be detected and reference picture and/or to be detected
Variance between image and reference picture carries out image quality evaluation to image to be detected.
Carrying out image quality evaluation to image to be detected in the present embodiment can foundation: image to be detected and with reference to figure
As between standard deviation or according to the variance between image to be detected and reference picture or according to image to be detected and ginseng
Examine the standard deviation and variance between image.Below then by taking the standard deviation between image to be detected and reference picture as an example, illustrate such as
What carries out image quality evaluation to image to be detected:
Formula can such as be used but be not limited to formula m=S (T, F, u) * S ' (T, F, u) and obtain the first spy of the decomposition level
Value indicative obtains fisrt feature value set { m after obtaining each the First Eigenvalue for decomposing level1,m2,......,mz, then it counts
The standard deviation of the fisrt feature value set is calculated, which is then the standard deviation between image to be detected and reference picture, works as mark
The value of quasi- difference is smaller, illustrates that image to be detected and reference picture are more similar, the distortion level of image to be detected is smaller, and then says
The picture quality of bright image to be detected is higher, otherwise illustrates that the picture quality of image to be detected is lower, and z is the total of decomposition level
Number.
In the present embodiment, the corresponding relationship of standard deviation and picture quality scoring can also be set, searched according to standard deviation
Corresponding picture quality scoring, evaluates the picture quality of image to be detected in a manner of using picture quality scoring.
From above-mentioned technical proposal it is found that being treated respectively after obtaining the reference picture of image to be detected and image to be detected
Detection image and reference picture are decomposed, and corresponding at least two first with different decomposition level of image to be detected are obtained
Image and corresponding at least two second images with different decomposition level of reference picture；To same the first figure for decomposing level
Picture and the second image: determining the first similarity between the first image of the decomposition level and the second image and determines the decomposition
The second similarity between the first image and the second image of level, according to each the first image and the second image for decomposing level
Between the first similarity and it is each decompose level the first image and the second image between the second similarity, to be detected
Image carries out image quality evaluation, wherein each the first similarity for decomposing level and the second similarity are based on being in same decomposition
Pixel interdependence between the first image and the second image of level obtains, so that commenting carrying out picture quality to image to be detected
The influence of pixel interdependence is considered when valence, and influence of the pixel interdependence to image quality evaluation is greater than the pixel difference opposite sex to image
The influence of quality evaluation, therefore commented by the picture quality of the first similarity and the second similarity that are obtained based on pixel interdependence
The accuracy of image quality evaluation can be improved in valence.
Referring to Fig. 3, can wrap it illustrates a kind of structure of image quality evaluation device provided in an embodiment of the present invention
It includes: acquiring unit 11, decomposition unit 12, similarity determining unit 13 and evaluation unit 14.
Acquiring unit 11, for obtaining the reference picture of image to be detected and image to be detected.It is to be understood that be checked
Altimetric image is the piece image for needing to carry out image quality evaluation, the reference picture of image to be detected be then to image to be detected into
Referenced image when row image quality evaluation, the reference picture of image to be detected can be image to be detected original image or
Person is some reference picture in preset reference image library, for how to determine the image to be detected in preset reference image library
Reference picture is referred to existing reference picture acquisition methods, no longer illustrates this present embodiment.
It is corresponding to obtain image to be detected for decomposing respectively to image to be detected and reference picture for decomposition unit 12
At least two first images and corresponding at least two second image of reference picture, wherein corresponding different point of each first image
Level is solved, each second image corresponds to different decomposition level, and for any first image: first image and all second figures
Second image decomposition level having the same as in.
It is so-called to decompose the image for referring to obtain the different resolution of piece image, as being 128*128 to a resolution ratio
For image, the image that the image and resolution ratio that available resolution ratio is 64*64 are 32*32, specific isolation and explanation
The related description in embodiment of the method is please referred to, this present embodiment is no longer illustrated.
Similarity determining unit 13, for same the first image and the second image for decomposing level: determining the decomposition layer
The first image and the second image of the first similarity and the determining decomposition level between the first image and the second image of grade
Between the second similarity, first similarity and first image and second image of second similarity based on the decomposition level
Between pixel interdependence obtain so that consider influence of the pixel interdependence to image quality evaluation in image quality evaluation,
To improve the accuracy of image quality evaluation.
In the present embodiment, similarity determining unit 13 structure as shown in figure 4, may include: segmentation subelement 131,
First determines that subelement 132, the first similarity determine that subelement 133, image block choose subelement 134, second and determine subelement
135 and second similarity determine subelement 136.
Divide subelement 131, for same the first image and the second image for decomposing level: to the of the decomposition level
One image and the second image are split respectively, obtain at least two image blocks and decomposition of the first image of the decomposition level
At least two image blocks of the second image of level, specific cutting procedure please refers to the related description in embodiment of the method, to this
The present embodiment no longer illustrates.
First determines subelement 132, i-th of image block X in the first image for determining the decomposition leveliWith this point
Solve i-th of image block X in the second image of leveliBetween similarity, the value of i is 1 to n, and n is the figure of the decomposition level
As the sum of block.As a kind of mode is:
Determine the discrete grey of discrete grey's collection of the first image of the decomposition level and the second image of the decomposition level
Collection, to i-th of image block X in the first image of the decomposition leveliWith i-th of image in the second image of the decomposition level
Block Xi, it is based on formula:
Obtain the decomposition level
I-th of image block X in first imageiWith i-th of image block X in the second image of the decomposition leveliBetween mutual information,
I-th of image block X mutual information being determined as in the first image of the decomposition leveliIn the second image of the decomposition level
I-th of image block XiBetween similarity.
Wherein T is the second image, and F is the first image, LTCollect for the discrete grey of the second image, LFFor first image from
Gray scale collection is dissipated, r is the pixel grey scale of the second image, and f is the pixel grey scale of the first image, and u is the corresponding deformation letter of the first image
Deformation parameter in number, p (r, f, u, Xi) be the first image and the second image joint probability density function, p (r, Xi) it is i-th
A image block XiRelative to the marginal probability density of the second image, p (f, u, Xi) it is i-th of image block XiRelative to the first image
Marginal probability density, the value of i is 1 to n, and n is the sum of the image block of the decomposition level, and detailed process please refers to method reality
The related description in example is applied, this present embodiment is no longer illustrated.
First similarity determines subelement 133, for i-th of image block X in the first image according to the decomposition leveli
With i-th of image block X in the second image of the decomposition leveliBetween similarity, obtain the first image of the decomposition level
And second the first similarity between image.Detailed process is as follows:
According to i-th of image block XiSimilarity between adjacent image block determines i-th of image block of the decomposition level
XiWeight.And according to i-th of image block X in the first image of the decomposition leveliIn the second image of the decomposition level
I-th of image block XiBetween similarity and i-th of image block XiWeight, obtain the first image and of the decomposition level
The first similarity between two images, detailed description are please referred in embodiment of the method, are no longer illustrated this present embodiment.
Image block chooses subelement 134, the local entropy of each image block in the first image for determining the decomposition level,
And according to the local entropy of each image block in the first image of the decomposition level, from all figures of the first image of the decomposition level
As selected part image block in block.Why using local entropy be because when being split to the first image and the second image,
Some image details are divided into image background, and influence of these contents to image quality evaluation is very small or even can ignore
Disregard, needs to select from all image blocks of the decomposition level to image quality evaluation shadow by the way of local entropy thus
Biggish image block is rung, improves the accuracy of image quality evaluation, and reduce calculation amount.
In the present embodiment, the calculation of local entropy may is thatObtain any figure
As the local entropy of block Ω, x ∈ Ω indicates that x is any pixel point in image block Ω, and p (Ω) indicates the image grayscale of image block Ω
Probability density.In the present embodiment, can be at least one image in the first image and the second image of the decomposition level
Standard determines the local entropy of each image block at least one image.
Second determines subelement 135, for determining j-th of image block Y of the decomposition level chosenjWith the decomposition level
The second image in j-th of image block YjSimilarity between corresponding image block, the value of j are 1 to m, and m is the decomposition layer
The sum of the image block selected in grade.
Second similarity determines subelement 136, for j-th of image block Y according to the decomposition level of selectionjWith this point
Solve level the second image in j-th of image block YjSimilarity and j-th of image block Y between corresponding image blockjPower
Weight, obtains the second similarity between the first image of the decomposition level and the second image, wherein the weight root of j-th of image block
It is obtained according to the similarity of the image block adjacent with j-th of image block.
Determine that subelement 135 and the second similarity determine for subelement 136 for second, implementation procedure and above-mentioned the
One determines that subelement 132 and the first similarity determine that subelement 133 is identical, no longer illustrates this present embodiment.
Evaluation unit 14, for according to it is each decompose level the first image and the second image between the first similarity with
And the second similarity between each the first image and the second image for decomposing level, picture quality is carried out to image to be detected and is commented
Valence.
Wherein evaluation unit 14 may include: that characteristic value determines that subelement, evaluation parameter obtain subelement and evaluation is single
Member.Characteristic value determines subelement, for each decomposition level: according between the first image and the second image of the decomposition level
The first similarity and the decomposition level the first image and the second image between the second similarity, determine the decomposition level
The First Eigenvalue.
Evaluation parameter obtains subelement, for according to it is each decompose level the First Eigenvalue, obtain image to be detected and
Standard deviation between reference picture and/or obtain the variance between image to be detected and reference picture.Subelement is evaluated, root is used for
According to the standard deviation between image to be detected and reference picture and/or the variance between image to be detected and reference picture, to be checked
Altimetric image carries out image quality evaluation.
Carrying out image quality evaluation to image to be detected in the present embodiment can foundation: image to be detected and with reference to figure
As between standard deviation or according to the variance between image to be detected and reference picture or according to image to be detected and ginseng
Examine the standard deviation and variance between image.Below then by taking the standard deviation between image to be detected and reference picture as an example, illustrate such as
What carries out image quality evaluation to image to be detected:
Formula can such as be used but be not limited to formula m=S (T, F, u) * S ' (T, F, u) and obtain the first spy of the decomposition level
Value indicative obtains fisrt feature value set { m after obtaining each the First Eigenvalue for decomposing level1,m2,......,mz, then it counts
The standard deviation of the fisrt feature value set is calculated, which is then the standard deviation between image to be detected and reference picture, works as mark
The value of quasi- difference is smaller, illustrates that image to be detected and reference picture are more similar, the distortion level of image to be detected is smaller, and then says
The picture quality of bright image to be detected is higher, otherwise illustrates that the picture quality of image to be detected is lower, and z is the total of decomposition level
Number.
In the present embodiment, the corresponding relationship of standard deviation and picture quality scoring can also be set, searched according to standard deviation
Corresponding picture quality scoring, evaluates the picture quality of image to be detected in a manner of using picture quality scoring.
From above-mentioned technical proposal it is found that being treated respectively after obtaining the reference picture of image to be detected and image to be detected
Detection image and reference picture are decomposed, and corresponding at least two first with different decomposition level of image to be detected are obtained
Image and corresponding at least two second images with different decomposition level of reference picture；To same the first figure for decomposing level
Picture and the second image: determining the first similarity between the first image of the decomposition level and the second image and determines the decomposition
The second similarity between the first image and the second image of level, according to each the first image and the second image for decomposing level
Between the first similarity and it is each decompose level the first image and the second image between the second similarity, to be detected
Image carries out image quality evaluation, wherein each the first similarity for decomposing level and the second similarity are based on being in same decomposition
Pixel interdependence between the first image and the second image of level obtains, so that commenting carrying out picture quality to image to be detected
The influence of pixel interdependence is considered when valence, and influence of the pixel interdependence to image quality evaluation is greater than the pixel difference opposite sex to image
The influence of quality evaluation, therefore commented by the picture quality of the first similarity and the second similarity that are obtained based on pixel interdependence
The accuracy of image quality evaluation can be improved in valence.
In addition, the embodiment of the present invention also provides a kind of storage medium, it is stored with computer program code on storage medium, counts
Calculation machine program code realizes above-mentioned image quality evaluating method when executing.
The embodiment of the present invention also provides a kind of image quality evaluation equipment, the image quality evaluation equipment include processor and
Memory realizes that above-mentioned image quality evaluating method, memory are used to store the figure of image to be detected when wherein processor is run
As quality evaluation result.
It should be noted that all the embodiments in this specification are described in a progressive manner, each embodiment weight
Point explanation is the difference from other embodiments, and the same or similar parts between the embodiments can be referred to each other.
For device class embodiment, since it is basically similar to the method embodiment, so being described relatively simple, related place ginseng
See the part explanation of embodiment of the method.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that the process, method, article or equipment for including a series of elements not only includes that
A little elements, but also including other elements that are not explicitly listed, or further include for this process, method, article or
The intrinsic element of equipment.In the absence of more restrictions, the element limited by sentence "including a ...", is not arranged
Except there is also other identical elements in the process, method, article or apparatus that includes the element.
The foregoing description of the disclosed embodiments can be realized those skilled in the art or using the present invention.To this
A variety of modifications of a little embodiments will be apparent for a person skilled in the art, and the general principles defined herein can
Without departing from the spirit or scope of the present invention, to realize in other embodiments.Therefore, the present invention will not be limited
It is formed on the embodiments shown herein, and is to fit to consistent with the principles and novel features disclosed in this article widest
Range.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (11)
1. a kind of image quality evaluating method, which is characterized in that the described method includes:
Obtain the reference picture of image to be detected and described image to be detected；
Described image to be detected and the reference picture are decomposed respectively, obtain described image to be detected corresponding at least two
A first image and corresponding at least two second image of the reference picture, wherein each first image corresponds to different decomposition layer
Grade, each second image correspond to different decomposition level, and for any first image: in first image and all second images
Second image decomposition level having the same；
To same the first image and the second image for decomposing level: determining between the first image of the decomposition level and the second image
The first similarity and determine the second similarity between the first image and the second image of the decomposition level, this is first similar
Degree and the pixel interdependence between the first image and the second image of second similarity based on the decomposition level obtain；
According to the first similarity between each the first image and the second image for decomposing level and each decompose the of level
The second similarity between one image and the second image carries out image quality evaluation to described image to be detected.
2. the method according to claim 1, wherein described to same the first image and the second figure for decomposing level
Picture: determine that the first similarity between the first image of the decomposition level and the second image includes:
The first image and the second image of the decomposition level are split respectively, obtain the first image of the decomposition level extremely
At least two image blocks of few two image blocks and the second image of the decomposition level；
Determine i-th of image block X in the first image of the decomposition leveliWith i-th of figure in the second image of the decomposition level
As block XiBetween similarity, the value of i is 1 to n, and n is the sum of the image block of the decomposition level；
According to i-th of image block X in the first image of the decomposition leveliWith i-th of figure in the second image of the decomposition level
As block XiBetween similarity, obtain the first similarity between the first image of the decomposition level and the second image.
3. according to the method described in claim 2, it is characterized in that, in the first image of the determination decomposition level i-th
A image block XiWith i-th of image block X in the second image of the decomposition leveliBetween similarity include:
Determine discrete grey's collection of discrete grey's collection of the first image of the decomposition level and the second image of the decomposition level；
To i-th of image block X in the first image of the decomposition leveliWith i-th of image in the second image of the decomposition level
Block Xi, it is based on formula:
Obtain the first of the decomposition level
I-th of image block X in imageiWith i-th of image block X in the second image of the decomposition leveliBetween mutual information, T
Two images, F are the first image, LTCollect for the discrete grey of the second image, LFCollect for the discrete grey of the first image, r is the second figure
The pixel grey scale of picture, f be the first image pixel grey scale, u be the corresponding deformation function of the first image in deformation parameter, p (r,
f,u,Xi) be the first image and the second image joint probability density function, p (r, Xi) it is i-th of image block XiRelative to second
The marginal probability density of image, p (f, u, Xi) it is i-th of image block XiRelative to the marginal probability density of the first image, i's is taken
Value is 1 to n, and n is the sum of the image block of the decomposition level；
By i-th of image block X in the first image of the decomposition leveliWith i-th of image in the second image of the decomposition level
Block XiBetween mutual information determine are as follows: i-th of image block X in the first image of the decomposition leveliWith the second of the decomposition level
I-th of image block X in imageiBetween similarity.
4. according to the method in claim 2 or 3, which is characterized in that in first image according to the decomposition level
I-th of image block XiWith i-th of image block X in the second image of the decomposition leveliBetween similarity, obtain the decomposition layer
Grade the first image and the second image between the first similarity include:
According to i-th of image block XiSimilarity between adjacent image block determines i-th of image block X of the decomposition leveli's
Weight；
According to i-th of image block X in the first image of the decomposition leveliWith i-th of figure in the second image of the decomposition level
As block XiBetween similarity and i-th of image block XiWeight, obtain the decomposition level the first image and the second image it
Between the first similarity.
5. the method according to claim 1, wherein the first image and the second figure of the determination decomposition level
As between the second similarity include:
The first image and the second image of the decomposition level are split respectively, obtain the first image of the decomposition level extremely
At least two image blocks of few two image blocks and the second image of the decomposition level, and determine the first image of the decomposition level
In each image block local entropy；
According to the local entropy of each image block in the first image of the decomposition level, from all of the first image of the decomposition level
Selected part image block in image block；
Determine j-th of image block Y of the decomposition level chosenjWith in the second image of the decomposition level with j-th of image block Yj
Similarity between corresponding image block, the value of j are 1 to m, and m is the sum of the image block selected in the decomposition level；
According to j-th of image block Y of the decomposition level of selectionjWith in the second image of the decomposition level with j-th of image block Yj
Similarity and j-th of image block Y between corresponding image blockjWeight, obtain the first image and of the decomposition level
The second similarity between two images, wherein phase of the weight of j-th of image block according to the image block adjacent with j-th of image block
It is obtained like degree.
It is described according to each the first image for decomposing level and the 6. according to the method described in claim 1, its feature is being
The second similarity between the first similarity and each the first image and the second image for decomposing level between two images is right
Described image to be detected carries out image quality evaluation
To each decomposition level: according to the first similarity and this point between the first image and the second image of the decomposition level
The second similarity between the first image and the second image of level is solved, determines the First Eigenvalue of the decomposition level；
According to each the First Eigenvalue for decomposing level, the standard deviation between described image to be detected and the reference picture is obtained
And/or obtain the variance between described image to be detected and the reference picture；
According to the standard deviation and/or described image to be detected and the reference between described image to be detected and the reference picture
Variance between image carries out image quality evaluation to described image to be detected.
7. a kind of image quality evaluation device, which is characterized in that described device includes:
Acquiring unit, for obtaining the reference picture of image to be detected and described image to be detected；
Decomposition unit obtains the mapping to be checked for decomposing respectively to described image to be detected and the reference picture
As corresponding at least two first image and corresponding at least two second image of the reference picture, wherein each first image
Corresponding different decomposition level, each second image correspond to different decomposition level, and for any first image: first image and
Second image decomposition level having the same in all second images；
Similarity determining unit, for same the first image and the second image for decomposing level: determining the of the decomposition level
It the first similarity between one image and the second image and determines between the first image and the second image of the decomposition level
Between second similarity, first similarity and second similarity the first image and the second image based on the decomposition level
Pixel interdependence obtains；
Evaluation unit, for according to the first similarity between each the first image and the second image for decomposing level and each
The second similarity between the first image and the second image of level is decomposed, picture quality is carried out to described image to be detected and is commented
Valence.
8. device according to claim 7, which is characterized in that the similarity determining unit, comprising:
Divide subelement, for same the first image and the second image for decomposing level: to the first image of the decomposition level
It is split respectively with the second image, obtains at least two image blocks and the decomposition level of the first image of the decomposition level
At least two image blocks of the second image；
First determines subelement, i-th of image block X in the first image for determining the decomposition leveliWith the decomposition level
I-th of image block X in second imageiBetween similarity, the value of i is 1 to n, and n is the total of the image block of the decomposition level
Number；
First similarity determines subelement, for i-th of image block X in the first image according to the decomposition leveliWith the decomposition
I-th of image block X in second image of leveliBetween similarity, obtain the first image and the second figure of the decomposition level
The first similarity as between；
Image block chooses subelement, the local entropy of each image block in the first image for determining the decomposition level, and according to
The local entropy of each image block in first image of the decomposition level, from all image blocks of the first image of the decomposition level
Selected part image block；
Second determines subelement, for determining j-th of image block Y of the decomposition level chosenjWith the second figure of the decomposition level
As in j-th of image block YjSimilarity between corresponding image block, the value of j are 1 to m, and m is to choose in the decomposition level
The sum of image block out；
Second similarity determines subelement, for j-th of image block Y according to the decomposition level of selectionjWith the decomposition level
In second image with j-th of image block YjSimilarity and j-th of image block Y between corresponding image blockjWeight, obtain
The second similarity between the first image and the second image of the decomposition level, wherein the weight of j-th of image block according to jth
The similarity of the adjacent image block of a image block obtains.
9. device according to claim 8, which is characterized in that described first determines subelement, for determining the decomposition layer
Discrete grey's collection of second image of the discrete grey's collection and decomposition level of the first image of grade；To the first of the decomposition level
I-th of image block X in imageiWith i-th of image block X in the second image of the decomposition leveli, it is based on formula:
Obtain the first of the decomposition level
I-th of image block X in imageiWith i-th of image block X in the second image of the decomposition leveliBetween mutual information, by institute
State i-th of image block X that mutual information is determined as in the first image of the decomposition leveliIn the second image of the decomposition level
I-th of image block XiBetween similarity；
Wherein T is the second image, and F is the first image, LTCollect for the discrete grey of the second image, LFFor the discrete ash of the first image
Degree collection, r are the pixel grey scale of the second image, and f is the pixel grey scale of the first image, and u is in the corresponding deformation function of the first image
Deformation parameter, p (r, f, u, Xi) be the first image and the second image joint probability density function, p (r, Xi) it is i-th of figure
As block XiRelative to the marginal probability density of the second image, p (f, u, Xi) it is i-th of image block XiSide relative to the first image
Edge probability density, the value of i are 1 to n, and n is the sum of the image block of the decomposition level.
10. device according to claim 8 or claim 9, which is characterized in that first similarity determines subelement, is used for root
According to i-th of image block XiSimilarity between adjacent image block determines i-th of image block X of the decomposition leveliWeight；And
According to i-th of image block X in the first image of the decomposition leveliWith i-th of image in the second image of the decomposition level
Block XiBetween similarity and i-th of image block XiWeight, obtain between the first image of the decomposition level and the second image
The first similarity.
11. device according to claim 7, which is characterized in that the evaluation unit, comprising:
Characteristic value determines subelement, for each decomposition level: according to the first image of the decomposition level and the second image it
Between the first similarity and the decomposition level the first image and the second image between the second similarity, determine the decomposition layer
The First Eigenvalue of grade；
Evaluation parameter obtains subelement, for according to it is each decompose level the First Eigenvalue, obtain described image to be detected and
Standard deviation between the reference picture and/or obtain the variance between described image to be detected and the reference picture；
Subelement is evaluated, for according to standard deviation between described image to be detected and the reference picture and/or described to be checked
Variance between altimetric image and the reference picture carries out image quality evaluation to described image to be detected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811352578.8A CN109523533B (en) | 2018-11-14 | 2018-11-14 | Image quality evaluation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811352578.8A CN109523533B (en) | 2018-11-14 | 2018-11-14 | Image quality evaluation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109523533A true CN109523533A (en) | 2019-03-26 |
| CN109523533B CN109523533B (en) | 2021-04-13 |
Family
ID=65777571
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811352578.8A Active CN109523533B (en) | 2018-11-14 | 2018-11-14 | Image quality evaluation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109523533B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114418972A (en) * | 2022-01-06 | 2022-04-29 | 腾讯科技（深圳）有限公司 | Picture quality detection method, device, equipment and storage medium |
| CN116385375A (en) * | 2023-03-17 | 2023-07-04 | 银河航天(北京)网络技术有限公司 | Forest defect area detection method and device based on remote sensing image and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101930607A (en) * | 2010-08-04 | 2010-12-29 | 深圳中兴力维技术有限公司 | Method for judging quality of image |
| CN103996188A (en) * | 2014-04-27 | 2014-08-20 | 嘉兴学院 | Full-reference-type image quality evaluation method based on Gabor weighted characteristics |
| CN104574399A (en) * | 2015-01-06 | 2015-04-29 | 天津大学 | Image quality evaluation method based on multi-scale vision significance and gradient magnitude |
| US20170109870A1 (en) * | 2015-10-16 | 2017-04-20 | Sogang University Research Foundation | Image processing device |
| CN107527349A (en) * | 2017-06-27 | 2017-12-29 | 广州城建职业学院 | Image partition method based on network dynamics evolutionary strategy |
| CN108269253A (en) * | 2018-01-11 | 2018-07-10 | 天津大学 | Stereo image quality evaluation method based on wavelet transformation and local structure feature |
-
2018
- 2018-11-14 CN CN201811352578.8A patent/CN109523533B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101930607A (en) * | 2010-08-04 | 2010-12-29 | 深圳中兴力维技术有限公司 | Method for judging quality of image |
| CN103996188A (en) * | 2014-04-27 | 2014-08-20 | 嘉兴学院 | Full-reference-type image quality evaluation method based on Gabor weighted characteristics |
| CN104574399A (en) * | 2015-01-06 | 2015-04-29 | 天津大学 | Image quality evaluation method based on multi-scale vision significance and gradient magnitude |
| US20170109870A1 (en) * | 2015-10-16 | 2017-04-20 | Sogang University Research Foundation | Image processing device |
| CN107527349A (en) * | 2017-06-27 | 2017-12-29 | 广州城建职业学院 | Image partition method based on network dynamics evolutionary strategy |
| CN108269253A (en) * | 2018-01-11 | 2018-07-10 | 天津大学 | Stereo image quality evaluation method based on wavelet transformation and local structure feature |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114418972A (en) * | 2022-01-06 | 2022-04-29 | 腾讯科技（深圳）有限公司 | Picture quality detection method, device, equipment and storage medium |
| CN116385375A (en) * | 2023-03-17 | 2023-07-04 | 银河航天(北京)网络技术有限公司 | Forest defect area detection method and device based on remote sensing image and storage medium |
| CN116385375B (en) * | 2023-03-17 | 2023-10-20 | 银河航天(北京)网络技术有限公司 | Forest defect area detection method and device based on remote sensing image and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109523533B (en) | 2021-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | Cross-based local stereo matching using orthogonal integral images | |
| Zeng et al. | 3D point cloud denoising using graph Laplacian regularization of a low dimensional manifold model | |
| Yitzhaky et al. | A method for objective edge detection evaluation and detector parameter selection | |
| CN104200240B (en) | A kind of Sketch Searching method based on content-adaptive Hash coding | |
| US8270724B2 (en) | Image signature matching device | |
| CN110189255A (en) | Method for detecting human face based on hierarchical detection | |
| CN108428229A (en) | It is a kind of that apparent and geometric properties lung's Texture Recognitions are extracted based on deep neural network | |
| CN1739119A (en) | Full depth map acquisition | |
| EP2528035B1 (en) | Apparatus and method for detecting a vertex of an image | |
| US7928978B2 (en) | Method for generating multi-resolution three-dimensional model | |
| CN109978871B (en) | Fiber bundle screening method integrating probability type and determination type fiber bundle tracking | |
| CN108805863B (en) | Method for detecting image change by combining deep convolutional neural network with morphology | |
| CN101630405A (en) | Multi-focusing image fusion method utilizing core Fisher classification and redundant wavelet transformation | |
| CN1322471C (en) | Comparing patterns | |
| CN106228528A (en) | A kind of multi-focus image fusing method based on decision diagram Yu rarefaction representation | |
| CN108876756A (en) | The measure and device of image similarity | |
| CN108122221A (en) | The dividing method and device of diffusion-weighted imaging image midbrain ischemic area | |
| CN102136063A (en) | Zernike moment-based rapid image comparison method | |
| CN106651792A (en) | Method and device for removing strip noise of satellite image | |
| CN109523533A (en) | A kind of image quality evaluating method and device | |
| Gao et al. | Augmenting LIDC dataset using 3D generative adversarial networks to improve lung nodule detection | |
| CN107506769A (en) | A kind of extracting method and system of urban water-body information | |
| CN109816618A (en) | A kind of region energy photon counting Image Fusion based on adaptive threshold | |
| Nunsong et al. | Modified differential box-counting method using weighted triangle-box partition | |
| Guo et al. | Evaluating the local visibility of geometric artifacts |
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 |