CN108156451A - A kind of 3-D view/video without reference mass appraisal procedure - Google Patents
A kind of 3-D view/video without reference mass appraisal procedure Download PDFInfo
- Publication number
- CN108156451A CN108156451A CN201711330503.5A CN201711330503A CN108156451A CN 108156451 A CN108156451 A CN 108156451A CN 201711330503 A CN201711330503 A CN 201711330503A CN 108156451 A CN108156451 A CN 108156451A
- Authority
- CN
- China
- Prior art keywords
- view
- appraisal procedure
- sub
- block
- video
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/004—Diagnosis, testing or measuring for television systems or their details for digital television systems
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
- H04N2013/0074—Stereoscopic image analysis
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
Abstract
The invention discloses a kind of 3-D view/video without reference mass appraisal procedure, the input that it receives is the 3-D view of MVD forms：Include left view, the depth map of left view, right view, each one of the depth map of right view.Quality evaluation process is as follows：（1）Using virtual view I1 of the DIBR algorithms generation left view in binocular center；Generate virtual view I2 of the right view in binocular center；（2）The structural similarity of I1 and I2 is calculated using SSIM algorithms, and is exported using this value as the quality assessment value of input 3-D view.
Description
Technical field
It is more particularly to a kind of to MVD forms the present invention relates to a kind of 3-D view/video without reference mass appraisal procedure
The 3-D view of presentation/video is carried out without reference mass appraisal procedure.
Background technology
Traditional two dimensional image and video quality evaluation method has been studied many years.With 3 D video application it is fast
Speed development, the correlative study assessed 3-D view and video quality is more and more, including with reference and without reference
Method for evaluating quality：M.J.Chen proposes a kind of method for evaluating quality referred to entirely of double vision 3-D view；Z.Sazzad is carried
A kind of method for evaluating quality without reference of double vision 3-D view is gone out, but cannot to meet 3-D view automatic for this appraisal procedure
The demand of broadcasting regards so that having to create necessary free view-point based on depth information using DIBR algorithms in the receiver
Figure, regrettably DIBR introduces completely new quality flaw；The full reference mass that E.Bosc proposes common 3-D view is commented
Estimate method, but due to being difficult to obtain reference picture in the receiver ends of 3d television systems and be difficult to functionization；M.Solh is proposed
A kind of method that the 3-D view that a depth map (2d+z) is added to present to single-frame images progress is assessed without reference mass, but the party
The shortcomings that method is that it cannot be carried out using the relationship between different views in multi-viewpoint 3 D image (MVD) form with depth map
Prediction.
Invention content
It is that the 3-D view that MVD forms are presented is carried out without with reference to matter for no one of these existing appraisal procedures
The phenomenon that amount assessment, the present invention provide a kind of 3-D view/video without reference mass appraisal procedure, MVD forms are presented
3-D view/video assessed without reference mass, is designed for the real-time quality monitoring in three-dimensional television system.
The present invention uses following technical scheme to solve above-mentioned technical problem：
The present invention provide a kind of 3-D view/video without reference mass appraisal procedure, the method for evaluating quality it is specific
Step is as follows：
(1) 3-D view of MVD forms is inputted；
(2) using DIBR algorithms, left view is generated respectively in the virtual view I1 of binocular center, right view in binocular
The virtual view I2 of center；
(3) structural similarity of the I1 generated in step (2) and I2 are calculated；
(4) it is the value of structural similarity being calculated in step (3) is defeated as the quality assessment value of input 3-D view
Go out.
As the further technical solution of the present invention, the 3-D view of MVD forms inputted in step (1) includes left view
The depth map of figure, the depth map of left view, right view, right view.
The I1 generated in step (2) is calculated using SSIM algorithms as the further technical solution of the present invention, in step (3)
With the structural similarity of I2.
As the further technical solution of the present invention, the structural similarity of I1 and I2 are：
Wherein, SSIM (I1, I2) represents the structural similarity of I1 and I2, and M represents the number of the sub-block of I1 and I2, Point
Not Biao Shi I1 i-th of sub-block, the mean value of i-th of sub-block of I2,Respectively represent I1 i-th of sub-block,
The standard deviation of i-th of sub-block of I2,It representsAssociation between i-th of sub-block of I1 and i-th of sub-block of I2
Variance, C1、C2、C3It is constant.
As the further technical solution of the present invention, C3=C2/2.
The present invention compared with prior art, has following technique effect using above technical scheme：The present invention proposes needle
To a kind of method for evaluating quality SVC of no reference of 3-D view/video of MVD forms, it is suitable in three-dimensional television system
Real-time quality monitoring.As the method for evaluating quality of no reference, SVC can receiving terminal there is no reference picture (original not
The image of distortion) in the case of work；Largely being calculated in SVC can complete to connect so as to be greatly reduced in three-dimensional television system
The calculating intensity of receiving end；Experiment shows that SVC has the not defeated accuracy rate in the method for evaluating quality referred to entirely.
Description of the drawings
Fig. 1 is three-dimensional television system schematic diagram；
Fig. 2 is the schematic diagram of SVC methods；
Fig. 3 is DIBR algorithm flow block diagrams；
Fig. 4 is that the scatter plot of different quality appraisal procedure compares, wherein, (a) is the method for the present invention, and (b) is SSIM side
Method, (c) are MS-SSIM methods, and (d) is VIFP methods；
Fig. 5 is flow chart of the method for the present invention.
Specific embodiment
Technical scheme of the present invention is described in further detail below in conjunction with the accompanying drawings：
This patent proposes a kind of method that 3-D view based on MVD forms carries out reference-free quality evaluation, this method
It is to be designed for the real-time quality monitoring in three-dimensional television system as shown in Figure 1.The appraisal procedure is named as conjunction by we
Into view comparative approach (Synthesized View Comparison Method, SVC), it generates left view using DIBR algorithms
The virtual view of figure and each same viewpoint of leisure of right view, and the difference between the two virtual views compares.
Two to be weighed using the structural similarity (SSIM) in conventional two-dimensional image method for evaluating quality virtual in our realization
Difference between view.If left and right view texture information and the perfect defect of depth information, base between the two virtual views
Architectural difference is not present in this.Therefore, by comparing both synthesis views, SVC methods can be measured by texture and depth information
In flaw caused by synthesis viewing quality decline.Compared with other video quality assessment schemes, SVC is a three-dimensional television
Workable lightweight solutions in system, it make full use of the virtual view that is generated in receiving terminal using DIBR algorithms and by
This design quality evaluation scheme, in this scheme, its most of computation-intensive task is complete via three-dimensional television system
Into.
It in this patent, can be with for more than two view it is assumed that 3 D video is using two views in left and right
Simple extension.3 D video is stored with MVD forms, it includes the depth of left view, the depth map of left view, right view and right view
Degree figure.
Fig. 2 describes the flow diagram of SVC methods.We using left view and right view in eyes middle position respectively
Virtual view is generated using DIBR, in the case that imperforate, the structure between the two virtual views is compared using SSIM
Sex differernce, and the assessed value of the quality in this, as the 3 D video.
2004, C.Fehn proposed a kind of Rendering algorithms based on depth map, i.e. DIBR.Present DIBR has become
The mainstream algorithm of free view-point view generation.Depth map is converted to disparity map, each picture by the algorithm according to objective circumstances first
Element is moved to new position by according to its situation in disparity map in newly-generated view.Finally, it will be more by merging
Multiple virtual view informations that a different cameras generates, complete the filling of hole, Fig. 3 gives the block diagram of this process.
The receiving terminal of one true three-dimensional television system, SVC can save a large amount of meter using the output that pixel in DIBR moves
Evaluation time.
Equally 2004, Z.Wang et al. by distorted image compared with the brightness of reference picture, contrast compares and structure
Result of the comparison is combined, and carries out image quality measure.The digital picture x and y of given two alignment, their comparison is base
In：
Wherein, μxAnd σx(μyAnd σy) be (x, y) signal mean value and standard deviation, σxyIt is the covariance between x and y, introduces
Constant C1、C2、C3It is the unstability in order to avoid numerical computations.
Finally, SSIM characterizes picture quality with the value being calculated as below：
Wherein, xi,yiIt is the picture material of serial number i positions, and M is the counting of picture material.
We used seven cycle tests to carry out subjective evaluation:" the Poznan that Poznan Polytechnics provides
Street ", by Nokia provide " Undo_Dancer " and " GT_Fly ", by Nagoya University offer " Kendo " and
It " Balloons ", " Newspaper " that is provided by Gwangju Universities of Science and Technology and is provided by Information ＆ Communication Technology research institute of South Korea
" Shark ".For image fault caused by analog compression and packet loss, we carry out texture and depth image using JM softwares
Compression (setting QP=28), then carries out packet loss simulation, packet loss 3%.Next, we generate the centre finally synthesized
View, it is formed by carrying out holes filling after left view and the virtual view of each self-generating eyes center of right view.It adopts
With the method for stratified random smapling, 210 width images in all medial views finally synthesized are carried out with subjective/objective comparison.
Finally, a subjective testing being made of 22 subjective marking persons has been carried out.Subjective scoring, because subject is unfamiliar with computer
The video content of synthesis, it is also possible to artistic effect and distortion are not differentiated between, so we employ double stimulation continuous mass scaling laws
(DSCQS), each image on data set obtains different average (DMOS).
Fig. 4 is that the scatter plot of different quality appraisal procedure compares, and the longitudinal axis and horizontal axis represent subjective point and objective point respectively, often
A point represents a sample as a result, wherein (a) is appraisal procedure SVC, PCC=proposed by the present invention in test set
0.5822, SROCC=0.6284；(b) it is SSIM, PCC=0.4635, SROCC=0.4730；(c) it is MS-SSIM, PCC=
0.7221, SROCC=0.7050；(d) it is VIFP, PCC=0.5528, SROCC=0.5655.
(a) in Fig. 4 shows the scatter plot of subjectivity/objective comparison that SVC is provided, and carries out nonlinear regression, Pearson came
Related coefficient (PCC) is 0.5822, and Spearman rank correlation coefficient (SROCC) is 0.6284.We also calculate final conjunction
Into virtual eyes center view and real camera shooting eyes center view between SSIM, MS-
SSIM (multiple dimensioned SSIM) and VIFP (the visual information fidelity rule in pixel domain), and use these popular full reference pictures
Performance comparison other of the quality evaluating method as SVC carries out nonlinear regression, as a result sees (b), (c), (d) in Fig. 4.I
It can be found that SVC, one for the 3-D view of MVD forms without reference mass appraisal procedure, performance is in this data set
It is upper to be better than traditional full reference index, SSIM and VIFP.
As the method for evaluating quality of no reference, reference picture (original undistorted figure can be not present in receiving terminal in SVC
Picture) in the case of work；Largely being calculated in SVC can complete that the meter of receiving terminal is greatly reduced in three-dimensional television system
Calculate intensity；Experiment shows that SVC has the not defeated accuracy rate in the method for evaluating quality referred to entirely.
SVC is a 3-D view/video quality evaluation method, and the input that it receives is the 3-D view of MVD forms：Packet
Containing left view, the depth map of left view, right view, each one of the depth map of right view.Quality evaluation process is as shown in Figure 5：
(1) using virtual view I1 of the DIBR algorithms generation left view in binocular center；Right view is generated in binocular
The virtual view I2 of center；
(2) structural similarity of I1 and I2 is calculated using SSIM algorithms, and using this value as the matter of input 3-D view
Measure assessed value output.
The above, the only specific embodiment in the present invention, but protection scope of the present invention is not limited thereto are appointed
What be familiar with the people of the technology disclosed herein technical scope in, it will be appreciated that the transformation or replacement expected should all be covered
Within the scope of the present invention, therefore, protection scope of the present invention should be subject to the protection domain of claims.
Claims (5)
1. a kind of 3-D view/video without reference mass appraisal procedure, which is characterized in that the specific step of the method for evaluating quality
It is rapid as follows：
(1) 3-D view of MVD forms is inputted；
(2) using DIBR algorithms, left view is generated respectively in the virtual view I1 of binocular center, right view at binocular center
The virtual view I2 of position；
(3) structural similarity of the I1 generated in step (2) and I2 are calculated；
(4) it is exported the value of structural similarity being calculated in step (3) as the quality assessment value of input 3-D view.
2. a kind of 3-D view/video according to claim 1 without reference mass appraisal procedure, which is characterized in that step
Suddenly the 3-D view of the MVD forms inputted in (1) includes the depth of left view, the depth map of left view, right view, right view
Figure.
3. a kind of 3-D view/video according to claim 1 without reference mass appraisal procedure, which is characterized in that step
Suddenly the structural similarity of the I1 generated in step (2) and I2 are calculated in (3) using SSIM algorithms.
4. a kind of 3-D view/video according to claim 3 without reference mass appraisal procedure, which is characterized in that I1
Structural similarity with I2 is：
Wherein, SSIM (I1, I2) represents the structural similarity of I1 and I2, and M represents the number of the sub-block of I1 and I2, Point
Not Biao Shi I1 i-th of sub-block, the mean value of i-th of sub-block of I2,Respectively represent I1 i-th of sub-block,
The standard deviation of i-th of sub-block of I2,It representsAssociation between i-th of sub-block of I1 and i-th of sub-block of I2
Variance, C1、C2、C3It is constant.
5. a kind of 3-D view/video according to claim 4 without reference mass appraisal procedure, which is characterized in that C3
=C2/2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711330503.5A CN108156451B (en) | 2017-12-11 | 2017-12-11 | A kind of 3-D image/video without reference mass appraisal procedure |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711330503.5A CN108156451B (en) | 2017-12-11 | 2017-12-11 | A kind of 3-D image/video without reference mass appraisal procedure |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108156451A true CN108156451A (en) | 2018-06-12 |
| CN108156451B CN108156451B (en) | 2019-09-13 |
Family
ID=62467192
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711330503.5A Active CN108156451B (en) | 2017-12-11 | 2017-12-11 | A kind of 3-D image/video without reference mass appraisal procedure |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108156451B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104853175A (en) * | 2015-04-24 | 2015-08-19 | 张艳 | Novel synthesized virtual viewpoint objective quality evaluation method |
| CN105407349A (en) * | 2015-11-30 | 2016-03-16 | 宁波大学 | No-reference objective three-dimensional image quality evaluation method based on binocular visual perception |
| CN106341677A (en) * | 2015-07-07 | 2017-01-18 | 中国科学院深圳先进技术研究院 | Virtual viewpoint video quality evaluation method |
| CN107371016A (en) * | 2017-07-25 | 2017-11-21 | 天津大学 | Based on asymmetric distortion without with reference to 3D stereo image quality evaluation methods |
-
2017
- 2017-12-11 CN CN201711330503.5A patent/CN108156451B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104853175A (en) * | 2015-04-24 | 2015-08-19 | 张艳 | Novel synthesized virtual viewpoint objective quality evaluation method |
| CN106341677A (en) * | 2015-07-07 | 2017-01-18 | 中国科学院深圳先进技术研究院 | Virtual viewpoint video quality evaluation method |
| CN105407349A (en) * | 2015-11-30 | 2016-03-16 | 宁波大学 | No-reference objective three-dimensional image quality evaluation method based on binocular visual perception |
| CN107371016A (en) * | 2017-07-25 | 2017-11-21 | 天津大学 | Based on asymmetric distortion without with reference to 3D stereo image quality evaluation methods |
Non-Patent Citations (2)
| Title |
|---|
| 李永生: "基于Contourlet变换的平面和立体图像质量评价算法研究", 《江南大学硕士学位论文》 * |
| 马允: "无参考立体图像客观质量评价方法研究", 《宁波大学硕士学位论文》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108156451B (en) | 2019-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103763552B (en) | Stereoscopic image non-reference quality evaluation method based on visual perception characteristics | |
| CN103152600B (en) | Three-dimensional video quality evaluation method | |
| CN101720047B (en) | Method for acquiring range image by stereo matching of multi-aperture photographing based on color segmentation | |
| CN108495110A (en) | A kind of virtual visual point image generating method fighting network based on production | |
| CN104756489B (en) | A kind of virtual visual point synthesizing method and system | |
| CN101902657A (en) | Method for generating virtual multi-viewpoint images based on depth image layering | |
| CN103780895B (en) | A kind of three-dimensional video quality evaluation method | |
| Lie et al. | 2D to 3D video conversion with key-frame depth propagation and trilateral filtering | |
| CN104036502B (en) | A kind of without with reference to fuzzy distortion stereo image quality evaluation methodology | |
| CN107240147B (en) | Image rendering method and system | |
| CN106408513A (en) | Super-resolution reconstruction method of depth map | |
| CN107360416A (en) | Stereo image quality evaluation method based on local multivariate Gaussian description | |
| CN109345502A (en) | A kind of stereo image quality evaluation method based on disparity map stereochemical structure information extraction | |
| CN108337504A (en) | A kind of method and device of evaluation video quality | |
| CN101662695B (en) | Method and device for acquiring virtual viewport | |
| CN104853175B (en) | Novel synthesized virtual viewpoint objective quality evaluation method | |
| CN112102504A (en) | Three-dimensional scene and two-dimensional image mixing method based on mixed reality | |
| CN109801273B (en) | Light field image quality evaluation method based on polar plane linear similarity | |
| EP4191539A1 (en) | Method for performing volumetric reconstruction | |
| CN104661014B (en) | The gap filling method that space-time combines | |
| CN105430397B (en) | A kind of 3D rendering Quality of experience Forecasting Methodology and device | |
| CN103745457A (en) | Stereo image objective quality evaluation method | |
| CN108848365A (en) | A kind of reorientation stereo image quality evaluation method | |
| CN108156451B (en) | A kind of 3-D image/video without reference mass appraisal procedure | |
| CN109257591A (en) | Based on rarefaction representation without reference stereoscopic video quality method for objectively evaluating |
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 |