CN108566547A - A kind of method for evaluating video quality of optimization - Google Patents
A kind of method for evaluating video quality of optimization Download PDFInfo
- Publication number
- CN108566547A CN108566547A CN201711004920.0A CN201711004920A CN108566547A CN 108566547 A CN108566547 A CN 108566547A CN 201711004920 A CN201711004920 A CN 201711004920A CN 108566547 A CN108566547 A CN 108566547A
- Authority
- CN
- China
- Prior art keywords
- frame
- ssim
- video
- image
- picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
Abstract
The present invention relates to multimedia messages (picture and video), machine vision, computation vision fields, it can be made up in objective evaluation method of video quality for the shortage in terms of user experience more particularly to a kind of, be more in line with the method for evaluating video quality of the optimization of practical application；The frame number of the image of SSIM ＜ 0.6 is taken out, the frame number calculated between any two is poor；If first frame has an impact viewing, poor according to interframe serial number, judge whether the damage of the picture has an impact viewing；If ▽ N >=N/2, a later frame have an impact；If ▽ N ＜ N/2, a later frame is without influence；Step 4：For the frame of no influence, its SSIM=0.6 is set, for influential picture, keeps SSIM constant；Step 5：The average value S2 of the SSIM of the image of all SSIM ＜ 0.6 is recalculated, theoretically, 0.0 ＜ S2 ＜ 0.6 represent the degree for influencing video-see；6th step：According to result above, created according to subjective 5 points of evaluation criterions processed《The double standards system tables of 9 points of user experiences and video impairment processed》, to use.
Description
Technical field
The present invention relates to multimedia messages (picture and video), machine vision, computation vision fields, and in particular to a kind of
The method for evaluating video quality of optimization.
Background technology
Current video quality evaluation method is mostly to be evaluated using for the video after lossy compression.Method
It is broadly divided into two major classes：Subjective evaluation method and method for objectively evaluating.The former matter with percipient subjective feeling evaluation object
Amount；The latter weighs video image quality according to the quantizating index that model provides.Objective Video evaluation method is by establishing mathematics
Model obtains quantizating index or parameter after progress is related and obtains video quality to weigh picture quality, then by mathematical model.
According to whether having original reference video, full reference video quality evaluation, the reference of matter drop and non-reference picture evaluation side can be divided into
Method.Full reference evaluation method is that the video sequence of the distortion video sequence with evaluation and test and original reference is carried out check analysis, root
The quality of distortion video sequence to be evaluated is calculated according to the result of analysis.Full reference evaluation method be current development more at
Ripe method for objectively evaluating, and more one kind is applied, wherein well-known structural similarity image measurement (SSIM)
It is exactly one kind of full reference method.
Full reference evaluation method is actually to analyze the picture in video, to the correspondence picture of two videos
To carrying out similarity analysis.This method is concluded that according to the average value per frame picture quality to judge video matter
Amount.But the quality of video, in addition to seeing picture quality, should more be associated with the drop of picture quality from essence and practicability
It is low whether to influence to watch the quality and availability that more fully evaluate video.
Invention content
The object of the present invention is to provide one kind can make up in objective evaluation method of video quality for user's body
The shortage in proved recipe face is more in line with the method for evaluating video quality of the optimization of practical application.
Technical solution is used by invention solves its technical problem：A kind of method for evaluating video quality of optimization, including
Following steps：
Step 1：
It is assumed that X and Y respectively represent original and distorted image signal.The calculating similarity of the task is divided by the algorithm
Three parts.It is brightness comparison function, contrast comparison function and structure respectively to define l (x, y), c (x, y) and s (x, y)
Comparison function.The mean value of brightness in piece image is calculated first：
Wherein, N indicates the number for the pixel that piece image is possessed, xiIndicate bright corresponding to each pixel in image X
Angle value.
The computational methods of standard deviation and covariance：
Brightness comparison function, contrast comparison function and structure comparison function can be calculate by the following formula to obtain respectively：
By formula (2-8) it is found that the range of the object quality score of single-frame images is 0-1 is scored at 1 and indicates the frame image
It is high-quality；And when being scored at 0, indicate the of poor quality of image；
Step 2：The frame number of the image of SSIM ＜ 0.6 is taken out, the frame number calculated between any two is poor；
Step 3：If first frame has an impact viewing, poor according to interframe serial number, the damage of the picture is judged
Whether wound has an impact viewing；
If ▽ N >=N/2, a later frame have an impact；If ▽ N ＜ N/2, a later frame is without influence；
Step 4：For the frame of no influence, its SSIM=0.6 is set, for influential picture, keeps SSIM constant；
Step 5：The average value S2 of the SSIM of the image of all SSIM ＜ 0.6 is recalculated, theoretically, 0.0 ＜ S2 ＜
0.6, represent the degree for influencing video-see；
6th step：According to result above, created according to subjective 5 points of evaluation criterions processed《9 points of user experience and video damages processed
The double standards system tables of wound》, to use.
The beneficial effects of the invention are as follows：The method of the fast video audit of the present invention can make up in video quality objective
For the shortage in terms of user experience in evaluation method, it is more in line with the method for objectively evaluating of practical application, practicability higher,
The evaluation method robustness higher simultaneously, evaluation video are more comprehensive.
Description of the drawings
Fig. 1 is the schematic diagram of the human eye video identification of the fast video audit of the present invention.
Specific implementation mode
Invention is described in further detail presently in connection with attached drawing.These attached drawings are simplified schematic diagram, only to show
Meaning mode illustrates the basic structure of invention, therefore it only shows and invents related composition.
As shown in Figure 1, human eye have video persistence characteristic, that is, in visual scene actual Loss Of View it
Afterwards, the picture that human eye is seen can't disappear immediately.On this basis, human eye also have temporal masking, refer to regarding
When the case where scene switching etc. occurred in frequency sequence leads to picture acute variation, the visual resolving power of human eye can be caused
Acutely decline suddenly.In some cases, visual resolving power can drop to original 1/10, even meeting under extreme case
It is lower.It means that when watching video, if occurring new scene suddenly, due to the decline of visual resolving power, human eye
The emerging scene of Chu is not seen substantially.Probably need the adjustment by 0.5s, visual resolving power that can just be gradually restored to
State originally.It can be seen that within this 0.5s time that resolving power declines, it is not necessary that play the very high picture of resolving power
Therefore there is plurality of pictures damage in 0.5s in face, be essentially not influence viewing.Assuming that the frame per second of video sequence is N
/ s, then averagely occur N/2 pictures in 0.5s, then occurs the picture of two damages, theoretically, second in N/2 pictures
It is not influence viewing to open damage picture, and dotted line represents damage picture in Fig. 1.
Based on principles above, according to the object quality score for calculating single frames picture based on SSIM
It is assumed that X and Y respectively represent original and distorted image signal.The calculating similarity of the task is divided by the algorithm
Three parts.It is brightness comparison function, contrast comparison function and structure respectively to define l (x, y), c (x, y) and s (x, y)
Comparison function.The mean value of brightness in piece image is calculated first：
Wherein, N indicates the number for the pixel that piece image is possessed, xiIndicate bright corresponding to each pixel in image X
Angle value.
The computational methods of standard deviation and covariance：
Brightness comparison function, contrast comparison function and structure comparison function can be calculate by the following formula to obtain respectively：
By formula (2-8) it is found that the range of the object quality score of single-frame images is [0,1] is scored at 1 and indicates the frame
The quality of image is very good；And when being scored at 0, indicate that the quality of image is excessively poor, the information almost all of image is broken
It is bad.
The average value S1 of SSIM (x, y) is calculated, but information above is merely able to indicate the picture extent of damage.According to subjectivity
The five-grade marking system evaluation criterion is evaluated, can be classified to image lesion program.
1 picture object quality score of table damages compareing for scale subjective assessment standard with the five-grade marking system
| Score value | Picture degree of injury | Objective score |
| 1 | Damage seriously affects viewing | 0.0~0.2 |
| 2 | Damage influence is watched | 0.2~0.4 |
| 3 | Damage is apparent, is slightly influenced on viewing | 0.4~0.6 |
| 4 | Subtle damage does not influence to watch | 0.6~0.8 |
| 5 | It can not discover damage | 0.8~1.0 |
If picture has damage, but does not influence picture viewing, then the viewing of video is not interfered in turn.It is bright for damaging
Aobvious, the image for influencing viewing is analyzed.
The image that SSIM values are more than 0.6 is assumed herein, does not influence the viewing of picture.
According to upper table and above-mentioned principle, it is as follows that analysis corrupted picture influences viewing degree analyzing method to video：
Step 2：Take out SSIM<The frame number of 0.6 image calculates frame number difference ▽ N between any two；
Step 3：If first frame has an impact viewing, poor according to interframe serial number, judge that the damage of the picture sees if there is sight
It influences.If ▽ N >=N/2, a later frame have an impact；If ▽ N ＜ N/2, a later frame is without influence；
Step 4：For the frame of no influence, its SSIM=0.6 is set, for influential picture, keeps SSIM constant；
Step 5：The average value S2 of the SSIM of the image of all SSIM ＜ 0.6 is recalculated, theoretically, 0.0 ＜ S2 ＜
0.6, represent the degree for influencing video-see.
S2, which is represented, influences viewing program, and S1 represents video and is damaged program.Following table is marked according to 5 points of subjective assessment systems
The double standards system tables of 9 points of systems of quasi- Establishing, weigh the quality of video in terms of user experience and video impairment two.
Table 29 divides user experience and video impairment processed double standards system tables
The method of the fast video audit of the present invention is two angles of damage and human eye viewing experience according to picture in video
Degree goes to weigh the method for objectively evaluating of video quality；The damage picture of user's viewing is not influenced simultaneously according to the specificity analysis of human eye
And establish the new standard system table of evaluation video quality.
The method of the fast video audit of the present invention can make up in objective evaluation method of video quality for user's body
The shortage in proved recipe face is more in line with the method for objectively evaluating of practical application, practicability higher, while the evaluation method robustness
Higher, evaluation video are more comprehensive.
It is enlightenment with the above-mentioned desirable embodiment according to invention, through the above description, relevant staff is complete
Can without departing from the scope of the technological thought of the present invention', carry out various changes and amendments, this invention it is technical
Range is not limited to the contents of the specification, it is necessary to determine its technical scope according to right.
Claims (1)
1. a kind of method for evaluating video quality of optimization, it is characterised in that：Include the following steps：
Step 1：
It is assumed that X and Y respectively represent original and distorted image signal.The calculating similarity of the task is divided into three by the algorithm
Part.It is that brightness comparison function, contrast comparison function and structure compare letter respectively to define l (x, y), c (x, y) and s (x, y)
Number.The mean value of brightness in piece image is calculated first：
Wherein, N indicates the number for the pixel that piece image is possessed, xiIndicate the brightness value corresponding to each pixel in image X.
The computational methods of standard deviation and covariance：
Brightness comparison function, contrast comparison function and structure comparison function can be calculate by the following formula to obtain respectively：
By formula (2-8) it is found that the range of the object quality score of single-frame images, which is 0-1, is scored at 1 matter for indicating the frame image
It measures；And when being scored at 0, indicate the of poor quality of image；
Step 2：The frame number of the image of SSIM ＜ 0.6 is taken out, the frame number calculated between any two is poor
Step 3：If first frame has an impact viewing, poor according to interframe serial number, judge that the damage of the picture sees if there is shadow to sight
It rings；IfThen a later frame has an impact；IfThen a later frame is without influence；
Step 4：For the frame of no influence, its SSIM=0.6 is set, for influential picture, keeps SSIM constant；
Step 5：The average value S2 of the SSIM of the image of all SSIM ＜ 0.6 is recalculated, theoretically, 0.0 ＜ S2 ＜ 0.6, generation
Table influences the degree of video-see；
6th step：According to result above, created according to subjective 5 points of evaluation criterions processed《The double marks of 9 points of user experiences and video impairment processed
Quasi- system table》, to use.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711004920.0A CN108566547A (en) | 2017-10-25 | 2017-10-25 | A kind of method for evaluating video quality of optimization |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711004920.0A CN108566547A (en) | 2017-10-25 | 2017-10-25 | A kind of method for evaluating video quality of optimization |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108566547A true CN108566547A (en) | 2018-09-21 |
Family
ID=63530275
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711004920.0A Pending CN108566547A (en) | 2017-10-25 | 2017-10-25 | A kind of method for evaluating video quality of optimization |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108566547A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101695141A (en) * | 2009-10-20 | 2010-04-14 | 浙江大学 | Method and device for evaluating video quality |
| CN103544708A (en) * | 2013-10-31 | 2014-01-29 | 南京邮电大学 | Image quality objective evaluation method based on MMTD |
| CN104123723A (en) * | 2014-07-08 | 2014-10-29 | 上海交通大学 | Structure compensation based image quality evaluation method |
| CN105979266A (en) * | 2016-05-06 | 2016-09-28 | 西安电子科技大学 | Interframe relevance and time slot worst based time domain information fusion method |
| CN106875389A (en) * | 2017-02-23 | 2017-06-20 | 天津大学 | Three-dimensional video quality evaluation method based on motion conspicuousness |
-
2017
- 2017-10-25 CN CN201711004920.0A patent/CN108566547A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101695141A (en) * | 2009-10-20 | 2010-04-14 | 浙江大学 | Method and device for evaluating video quality |
| CN103544708A (en) * | 2013-10-31 | 2014-01-29 | 南京邮电大学 | Image quality objective evaluation method based on MMTD |
| CN104123723A (en) * | 2014-07-08 | 2014-10-29 | 上海交通大学 | Structure compensation based image quality evaluation method |
| CN105979266A (en) * | 2016-05-06 | 2016-09-28 | 西安电子科技大学 | Interframe relevance and time slot worst based time domain information fusion method |
| CN106875389A (en) * | 2017-02-23 | 2017-06-20 | 天津大学 | Three-dimensional video quality evaluation method based on motion conspicuousness |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100559880C (en) | A kind of highly-clear video image quality evaluation method and device based on self-adapted ST area | |
| EP0940996B1 (en) | A method, an arrangement, an ASIC and their use for objective assessment of video quality | |
| Jayaraman et al. | Objective quality assessment of multiply distorted images | |
| EP1874055A2 (en) | Pulldown correction for progressive scan display of audiovisual recordings | |
| Min et al. | Visual attention data for image quality assessment databases | |
| CN114598864A (en) | Full-reference ultrahigh-definition video quality objective evaluation method based on deep learning | |
| Patil et al. | Survey on image quality assessment techniques | |
| Opozda et al. | The survey of subjective and objective methods for quality assessment of 2D and 3D images | |
| Rousselot et al. | Quality metric aggregation for HDR/WCG images | |
| CN109413413B (en) | Video quality evaluation method and device | |
| Devnani et al. | Comparative analysis of image quality measures | |
| CN109360189B (en) | Method for detecting pixel defect point of image of uncooled infrared movement | |
| Gao et al. | A content-based image quality metric | |
| CN108566547A (en) | A kind of method for evaluating video quality of optimization | |
| CN103763498B (en) | HDMI signal processing systems and processing method | |
| Fry et al. | Bridging the gap between imaging performance and image quality measures | |
| CN114693553A (en) | Mobile intelligent terminal image processing method and system | |
| CN112581461B (en) | No-reference image quality evaluation method and device based on generation network | |
| CN112950479A (en) | Image gray level region stretching algorithm | |
| CN111179197B (en) | Contrast enhancement method and device | |
| US20050105802A1 (en) | Method and an arrangement for objective assessment of video quality | |
| Fang et al. | A spatial-temporal weighted method for asymmetrically distorted stereo video quality assessment | |
| Attar et al. | Ebiqa: An edge based image quality assessment | |
| CN109587474A (en) | No-reference video quality evaluating method and device based on distortion restoring degree | |
| Wei et al. | Non-invasive image quality assessment based on eye-tracking |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180921 |
|
| RJ01 | Rejection of invention patent application after publication |