CN105721863B - Method for evaluating video quality - Google Patents
Method for evaluating video quality Download PDFInfo
- Publication number
- CN105721863B CN105721863B CN201610072995.1A CN201610072995A CN105721863B CN 105721863 B CN105721863 B CN 105721863B CN 201610072995 A CN201610072995 A CN 201610072995A CN 105721863 B CN105721863 B CN 105721863B
- Authority
- CN
- China
- Prior art keywords
- image
- notable
- source images
- target image
- algorithms
- 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.)
- Expired - Fee Related
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
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)
- Image Analysis (AREA)
Abstract
The present invention proposes a kind of full reference image quality appraisement method based on picture structure and human-eye visual characteristic, first obtain the source images and target image of multi-view point video, the notable figure of the area-of-interest of target image is extracted with notable figure extracting tool case, Region Matching is carried out to notable figure and target image, optimum weighting coefficient is tried to achieve, evaluating objective quality is carried out to it using the full reference mass evaluation method based on picture structure and human eye vision.Compared to based on the histogrammic color calibration methods of 3DGMM, calibration result of the present invention is more preferably.And evaluation result is obtained according to method for evaluating objective quality, more meets subjective and objective uniformity.
Description
Technical field
The present invention relates to a kind of multi-viewpoint video image processing method, more particularly to a kind of multi-view point video area-of-interest
Image quality evaluating method.
Background technology
With developing rapidly and extensive use for digital video, the real world described with 2D can not meet people increasingly
The visual demand of growth；But 3D stereo video data amounts are huge, this is proposed sternly to the bandwidth and memory space in communication system
High challenge, therefore the Efficient Compression of 3D video datas is significant.
In order to reduce data shared bandwidth in storage and transmitting procedure, video matter is often reduced in quantizing process
Amount, this can cause the distortion of digital of digital video data.Change compression algorithm because the quality of the video after compressed encoding is directly reflected
Or the performance of compression algorithm, video service system allows for holding the situation of simultaneously quantitation video Quality Down in time, therefore regards
The problem of evaluation of frequency information quality merits attention as one.
Video quality evaluation includes subjective quality assessment (subjective sensation evaluation assessment) and evaluating objective quality (objective evaluation
Method).Current most of image processing systems using human eye as terminal system, it is most directly also that most reliable picture quality is commented that it, which is,
Valency method, but this subjective quality assessment complex operation, by the background difference of observing environment and observer influenceed greatly, real-time
Difference, it is impossible to adapt to Most current video service system.
Evaluating objective quality includes full reference, partly referred to and the video quality metric without reference data.ITU-R expert groups
Using Y-PSNR (PSNR) and root-mean-square error (MSE) as two kinds of traditional objective effective evaluations, but practice is demonstrate,proved
Often there is the situation inconsistent with the subjective sensation of people in bright both approaches.Both full reference methods, which are assumed that, to be judged to regard
Undistorted original video can be obtained during frequency quality, to contrast distortion video and original video so as to evaluate the matter of distortion video
Amount, gained is not real picture quality, but similarity degree or fidelity.
In most of Video service application, people is final video reception person, for various picture quality objective evaluation sides
Method, its purpose studied must the evaluation result of objective quality is consistent with the subjective sensation of people, therefore, it is still necessary to which new regards
Frequency quality evaluating method.
The content of the invention
The deficiency existed for current method for evaluating video quality, the invention provides a kind of method for evaluating video quality.
Method for evaluating video quality of the present invention, including：
--- the source images and target image of multi-view point video are obtained, notable figure is gathered in the target image of acquisition, its
The gray value of (x, y) position is S (x, y)；
The region of same position in the notable figure and source images of acquisition is matched, formula is：
bj=cj×S(x,y)；
Wherein, Iref(x, y) is the component value of source images a certain color in pixel R, G, B of (x, y) position, source images
It is (Δ x, Δ y), I with the difference on matched pixel point in target image in the picture positiontar(x+ Δs x, y+ Δ y) is target
The gray value of correspondence position, a in imagejAnd bjRepresent respectively and multiply sex factor and add factor, cjFor Saliency maps multiplying property because
Son；
Optimum weighting coefficient a is solved by asking source images and target image brightness histogramj、bjAnd cj；By source images and
Target image is matched, and carries out the multiple views color correction of area-of-interest；
--- the image after output calibration, carry out quality evaluation using full reference mass evaluation method.
In an advantageous embodiment, the full reference mass evaluation method is included such as any one in the following group or several
Kind：PSNR algorithms, MSE algorithms, characteristic similarity FSIM algorithms, the characteristic similarity RFSIM algorithms converted based on Riesz, base
Significantly sense VSI algorithms in spectrum residual SR-SIM algorithms, view-based access control model.
In an advantageous embodiment, the calculating to FSIM indexes is divided into two steps：Local similar diagram is calculated first, so
Similar diagram is mapped to a single similarity score afterwards.
In an advantageous embodiment, the notable figure behaviour area-of-interest.People's area-of-interest refers to：Attract
The object of human eye notice.
Wherein, the collection in people region interested can be by the attention computation model based on significance or multiple
Miscellaneous static natural image significance test model is realized.
In an advantageous embodiment, the image after the correction exported, can include coloured image or coloured image
With the gray-scale map corresponding to the coloured image.
Compared with prior art, the advantage of the method for evaluating video quality proposed by the present invention based on area-of-interest exists
In：
By the way that vision noticing mechanism to be applied to the extraction of interesting image regions, the Saliency maps of image are extracted, can be with
Greatly improve efficiency and the degree of accuracy, it is to avoid unnecessary computing resource waste, while good subjective perception figure can be obtained, and
There is good uniformity with evaluating objective quality.
Brief description of the drawings
Fig. 1 is method for evaluating video quality schematic flow sheet of the present invention；
Fig. 2 is the experimental result picture that the inventive method is tested under MATLAB environment, wherein, Fig. 2A is car racing video
Image sequence experimental result, Fig. 2 B are dancing sequence of video images experimental result；With runic frame for reference to figure in figure.
Embodiment
The present invention proposes a kind of full reference image quality appraisement method based on picture structure and human-eye visual characteristic, first
The source images and target image of multi-view point video are obtained, the area-of-interest of target image is extracted with notable figure extracting tool case
Notable figure, carries out Region Matching to notable figure and source images, tries to achieve optimum weighting coefficient, and subjectivity is carried out to the coloured image of synthesis
Quality evaluation, recycles the full reference mass evaluation method based on picture structure and human eye vision to carry out objective quality to it and comments
Valency.
Referring to Figures 1 and 2, using 640x480 car racing video image (race1) sequence visual point image and dancing video figure
As (flamenco2) sequence visual point image is tested under MATLAB environment, area-of-interest is based on by described herein
Color calibration method with being contrasted based on the histogrammic color calibration methods of 3DGMM.The inventive method is as follows：
The source images and target image of multi-view point video are obtained, sense is extracted in the target image of the multi-view point video of acquisition
The notable figure (its gray value is S (x, y)) in interest region, then the notable figure and source images of acquisition are matched；Formula can table
It is shown as：
Iref(x, y)=aj×Itar(x+Δx,y+Δy)+bj；
bj=cj×S(x,y)；
Wherein, image is subjected to triple channel division, Iref(x, y) be source images in pixel R, G, B of (x, y) position certain
The component value of one color, source images are (Δ x, Δ y), I with the difference of respective pixel in position in target imagetar(x+Δx,
Y+ Δs y) is the gray value of correspondence position in target image, ajAnd bjRepresent respectively and multiply sex factor and add factor；cjFor notable figure
Multiply sex factor.
Optimum weighting coefficient a is solved by asking source images and target image brightness histogramj、bjAnd cj；By source images and
Target image is matched, the multiple views color correction of area-of-interest.
Image after output calibration, and quality evaluation is carried out to area-of-interest.Wherein, quality evaluation includes subjective quality
Evaluate.Wherein, quality evaluation includes subjective quality assessment and utilizes the full reference mass based on picture structure and human eye vision
Evaluation method carries out evaluating objective quality to it.
From Fig. 1 and Fig. 2 it can be seen that set forth herein the color calibration method based on area-of-interest subjective
Calibration result is good.
The present invention is using several efficient quality evaluating methods such as PSNR, MSE, FSIM, RFSIM, SR-SIM, VSI to carrying
The result of the color correction based on area-of-interest gone out carries out evaluating objective quality, and experimental result is as shown in table 1, table 2.
The evaluation result of the sequence race1 of table 1 the 0th viewpoint
The evaluation result of the sequence flamenco2 of table 2 the 0th viewpoint
Wherein, the calculating to FSIM indexes calculates local similar diagram in two steps, first, and similar diagram is then mapped to one
Single similarity score.The evaluation criterion RFSIM based on the Riesz characteristic similarities converted is calculated again, based on spectrum residual
Evaluation number SR-SIM, the evaluation number VSI that significantly senses of view-based access control model, the objective of various criterion is carried out to the image of synthesis
Quality evaluation.
Various evaluating objective quality indexs are finally seen whether they meet subjective and objective matter compared with subjective quality assessment
Measure the uniformity evaluated.
Correction effect of the color calibration method in objective indicator in area-of-interest is can be seen that from above table data
Fruit is consistent with subjective perceptual quality.
It can see from experimental result, compared to based on the histogrammic color calibration methods of 3DGMM (b in Fig. 2) portion
Point), in the embodiment of the present invention method based on area-of-interest in subjective calibration result more preferably.And according to objective quality
The result that evaluation method is obtained, compared to based on the histogrammic color calibration methods of 3DGMM (b in Fig. 2) part), base of the present invention
There is preferable effect in objective indicator in the color calibration method of area-of-interest, i.e., method proposed by the present invention meets master
Objective uniformity.
The specific embodiment of the present invention is described in detail above, but it is intended only as example, and the present invention is not limited
It is formed on particular embodiments described above.To those skilled in the art, it is any to the equivalent modifications that carry out of the present invention and
Substitute also all among scope of the invention.Therefore, the impartial conversion made without departing from the spirit and scope of the invention and
Modification, all should be contained within the scope of the invention.
Claims (6)
1. a kind of method for evaluating video quality, it is characterised in that including：
--- the source images and target image of multi-view point video are obtained, notable figure are gathered in the target image of acquisition, its (x, y)
The gray value of position is S (x, y)；
The region of same position in the notable figure and source images of acquisition is matched, formula is：
Iref(x, y)=aj×Itar(x+Δx,y+Δy)+bj；
bj=cj×S(x,y)；
Wherein, Iref(x, y) is the component value of source images a certain color in pixel R, G, B of (x, y) position, source images and mesh
The difference of matched pixel point in the picture on position is (Δ x, Δ y), I in logo imagetar(x+ Δs x, y+ Δ y) is target image
The gray value of middle correspondence position, ajAnd bjRepresent respectively and multiply sex factor and add factor, cjMultiply sex factor for notable figure；
Optimum weighting coefficient a is solved by asking source images and target image brightness histogramj、bjAnd cj；By source images and target
Images match, carries out the multiple views color correction of area-of-interest；
--- the image after output calibration, carry out quality evaluation using full reference mass evaluation method.
2. method for evaluating video quality according to claim 1, it is characterised in that the full reference mass evaluation method bag
Include such as any one or a few in the following group：PSNR algorithms, MSE algorithms, characteristic similarity FSIM algorithms, based on Riesz conversion
Characteristic similarity RFSIM algorithms, VSI algorithms are significantly sensed based on spectrum residual SR-SIM algorithms, view-based access control model.
3. method for evaluating video quality according to claim 2, it is characterised in that the calculating to FSIM indexes is divided into two
Step：Local similar diagram is calculated first, and similar diagram is then mapped to a single similarity score.
4. method for evaluating video quality according to claim 1, it is characterised in that the notable figure behaviour region of interest
Domain.
5. method for evaluating video quality according to claim 4, it is characterised in that the collection in people region interested
It is to be realized by the attention computation model based on significance or complicated static natural image significance test model.
6. method for evaluating video quality according to claim 1, it is characterised in that the image after the correction exported, bag
Include coloured image or coloured image and the gray-scale map corresponding to the coloured image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610072995.1A CN105721863B (en) | 2016-02-02 | 2016-02-02 | Method for evaluating video quality |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610072995.1A CN105721863B (en) | 2016-02-02 | 2016-02-02 | Method for evaluating video quality |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105721863A CN105721863A (en) | 2016-06-29 |
| CN105721863B true CN105721863B (en) | 2017-11-07 |
Family
ID=56155534
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610072995.1A Expired - Fee Related CN105721863B (en) | 2016-02-02 | 2016-02-02 | Method for evaluating video quality |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105721863B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109934786B (en) * | 2019-03-14 | 2023-03-17 | 河北师范大学 | Image color correction method and system and terminal equipment |
| CN112383829B (en) * | 2019-11-06 | 2022-06-24 | 致讯科技(天津)有限公司 | Experience quality evaluation method and device |
| CN111696081B (en) * | 2020-05-18 | 2024-04-09 | 南京大学 | Method for reasoning panoramic video quality from visual field video quality |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101047867A (en) * | 2007-03-20 | 2007-10-03 | 宁波大学 | Method for correcting multi-viewpoint vedio color |
| CN101729911A (en) * | 2009-12-23 | 2010-06-09 | 宁波大学 | Multi-view image color correction method based on visual perception |
-
2016
- 2016-02-02 CN CN201610072995.1A patent/CN105721863B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101047867A (en) * | 2007-03-20 | 2007-10-03 | 宁波大学 | Method for correcting multi-viewpoint vedio color |
| CN101729911A (en) * | 2009-12-23 | 2010-06-09 | 宁波大学 | Multi-view image color correction method based on visual perception |
Non-Patent Citations (1)
| Title |
|---|
| 多视点视频颜色校正的研究;崔斌;《CNKI》;20151115;说明书3.2、4.1，第5章 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105721863A (en) | 2016-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107578404B (en) | View-based access control model notable feature is extracted complete with reference to objective evaluation method for quality of stereo images | |
| CN105208374B (en) | A kind of non-reference picture assessment method for encoding quality based on deep learning | |
| CN102333233B (en) | Stereo image quality objective evaluation method based on visual perception | |
| CN108010024B (en) | Blind reference tone mapping image quality evaluation method | |
| CN111127374B (en) | Pan-sharing method based on multi-scale dense network | |
| CN101610425B (en) | Method for evaluating stereo image quality and device | |
| CN109523506B (en) | Full-reference stereo image quality objective evaluation method based on visual salient image feature enhancement | |
| CN110516716B (en) | No-reference image quality evaluation method based on multi-branch similarity network | |
| CN108765414B (en) | No-reference stereo image quality evaluation method based on wavelet decomposition and natural scene statistics | |
| CN109831664B (en) | Rapid compressed stereo video quality evaluation method based on deep learning | |
| CN109255358B (en) | 3D image quality evaluation method based on visual saliency and depth map | |
| CN109978854B (en) | Screen content image quality evaluation method based on edge and structural features | |
| CN103426173B (en) | Objective evaluation method for stereo image quality | |
| CN105338343A (en) | No-reference stereo image quality evaluation method based on binocular perception | |
| CN109919959A (en) | Tone mapping image quality evaluating method based on color, naturality and structure | |
| CN103841410B (en) | Based on half reference video QoE objective evaluation method of image feature information | |
| CN105721863B (en) | Method for evaluating video quality | |
| CN102547368A (en) | Objective evaluation method for quality of stereo images | |
| Chen et al. | Blind quality index for tone-mapped images based on luminance partition | |
| CN112950596A (en) | Tone mapping omnidirectional image quality evaluation method based on multi-region and multi-layer | |
| CN110570435A (en) | method and device for carrying out damage segmentation on vehicle damage image | |
| CN102722888A (en) | Stereoscopic image objective quality evaluation method based on physiological and psychological stereoscopic vision | |
| CN103780895A (en) | Stereoscopic video quality evaluation method | |
| CN107146220A (en) | A kind of universal non-reference picture quality appraisement method | |
| CN106709504A (en) | Detail-preserving high fidelity tone mapping method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171107 Termination date: 20200202 |