CN104346809A - Image quality evaluation method for image quality dataset adopting high dynamic range - Google Patents
Image quality evaluation method for image quality dataset adopting high dynamic range Download PDFInfo
- Publication number
- CN104346809A CN104346809A CN201410490258.4A CN201410490258A CN104346809A CN 104346809 A CN104346809 A CN 104346809A CN 201410490258 A CN201410490258 A CN 201410490258A CN 104346809 A CN104346809 A CN 104346809A
- Authority
- CN
- China
- Prior art keywords
- image
- image quality
- dynamic range
- high dynamic
- images
- 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
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/02—Diagnosis, testing or measuring for television systems or their details for colour television signals
-
- 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)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an image quality evaluation method for an image quality dataset adopting a high dynamic range. The image quality evaluation method for the image quality dataset adopting the high dynamic range comprises the steps of 1) selecting 8 reference images and correspondingly using 192 images formed by distorted images which are at 8 distortion levels and are subjected to JEPG/JEPG2000 compression and white noise and Gaussian blurring processing; 2) conducting subjective visual testing, using a pair of corrected 8-bit LDR and 10-bit HDR displays to display the images and recording subjective scores; 3) testing some IQA methods which are the most advanced at present on the obtained HDR2014 to obtain final subjective quality scores, wherein the stronger the correlation between subjective quality scores and MOS/DMOS values is, the more accurate the evaluation methods are. According to results, compared with LDR, the image quality evaluation method for the image quality dataset adopting the high dynamic range has the advantages that the sensing quality of visual irritation is improved; and according to testing by using the current several IQA methods, except a small part of testing methods with performance which is obviously decreased, the testing performance of other methods is very good.
Description
Technical field
The present invention relates to a kind of image quality evaluating method, particularly, relate to a kind of image quality evaluating method adopting the image quality data collection of high dynamic range.
Background technology
The information of the mankind 80% is obtained by vision, and especially at the society of information prosperity, the flourishing method of the various medium such as Digital Television, digital image, video conference, social media, makes the high-quality demand of image and video increasing.Image quality evaluation (IQA) is a classical research direction, and it plays an important role at the numerous areas of Digital Image Processing, the development of such as compression of images, storage and transmission etc. and optimization.
It is the image of low-dynamic range (LDR) that existing image quality evaluation (IQA) method is all studied.But in real world, our vision system is can a wide range of visible ray of perception, and can cross over absolute range from direct sunlight to weak sunlight about have 10 orders of magnitude.So the high dynamic range (HDR) of expression real scene is imaged in iconography has milestone significance.Theoretically, the visual quality of HDR image is more much higher than conventional LDR image, because it contains the contrast of wider scope, and details, brightness and color.The progress of sensor technology makes the availability of HDR image become possibility.It is at digital photography, visual art perform, amusement and game, or even medical treatment and safe imaging create revolutionary change.
Although HDR image is just becoming more and more general and important in computer graphics, the dynamic range of various conventional equipment (display, printer etc.) is more much smaller than the dynamic range in the scene of real world.Therefore, tone mapping operator (TMOS) is rendered into LDR image HDR is an emerging field a long time ago.Infusive, we can see the huge advance of display technique recently.10 display technique innovation and creation critical point, makes the directly visual of HDR image to become possibility.
As everyone knows, three kinds can be divided into based on the objective IQA of original image: full ginseng (FR), half joins (RR) and (FR refers to that whole availability is assumed that known with reference to image without ginseng (NR); RR is this reference picture is that part is available; NR represents that original image is unavailable).In eighties of last century, the Y-PSNR (PSNR) of square error (MSE) and equivalence thereof is the standard method of test commonly used for a pair, because it is simple, portability is strong and and have clear and definite physical significance.But, the two does not all consider the impact of picture material, thus causes the correlativity of the quality assessment of they and people poor, i.e. mean opinion score (MOS).For the deficiency of MSE and PSNR, (the Image Quality Assessment:From Error Visibility to Structural Similarity such as Wang, belonging to periodical: Transaction on Image Processing) propose a kind of effective structural similarity index (SSIM) based on the hypothesis of the great attention structural information of human visual system (HVS), and SSIM achieves the brightness between reference picture and distorted image, the comparison of contrast and structural information.Inspire by this, propose numerous variant based on SSIM.Except full ginseng IQA method above-mentioned, increasing researchist is devoted to for the IQA index method without ginseng and half ginseng.
Summary of the invention
For defect of the prior art, the object of this invention is to provide a kind of image quality evaluating method adopting the image quality data collection of high dynamic range, its complexity is low, and has obvious improvement than legacy data collection, can be used for evaluate image quality more exactly.
For achieving the above object, the image quality evaluating method of the image quality data collection of employing high dynamic range of the present invention, comprises the following steps:
The first step, compressed by JPEG and JPEG2000 of use 8 level of distortion of 6 reference pictures and correspondence thereof, the distorted image of white noise and Gaussian Blur process forms 192 images;
Second step, participate in subjective vision by the personnel of non-image process specialty and test, and with the LDR and 10 of 8 after a pair calibration HDR display display image and the subjectivity of record give a mark;
3rd step, propose data set HDR2014 on use image quality evaluating method to test, draw final objective quality scores, the correlativity of objective quality scores and MOS/DMOS value is stronger, and evaluation method is more accurate.
Preferably, in the described first step: selection size is six harmless natural HDR image of 512x384 size, and all images are all converted to 8 bit images by Adobe Photoshop; Adopt distortion JPEG and JPEG2000 respectively again, Gaussian Blur and white noise process obtain other 192 images, thus composition data collection.
Principle of the present invention is: the visual quality of high dynamic range (HDR) image is more much higher than conventional low-dynamic range (LDR) image, because it contains the contrast of wider scope, details, brightness and color.The Objective image quality evaluation method square error (MSE) of full ginseng (FR) and Y-PSNR (PSNR) mean opinion score (MOS) structural similarity index (SSIM) of equivalence and half thereof are joined (RR) and on the image data set of low-dynamic range (LDR), are had good performance without many methods of ginseng (NR).The new HDR image qualitative data collection that the present invention adopts test performance good, and describe the efficiency and applicability of said method.
Compared with prior art, the present invention has following beneficial effect:
The present invention adopts high dynamic range images qualitative data collection, data set tests some ubiquitous state-of-the-art IQA methods, experimental result confirms, HDR display compares the perceived quality that really improve visual stimulus of LDR, still performance is fine in HDR image for existing several IQA method, but the performance of certain methods significantly declines.The present invention can be widely used in various HDR image quality assessment, and complexity is low, and has obvious improvement than legacy data collection.
Accompanying drawing explanation
By reading the detailed description done non-limiting example with reference to the following drawings, other features, objects and advantages of the present invention will become more obvious:
Fig. 1 is the implementing procedure figure of one embodiment of the invention;
Fig. 2 is the original harmless coloured image of 6 512x384 sizes that one embodiment of the invention obtains from network, wherein: (a)-(f) is 8 bitmaps of scene 1-6 respectively;
Fig. 3 is each scene distortion map of one embodiment of the invention, wherein: the Gaussian Blur distortion map that (a)-(c) is scene 2; D JPEG distortion map that ()-(f) is scene 3; G JP2K distortion map that ()-(i) is scene 4; J white noise distortion map that ()-(l) is scene 5.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art and understand the present invention further, but not limit the present invention in any form.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, some distortion and improvement can also be made.These all belong to protection scope of the present invention.
The present embodiment provides a kind of image quality data collection of high dynamic range, and implementing procedure as shown in Figure 1, comprises the steps:
The generation of the first step, HDR2014 data set
JPEG: HDR image is compressed into jpeg image by the imwrite function calling Matlab, uses Q parameter to be (70,60,50,40,30,20,10,5), then uses hdrwrite function to generate HDR image from JPEG;
JP2K: HDR image is compressed into JP2K image according to different ratio of compression parameters by the imwrite function calling Matlab, the ratio of compression of use is 20,40,60,80,100,200,300 and 400, then uses hdrwrite function to generate HDR image from JP2K;
White noise: call the imnoise function in Matlab, respectively to interpolation 8 kinds of normal noise of R, G, channel B, μ=0, σ
2=(0.002,0.004,0.008,0.01,0.04,0.08,0.1,0.3);
Gaussian Blur: call fspecial and the imfilter function in Matlab, uses window to be l to R, G, channel B respectively
g× l
g(l
g=20), standard deviation is σ
gthe gaussian kernel function of=(1,2,3,4,5,6,7,8) carries out Fuzzy Processing.
More than obtain HDR2014 data set after process.
Second step, subjective testing to HDR2014 data set
The image of HDR2014 data centralization is shown respectively on liquid crystal LCD (8) display and HDR (10) display (all being shown by Adobe Photoshop software); The resolution of LCD and HDR display is 1920x1080 and 1920x1200, and refresh rate is 60Hz and 59Hz; (SS) method is stimulated according to the single that ITU-R BT.500-12 carries out, 25 observer (14 men, 11 female, age 20-30) participate in, wherein major part is student enrollment, from different majors (computer science, electronic engineering, chemistry etc.), and everyone has normal or is remedied to twenty-twenty vision; On two displays, show all HDR image respectively to each participant, in experiment, do not have the interference of other members; In order to eliminate the impact of order, picture presentation order is random; After observing image in a few second, participant is required to grade to image, and continuous mass yardstick is from 0 to 1, and precision is 0.01%; The method of this subjective testing carrys out strict implement according to international standard BT500.Test result has a lot in addition, can get rid of completely because the deviation brought at random of certain individuality.
Gamma correction is have employed during the HDR image shown in Adobe Photoshop; Therefore, needed to carry out Gamma correction pre-service to these images before use IQA index; Conventional gamma factor is γ=2.2, and Gamma correction function is:
Wherein: L
wactual illumination, L
ddisplay illumination;
List the mean opinion score (MOS) of original image on 8 displays and 10 displays, and eliminate the mark (differing larger with other marks) of some exceptions, and calculate original image and different Opinion Score (DMOS) value of distorted image thereof.
3rd step, to full ginseng IQA index, continue original index MSE and PSNR, and also using structural similarity index (SSIM) classical way proposed for 2004, is the combination of brightness comparison function l, contrast comparison function c and structural similarity function s; Then, according to the multiple dimensioned SSIM (MS-SSIM) of NSS model, image gradient, people's brain science etc., the weighting SSIM (IW-SSIM) of the information content, state-of-the-art characteristic similarity (FSIM), gradient similar (GSM), and inner generation model (IGM); In addition the visual information fidelity (VIF) of statistical model based on natural views and information theory setting also having H.R.Sheikh to propose; Calculate and calculate PLCC under these 9 kinds full ginseng methods under HDR2014, the assessed value of SROCC, KROCC and RMSE and the SROCC value under the clear and definite image (scene 1-6) of respective content;
To without ginseng IQA index, consider its widely using in most applications, therefore more valuable than full ginseng IQA method when original image signal can not be obtained; Use three kinds of early stage popular image's authenticity based on distortion identification and Integrity Assessment (DIIVINE), mark (BLIINDS-II) and blind/nothing ginseng image space quality assessment (BRISQUE) based on DCT statistics without ginseng figure integrality, on DCT, DWT and spatial domain, use support vector machine (SVM) to have good performance respectively; Also has the nothing ginseng method that proposition two kinds in " Making a completely blind image quality analyzer " (A.Mittal is published in IEEE Signal Processing Letters2013) and " Learning without human scores for blind image quality assessment " (W.Xue is published in CVPR2013) article is new in addition, the image that need not manually give a mark, picture material and distortion category prior knowledge.
As full ginseng and the compromise without ginseng IQA, propose many methods having very high IQA performance half ginseng.Such as, based on distortion metrics free energy method (FEDM), can estimate the inside generation model of human brain when perceiving the visual signal of input, and structure degradation model (SDM) is successfully modified into effective half ginseng IQA method SSIM.
Under HDR2014 data set, use above-mentioned quality evaluating method to evaluate, and calculate this 4 kind of half ginseng method and 5 kinds without ginseng method calculate the clear and definite image (scene 1-6) of respective content and distortion map (fuzzy, JP2K, JPEG, white noise) under PLCC, SROCC, KROCC, and the assessed value of RMSE, and the clear and definite image (scene 1-6) of respective content and distortion map (fuzzy, JP2K, JPEG, white noise) under SROCC value.
Implementation result
According to above-mentioned steps, HDR2014 image data set is tested respectively.All experiments all realize on PC computing machine, and the major parameter of this PC computing machine is: central processing unit
core
tM2Duo CPU E6600@2.40GHz, internal memory 3GB; Software platform: MATLAB.
Some important parameters in the subjective testing of HDR2014 data set listed by table 1;
The mean opinion score (MOS) of original image on 8 displays and 10 displays listed by table 2;
Be: sort related coefficient (SRCC), Kendall of Pearson linear correlation (PLCC), Sperman sorts related coefficient (KROCC) and root-mean-square error (RMSE) be used for assessing and compare these full ginsengs of vying each other, half ginseng and without the IQA method of joining according to four kinds of methods that the performance measurement of VQEG suggestion is conventional.
Table 3 illustrates 9 kinds of full ginseng method PLCC under HDR2014 data set, the assessed value of SROCC, KROCC and RMSE.
Table 4 be 4 kind of half ginseng method and 5 kinds without the PLCC of ginseng method under HDR2014 data set, the assessed value of SROCC, KROCC and RMSE.
Table 5 represents the SROCC value of 9 kinds of full ginseng methods under the clear and definite image (scene 1-6) of respective content.
Table 6 is 4 kind of half ginseng method and 5 kinds without the SROCC value of ginseng method under the clear and definite image (scene 1-6) of respective content and distortion map (fuzzy, JP2K, JPEG, white noise).
The condition of table 1 subjective testing and parameter
The MOS value of table 2 original image on 8 and 10 displays, overstriking is higher MOS value
The full ginseng method PSNR of table 3, IW-PSNR, SSIM, MS-SSIM, IW-SSIM, VIF, FSIM, GSM, the PLCC of IGM under HDR2014 data set (198 images), the assessed value (after linear regression) of SROCC, KROCC and RMSE, overstriking is particular value
Table 4 half ginseng method FEDM, RRED, QFTB, SDM and without ginseng method DIIVINE, BLIINDS-II, BRISQUE, NIQE, the QAC PLCC under HDR2014 data set (198 images), the assessed value (after linear regression) of SROCC, KROCC and RMSE, overstriking is particular value
The full ginseng method PSNR of table 5, the SROCC (after linear regression) of IW-PSNR, SSIM, MS-SSIM, IW-SSIM, VIF, FSIM, GSM, IGM each scene under HDR2014 data set (198 images), overstriking is single best values
Can be found by table 2, the massfraction of the reference picture on 10 displays is than higher on 8 displays.In order to this point is described, for scene 2, scene 3 and scene 5, brightness when showing on 10 displays and color are continuous print, but can see that obvious color is jumped on 8 displays, on high (as in Fig. 2 (b), shown in (c)) and the earth (as Suo Shi (e) in Fig. 2) carry out detecting and can clearly see.For scene 4, the effect on 10 displays has more details (on plate more decorative pattern) than 8 displays in bright region.For (a) in Fig. 2 and (f), the difference in demonstration performance is also little.All these phenomenons confirm that HDR image has color, contrast and the intensity more wider than LDR image.But as can be seen from Figure 3, distortion level is higher, the HDR image of the distortion of more difficult differentiation on 10 and 8 displays.Then, eliminate the mark (differing larger with other marks) of some exceptions, and calculate original image and different Opinion Score (DMOS) value of distorted image thereof.
Above specific embodiments of the invention are described.It is to be appreciated that the present invention is not limited to above-mentioned particular implementation, those skilled in the art can make various distortion or amendment within the scope of the claims, and this does not affect flesh and blood of the present invention.
Claims (4)
1. adopt an image quality evaluating method for the image quality data collection of high dynamic range, it is characterized in that, comprise the following steps:
The first step, compressed by the JPEG/JPEG2000 of use 8 level of distortion of 6 reference pictures and correspondence thereof, the distorted image of white noise and Gaussian Blur process forms 192 images, composition data collection;
Second step, to be participated in subjective vision test by non-professional personnel, and with the LDR and 10 of 8 after a pair calibration HDR display display image and the subjectivity of record give a mark;
3rd step, the data set that obtains in the first step test existing image quality evaluating method, and draw final objective quality scores, the correlativity of objective quality scores and MOS/DMOS value is stronger, and evaluation method is more accurate.
2. the image quality evaluating method of the image quality data collection of employing high dynamic range according to claim 1, is characterized in that, in the described first step, concrete:
Selection size is six harmless natural HDR image of 512x384 size, and all images are all converted to 8 bit images by AdobePhotoshop; Carry out distortion JPEG and JPEG2000 respectively again, Gaussian Blur and white noise process obtain other 192 images, thus composition data collection.
3. the image quality evaluating method of the image quality data collection of employing high dynamic range according to claim 2, is characterized in that, described data set, and concrete acquisition is as follows:
JPEG: HDR image is compressed into jpeg image by the imwrite function calling Matlab, uses Q parameter to be (70,60,50,40,30,20,10,5), then uses hdrwrite function to generate HDR image from JPEG;
JP2K: HDR image is compressed into JP2K image according to different ratio of compression parameters by the imwrite function calling Matlab, the ratio of compression of use is 20,40,60,80,100,200,300 and 400, then uses hdrwrite function to generate HDR image from JP2K;
White noise: call the imnoise function in Matlab, respectively to interpolation 8 kinds of normal noise of R, G, channel B, μ=0, σ
2=(0.002,0.004,0.008,0.01,0.04,0.08,0.1,0.3);
Gaussian Blur: call fspecial and the imfilter function in Matlab, uses window to be l to R, G, channel B respectively
g× l
g(l
g=20), standard deviation is σ
gthe gaussian kernel function of=(1,2,3,4,5,6,7,8) carries out Fuzzy Processing;
Data set is obtained after above-mentioned process.
4. the image quality evaluating method of the image quality data collection of employing high dynamic range according to claim 1 and 2, it is characterized in that, in described second step, concrete: in subjective vision test, to calculate the different Opinion Score value of mean opinion score, original image and distorted image thereof.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410490258.4A CN104346809A (en) | 2014-09-23 | 2014-09-23 | Image quality evaluation method for image quality dataset adopting high dynamic range |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410490258.4A CN104346809A (en) | 2014-09-23 | 2014-09-23 | Image quality evaluation method for image quality dataset adopting high dynamic range |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104346809A true CN104346809A (en) | 2015-02-11 |
Family
ID=52502349
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410490258.4A Pending CN104346809A (en) | 2014-09-23 | 2014-09-23 | Image quality evaluation method for image quality dataset adopting high dynamic range |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104346809A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105100789A (en) * | 2015-07-22 | 2015-11-25 | 天津科技大学 | Method for evaluating video quality |
| KR20170112352A (en) * | 2016-03-31 | 2017-10-12 | 위니베르시떼 드 낭뜨 | Method for determining a visual quality index of a high dynamic range video sequence |
| CN109345525A (en) * | 2018-09-27 | 2019-02-15 | 方玉明 | One kind removing ghost high dynamic range images quality evaluating method |
| CN109416830A (en) * | 2016-07-08 | 2019-03-01 | 深圳市大疆创新科技有限公司 | System and method for image procossing |
| CN110415207A (en) * | 2019-04-30 | 2019-11-05 | 杭州电子科技大学 | A method of the image quality measure based on image fault type |
| WO2021036550A1 (en) * | 2019-08-23 | 2021-03-04 | 上海交通大学 | Visual task-based method and system for evaluating compression quality of point cloud data |
| CN114429476A (en) * | 2022-01-25 | 2022-05-03 | 惠州Tcl移动通信有限公司 | Image processing method, image processing apparatus, computer device, and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103578104A (en) * | 2013-10-25 | 2014-02-12 | 宁波大学 | Method for objective quality evaluation of partial reference image of Gaussian blur image |
| CN103745466A (en) * | 2014-01-06 | 2014-04-23 | 北京工业大学 | Image quality evaluation method based on independent component analysis |
| CN103927734A (en) * | 2013-01-11 | 2014-07-16 | 华中科技大学 | Method for evaluating quality of blurred images based on no-reference |
| CN103996192A (en) * | 2014-05-12 | 2014-08-20 | 同济大学 | Non-reference image quality evaluation method based on high-quality natural image statistical magnitude model |
-
2014
- 2014-09-23 CN CN201410490258.4A patent/CN104346809A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103927734A (en) * | 2013-01-11 | 2014-07-16 | 华中科技大学 | Method for evaluating quality of blurred images based on no-reference |
| CN103578104A (en) * | 2013-10-25 | 2014-02-12 | 宁波大学 | Method for objective quality evaluation of partial reference image of Gaussian blur image |
| CN103745466A (en) * | 2014-01-06 | 2014-04-23 | 北京工业大学 | Image quality evaluation method based on independent component analysis |
| CN103996192A (en) * | 2014-05-12 | 2014-08-20 | 同济大学 | Non-reference image quality evaluation method based on high-quality natural image statistical magnitude model |
Non-Patent Citations (1)
| Title |
|---|
| MIN LIU ET AL: "HDR2014 - A HIGH DYNAMIC RANGE IMAGE QUALITY DATABASE", 《ICMEW 2014》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105100789A (en) * | 2015-07-22 | 2015-11-25 | 天津科技大学 | Method for evaluating video quality |
| CN105100789B (en) * | 2015-07-22 | 2018-05-15 | 天津科技大学 | A kind of method for evaluating video quality |
| KR20170112352A (en) * | 2016-03-31 | 2017-10-12 | 위니베르시떼 드 낭뜨 | Method for determining a visual quality index of a high dynamic range video sequence |
| KR102573982B1 (en) | 2016-03-31 | 2023-09-04 | 낭트 유니베르시테 | Method for determining a visual quality index of a high dynamic range video sequence |
| CN109416830A (en) * | 2016-07-08 | 2019-03-01 | 深圳市大疆创新科技有限公司 | System and method for image procossing |
| CN109345525A (en) * | 2018-09-27 | 2019-02-15 | 方玉明 | One kind removing ghost high dynamic range images quality evaluating method |
| CN109345525B (en) * | 2018-09-27 | 2022-03-29 | 方玉明 | Ghost-removed high-dynamic-range image quality evaluation method |
| CN110415207A (en) * | 2019-04-30 | 2019-11-05 | 杭州电子科技大学 | A method of the image quality measure based on image fault type |
| WO2021036550A1 (en) * | 2019-08-23 | 2021-03-04 | 上海交通大学 | Visual task-based method and system for evaluating compression quality of point cloud data |
| CN114429476A (en) * | 2022-01-25 | 2022-05-03 | 惠州Tcl移动通信有限公司 | Image processing method, image processing apparatus, computer device, and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104346809A (en) | Image quality evaluation method for image quality dataset adopting high dynamic range | |
| Qureshi et al. | Towards the design of a consistent image contrast enhancement evaluation measure | |
| CN102333233B (en) | Stereo image quality objective evaluation method based on visual perception | |
| Liu et al. | A no-reference metric for perceived ringing artifacts in images | |
| CN103996192B (en) | Non-reference image quality evaluation method based on high-quality natural image statistical magnitude model | |
| Zhu et al. | Image quality assessment by visual gradient similarity | |
| CN111079740A (en) | Image quality evaluation method, electronic device, and computer-readable storage medium | |
| CN109978854B (en) | Screen content image quality evaluation method based on edge and structural features | |
| CN106920232A (en) | Gradient similarity graph image quality evaluation method and system based on conspicuousness detection | |
| Tourancheau et al. | Impact of subjective dataset on the performance of image quality metrics | |
| Liu et al. | A high-definition diversity-scene database for image quality assessment | |
| CN102595185A (en) | Stereo image quality objective evaluation method | |
| CN112950596B (en) | Tone mapping omnidirectional image quality evaluation method based on multiple areas and multiple levels | |
| Chen et al. | Perceptual quality assessment of cartoon images | |
| Ni et al. | SCID: A database for screen content images quality assessment | |
| Chen et al. | Blind quality index for tone-mapped images based on luminance partition | |
| CN110706196B (en) | Clustering perception-based no-reference tone mapping image quality evaluation algorithm | |
| CN107146220A (en) | A kind of universal non-reference picture quality appraisement method | |
| CN110910347B (en) | Tone mapping image non-reference quality evaluation method based on image segmentation | |
| Zhao et al. | No-reference quality index of tone-mapped images based on authenticity, preservation, and scene expressiveness | |
| Xu et al. | Remote-sensing image usability assessment based on ResNet by combining edge and texture maps | |
| CN106934770A (en) | A kind of method and apparatus for evaluating haze image defog effect | |
| Amirshahi et al. | How do image quality metrics perform on contrast enhanced images? | |
| CN106023214A (en) | Image quality evaluation method and system based on foveal vision gradient structural similarity | |
| Liu et al. | Fog removal algorithms: survey and perceptual evaluation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150211 |