CN109978859A - A kind of image display adaptation method for evaluating quality based on visible distortion pond - Google Patents
A kind of image display adaptation method for evaluating quality based on visible distortion pond Download PDFInfo
- Publication number
- CN109978859A CN109978859A CN201910237280.0A CN201910237280A CN109978859A CN 109978859 A CN109978859 A CN 109978859A CN 201910237280 A CN201910237280 A CN 201910237280A CN 109978859 A CN109978859 A CN 109978859A
- Authority
- CN
- China
- Prior art keywords
- image
- display
- localized mass
- adapts
- mass
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30168—Image quality inspection
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
- Testing, Inspecting, Measuring Of Stereoscopic Televisions And Televisions (AREA)
Abstract
The present invention relates to a kind of image displays based on visible distortion pond to adapt to method for evaluating quality.The following steps are included: S1: reading original image, generate Saliency maps S using obvious object detection algorithm;S2: original image is established using method for registering images and display adapts to the Pixel-level mapping relations of image;S3: the block grade fidelity F of image is calculated using Pixel-level mapping relations;S4: using the tactful pond Saliency maps S and eyefidelity F of visible distortion pondization, the overall objective quality of evaluation Q that display adapts to image is calculated.The present invention effectively can adapt to the human eye evaluation mechanism in image objective quality assessment by simulative display, solve deficiency of the different degree weighting pondization strategy in comparison among groups, and help to promote the consistency between the assessment score of objective evaluation method and subjective scoring, it can be applied to objective quality assessment field and other application fields for needing that pondization strategy is used to combine local message assessment total quality that display adapts to image.
Description
Technical field
It is especially a kind of to be based on visible distortion pond the present invention relates to image and video processing and computer vision field
Image display adapt to method for evaluating quality.
Background technique
In recent years, researchers propose many images and show adaptive method to be operated the content of image to be adapted to
The various targets with different sizes and the ratio of width to height show screen.However, still showing adaptive method without a kind of image at present
Various application scenarios can be perfectly suitable for, and generates satisfactory display and adapts to image.It restricts image and shows adaptation side
Method further develops many because being known as, one of them important factor is exactly still to lack at present and human subject's perception more one
The image display of cause adapts to quality evaluation (IRQA) method.It is small on data set that most of images show that adaptive method is often used
Scale subjective testing carrys out the validity of verification method.Although the result of subjective testing can intuitively illustrate that a display adapts to very much
The superiority and inferiority of the visual quality of image, but method for subjective testing is usually nonautomatic, has particular requirement to tester's quality, and
And it can not be effectively treated in face of large-scale data.Therefore, the objective effective IRQA method of one kind is developed for promoting image to show
The further development of adaptive method is vital.
The IRQA of early stage is measured, such as two-way similitude (BDS), color layout (CL) and land mobile distance (EMD) etc.
By simply calculating image distance or display adaptation image and reference picture being converted into the feature with identical dimensional and retouched
Son is stated, to assess the objective quality that display adapts to image.However, these methods do not account for human subject's assessment to different images
Content deformation sensitivity difference, therefore the consistency of objective assessment result and the subjective perception of the mankind is very low.
In recent years, the difficult point that researcher is directed to IRQA both at home and abroad proposes many solutions.Liu et al. people is using entirely
Office's structural similarity and local corresponding relationship assessment display adapt to image.They are first by from thick scale to the traversal of thin scale
Global geometry is extracted, and establishes local pixel corresponding relationship in a manner of top-down.Secondly, in conjunction with rudimentary conspicuousness
Information and advanced face information generate different degree figure.Finally, being based on corresponding relationship, similarity measurement is weighted using different degree
Assessment display adapts to the visual quality of image.Fang et al. establishes the intensive corresponding relationship between image using SIFT-Flow, and
It calculates structural similarity (SSIM) figure of different scale and adapts to the structural information retained in image to measure display.They combine certainly
Then upper and top-down saliency information generation visual importance figure down passes through the SSIM of calculating different degree weighting
Graph evaluation display adapts to the whole visual quality of image.Zhang et al. proposes a kind of backward registration Algorithm based on SIFT-Flow
To simulate original image Geometrical change experienced during showing adaptation.Then, the ratio of width to height phase is weighted by calculating different degree
The visual quality for adapting to image is shown like property (ARS) metric evaluation.These methods require to combine visual attention model, using pond
Change method obtains whole visual quality according to local image quality pond.However the similarity measurement of local quality is measured at present
Still there is deficiency, common importance weighting pond (IWP) strategy is also easy to appear the change of different degree rule in comparison among groups and asks
Topic, therefore the consistency between objective evaluation result and subjective scores is not still high.
Summary of the invention
In view of this, the purpose of the present invention is to propose to a kind of image displays based on visible distortion pond to adapt to quality evaluation
Method, helps to improve the consistency of objective quality assessment result and subjective perception, and effect of optimization is obvious.
The present invention is realized using following scheme: a kind of image display adaptation quality evaluation side based on visible distortion pond
Method, comprising the following steps:
Step S1: it obtains original image and reads, and generate Saliency maps S using obvious object detection algorithm;
Step S2: it obtains display and adapts to image, establish the original graph using the method for registering images based on SIFT-Flow
As adapting to the Pixel-level mapping relations between image with the display;
Step S3: the original image is averagely divided into several localized masses;The display is adapted to image averaging to divide
For several localized masses corresponding with original image;
Step S4: calculating the block grade fidelity F that display adapts to image using Pixel-level mapping relations obtained in step S2,
It obtains the display and adapts to eyefidelity F of the image relative to the original image;
Step S5: it using the visible distortion pondization tactful pondization Saliency maps S and described image fidelity F, and calculates
The display adapts to the overall objective quality of evaluation Q of imageNBP。
Further, the step S2 specifically includes the following contents:
The original image is established using the method for registering images based on SIFT-Flow and the display adapts between image
Pixel-level mapping relations, from it is described display adapt to image to the original image SIFT-Flow vector field can be by asking
Following energy minimization problem is solved to calculate:
Wherein, IoAnd IrIt respectively represents original image and display adapts to image;prRepresent a picture in display adaptation image
Plain coordinate;qrIt is prAn adjacent pixel coordinate;ε represents the neighborhood collection of one four connection;w(pr)=(u (pr),v(pr))
Represent prSIFT-Flow displacement vector;U and v respectively represents the horizontal and vertical component of SIFT-Flow displacement vector;D is one
A threshold value, default value 40;α is the weighted factor of Section 2, default value 2.
Further, the step S4 specifically includes the following steps:
Step S41: original image is calculated using the Pixel-level mapping relations and adapts to corresponding office between image with display
The ratio of width to height similitude of portion's block, to calculate the geometric distortion of localized mass, calculation formula is as follows:
Wherein, k represents the index number of localized mass in original image;The width of localized mass and high variation ratio are respectively rw
(k)=w*(k)/N and rh(k)=h*(k)/N;w*(k) and h*(k) respectively be reconstruct display adapt to image localized mass width and
Height, by calculating, maximum horizontal and vertical distance is obtained between pixel in localized mass;When wide identical with high variation ratio
When, the localized mass for being equivalent to original image is waited than scaling, then AR measurement reaches maximum value, i.e. SAR=1;
Step S42: the area similitude of corresponding localized mass is calculated using Pixel-level mapping relations, to measure localized mass
Information is lost, and calculation formula is as follows:
Wherein, ra(k)=a*(k)/N2It is the area change ratio of k-th of localized mass;η is a positive number, and default value is
0.3, the weight in eyefidelity measurement is lost in for equilibrium geometry distortion and information;
Step S43: the localized mass of image is adapted to using the ratio of width to height similitude and area Similarity measures display
Eyefidelity, calculation formula is as follows:
F (k)=SAR(k)·SA(k)。
Further, the step S5 specifically includes the following steps:
Step S51: the distortion for defining each localized mass according to the average value of each localized mass in the Saliency maps S is visible
Degree calculates the block objective quality scores according to each piece of fidelity and distortion visibility:
Wherein,Represent the average significance value of k-th of localized mass of Saliency maps S;If eyefidelity F (k)
It is normalized to [0,1], then the mass range of k-th of localized mass isTherefore, whenWhen, corresponding part
The mass range of block is [0,1];And withIncrease, the quality radix of each localized mass constantly becomes larger, and range is continuous
It reduces;
Step S52: finally, it is whole to obtain display adaptation image by the average value for the quality score for calculating all localized masses
The objective quality scores of body;The block objective quality scores are calculated according to the fidelity of each localized mass and distortion visibility:
Wherein, QNBPThe objective quality scores obtained by visible distortion pond policy calculation are represented, M is represented in original image
The number of localized mass.
Further, each part is defined according to the average value of each localized mass in the Saliency maps S described in step S51
The distortion visibility of block specifically: localized mass definition one radix high for conspicuousness is small, the big mass range in section;Together
When, localized mass definition one radix low for conspicuousness is big, the small mass range in section.
Compared with prior art, the invention has the following beneficial effects:
(1) present invention is suitable for showing the objective quality assessment for adapting to image, helps to improve objective quality assessment result
With the consistency of subjective perception, effect of optimization is obvious.
(2) display that can be perfectly suitable for of the invention adapts to image quality measure, makes objective evaluation result and the mankind
Subjective scoring keeps better consistency, can be used for showing that adaptation image quality measure and image show adaptive method optimization etc.
Field.
Detailed description of the invention
Fig. 1 is the flow chart of the embodiment of the present invention.
Fig. 2 is the implementation flow chart of the holistic approach of the embodiment of the present invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
As shown in Figure 1, 2, a kind of image display adaptation quality evaluation based on visible distortion pond is present embodiments provided
Method, comprising the following steps:
Step S1: it obtains original image and reads, and generate Saliency maps S using obvious object detection algorithm;
Step S2: it obtains display and adapts to image, establish the original graph using the method for registering images based on SIFT-Flow
As adapting to the Pixel-level mapping relations between image with the display;
Step S3: the original image is averagely divided into several localized masses;The display is adapted to image averaging to divide
For several localized masses corresponding with original image;(display is adapted to image in the present embodiment all to have divided, and is drawn
It is divided into the image of the localized mass such as 9*9 of same size, is averagely divided into the 3*3 image block of 9 same sizes)
Step S4: the ratio of width to height and area phase of corresponding localized mass are calculated using Pixel-level mapping relations obtained in step S2
Block grade fidelity F is obtained like property;
Step S5: it using the visible distortion pondization tactful pondization Saliency maps S and described image fidelity F, and calculates
Display adapts to the overall objective quality of evaluation Q of imageNBP。
In the present embodiment, the step S2 specifically includes the following contents:
The picture between the original image and display adaptation image is established using the method for registering images based on SIFT-Flow
Plain grade mapping relations, adapting to image from display can be by solving following energy to the SIFT-Flow vector field of the original image
Minimization problem is measured to calculate:
Wherein, IoAnd IrIt respectively represents original image and display adapts to image;prRepresent a picture in display adaptation image
Plain coordinate;qrIt is prAn adjacent pixel coordinate;ε represents the neighborhood collection of one four connection;w(pr)=(u (pr),v(pr))
Represent prSIFT-Flow displacement vector;U and v respectively represents the horizontal and vertical component of SIFT-Flow displacement vector;D is one
A threshold value, default value 40;α is the weighted factor of Section 2, default value 2.
In the present embodiment, the step S4 specifically includes the following steps:
Step S41: original image is calculated using Pixel-level mapping relations and adapts to corresponding localized mass between image with display
The ratio of width to height similitude, to calculate the geometric distortion of localized mass, calculation formula is as follows:
Wherein, k represents the index number of localized mass in original image;The width of localized mass and high variation ratio are respectively rw
(k)=w*(k)/N and rh(k)=h*(k)/N;w*(k) and h*(k) respectively be reconstruct display adapt to image localized mass width and
Height, by calculating, maximum horizontal and vertical distance is obtained between pixel in localized mass;When wide identical with high variation ratio
When, the localized mass for being equivalent to original image is waited than scaling, then AR measurement reaches maximum value, i.e. SAR=1;
Step S42: the area similitude of corresponding localized mass is calculated using Pixel-level mapping relations, to measure localized mass
Information is lost, and calculation formula is as follows:
Wherein, ra(k)=a*(k)/N2It is the area change ratio of k-th of localized mass;η is a positive number, and default value is
0.3, the weight in eyefidelity measurement is lost in for equilibrium geometry distortion and information;
Step S43: the localized mass of image is adapted to using the ratio of width to height similitude and area Similarity measures display
Eyefidelity, calculation formula is as follows:
F (k)=SAR(k)·SA(k)。
In the present embodiment, the step S5 specifically includes the following steps:
Step S51: each office is defined according to the average value of each localized mass in the Saliency maps S
The distortion visibility of portion's block calculates the block according to each piece of fidelity and distortion visibility
Objective quality scores:
Wherein,Represent the average significance value of k-th of localized mass of Saliency maps S;If eyefidelity F (k)
It is normalized to [0,1], then the mass range of k-th of localized mass isTherefore, whenWhen, corresponding part
The mass range of block is [0,1];And withIncrease, the quality radix of each localized mass constantly becomes larger, and range is continuous
It reduces;
Step S52: finally, it is whole to obtain display adaptation image by the average value for the quality score for calculating all localized masses
The objective quality scores of body;The block objective quality scores are calculated according to each piece of fidelity and distortion visibility:
Wherein, QNBPThe objective quality scores obtained by visible distortion pond policy calculation are represented, M is represented in original image
The number of localized mass.
In the present embodiment, it is defined described in step S51 according to the average value of each localized mass in the Saliency maps S each
The distortion visibility of localized mass specifically: localized mass definition one radix high for conspicuousness is small, the big mass range in section,
The influence of overall picture quality is adapted to emphasize significant picture material to display.Meanwhile the localized mass low for conspicuousness
It is big to define a radix, the small mass range in section, to simulate the unwise of distortion of the human visual system to not significant content
Perception.
Particularly, show that adapting to image is a kind of distorted image in the present embodiment, it is obtained by modification picture size
It arrives, such as stretching and compressive deformation, is obscured similar to common, the distortion such as noise also belongs to one kind of distorted image.In this reality
It applies in example, which, which adapts to image, is obtained from MIT RatargetedMe and CUHK data set.
Preferably, the present embodiment adapts to introduce the ratio of width to height and area similarity measurement part in image quality measure in display
Quality, and the overall objective quality of image is obtained using the tactful pond local quality of visible distortion pondization and Saliency maps.Using
The method that the ratio of width to height similitude and area similitude combine measures image local quality, and proposes new visible distortion Chi Huafang
Method preferably can also adapt to the human eye assessment in image quality measure by simulative display while sufficiently calculating localized distortion
Mechanism.This method is suitable for display and adapts to image quality measure method, helps to improve objective quality assessment result and subjectivity is felt
The consistency known, effect of optimization are obvious.
Particularly, the distortion level that the present embodiment passes through calculating the ratio of width to height and area similitude sufficient metric topography.
Secondly, the visible distortion pondization strategy of design one is adapted to Chi Hua topography fidelity and Saliency maps to calculate display
The overall objective quality of image.It can be perfectly suitable for display and adapt to image quality measure, make objective evaluation result and the mankind
Subjective scoring keeps better consistency, can be used for showing that adaptation image quality measure and image show adaptive method optimization etc.
Field.
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, is all covered by the present invention.
Claims (5)
1. a kind of image display based on visible distortion pond adapts to method for evaluating quality, it is characterised in that: the following steps are included:
Step S1: it obtains original image and reads, and generate Saliency maps S using obvious object detection algorithm;
Step S2: obtain display adapt to image, using the method for registering images based on SIFT-Flow establish the original image with
The display adapts to the Pixel-level mapping relations between image;
Step S3: the original image is averagely divided into several localized masses;By the display adaptation image averaging be divided into
The corresponding several localized masses of original image;
Step S4: the block grade fidelity F that display adapts to image is calculated using Pixel-level mapping relations obtained in step S2, is obtained
The display adapts to eyefidelity F of the image relative to the original image;
Step S5: using the visible distortion pondization tactful pondization Saliency maps S and described image fidelity F, and described in calculating
Display adapts to the overall objective quality of evaluation Q of imageNBP。
2. a kind of image display based on visible distortion pond according to claim 1 adapts to method for evaluating quality, special
Sign is: the step S2 specifically includes the following contents:
The picture between the original image and the display adaptation image is established using the method for registering images based on SIFT-Flow
Plain grade mapping relations, adapting to image from the display can be by solving such as to the SIFT-Flow vector field of the original image
Lower energy minimization problem calculates:
Wherein, IoAnd IrIt respectively represents original image and display adapts to image;prThe pixel that display adapts in image is represented to sit
Mark;qrIt is prAn adjacent pixel coordinate;ε represents the neighborhood collection of one four connection;w(pr)=(u (pr),v(pr)) represent
prSIFT-Flow displacement vector;U and v respectively represents the horizontal and vertical component of SIFT-Flow displacement vector;D is a threshold
Value, default value 40;α is the weighted factor of Section 2, default value 2.
3. a kind of image display based on visible distortion pond according to claim 1 adapts to method for evaluating quality, special
Sign is: the step S4 specifically includes the following steps:
Step S41: it is calculated using the Pixel-level mapping relations corresponding between the original image and display adaptation image
Localized mass the ratio of width to height similitude, to calculate the geometric distortion of localized mass, calculation formula is as follows:
Wherein, k represents the index number of localized mass in original image;The width of localized mass and high variation ratio are respectively rw(k)=
w*(k)/N and rh(k)=h*(k)/N;w*(k) and h*It (k) is the width and height for adapting to the localized mass of image that show reconstructed respectively, it is logical
It crosses in calculating localized mass maximum horizontal and vertical distance between pixel and obtains;When wide identical with high variation ratio, phase
When the localized mass in original image is waited than scaling, then AR measurement reaches maximum value, i.e. SAR=1;
Step S42: the area similitude of corresponding localized mass is calculated using Pixel-level mapping relations, to measure the information of localized mass
It loses, calculation formula is as follows:
Wherein, ra(k)=a*(k)/N2It is the area change ratio of k-th of localized mass;η is a positive number, default value 0.3,
The weight in eyefidelity measurement is lost in for equilibrium geometry distortion and information;
Step S43: the figure of the localized mass of image is adapted to using the ratio of width to height similitude and area Similarity measures display
As fidelity, calculation formula is as follows:
F (k)=SAR(k)·SA(k)。
4. a kind of image display based on visible distortion pond according to claim 1 adapts to method for evaluating quality, special
Sign is: the step S5 specifically includes the following steps:
Step S51: the distortion visibility of each localized mass, root are defined according to the average value of each localized mass in the Saliency maps S
The block objective quality scores are calculated according to each piece of fidelity and distortion visibility:
Wherein,Represent the average significance value of k-th of localized mass of Saliency maps S;If eyefidelity F (k) is returned
One turns to [0,1], then the mass range of k-th of localized mass isTherefore, whenWhen, corresponding localized mass
Mass range is [0,1];And withIncrease, the quality radix of each localized mass constantly becomes larger, and range constantly reduces;
Step S52: the average value by calculating the quality score of all localized masses adapts to the objective of image entirety to obtain display
Mass fraction;The block objective quality scores are calculated according to the fidelity of each localized mass and distortion visibility:
Wherein, QNBPThe objective quality scores obtained by visible distortion pond policy calculation are represented, M represents part in original image
The number of block.
5. a kind of image display based on visible distortion pond according to claim 1 adapts to method for evaluating quality, special
Sign is: can according to the distortion that the average value of each localized mass in the Saliency maps S defines each localized mass described in step S51
Degree of opinion specifically: localized mass definition one radix high for conspicuousness is small, the big mass range in section;Meanwhile for significant
Property low localized mass big, the small mass range in section that defines a radix.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910237280.0A CN109978859B (en) | 2019-03-27 | 2019-03-27 | Image display adaptation quality evaluation method based on visible distortion pooling |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910237280.0A CN109978859B (en) | 2019-03-27 | 2019-03-27 | Image display adaptation quality evaluation method based on visible distortion pooling |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109978859A true CN109978859A (en) | 2019-07-05 |
CN109978859B CN109978859B (en) | 2022-06-14 |
Family
ID=67080897
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910237280.0A Active CN109978859B (en) | 2019-03-27 | 2019-03-27 | Image display adaptation quality evaluation method based on visible distortion pooling |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109978859B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111127460A (en) * | 2019-12-31 | 2020-05-08 | 福州大学 | Image display adaptation evaluation method based on block connection |
CN111242916A (en) * | 2020-01-09 | 2020-06-05 | 福州大学 | Image display adaptation evaluation method based on registration confidence measurement |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160212432A1 (en) * | 2013-09-06 | 2016-07-21 | Zhou Wang | Method and system for objective perceptual video quality assessment |
CN106780452A (en) * | 2016-12-07 | 2017-05-31 | 华侨大学 | A kind of combination small echo and the complete of Corner Feature refer to screen image quality measure method |
CN106991669A (en) * | 2017-03-14 | 2017-07-28 | 北京工业大学 | A kind of conspicuousness detection method based on depth-selectiveness difference |
-
2019
- 2019-03-27 CN CN201910237280.0A patent/CN109978859B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160212432A1 (en) * | 2013-09-06 | 2016-07-21 | Zhou Wang | Method and system for objective perceptual video quality assessment |
CN106780452A (en) * | 2016-12-07 | 2017-05-31 | 华侨大学 | A kind of combination small echo and the complete of Corner Feature refer to screen image quality measure method |
CN106991669A (en) * | 2017-03-14 | 2017-07-28 | 北京工业大学 | A kind of conspicuousness detection method based on depth-selectiveness difference |
Non-Patent Citations (3)
Title |
---|
HISHAM CHOLAKKAL 等: "Backtracking Spatial Pyramid Pooling-Based Image Classifier for Weakly Supervised Top–Down Salient Object Detection", 《IEEE TRANSACTIONS ON IMAGE PROCESSING》 * |
贾惠珍: "基于视觉特性和自然场景统计特性的图像质量评价研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
闫钧华等: "基于显著性图像边缘的全参考图像质量评价", 《仪器仪表学报》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111127460A (en) * | 2019-12-31 | 2020-05-08 | 福州大学 | Image display adaptation evaluation method based on block connection |
CN111127460B (en) * | 2019-12-31 | 2022-06-14 | 福州大学 | Image display adaptation evaluation method based on block connection |
CN111242916A (en) * | 2020-01-09 | 2020-06-05 | 福州大学 | Image display adaptation evaluation method based on registration confidence measurement |
CN111242916B (en) * | 2020-01-09 | 2022-06-14 | 福州大学 | Image display adaptation evaluation method based on registration confidence measurement |
Also Published As
Publication number | Publication date |
---|---|
CN109978859B (en) | 2022-06-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106327532A (en) | Three-dimensional registering method for single image | |
CN105046708B (en) | A kind of color correction objective evaluation method consistent with subjective perception | |
CN105279787A (en) | Method for generating three-dimensional (3D) building model based on photographed house type image identification | |
CN107240096A (en) | A kind of infrared and visual image fusion quality evaluating method | |
CN109919959A (en) | Tone mapping image quality evaluating method based on color, naturality and structure | |
CN103605716B (en) | Data processing method and device used for webpage click display | |
CN103177451B (en) | Based on the self-adapting window of image border and the Stereo Matching Algorithm of weight | |
CN106504230B (en) | It is complete with reference to color/graphics image quality measure method based on phase equalization | |
CN101976444A (en) | Pixel type based objective assessment method of image quality by utilizing structural similarity | |
CN108109147A (en) | A kind of reference-free quality evaluation method of blurred picture | |
CN102081742B (en) | Method for automatically evaluating writing ability | |
CN109978859A (en) | A kind of image display adaptation method for evaluating quality based on visible distortion pond | |
CN110490807A (en) | Image rebuilding method, device and storage medium | |
CN115546273A (en) | Scene structure depth estimation method for indoor fisheye image | |
CN105488798B (en) | SAR image method for measuring similarity based on point set contrast | |
CN104933719A (en) | Method for detecting image edge by integral image interblock distance | |
CN112132774A (en) | Quality evaluation method of tone mapping image | |
CN105513071B (en) | A kind of topographic map symbols quality evaluating method | |
US20230386023A1 (en) | Method for detecting medical images, electronic device, and storage medium | |
CN111369435A (en) | Color image depth up-sampling method and system based on self-adaptive stable model | |
CN114494262B (en) | Method and device for evaluating image contrast | |
CN106204564A (en) | A kind of laser photocentre extracting method | |
CN106023150A (en) | Real-time objective quality evaluation method for digital hologram noise interference | |
CN110910497A (en) | Method and system for realizing augmented reality map | |
CN102982532A (en) | Stereo image objective quality evaluation method base on matrix decomposition |
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 |