CN106055553A - Foggy-day image database used for significance detection and quality evaluation - Google Patents
Foggy-day image database used for significance detection and quality evaluation Download PDFInfo
- Publication number
- CN106055553A CN106055553A CN201610266010.9A CN201610266010A CN106055553A CN 106055553 A CN106055553 A CN 106055553A CN 201610266010 A CN201610266010 A CN 201610266010A CN 106055553 A CN106055553 A CN 106055553A
- Authority
- CN
- China
- Prior art keywords
- image
- quality evaluation
- mist
- misty
- foggy
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Library & Information Science (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a foggy-day image database used for significance detection and quality evaluation. The image database is established by the steps that firstly, foggy-day images are collected through real shooting and network search, and 500 images of Ground Truth of targets in fog are obtained through Photoshop matting; 15 volunteers are selected (five undergraduate students are selected from each grade from grade one to grade three respectively) to make subjective evaluation of the 500 images, so that 15 groups of MOS values are obtained; and finally, a mean error score is computed for each group of MOS values, so that a DMOS value of each image can be obtained.
Description
Technical field
The present invention relates to the data base of a kind of technical field of image processing, specifically a kind of for significance detection and quality
The Misty Image data base evaluated.
Background technology
Under the conditions of greasy weather, atmospheric turbulance, air to scattering of light and absorption, various weather conditions (such as air pressure, wind speed, gas
Temperature, wind direction, cloud layer, relative humidity etc.) all imaging can be produced certain impact, wherein atmospheric turbulance and background radiation make image
Obfuscation, atmospheric extinction causes picture contrast and brightness to decline, and aerocolloidal forward scattering causes image resolution ratio to decline, carefully
Joint information dropout.In recent years, along with increasing of haze weather, the quality requirement of outdoor visual system is increased by people the most day by day,
Make the study hotspot being treated as computer vision field of Misty Image.
Directly objective evaluation Misty Image quality has important meaning for real-time prediction, the research etc. of mist elimination algorithm of mist
Justice.Misty Image quality evaluation has great significance in image procossing, and it gives the process of Misty Image such as: mist elimination, enhancing
Etc. having established important basis.Meanwhile, detect by means of the saliency under the conditions of the greasy weather, not only can obtain target in mist
Significance change key character, can be also that successive image quality evaluation processes, with other, the prior information providing important.But mesh
Front scholars are less in the research of these two aspects, mainly have three reasons: the randomness of (1) mist and mobility, cause mist to figure
The impact that degrades of picture is a kind of to be different from other distortions " special distortion ".(2) due to the particularity of mist, find and be used for representing it
Feature relatively difficult.(3) existing image data base does not comprise this type of distortion.
Along with the rise of image quality evaluation research, multiple research team of the world has released one after another multiple for picture quality
The data base evaluated, for proof of algorithm and relative analysis, in general image quality evaluation data base both provides the tightest
The background knowledge of lattice, decent with the original reference image of series and the figure of every width reference picture difference type of distortion and degree
This, it is considered to subjective human visual experience, each image is also labelled with Mean Opinion Score (Mean Opinion Scores, MOS)
Or difference subjective scoring (Differential Mean Opinion Scores, DMOS), MOS value the biggest (DMOS value is the least) table
Bright correspondence image quality is the best, otherwise MOS value the least (DMOS value is the biggest) shows that correspondence image quality is the poorest.As space is limited, no
Enumerating, Common database and related introduction refer to shown in table 1 again.
22 kinds of image quality evaluation data base's summary sheets that table 1 is common
The several frequently seen type of distortion such as current image quality evaluation data base is the fuzzyyest, noise, compression, special to some
Image such as Misty Image then lack data base, so make algorithm training and test time lack foundation accurately.
Summary of the invention
The present invention relates to the main contents of two aspects:
(1) in Misty Image, target significance Test database builds
The most less about the research of Misty Image Feature Selection, how to choose one or more features, it is possible to maximum
The change of the reflection mistiness degree of limit and structure will be the main contents invented for the code book of Misty Image quality evaluation.
(2) Misty Image quality assessment database builds
For conventional images quality evaluation algorithm deficiency in terms of Misty Image quality evaluation, invent based on mixing spy
Levying the quality evaluating method of the applicable Misty Image of code book, experiment shows the method superior performance of invention.
Accompanying drawing explanation
Accompanying drawing 1 is that a kind of Misty Image database sharing process for significance detection and quality evaluation of the present invention is shown
It is intended to
Detailed description of the invention
A kind of for significance detection and the Misty Image data base of quality evaluation, building process comprises three below step
Rapid: image acquisition, stingy figure, subjective assessment.
(1) image acquisition
Use actual photographed (Canon EOS 6D) and web search (Baidu's picture, key word is mist, middle mist and thick fog)
Mode acquire 500 original Misty Image, scene covers mist, middle mist and thick fog environment, and well-marked target in mist is contained
Having covered people, building, automobile and plant etc., file format is bmp.
(2) figure is scratched
Use Photoshop2015 to scheme having carried out artificial scratching on 500 width images, be extracted the wheel of well-marked target in image
Exterior feature, and store with the form of black and white binary map, file format is bmp.
(3) subjective assessment
Pick 15 volunteers (a year to junior undergraduate each 5) and 500 width images carried out subjective evaluation and test,
Obtain 15 groups of MOS values, finally calculated mean value error mark and obtain the DMOS value of each image, detailed process often organizing MOS value
As follows.
Test computer model is Dell XPS 8900, and monitor resolution is 1600 × 900, and video card model is NVIDIA
GeForce GTX 960, volunteer's sight line becomes plumbness with display angle.
Before testing, volunteer through strict physiology inspection, the eye of all personnel do not have any disease (as achromatopsia,
Color weakness).And all volunteers are not the most the professionals of image or field of video processing, or routine work life is not related to
Note image or the people of video quality, and it is not engaged in the experience of the subjective assessment being similar to.The most also these volunteers are carried out
Relevant training, makes them understand all links of whole test process.
In order to ensure science and objectivity, experiment have employed single excitation continuous mass scaling law (Single Stimulus
Continuous Quality Evaluation, DSCQE), obtain subjectivity by playing image to be evaluated over the display
Quality score.
When image is carried out subjective assessment, subjective scoring is expressed by mark scale.In actual experiment process
Middle discovery, allows observer with comparalive ease by the quality of the micro-judgment image of oneself, and relate to give a mark to image
Shi Ze is the most difficult, and owing to the cognizance hierarchy etc. of image can be caused the error of evaluation result by observer, therefore to obtain
Subjective quality scores the most accurately, have employed following test process: band is evaluated and tested picture and contrasted sequence two-by-two by observer
After, in [0 100], carry out, according to the order of arrangement, give a mark (100 points of representation qualities are best, and 0 point of representation quality is worst).
In order to ensure the correctness of subjective experiment result, subjective assessment is retrained: experiment limits every sight
Survey person observes image temporal and is less than 10 seconds, and observer's distance computer screen distance is 70 centimetres, sight line and image center
Angle is less than 30 degree.
After obtaining the subjective scores of observer, finally experimental data it is analyzed and processes.First calculate each width to treat
The difference scores (Difference Opinion Score, DOS) of altimetric image, calculating process such as following formula:
dij=qiref-qij (1)
Wherein, qijFor the i-th observer observation to jth width image；qirefRepresent that i-th observer is to reference picture
Observation mark, dijThen represent the i-th observer difference scores to jth image observation value.And Mean difference scores
(Difference Mean Opinion Score, DMOS) is exactly the flat of the observation of the multiple observers to every piece image
Equal:
The most then for the DMOS value of test image.Table gives the subjective test results of parts of images.
Claims (1)
1. the Misty Image data base for significance detection with quality evaluation, it is characterised in that: described image library comprises
The Ground Truth of target in 500 width Misty Image and DMOS value, 500 width mists.Scene covers mist, middle mist and dense
Mist environment；Well-marked target in described mist covers people, building, automobile, plant；Described image data base is in Misty Image
The experimental verification of target significance detection algorithm and Misty Image quality evaluation algorithm provides sample data, is also Misty Image
Scene analysis, object detection and recognition provide data base to support.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610266010.9A CN106055553A (en) | 2016-04-27 | 2016-04-27 | Foggy-day image database used for significance detection and quality evaluation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610266010.9A CN106055553A (en) | 2016-04-27 | 2016-04-27 | Foggy-day image database used for significance detection and quality evaluation |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106055553A true CN106055553A (en) | 2016-10-26 |
Family
ID=57176278
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610266010.9A Pending CN106055553A (en) | 2016-04-27 | 2016-04-27 | Foggy-day image database used for significance detection and quality evaluation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106055553A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107977970A (en) * | 2016-12-30 | 2018-05-01 | 北京联合大学 | A kind of evaluating method of saliency data collection |
| CN108648209A (en) * | 2018-04-08 | 2018-10-12 | 北京联合大学 | A kind of evaluating method of the centre deviation of saliency data collection |
-
2016
- 2016-04-27 CN CN201610266010.9A patent/CN106055553A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107977970A (en) * | 2016-12-30 | 2018-05-01 | 北京联合大学 | A kind of evaluating method of saliency data collection |
| CN107977970B (en) * | 2016-12-30 | 2019-10-29 | 北京联合大学 | A kind of evaluating method of saliency data collection |
| CN108648209A (en) * | 2018-04-08 | 2018-10-12 | 北京联合大学 | A kind of evaluating method of the centre deviation of saliency data collection |
| CN108648209B (en) * | 2018-04-08 | 2021-06-29 | 北京联合大学 | Method for evaluating central deviation of significance data set |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Dupont et al. | Comparing saliency maps and eye-tracking focus maps: The potential use in visual impact assessment based on landscape photographs | |
| Sjaardema et al. | History and Evolution of the Johnson Criteria. | |
| Bergen et al. | The validity of computer-generated graphic images of forest landscape | |
| CN107729830B (en) | Camouflage effect detection and calculation method based on background features | |
| CN109191460A (en) | A kind of quality evaluating method for tone mapping image | |
| Cathcart et al. | Target detection in urban clutter | |
| Van Etten et al. | The spacenet multi-temporal urban development challenge | |
| CN109741285A (en) | A kind of construction method and system of underwater picture data set | |
| Yang et al. | MF-CFI: A fused evaluation index for camouflage patterns based on human visual perception | |
| CN104966310A (en) | Evaluation method for pattern painting camouflage effect | |
| CN106055553A (en) | Foggy-day image database used for significance detection and quality evaluation | |
| Gorte et al. | Scoring Antarctic surface mass balance in climate models to refine future projections | |
| CN104050678A (en) | Underwater monitoring color image quality measurement method | |
| Kohler et al. | Supporting image geolocation with diagramming and crowdsourcing | |
| Yan et al. | Evaluating simulated visible greenness in urban landscapes: An examination of a midsize US city | |
| CN107944387A (en) | A kind of analysis method of the urban heat island special heterogeneity based on semivariation theory | |
| Marušić et al. | Visual search on aerial imagery as support for finding lost persons | |
| Boori et al. | Vulnerability analysis on Hyderabad city, India | |
| CN114708543B (en) | Examination student positioning method in examination room monitoring video image | |
| CN111783678A (en) | Financial big data system based on online deep learning and market operation value quantification | |
| Chen et al. | Generating synthetic photogrammetric data for training deep learning based 3D point cloud segmentation models | |
| CN116309652A (en) | Analysis and evaluation method and system based on camouflage painting pattern | |
| CN110473182A (en) | A kind of subjective assessment processing method, device, electronic equipment and medium towards the full link simulation image of visible light | |
| CN110443277A (en) | A small amount of sample classification method based on attention model | |
| Kozoderov et al. | Cognitive technologies for processing optical images of high spatial and spectral resolution |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| DD01 | Delivery of document by public notice |
Addressee: PLA Military Academy Document name: Notification of before Expiration of Request of Examination as to Substance |
|
| DD01 | Delivery of document by public notice | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20161026 |
|
| WD01 | Invention patent application deemed withdrawn after publication |