CN108492280A - A kind of device and method of automatic decision digital picture quality - Google Patents
A kind of device and method of automatic decision digital picture quality Download PDFInfo
- Publication number
- CN108492280A CN108492280A CN201810174477.XA CN201810174477A CN108492280A CN 108492280 A CN108492280 A CN 108492280A CN 201810174477 A CN201810174477 A CN 201810174477A CN 108492280 A CN108492280 A CN 108492280A
- Authority
- CN
- China
- Prior art keywords
- picture
- unit
- quality
- verification
- pictures
- 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
- G06T7/90—Determination of colour characteristics
-
- 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
- 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)
- Image Processing (AREA)
Abstract
The present invention provides a kind of device and method of automatic decision digital picture quality, wherein, include picture quality pre-detection unit, dimension of picture verification unit, weight setting unit, recognition of face verification unit, picture quality judging unit, colour cast verification unit and output picture element unit cell including a kind of device of automatic decision digital picture quality, the invention also discloses a kind of methods of automatic decision number format picture quality.This method can be applied to Digital Media, as the Appreciation gist of picture quality, based on the acceptable learning of most of user, filter out the picture for meeting most users acceptable learning.The target of this method also resides in, the picture filtered out, with regard to its size, for color saturation, is more preferably shown on Digital Media screen.
Description
Technical field
The invention belongs to image processing techniques, and in particular to a kind of device and method of automatic decision digital picture quality.
Background technology
Image quality evaluation technology has very important effect in many image processing applications, picture quality now
Evaluation method can be divided into subjective assessment and objective evaluation, and subjective assessment is the ultimate criterion of picture quality, but subjective assessment
It is usually unrealistic for actual image processing system.Many researchers put into Objective image quality evaluation in recent years
In method, in the Digital Media epoch, a picture for meeting most users acceptable learning have the function of it is great, it is existing to comment
Valence and screening belong to artificially evaluate, and subjectivity is strong.
Invention content
In view of this, the main object of the present invention is to provide a kind of device and method of automatic decision digital picture quality.
The technical solution adopted by the present invention is：
A kind of device of automatic decision digital picture quality, including
Picture quality pre-detection unit obtains the initial relevant parameter of each pictures, according to the initial related ginseng obtained
It is several that an initial benchmark score value is set to every pictures, for evaluating its quality；
Dimension of picture verification unit detaches every figure according to the initial parameter of the every pictures obtained using screening module
The dimensional parameters of piece verify dimension of picture ratio according to dimensional parameters, and carry out utilizing optimization to each pictures
Unit obtains size of each pictures in setting range；
Weight setting unit receives the image parameters after optimization unit optimization, using initial benchmark score value to each
Picture carries out weight setting；
Recognition of face verification unit carries out recognition of face verification to picture, identifies face according to the weight per pictures
Feature, and its position in picture is verified；
Picture quality judging unit reduces it for the Target Photo of location determination, after gray scale, retains it
Structure and light and shade data, and calculated by hash computing modules, filter out low-quality picture；
Colour cast verification unit, the result of calculation according to hash computing modules carry out colour cast verification, the target of verification to picture
It is to filter out partially white picture；
Export picture element unit cell, determine the screening criteria of picture quality coefficient value, filter out more than screening criteria picture simultaneously
Output.
Preferably, further include storage unit and deleting unit, what the storage unit was used to export output picture element unit cell
Picture is stored, and the deleting unit deletes the partially white picture that colour cast verification unit filters out.
Preferably, low quality picture similarity verification unit, the low-quality are provided in the picture quality judging unit
Spirogram piece similarity verification unit carries out the verification of Hamming distance to filtering out low-quality picture, confirm its whether with setting low-quality
It is similar to measure image parameters.
Preferably, RGB color processing unit is provided in the colour cast verification unit, at the RGB color
Reason unit detects its inclined white pixel ratio.
The present invention also provides a kind of methods of automatic decision digital picture quality, include the following steps：
The first step obtains the initial relevant parameter of each pictures according to picture quality pre-detection unit, according to what is obtained
Initial relevant parameter sets an initial benchmark score value R to every pictures, for evaluating its quality；
Second step utilizes the screening module in dimension of picture verification unit according to the initial parameter of the every pictures obtained
Dimensional parameters of the separation per pictures, verify dimension of picture ratio according to dimensional parameters, and to each pictures into
Row obtains size of each pictures in setting range using optimization unit；Picture of the ratio in setting range is to its R value
Increase operation is carried out, ratio carries out reducing beyond the picture in setting range to its R value, and the coefficient of increase and decrease can be adjusted
School；
Third walks：Receive optimization unit optimization after image parameters, using initial benchmark score value to each pictures according to
Weight setting is carried out according to weight setting unit；
4th step carries out recognition of face verification to picture by recognition of face verification unit, identifies facial characteristics, and right
Its position in picture is verified；If there is identifying facial characteristics, then the position to it in picture verifies, and has
Facial characteristics and close to picture intermediate region then to its R value carry out add operation；There is facial characteristics and other than intermediate region
Reducing then is carried out to its R value；No facial characteristics does not operate its R value then；
5th step reduces it using picture quality judging unit for Target Photo, after gray scale, retains it
Structure and light and shade data, and calculating hash values are carried out by hash computing modules；For Target Photo, with known low quality picture
The verification for carrying out Hamming distance, confirms whether it is similar to known spam picture, and hit is similar then to carry out reducing to its R value；
It is dissimilar then its R value is not operated；
6th step carries out colour cast verification using colour cast verification unit to picture, and the target of verification filters out white partially
Picture；Analyzing processing is carried out for Target Photo RGB color, detects that its inclined white pixel ratio, this ratio are more than
Reducing then is carried out to its R value when 60%；
7th step determines the screening criteria of picture quality coefficient value, and the picture filtered out more than screening criteria utilizes output
Picture element unit cell exports.
The invention discloses a kind of device and method of automatic decision number format picture quality.The device and method can be with
Applied to Digital Media, as the Appreciation gist of picture quality, based on the acceptable learning of most of user, filters out and meet mostly
The picture of number user's acceptable learning.The target of this method also resides in, the picture filtered out, and with regard to its size, color saturation comes
It says, is more preferably shown on Digital Media screen.
Description of the drawings
Fig. 1 is the frame principle of device in the present invention；
Fig. 2 is flow chart of the method for the present invention.
Specific implementation mode
Below in conjunction with attached drawing and specific embodiment, the present invention will be described in detail, herein illustrative examples of the invention
And explanation is used for explaining the present invention, but it is not as a limitation of the invention.
Referring to Fig.1, a kind of device of automatic decision digital picture quality, including
Picture quality pre-detection unit 100 obtains the initial relevant parameter of each pictures, according to the initial correlation obtained
Parameter sets an initial benchmark score value to every pictures, for evaluating its quality；
Dimension of picture verification unit 102 is detached according to the initial parameter of the every pictures obtained using screening module 1020
Dimensional parameters per pictures verify dimension of picture ratio according to dimensional parameters, and carry out profit to each pictures
Size of each pictures in setting range is obtained with optimization unit；
Weight setting unit 103 receives the image parameters after optimization unit optimization, using initial benchmark score value to each
Pictures carry out weight setting；
Recognition of face verification unit 104 carries out recognition of face verification to picture, identifies according to the weight per pictures
Facial characteristics, and its position in picture is verified；
Picture quality judging unit 105 reduces it for the Target Photo of location determination, after gray scale, retains
Its structure and light and shade data, and calculated by hash computing modules 1050, filter out low-quality picture；
Colour cast verification unit 106, the result of calculation according to hash computing modules carry out colour cast verification, the mesh of verification to picture
Mark is to filter out partially white picture；
Picture element unit cell 109 is exported, the screening criteria of picture quality coefficient value is determined, filters out the picture more than screening criteria
And it exports.
Further include storage unit 108 and deleting unit 107, the storage unit is used for the figure to output picture element unit cell output
Piece is stored, and the deleting unit deletes the partially white picture that colour cast verification unit filters out.
Low quality picture similarity verification unit 1051, the low-quality are provided in the picture quality judging unit 105
Spirogram piece similarity verification unit 1051 carries out the verification of Hamming distance to filtering out low-quality picture, confirm its whether with setting
Low quality image parameters are similar.
RGB color processing unit 1060 is provided in the colour cast verification unit 106, at the RGB color
Reason unit 1060 detects its inclined white pixel ratio.
With reference to Fig. 2, the present invention also provides a kind of methods of automatic decision digital picture quality, include the following steps：
The first step obtains the initial relevant parameter of each pictures according to picture quality pre-detection unit 100, according to acquisition
Initial relevant parameter to every pictures set an initial benchmark score value R, for evaluating its quality；
Second step utilizes the screening mould in dimension of picture verification unit 102 according to the initial parameter of the every pictures obtained
Block 1020 detaches the dimensional parameters of every pictures, is verified to dimension of picture ratio according to dimensional parameters, and to each
Picture carries out obtaining size of each pictures in setting range using optimization unit；Picture pair of the ratio in setting range
Its R value carries out increase operation, and ratio carries out reducing beyond the picture in setting range to its R value, and the coefficient of increase and decrease can be into
Row adjustment；
Third walks：The image parameters after optimization unit 103 optimizes are received, using initial benchmark score value to each pictures
Weight setting is carried out according to weight setting unit；
4th step carries out recognition of face verification to picture by recognition of face verification unit 104, identifies facial characteristics,
And its position in picture is verified；If there is identifying facial characteristics, then the position to it in picture carries out school
It tests, there is facial characteristics and add operation then is carried out to its R value close to picture intermediate region；Have facial characteristics and intermediate region with
Outer then carries out reducing to its R value；No facial characteristics does not operate its R value then；
5th step reduces it using picture quality judging unit 105 for Target Photo, after gray scale, retains
Its structure and light and shade data, and calculating hash values are carried out by hash computing modules 1050；For Target Photo, with known low-quality
Spirogram piece carries out the verification of Hamming distance, confirms whether it is similar to known spam picture, and hit is similar, subtracts to its R value
Operation；It is dissimilar then its R value is not operated；
6th step carries out colour cast verification using colour cast verification unit 106 to picture, and the target of verification is to filter out white partially
Picture；Analyzing processing is carried out for Target Photo RGB color, detects that its inclined white pixel ratio, this ratio are more than
Reducing then is carried out to its R value when 60%；
7th step determines the screening criteria of picture quality coefficient value, and the picture filtered out more than screening criteria utilizes output
Picture element unit cell 109 exports.
Referring to pixel-parameters 30*30,40*40 ..., the picture of 300*300 is described in detail as embodiment.
In the present embodiment, for selecting the outputting standard that pixel range is best for 100*100~150*150.
The first step obtains the pixel-parameters of each pictures, (pixel-parameters 30* according to picture quality pre-detection unit 100
30,40*40 ..., 300*300) according to obtain pixel to every pictures set an initial benchmark score value R, for example, 30*
The benchmark score value of 30 pixel pictures is that the benchmark score value of 30,40*40 pixel pictures is 40, and so on, to according to for evaluating
Its quality；
Second step, the pixel-parameters according to the every pictures obtained are converted into dimensional parameters, are verified using dimension of picture single
Screening module 1020 in member 102 detaches the dimensional parameters of every pictures, and school is carried out to dimension of picture ratio according to dimensional parameters
Test, and to each pictures carry out using optimization unit obtain each pictures in setting range size (100*100~
150*150)；Picture of the ratio in setting range carries out increase operation to its R value, and ratio is beyond the picture pair in setting range
Its R value carries out reducing, and the coefficient of increase and decrease can carry out adjustment, such as more than the coefficient of reducing be 0.7, increase operation
Coefficient is 1.1；
Third walks：The image parameters after optimization unit 103 optimizes are received, at this point, using initial benchmark score value to each
Pictures carry out weight setting according to weight setting unit,
4th step carries out recognition of face verification to picture by recognition of face verification unit 104, identifies facial characteristics,
And its position in picture is verified；If there is identifying facial characteristics, then the position to it in picture carries out school
It tests, there is facial characteristics and add operation then is carried out to its R value close to picture intermediate region；Have facial characteristics and intermediate region with
Outer then carries out reducing to its R value；No facial characteristics does not operate its R value then；
5th step reduces it using picture quality judging unit 105 for Target Photo, after gray scale, retains
Its structure and light and shade data, and calculating hash values are carried out by hash computing modules 1050；For Target Photo, with known low-quality
Spirogram piece carries out the verification of Hamming distance, confirms whether it is similar to known spam picture, and hit is similar, subtracts to its R value
Operation；It is dissimilar then its R value is not operated；
6th step carries out colour cast verification using colour cast verification unit 106 to picture, and the target of verification is to filter out white partially
Picture；Analyzing processing is carried out for Target Photo RGB color, detects that its inclined white pixel ratio, this ratio are more than
Reducing then is carried out to its R value when 60%；
7th step determines the screening criteria of picture quality coefficient value, and the picture filtered out more than screening criteria utilizes output
Picture element unit cell 109 exports.The photo that object pixel is (100*100~150*150) is obtained to export.
The technical solution disclosed in the embodiment of the present invention is described in detail above, specific implementation used herein
Example is expounded the principle and embodiment of the embodiment of the present invention, and the explanation of above example is only applicable to help to understand
The principle of the embodiment of the present invention；Meanwhile for those of ordinary skill in the art, embodiment, is being embodied according to the present invention
There will be changes in mode and application range, in conclusion the content of the present specification should not be construed as the limit to the present invention
System.
Claims (5)
1. a kind of device of automatic decision digital picture quality, which is characterized in that including
Picture quality pre-detection unit obtains the initial relevant parameter of each pictures, according to the initial relevant parameter pair obtained
An initial benchmark score value is set per pictures, for evaluating its quality；
Dimension of picture verification unit is detached using screening module per pictures according to the initial parameter of the every pictures obtained
Dimensional parameters verify dimension of picture ratio according to dimensional parameters, and carry out utilizing optimization unit to each pictures
Obtain size of each pictures in setting range；
Weight setting unit receives the image parameters after optimization unit optimization, using initial benchmark score value to each pictures
Carry out weight setting；
Recognition of face verification unit carries out recognition of face verification according to the weight per pictures to picture, identifies facial spy
Sign, and its position in picture is verified；
Picture quality judging unit reduces it for the Target Photo of location determination, after gray scale, retains its structure
It with light and shade data, and is calculated by hash computing modules, filters out low-quality picture；
Colour cast verification unit, the result of calculation according to hash computing modules carry out colour cast verification to picture, and the target of verification was
Filter partially white picture；
Picture element unit cell is exported, the screening criteria of picture quality coefficient value is determined, filters out the picture more than screening criteria and output.
2. the device of automatic decision digital picture quality according to claim 1, which is characterized in that further include storage unit
And deleting unit, the storage unit are used to store the picture of output picture element unit cell output, the deleting unit will be inclined
The partially white picture that color verification unit filters out is deleted.
3. the device of automatic decision digital picture quality according to claim 1, which is characterized in that the picture quality is sentenced
It is provided with low quality picture similarity verification unit in disconnected unit, the low quality picture similarity verification unit is to filtering out low-quality
The picture of amount carries out the verification of Hamming distance, confirms whether it is similar to setting low quality image parameters.
4. the device of automatic decision digital picture quality according to claim 1, which is characterized in that the colour cast verification is single
RGB color processing unit is provided in member, the RGB color processing unit detects its inclined white pixel ratio.
5. a kind of method of automatic decision digital picture quality, which is characterized in that include the following steps：
The first step obtains the initial relevant parameter of each pictures according to picture quality pre-detection unit, initial according to what is obtained
Relevant parameter sets an initial benchmark score value R to every pictures, for evaluating its quality；
Second step is detached according to the initial parameter of the every pictures obtained using the screening module in dimension of picture verification unit
Dimensional parameters per pictures verify dimension of picture ratio according to dimensional parameters, and carry out profit to each pictures
Size of each pictures in setting range is obtained with optimization unit；Picture of the ratio in setting range carries out its R value
Increase operation, ratio carries out reducing beyond the picture in setting range to its R value, and the coefficient of increase and decrease can carry out adjustment；
Third walks：The image parameters after optimization unit optimization are received, using initial benchmark score value to each pictures according to power
Weight setup unit carries out weight setting；
4th step carries out recognition of face verification to picture by recognition of face verification unit, identifies facial characteristics, and to its
Position in picture is verified；If there is identifying facial characteristics, then the position to it in picture verifies, and has face
Feature and close to picture intermediate region then to its R value carry out add operation；There is facial characteristics and then right other than intermediate region
Its R value carries out reducing；No facial characteristics does not operate its R value then；
5th step reduces it using picture quality judging unit for Target Photo, after gray scale, retains its structure
With light and shade data, and by hash computing modules carry out calculate hash values；For Target Photo, carried out with known low quality picture
The verification of Hamming distance confirms whether it is similar to known spam picture, and hit is similar then to carry out reducing to its R value；Not phase
Its R value is not operated then seemingly；
6th step carries out colour cast verification using colour cast verification unit to picture, and the target of verification is to filter out partially white picture；
Analyzing processing is carried out for Target Photo RGB color, detects its inclined white pixel ratio, when this ratio is more than 60% then
Reducing is carried out to its R value；
7th step determines the screening criteria of picture quality coefficient value, and the picture filtered out more than screening criteria utilizes output picture
Unit exports.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810174477.XA CN108492280A (en) | 2018-03-02 | 2018-03-02 | A kind of device and method of automatic decision digital picture quality |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810174477.XA CN108492280A (en) | 2018-03-02 | 2018-03-02 | A kind of device and method of automatic decision digital picture quality |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108492280A true CN108492280A (en) | 2018-09-04 |
Family
ID=63341168
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810174477.XA Pending CN108492280A (en) | 2018-03-02 | 2018-03-02 | A kind of device and method of automatic decision digital picture quality |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108492280A (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015010559A1 (en) * | 2013-07-24 | 2015-01-29 | Tencent Technology (Shenzhen) Company Limited | Devices, terminals and methods for image processing |
| CN104834898A (en) * | 2015-04-09 | 2015-08-12 | 华南理工大学 | Quality classification method for portrait photography image |
| CN105139404A (en) * | 2015-08-31 | 2015-12-09 | 广州市幸福网络技术有限公司 | Identification camera capable of detecting photographing quality and photographing quality detecting method |
| CN106202089A (en) * | 2015-05-04 | 2016-12-07 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus determining picture quality and web page display |
| CN106570155A (en) * | 2016-10-28 | 2017-04-19 | 努比亚技术有限公司 | Image management device and method |
| CN106650829A (en) * | 2017-01-04 | 2017-05-10 | 华南理工大学 | Picture similarity calculation method |
-
2018
- 2018-03-02 CN CN201810174477.XA patent/CN108492280A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015010559A1 (en) * | 2013-07-24 | 2015-01-29 | Tencent Technology (Shenzhen) Company Limited | Devices, terminals and methods for image processing |
| CN104834898A (en) * | 2015-04-09 | 2015-08-12 | 华南理工大学 | Quality classification method for portrait photography image |
| CN106202089A (en) * | 2015-05-04 | 2016-12-07 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus determining picture quality and web page display |
| CN105139404A (en) * | 2015-08-31 | 2015-12-09 | 广州市幸福网络技术有限公司 | Identification camera capable of detecting photographing quality and photographing quality detecting method |
| CN106570155A (en) * | 2016-10-28 | 2017-04-19 | 努比亚技术有限公司 | Image management device and method |
| CN106650829A (en) * | 2017-01-04 | 2017-05-10 | 华南理工大学 | Picture similarity calculation method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105187810B (en) | A kind of auto white balance method and electronic medium device based on face color character | |
| CN108615226B (en) | Image defogging method based on generation type countermeasure network | |
| CN105741328B (en) | The shooting image quality evaluating method of view-based access control model perception | |
| CN108537215A (en) | A kind of flame detecting method based on image object detection | |
| CN110675328A (en) | Low-illumination image enhancement method and device based on condition generation countermeasure network | |
| CN111401324A (en) | Image quality evaluation method, device, storage medium and electronic equipment | |
| CN107451969A (en) | Image processing method, device, mobile terminal and computer-readable recording medium | |
| CN107657619B (en) | A kind of low-light (level) Forest fire image dividing method | |
| CN105608700B (en) | Photo screening method and system | |
| CN108401154B (en) | Image exposure degree non-reference quality evaluation method | |
| CN101738394A (en) | Method and system for detecting indoor smog | |
| CN107993209A (en) | Image processing method, device, computer-readable recording medium and electronic equipment | |
| CN110059675A (en) | A kind of robot identifies road traffic law enforcement behavior and provides the method for standardization auxiliary | |
| CN110503092B (en) | Improved SSD monitoring video target detection method based on field adaptation | |
| CN110782448A (en) | Rendered image evaluation method and device | |
| CN107909542A (en) | Image processing method, device, computer-readable recording medium and electronic equipment | |
| CN114359235A (en) | Wood surface defect detection method based on improved YOLOv5l network | |
| CN111291778B (en) | Training method of depth classification model, exposure anomaly detection method and device | |
| CN104915933A (en) | Foggy day image enhancing method based on APSO-BP coupling algorithm | |
| CN113344843B (en) | Image quality evaluation method, device and system | |
| CN103207995A (en) | PCA (Principal Component Analysis)-based 3D (three dimensional) face identification method | |
| CN111898525B (en) | Construction method of smoke identification model, and method and device for detecting smoke | |
| CN108492280A (en) | A kind of device and method of automatic decision digital picture quality | |
| CN112365425A (en) | Low-illumination image enhancement method and system | |
| CN115690100B (en) | Semi-supervised signal point detection model training method, signal point detection method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180904 |
|
| RJ01 | Rejection of invention patent application after publication |