CN109145930A - It is a kind of based on semi-supervised image significance detection method - Google Patents
It is a kind of based on semi-supervised image significance detection method Download PDFInfo
- Publication number
- CN109145930A CN109145930A CN201810969685.9A CN201810969685A CN109145930A CN 109145930 A CN109145930 A CN 109145930A CN 201810969685 A CN201810969685 A CN 201810969685A CN 109145930 A CN109145930 A CN 109145930A
- Authority
- CN
- China
- Prior art keywords
- coordinate system
- semi
- region unit
- detection method
- brightness
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
It is disclosed by the invention to belong to saliency detection technique field, it is specially a kind of based on semi-supervised image significance detection method, the specific detecting step based on semi-supervised image significance detection method is as follows: S1: establishing coordinate system: establishing plane coordinate system, plane coordinate system is established based on operating system, plane coordinate system horizontal axis X, longitudinal axis Y, coordinate origin O；S2: insertion object divides region；S3: object handles in territory element；S4: significant point is highlighted, background blurring；S5: regions moduleization combination saves, this programme is individually operated by the division mode progress of region unit by will test object, amplifying of region unit can be handled, it is individual using each region block as unit, so that not influencing each other between the block of each region, BORDER PROCESSING is more complete, and region unit is individually operated, reduces the case where malfunctioning to the identification of identification object.
Description
Technical field
It is specially a kind of to be detected based on semi-supervised saliency the present invention relates to saliency detection technique field
Method.
Background technique
In recent years, with the fast development of multimedia technology and internet communication, image classification problem, which receives, is much ground
The concern for the person of studying carefully, various image classification algorithms also emerge one after another.However, many traditional image classification algorithms are all based on supervision
Learn to be studied, this needs collects a large amount of markd samples before training could establish accurate sorter model
And achieve the purpose that correctly to classify.And the markers work of this repeatability both time-consuming consumption wealth, but a large amount of unmarked samples of collection
But it is easy to.It, can be from a large amount of medical image of infection from hospital as training such as in computer-aided medical science image analysis
Example, it is but if requiring medical expert that the lesion in these images is all marked, then often unpractical.Furthermore with
The development of present big data, classifies to the webpage information to magnanimity, and manually each and every one marking then to one is even more to add in hardly possible
It is difficult.
Based on such problems, researcher, which starts to only use a small amount of valuable marked sample, to be trained, and is led to
Cross the learning performance for gradually improving classifier using a large amount of unmarked sample.
Active Learning and semi-supervised learning are the popular algorithms in this field at present.
In semi-supervised learning, learner utilizes unmarked sample automatically, is not necessarily to manual intervention in whole process, it is thus only necessary to
Classification results are not marked accurately most to sample and its predict that obtained label is added in marked training set.Have at present it is many from
Learning art and its innovatory algorithm are all the classical semi-supervised learnings of comparison.
Salient region in image refers to the target area most paid close attention in human vision in a secondary picture, conspicuousness detection
Result inhibit background while target area uniformly can be highlighted prominent, be important at present convenient for the post-processing of image
One of research field.Conspicuousness detection in recent years has been widely used in the retrieval to close image, image and video pressure
Many field of image processings such as detection, image and Video segmentation of specific objective in contracting, image, and these necks are promoted well
The development in domain.Original saliency detection mode be easy to cause boundary imperfect, and test object identification error is be easy to cause
The situation of prominent target inaccuracy.
Summary of the invention
The purpose of the present invention is to provide a kind of based on semi-supervised image significance detection method, to solve above-mentioned background
The original saliency detection mode proposed in technology be easy to cause boundary imperfect, and test object identification error is easy
Cause the problem of prominent target inaccuracy.
To achieve the above object, the invention provides the following technical scheme: a kind of detected based on semi-supervised saliency
It is as follows to be somebody's turn to do the specific detecting step based on semi-supervised image significance detection method for method:
S1: it establishes coordinate system: establishing plane coordinate system, plane coordinate system is established based on operating system, and plane coordinate system is horizontal
Axis X, longitudinal axis Y, coordinate origin O；
S2: insertion object divides region: will test in the plane coordinate system established in object inserting step S1, adjustment inspection
Object is surveyed to be fully located in plane coordinate system, and a corner points of test object are overlapped with coordinate origin O, to test object into
Row region unit divides, and will test object equal proportion and is divided into H parts, the central point F (x of each region unitf, yf) it is used as the region unit
Identification code name；
S3: object handles in territory element: individually handling the step S2 individual region unit for being always divided into H parts,
The grade of brightness is set, high, normal, basic three grades is divided into, the brightness degree of the pixel in each region unit is detected, to region unit
In the brightness degree of each pixel be marked；
S4: significant point is highlighted, background blurring: marking according in step S3 to the brightness degree of pixel each in region unit
Value is handled:
Raising brightness is carried out for high-grade pixel；
For in, the pixel of inferior grade carry out reduction brightness；
S5: regions moduleization combination saves: by treated in step S4, region unit is kept separately and by all areas block
After treatment carries out original position combination, in conjunction with rear preservation.
Preferably, the operating system in the step S1 is LINUX or Windows operating system.
It preferably, is processing or synchronization process one by one to the processing of individual region unit in the step S3.
Preferably, it is described it is insufficient after the label that marks in S3 include three kinds, and three of the labels of three kinds of labels and brightness
Grade corresponds.
Preferably, brightness is adjusted to be directed to pixel by exposure in the step S4.
Compared with prior art, the beneficial effects of the present invention are: this programme is by will test object drawing by region unit
Point mode carries out individually operated, can handle amplifying of region unit, individual using each region block as unit, so that respectively
It not influencing each other between a region unit, BORDER PROCESSING is more complete, and region unit is individually operated, reduce the identification to identification object
The case where error.
Detailed description of the invention
Fig. 1 is detection method flow chart.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Referring to Fig. 1, the present invention provides a kind of technical solution: it is a kind of based on semi-supervised image significance detection method,
The specific detecting step based on semi-supervised image significance detection method is as follows:
S1: it establishes coordinate system: establishing plane coordinate system, plane coordinate system is established based on operating system, and plane coordinate system is horizontal
Axis X, longitudinal axis Y, coordinate origin O establish plane coordinate system on photoshop CS5, and coordinate points are shown in photoshop
In the operation interface of CS5；
S2: insertion object divides region: picture to be detected is inserted into the plane coordinate system of photoshop CS5,
Picture to be detected is shown by way of rectangle, and an angle of picture to be detected is overlapped with coordinate origin O, will be to be detected
Picture be uniformly divided into 300 parts in such a way that rectangle is divided in plane coordinate system, it is every it is a be used as a territory element,
One territory element is proposed and amplified to it 300 times of progress light sensation information collections, the pixel in the territory element is carried out
Brightness acquisition, the range of senior middle school's inferior grade of brightness be respectively as follows: high-grade > 350nit, 180nit≤middle grade≤256nit,
Inferior grade < 180nit；
The brightness value of each pixel of acquisition is compareed with the codomain range of each grade, so that it is determined that it belonged to etc.
Grade improve for high-grade pixel the brightness value of 30nit；For in, the pixel of inferior grade carry out reduction 50nit
Brightness value so that during treated, the brightness of the pixel of inferior grade be not less than 30nit；
It will test in the plane coordinate system established in object inserting step S1, adjustment test object is fully located at plane coordinates
In system, and a corner points of test object are overlapped with coordinate origin O, are carried out region unit division to test object, be will test pair
As equal proportion is divided into H parts, the central point F (x of each region unitf, yf) identification code name as the region unit, the F in H region
(xf, yf) it is considered as a set M；
S3: object handles in territory element: the individual region unit of H part that step S2 is always divided into set M is carried out independent
Processing, is arranged the grade of brightness, is divided into high, normal, basic three grades, detect the brightness degree of the pixel in each region unit, right
The brightness degree of each pixel is marked in region unit, is converted to supervised learning or unsupervised according to measurement result
It practises, image procossing is carried out by traditional mode；
S4: significant point is highlighted, background blurring: marking according in step S3 to the brightness degree of pixel each in region unit
Value is handled:
Raising brightness is carried out for high-grade pixel；
For in, the pixel of inferior grade carry out reduction brightness；
S5: regions moduleization combination saves: by treated in step S4, region unit is kept separately and by all areas block
After treatment carries out original position combination, in conjunction with rear preservation.
Wherein, the operating system in the step S1 is LINUX or Windows operating system, right in the step S3
The processing of individual region unit is processing or synchronization process one by one, and the label marked in S3 after the deficiency includes three kinds, and
Three kinds of labels marked and the three grades of brightness correspond, and are directed to pixel progress in the step S4 by exposure
Adjust brightness.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
A variety of variations, modification, replacement can be carried out to these embodiments without departing from the principles and spirit of the present invention by understanding
And modification, the scope of the present invention is defined by the appended.
Claims (5)
1. a kind of based on semi-supervised image significance detection method, it is characterised in that: should be based on semi-supervised saliency
The specific detecting step of detection method is as follows:
S1: it establishes coordinate system: establishing plane coordinate system, plane coordinate system is established based on operating system, plane coordinate system horizontal axis X,
Longitudinal axis Y, coordinate origin O；
S2: insertion object divides region: will test in the plane coordinate system established in object inserting step S1, adjustment detection pair
As being fully located in plane coordinate system, and a corner points of test object are overlapped with coordinate origin O, carry out area to test object
Domain block divides, and will test object equal proportion and is divided into H parts, the central point F (x of each region unitf, yf) knowledge as the region unit
Other code name；
S3: object handles in territory element: individually handling the step S2 individual region unit for being always divided into H parts, setting
The grade of brightness is divided into high, normal, basic three grades, detects the brightness degree of the pixel in each region unit, to each in region unit
The brightness degree of a pixel is marked；
S4: significant point is highlighted, background blurring: according in step S3 to the brightness degree mark value of pixel each in region unit into
Row processing:
Raising brightness is carried out for high-grade pixel；
For in, the pixel of inferior grade carry out reduction brightness；
S5: regions moduleization combination saves: by treated in step S4, region unit is kept separately and handles all areas block
After carry out original position combination, in conjunction with rear preservation.
2. according to claim 1 a kind of based on semi-supervised image significance detection method, it is characterised in that: the step
Operating system in rapid S1 is LINUX or Windows operating system.
3. according to claim 1 a kind of based on semi-supervised image significance detection method, it is characterised in that: the step
It is processing or synchronization process one by one to the processing of individual region unit in rapid S3.
4. according to claim 1 a kind of based on semi-supervised image significance detection method, it is characterised in that: it is described not
The label marked in S3 after foot includes three kinds, and three kinds of labels marked and the three grades of brightness correspond.
5. according to claim 1 a kind of based on semi-supervised image significance detection method, it is characterised in that: the step
Brightness is adjusted to be directed to pixel by exposure in rapid S4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810969685.9A CN109145930A (en) | 2018-08-23 | 2018-08-23 | It is a kind of based on semi-supervised image significance detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810969685.9A CN109145930A (en) | 2018-08-23 | 2018-08-23 | It is a kind of based on semi-supervised image significance detection method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109145930A true CN109145930A (en) | 2019-01-04 |
Family
ID=64827715
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810969685.9A Pending CN109145930A (en) | 2018-08-23 | 2018-08-23 | It is a kind of based on semi-supervised image significance detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109145930A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110602384A (en) * | 2019-08-27 | 2019-12-20 | 维沃移动通信有限公司 | Exposure control method and electronic device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1536531A (en) * | 2003-04-10 | 2004-10-13 | ������������ʽ���� | Image processing device and image processing method and processing program |
| CN1770254A (en) * | 2004-10-15 | 2006-05-10 | 创世纪微芯片公司 | Method of generating transfer curves for adaptive contrast enhancement |
| CN101485191A (en) * | 2006-07-04 | 2009-07-15 | 三星电子株式会社 | Image compensation apparatus and method |
| CN101582991A (en) * | 2008-05-13 | 2009-11-18 | 深圳华为通信技术有限公司 | Method and device for processing image |
| CN103631568A (en) * | 2013-12-20 | 2014-03-12 | 厦门大学 | Medical-image-oriented multi-thread parallel computing method |
| CN107895352A (en) * | 2017-10-30 | 2018-04-10 | 维沃移动通信有限公司 | A kind of image processing method and mobile terminal |
| CN108009522A (en) * | 2017-12-21 | 2018-05-08 | 海信集团有限公司 | A kind of Approach for road detection, device and terminal |
-
2018
- 2018-08-23 CN CN201810969685.9A patent/CN109145930A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1536531A (en) * | 2003-04-10 | 2004-10-13 | ������������ʽ���� | Image processing device and image processing method and processing program |
| CN1770254A (en) * | 2004-10-15 | 2006-05-10 | 创世纪微芯片公司 | Method of generating transfer curves for adaptive contrast enhancement |
| CN101485191A (en) * | 2006-07-04 | 2009-07-15 | 三星电子株式会社 | Image compensation apparatus and method |
| CN101582991A (en) * | 2008-05-13 | 2009-11-18 | 深圳华为通信技术有限公司 | Method and device for processing image |
| CN103631568A (en) * | 2013-12-20 | 2014-03-12 | 厦门大学 | Medical-image-oriented multi-thread parallel computing method |
| CN107895352A (en) * | 2017-10-30 | 2018-04-10 | 维沃移动通信有限公司 | A kind of image processing method and mobile terminal |
| CN108009522A (en) * | 2017-12-21 | 2018-05-08 | 海信集团有限公司 | A kind of Approach for road detection, device and terminal |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110602384A (en) * | 2019-08-27 | 2019-12-20 | 维沃移动通信有限公司 | Exposure control method and electronic device |
| CN110602384B (en) * | 2019-08-27 | 2022-03-29 | 维沃移动通信有限公司 | Exposure control method and electronic device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107767376B (en) | X-ray bone age prediction method and system based on deep learning | |
| CN109859203B (en) | Defect tooth image identification method based on deep learning | |
| Quelhas et al. | Cell nuclei and cytoplasm joint segmentation using the sliding band filter | |
| WO2022012110A1 (en) | Method and system for recognizing cells in embryo light microscope image, and device and storage medium | |
| CN113723573B (en) | Tumor tissue pathological classification system and method based on adaptive proportion learning | |
| CN111368643B (en) | Escherichia coli dynamic growth monitoring method | |
| CN112614128B (en) | System and method for assisting biopsy under endoscope based on machine learning | |
| WO2022042348A1 (en) | Medical image annotation method and apparatus, device, and storage medium | |
| CN111754453A (en) | Pulmonary tuberculosis detection method and system based on chest radiography image and storage medium | |
| CN108629762B (en) | Image preprocessing method and system for reducing interference characteristics of bone age evaluation model | |
| CN111508016B (en) | Vitiligo region chromaticity value and area calculation method based on image processing | |
| CN110415212A (en) | Abnormal cell detection method, device and computer readable storage medium | |
| Huang et al. | Automated hemorrhage detection from coarsely annotated fundus images in diabetic retinopathy | |
| CN106570515A (en) | Method and system for treating medical images | |
| US11804029B2 (en) | Hierarchical constraint (HC)-based method and system for classifying fine-grained graptolite images | |
| CN111126143A (en) | Deep learning-based exercise judgment guidance method and system | |
| CN110674726A (en) | Skin disease auxiliary diagnosis method and system based on target detection and transfer learning | |
| CN115526834A (en) | Immunofluorescence image detection method and device, equipment and storage medium | |
| CN105354405A (en) | Machine learning based immunohistochemical image automatic interpretation system | |
| Tang et al. | Leaf extraction from complicated background | |
| CN110197483A (en) | Deep basal pit crack detection method based on vision signal | |
| CN110415246B (en) | Analysis method of abdomen fat component | |
| CN109145930A (en) | It is a kind of based on semi-supervised image significance detection method | |
| CN113658129A (en) | Position extraction method combining visual saliency and line segment strength | |
| Gurunathan et al. | Plant leaf diseases detection using KNN classifier |
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: 20190104 |
|
| RJ01 | Rejection of invention patent application after publication |