CN105930391A - Update method and image server of image sample database of super-resolution image system - Google Patents
Update method and image server of image sample database of super-resolution image system Download PDFInfo
- Publication number
- CN105930391A CN105930391A CN201610232700.2A CN201610232700A CN105930391A CN 105930391 A CN105930391 A CN 105930391A CN 201610232700 A CN201610232700 A CN 201610232700A CN 105930391 A CN105930391 A CN 105930391A
- Authority
- CN
- China
- Prior art keywords
- image
- sample data
- original image
- data storehouse
- original
- 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
- 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
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
Abstract
One embodiment of the invention provides an update method and an image server of an image sample database of a super-resolution image system and relates to the field of image processing, aiming at increasing update efficiency of the sample database and reducing waste of system resources. The method comprises following steps: obtaining at least one original image and extracting image features from the original image; determining whether de-duplication is performed on image features of the original image according to the image features of the original image and image features of saved images in the image sample database; performing de-duplication on the image features of the original image if the image features of the original image exist in the image sample database; extracting the image features of the original image and updating the image sample database according to the original image if the image features of the original image do not exist in the image sample database. The embodiment of the invention is used for updating an image sample database.
Description
Technical field
Embodiments of the invention belong to image processing field, particularly relate to a kind of supersolution as image in system
The update method of sample database and image server.
Background technology
The high-resolution of image is to point to the picture element density height that image contains, and is provided that abundant details letter
Breath, the description to objective scene is the most careful.High-definition picture the information age demand very
Extensively, take photo by plane field, medical digital are investigated in such as satellite remote sensing images, Video security monitoring, military affairs
The field such as image and video format conversion all has highly important application.Wherein, super image resoluting technique can
More higher high-definition picture than original image resolution to generate, the most generally it is widely used in height
Image in different resolution processes.
Super image resoluting technique is broadly divided into two classes: based on the super image resoluting technique rebuild and supersolution based on study
As technology.Refer to that all information all can only be all from input image data based on the super image resoluting technique rebuild
Middle acquisition, do not have any additional background knowledge, whole solution process be equivalent to an information retrieval and
The problem of information fusion.Along with differentiate amplification coefficient increase, it is desirable to provide input picture sample number
Amount sharply increases, until after reaching the upper limit of amplification coefficient, no matter increasing how many input picture sample,
All cannot improve reconstruction effect again.For the limitation of algorithm for reconstructing, super image resoluting technique based on study
Research field as a forward position is arisen at the historic moment.
At present, based on study super image resoluting technique, mainly image server according to obtain wait train
Image, hereinafter referred to as original image, be trained, obtain the low resolution figure with same characteristic features
With corresponding high resolution graphics.The corresponding picture of one feature is to (low resolution and high-resolution
Picture to), then image server storage has multiple pictures pair that multiple feature is corresponding, such picture
(or feature to) is constituted a sample database.This sample database is provided to terminal, for end
End, when getting the image of low resolution, by this sample database, determines and comprises same characteristic features
High-resolution image, thus realize terminal and carry out high-resolution image and show.
Along with the development of Internet technology, new video image kind and content (and video source figure
The characteristics of image of picture) constantly occur, in order to ensure terminal play video time, as much as possible by low
The video image of resolution is converted to high-resolution video image, and this is accomplished by sample database
Picture is to enough real-time, so, it is necessary to the image in real-time update sample database.At present, more
During image in new images sample database, it is to upload image to be trained by artificial timing, passes through
The image uploaded updates, and so, picture is inefficient to updating.Additionally, prior art is updating
During original image, it is to update by the way of directly increasing original image.That is, as long as server connects
Receive that artificial timing uploads when the image trained, all this image can be trained, obtain training
After the image pair that obtains, increase in sample database.But, it is contemplated that in actual application scenarios,
There may be the image uploaded the most in the same time, there is identical feature, if this feature is instructed again
When practicing, the renewal efficiency of sample database can be reduced, but also system resource can be wasted.
Summary of the invention
Embodiments of the invention provide a kind of supersolution as the update method in image sample data storehouse in system
And image server, it is possible to increase the renewal efficiency of sample database, reduces system resource waste.
First aspect, it is provided that a kind of supersolution is as the update method in the image sample data storehouse in system, bag
Include:
Obtain at least one original image, the most each described original image include low-resolution image and
The image pair of corresponding high-definition picture composition；
Characteristics of image is extracted in described original image；
The figure of the prestored digital image in characteristics of image according to described original image and image sample data storehouse
As feature judges whether the characteristics of image duplicate removal in described original image；
If the characteristics of image of original image described in the existence in described image sample data storehouse, then to described
The characteristics of image duplicate removal of original image；
If the characteristics of image that there is not described original image in described image sample data storehouse, then extract
The characteristics of image of described original image also updates described image sample data storehouse according to described original image.
On the other hand, it is provided that a kind of image server, including:
Acquiring unit, is used for obtaining at least one original image, and the most each described original image includes
Low-resolution image and the image pair of corresponding high-definition picture composition；
Sampling unit, for extracting characteristics of image in the original image that described acquiring unit obtains；
Processing unit, in the characteristics of image according to described original image and image sample data storehouse
The characteristics of image of prestored digital image judges whether the characteristics of image duplicate removal in described original image；If it is described
The characteristics of image of original image described in existence in image sample data storehouse, then to described original image
Characteristics of image duplicate removal；If the image that there is not described original image in described image sample data storehouse is special
Levy, then extract described original image characteristics of image and according to described original image update described figure decent
Database.
Embodiments of the invention provide supersolution as in system image sample data storehouse update method and
Image server, obtains at least one original image, and the most each described original image includes low resolution
Rate image and the image pair of corresponding high-definition picture composition；Image is extracted in described original image
Feature；The figure of the prestored digital image in characteristics of image according to described original image and image sample data storehouse
As feature judges whether the characteristics of image duplicate removal in described original image；If described image sample data
The characteristics of image of original image described in existence in storehouse, then go the characteristics of image of described original image
Weight；If the characteristics of image that there is not described original image in described image sample data storehouse, then extract
The characteristics of image of described original image also updates described image sample data storehouse according to described original image；
When therefore obtaining new original image in image sample data storehouse, it is possible to decent to figure according to characteristics of image
Database is updated, it is to avoid by the characteristics of image update all of all original images to image pattern
Data base, it is possible to increase the renewal efficiency of sample database, reduces system resource waste.
Accompanying drawing explanation
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below by right
In embodiment description, the required accompanying drawing used is briefly described, it should be apparent that, in describing below
Accompanying drawing be only some embodiments of the present invention, for those of ordinary skill in the art, not
On the premise of paying creative work, it is also possible to obtain other accompanying drawing according to these accompanying drawings.
Fig. 1 provides a kind of supersolution as the structural representation of system for embodiments of the invention；
Fig. 2 provides a kind of supersolution as the image sample data storehouse in system more for embodiments of the invention
The schematic flow sheet of new method；
Fig. 3 provides a kind of supersolution as the image sample data storehouse in system for another embodiment of the present invention
The schematic flow sheet of update method；
Fig. 4 provides a kind of supersolution as the image sample data storehouse in system for one more embodiment of the present invention
The schematic flow sheet of update method；
Fig. 5 provides the schematic diagram of a kind of original image for embodiments of the invention；
Fig. 6 provides the structural representation of a kind of image server for embodiments of the invention；
Fig. 7 provides the structural representation of a kind of image server for another embodiment of the present invention.
Detailed description of the invention
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is entered
Row clearly and completely describes, it is clear that described embodiment is only a part of embodiment of the present invention,
Rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not having
Have and make the every other embodiment obtained under creative work premise, broadly fall into present invention protection
Scope.
The principle that embodiments of the invention are used is: in supersolution as, in system, updating image pattern number
According to no longer using the mode directly increasing original image to update during storehouse, but after obtaining original image,
The characteristics of image of the prestored digital image in characteristics of image according to original image and image sample data storehouse is sentenced
Disconnected whether to the characteristics of image duplicate removal in original image, if the existence institute in described image sample data storehouse
State the characteristics of image of original image, then the characteristics of image duplicate removal to described original image；If described image
The characteristics of image that there is not original image in sample database, then extract the characteristics of image of original image
And update image sample data storehouse according to original image；Avoid the characteristics of image by all original images complete
Portion is updated to image sample data storehouse, it is possible to increase the renewal efficiency in image sample data storehouse, reduces system
The system wasting of resources.
As it is shown in figure 1, the supersolution applied of the present invention is as system is by the image server in high in the clouds and movement
Terminal is constituted, and Fig. 1 provides this supersolution as the Organization Chart of system, wherein:
Acquisition for mobile terminal carries out pretreatment to remove interference noise, then after the image of signal source
The image of low resolution carries out a liter sampling, obtains the image of high-resolution, (wherein up-samples permissible
By the picture element signal of low resolution is carried out interpolation, obtain more picture element signal and form relatively high-resolution
The image of rate, carries out supersolution picture by the image of the high-resolution obtained after interpolation, obtains high-resolution
Image)；Then, extract the characteristics of image of the image of this high-resolution, according to this characteristics of image and
Image sample data storehouse carries out template matching, is melted by the image of the image matched and low resolution
Close (merging) and export.
The image server in high in the clouds, is existed by network end user (e.g., attendant or developer etc.)
Image in memorizer, by uploading or downloading original image, is updated, then to storage by high in the clouds
Original image in device carries out sample and chooses, and removes what original image was brought by noise etc. by pretreatment
Interference, then extracts characteristics of image, and sub-category training image sample database, to original image sample number
It is updated according to storehouse.So effective dynamically lifting super-resolution display effect, can allow picture more
Approaching to reality effect.Based on above-mentioned supersolution as system, prior art, when updating original image, it is
Update by the way of directly increasing original image.That is, as long as image server receives and manually determines
Shi Shangchuan when the image trained, all this image can be trained, obtain the figure that obtains after training
As right, increase in image sample data storehouse.But, it is contemplated that in actual application scenarios, Ke Nengcun
At the image uploaded the most in the same time, there is identical feature, if this feature is trained again,
The renewal efficiency of sample database can be reduced, but also system resource can be wasted.
For solving above-mentioned technical problem, based on above-mentioned supersolution as system, shown in reference Fig. 2, this
Bright enforcement a kind of supersolution of offer is as the update method in the image sample data storehouse in system, including as follows
Step:
101, image server obtains at least one original image.
The most each original image includes what low-resolution image and corresponding high-definition picture formed
Image pair.Wherein in step 101, obtain at least one original image, specifically include following two side
Formula:
Mode one: obtain low-resolution image；Described low-resolution image is transformed to high resolution graphics
Picture.
Mode two: obtain high-definition picture；Described high-definition picture is transformed to low resolution figure
Picture.
102, image server extracts characteristics of image in original image.
103, image server is according in the characteristics of image of described original image and image sample data storehouse
The characteristics of image of prestored digital image judge whether the characteristics of image duplicate removal in original image.
If 104, the characteristics of image of the existence original image in image server image sample data storehouse,
The then characteristics of image duplicate removal to original image.
If 105, there is not described original image in image server described image sample data storehouse
Characteristics of image, then extract the characteristics of image of original image and update image sample data according to original image
Storehouse.
The supersolution that embodiments of the invention provide is as the update method in image sample data storehouse in system, logical
Crossing and obtain at least one original image, the most each original image includes low-resolution image and correspondence
The image pair of high-definition picture composition；Characteristics of image is extracted in original image；According to original image
Characteristics of image and image sample data storehouse in the characteristics of image of prestored digital image judge whether original
Characteristics of image duplicate removal in image；If the image of the existence original image in image sample data storehouse is special
Levy, then the characteristics of image duplicate removal to described original image；If not existing in image sample data storehouse is former
The characteristics of image of beginning image, then extract the characteristics of image of original image and according to original image more new images
Sample database；When therefore obtaining new original image in image sample data storehouse, it is possible to according to image
Image sample data storehouse is updated by feature, it is to avoid by the characteristics of image of all original images the most more
New to image sample data storehouse, it is possible to increase the renewal efficiency of sample database, reduce system resource wave
Take.
With reference to shown in Fig. 3, it is provided that another supersolution as the update method in the image sample data storehouse in system,
Comprise the steps:
201, image server obtains at least one original image.
The most each original image includes what low-resolution image and corresponding high-definition picture formed
Image pair, wherein obtains in the method present invention of image and does not limits, and can be attendant or user
Directly upload pre-prepd image, or the real time imaging downloaded from client terminal web page, wherein step
In 201, obtain at least one original image, specifically include following two mode:
Mode one: obtain low-resolution image；Described low-resolution image is transformed to high resolution graphics
Picture.
Mode two: obtain high-definition picture；Described high-definition picture is transformed to low resolution figure
Picture.
202, image server carries out filtration treatment to high-definition picture, removes high-definition picture
In low resolution feature.
In view of in actual application scenarios, full resolution pricture there may be low resolution feature, wherein
Low resolution feature can be: mosaic；These low resolution features can be in high-definition picture
Image characteristics extraction produces impact, therefore can avoid low-resolution image feature pair by step 202
The impact of high-definition picture.
203, image server extracts characteristics of image in original image.
Concrete a kind of in step 203 optional mode is: obtains in high-definition picture and differentiates
Rate is higher than the characteristics of image of predetermined threshold value.Wherein low-resolution image can also be avoided special by which
Levying the impact on high-definition picture, therefore which and step 202 can be two kinds of optional sides
Formula, i.e. can filter out the low resolution feature in high-definition picture by step 202, it is also possible to
Use and remove the low resolution feature in high-definition picture such as the mode in step 203.Characteristics of image
At least include: one or more in color characteristic, Gradient Features, frequency domain character and architectural feature.
204, image server according in the characteristics of image of original image and image sample data storehouse
The characteristics of image depositing image judges whether the characteristics of image duplicate removal in original image.
Prestored digital image in image sample data storehouse stores according to affiliated image category, and image category is at least
Including: landscape class, figure kind；In order to improve lookup speed, can be in advance to image sample data storehouse
In feature or correspondence image classification, e.g., figure kind, landscape class.Can after extracting characteristics of image
To judge the classification of the image residing for this characteristics of image, if belonging to figure kind, just search from figure kind,
Lookup speed can be effectively improved.Concrete the most also includes: judge the figure of original image
As the image category belonging to feature.
Concrete step 204 includes: the described characteristics of image according to described original image and image pattern
The characteristics of image of the prestored digital image in data base judges the figure of the prestored digital image in image sample data storehouse
As whether feature exists the similar or identical characteristics of image of the characteristics of image of described original image, if
Then to the characteristics of image duplicate removal in described original image.
Following two judgment mode is used for the duplicate removal mode in above-mentioned steps 204:
Mode one:
S1, the first Random Maps of the characteristics of image obtaining described original image represent；
S2, second the reflecting at random of characteristics of image of the prestored digital image obtained in described image sample data storehouse
The sequencing table that representation by arrows is constituted；
S3, according to sequencing table described in described sequencing table sorted search, whether inquiry sequencing table exists and the
The similarity that one Random Maps represents is more than the characteristics of image of first threshold, if then it is determined that the presence of identical
Characteristics of image.
Mode two:
S1, obtain original image characteristics of image dimension normalization after global image represent；
The first Hash sequence that S2, generation global image represent；
If the interval of S3 the first Hash sequence mapping has comprised at least one in image sample data storehouse
Deposit the second Hash sequence that the characteristics of image of image is corresponding；Then it is determined that the presence of similar image features.
If the characteristics of image of the existence original image in 205 image sample data storehouses, then images serve
The device characteristics of image duplicate removal to original image.
If the characteristics of image that there is not original image in 206 described image sample data storehouses, then scheme
As server extracts the characteristics of image of original image and updates image sample data storehouse according to original image.
With reference to the example shown in Fig. 4, the specific implementation of this above-described embodiment is described as follows:
301, image server obtains image PC1 and image PC2.
Wherein image PC1 is high-definition picture, and image PC2 is high-definition picture, image PC1
Characteristics of image 1 (circular) and characteristics of image 2 (rhombus) is all comprised, as shown in Figure 5 with PC2.
302, image server extracts characteristics of image 1 and 2 respectively in image PC1 and PC2；
The characteristics of image of the prestored digital image during 303, image server judges whether image sample data storehouse
In whether there is characteristics of image 1 and 2.
As shown in table 1, image sample data storehouse is deposited with the corresponding relation of characteristics of image and original image
Storage characteristics of image and original image；
| Characteristics of image | Low-resolution image | High-definition picture |
| T1 | P11 | P21 |
| T2 | P12 | P22 |
| Characteristics of image 1 | P13 | P23 |
304, image server judgement table 1 exists characteristics of image 1, then to characteristics of image 1 duplicate removal
Process.
305, image server judgement table 1 does not exist characteristics of image 2, extract characteristics of image 2 also
Image sample data storehouse is updated according to PC1 and PC2.
With reference to shown in Fig. 6, embodiments of the invention provide a kind of image server, including:
Acquiring unit 61, is used for obtaining at least one original image, the most each described original image
Including low-resolution image and the image pair of corresponding high-definition picture composition；
Sampling unit 62, special for extracting image in the original image that described acquiring unit 61 obtains
Levy；
Processing unit 63, for the characteristics of image according to described original image and image sample data storehouse
In the characteristics of image of prestored digital image judge whether the characteristics of image duplicate removal in described original image；If
The characteristics of image of original image described in existence in described image sample data storehouse, then to described original graph
The characteristics of image duplicate removal of picture；If the figure that there is not described original image in described image sample data storehouse
As feature, then extract the characteristics of image of described original image and update described figure according to described original image
Decent database.
Optionally, shown in reference Fig. 7, described acquiring unit 61, including:
Obtain subelement 611, be used for obtaining low-resolution image；
Conversion subelement 612, for being transformed to the described low-resolution image obtaining subelement acquisition
High-definition picture.
Optionally, described acquiring unit 61, including:
Obtain subelement 611, be used for obtaining high-definition picture；
Conversion subelement 612, for being transformed to the described high-definition picture obtaining subelement acquisition
Low-resolution image.
Optionally, image server also includes: filter element 64, for described high resolution graphics
As carrying out filtration treatment, remove the low resolution feature in described high-definition picture.In view of reality
In application scenarios, there may be low resolution feature in full resolution pricture, wherein low resolution feature can
Think: mosaic；Image characteristics extraction in high-definition picture can be produced by these low resolution features
Raw impact, therefore can avoid low point by the low resolution feature in the described high-definition picture of removal
The impact on high-definition picture of the resolution characteristics of image.
Optionally, sampling unit 62, it is higher than specifically for obtaining resolution in high-definition picture
The characteristics of image of predetermined threshold value.Wherein by obtaining resolution in high-definition picture higher than presetting threshold
The characteristics of image of value, can avoid the impact on high-definition picture of the low-resolution image feature.
Optionally, described characteristics of image at least includes: color characteristic, Gradient Features, frequency domain character and
One or more in architectural feature.
Optionally, the prestored digital image in described image sample data storehouse stores according to affiliated image category,
Described image category at least includes: landscape class, figure kind；
Described processing unit 63 is additionally operable to the image class judged belonging to the characteristics of image of described original image
Not.
Optionally, described processing unit 63 specifically for the characteristics of image according to described original image and
The characteristics of image of the prestored digital image in image sample data storehouse judges depositing in image sample data storehouse
Whether the characteristics of image of image exists the similar or identical image of the characteristics of image of described original image
Feature, if then to the characteristics of image duplicate removal in described original image.
Optionally, described processing unit 63 is specifically for obtaining the characteristics of image of described original image
First Random Maps represents；Obtain the characteristics of image of prestored digital image in described image sample data storehouse
Second Random Maps represents the sequencing table of composition；According to sequencing table described in described sequencing table sorted search,
Inquire about whether described sequencing table exists and state similarity that the first Random Maps represents more than first threshold
Characteristics of image, if then it is determined that the presence of identical image feature.
Optionally, described processing unit 63 is specifically for obtaining the characteristics of image chi of described original image
Global image after degree normalization represents；Generate the first Hash sequence that described global image represents；If
The interval of described first Hash sequence mapping has comprised at least one in described image sample data storehouse
Deposit the second Hash sequence that the characteristics of image of image is corresponding；Then it is determined that the presence of similar image features.
The image server that embodiments of the invention provide, it is possible to obtain at least one original image, its
In each described original image include low-resolution image and the figure of corresponding high-definition picture composition
As right；Characteristics of image is extracted in described original image；Characteristics of image according to described original image and
The characteristics of image of the prestored digital image in image sample data storehouse judges whether in described original image
Characteristics of image duplicate removal；If the image of original image described in the existence in described image sample data storehouse is special
Levy, then the characteristics of image duplicate removal to described original image；If not depositing in described image sample data storehouse
At the characteristics of image of described original image, then extract the characteristics of image of described original image and according to described
Original image updates described image sample data storehouse；Therefore obtain new original in image sample data storehouse
During image, it is possible to according to characteristics of image, image sample data storehouse is updated, it is to avoid by all original
The characteristics of image update all of image is to image sample data storehouse, it is possible to increase the renewal of sample database
Efficiency, reduces system resource waste.
Those skilled in the art is it can be understood that arrive, for convenience and simplicity of description, above-mentioned
The specific works process of the system, device and the unit that describe, is referred in preceding method embodiment
Corresponding process, does not repeats them here.
In several embodiments provided herein, it should be understood that disclosed system, terminal
And method, can realize by another way.Such as, terminal embodiment described above is only
It is schematic, such as, the division of described unit, it is only a kind of logic function and divides, actual real
Can have now other dividing mode, the most multiple unit or assembly can in conjunction with or can be integrated
To another system, or some features can be ignored, or does not performs.Another point, shown or discussed
Coupling each other direct-coupling or communication connection can be by some interfaces, terminal or list
The INDIRECT COUPLING of unit or communication connection, can be electrical, machinery or other form.
It addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit
In, it is also possible to it is that unit is individually physically present, it is also possible to two or more unit are integrated in
In one unit.
If described function realizes using the form of SFU software functional unit and as independent production marketing or
During use, can be stored in a computer read/write memory medium.Based on such understanding, this
The part that the most in other words prior art contributed of technical scheme of invention or this technical side
The part of case can embody with the form of software product, and this computer software product is stored in one
In storage medium, including some instructions with so that computer equipment (can be personal computer,
Server, or the network equipment etc.) perform all or part of of method described in each embodiment of the present invention
Step.And aforesaid storage medium includes: USB flash disk, portable hard drive, read only memory (English abbreviation:
ROM, English full name: Read-Only Memory), random access memory (English abbreviation:
RAM, English full name: Random Access Memory), magnetic disc or CD etc. various permissible
The medium of storage program code.
The above, the only detailed description of the invention of the present invention, but protection scope of the present invention not office
Being limited to this, any those familiar with the art, can in the technical scope that the invention discloses
Readily occur in change or replace, all should contain within protection scope of the present invention.Therefore, the present invention
Protection domain should described be as the criterion with scope of the claims.
Claims (10)
1. supersolution exists as the update method in the image sample data storehouse in system, its feature
In, including:
Obtaining at least one original image, the most each described original image includes low resolution figure
Picture and the image pair of corresponding high-definition picture composition；
Characteristics of image is extracted in described original image；
Prestored digital image in characteristics of image according to described original image and image sample data storehouse
Characteristics of image judges whether the characteristics of image duplicate removal in described original image；
If the characteristics of image of original image, the most right described in the existence in described image sample data storehouse
The characteristics of image duplicate removal of described original image；
If the characteristics of image that there is not described original image in described image sample data storehouse, then
Extract the characteristics of image of described original image and update described image pattern according to described original image
Data base.
Method the most according to claim 1, it is characterised in that described acquisition at least one
Original image, including:
Obtain low-resolution image, described low-resolution image is transformed to high-definition picture；
And/or
Obtain high-definition picture, described high-definition picture is transformed to low-resolution image.
3. according to the method described in any one of claim 1-2, it is characterised in that described former
Before beginning image extracts characteristics of image, also include:
Described high-definition picture is carried out filtration treatment, removes in described high-definition picture
Low resolution feature.
4. according to the method described in any one of claim 1-2, it is characterised in that described former
Beginning image extracts characteristics of image, also includes:
The resolution characteristics of image higher than predetermined threshold value is obtained in described high-definition picture.
Method the most according to claim 1, it is characterised in that described characteristics of image includes:
One or more in color characteristic, Gradient Features, frequency domain character and architectural feature.
Method the most according to claim 1, it is characterised in that described image sample data
Prestored digital image in storehouse stores according to affiliated image category, and described image category at least includes: landscape
Class, figure kind；
The described characteristics of image according to described original image and image sample data storehouse deposit figure
Before the characteristics of image of picture judges whether the characteristics of image duplicate removal in described original image, also include:
Judge the image category belonging to the characteristics of image of described original image.
Method the most according to claim 1, it is characterised in that described according to described original
The characteristics of image of the prestored digital image in the characteristics of image of image and image sample data storehouse judges whether
To the characteristics of image duplicate removal in described original image, including:
The described characteristics of image according to described original image and image sample data storehouse deposit figure
The characteristics of image of picture judges whether deposit in the characteristics of image of the prestored digital image in image sample data storehouse
At the similar or identical characteristics of image of the characteristics of image of described original image, if then to described original
Characteristics of image duplicate removal in image.
Method the most according to claim 7, it is characterised in that according to described original image
Characteristics of image and image sample data storehouse in the characteristics of image of prestored digital image judge image pattern
Whether the characteristics of image of the prestored digital image in data base exists the characteristics of image of described original image
Identical image feature, including:
First Random Maps of the characteristics of image obtaining described original image represents；
Obtain second the reflecting at random of characteristics of image of the prestored digital image in described image sample data storehouse
The sequencing table that representation by arrows is constituted；
According to sequencing table described in described sequencing table sorted search, inquire about whether described sequencing table exists
With state similarity that the first Random Maps the represents characteristics of image more than first threshold, if then judging
There is identical image feature.
Method the most according to claim 7, it is characterised in that according to described original image
Characteristics of image and image sample data storehouse in the characteristics of image of prestored digital image judge image pattern
Whether the characteristics of image of the prestored digital image in data base exists the characteristics of image of described original image
Similar image features, including:
Obtain the global image after the characteristics of image dimension normalization of described original image to represent；
Generate the first Hash sequence that described global image represents；
If the interval of described first Hash sequence mapping comprises in described image sample data storehouse extremely
The second Hash sequence that the characteristics of image of a prestored digital image is corresponding less；Then it is determined that the presence of similar image
Feature.
10. an image server, it is characterised in that including:
Acquiring unit, is used for obtaining at least one original image, the most each described original image
Including low-resolution image and the image pair of corresponding high-definition picture composition；
Sampling unit, for extracting characteristics of image in the original image that described acquiring unit obtains；
Processing unit, for the characteristics of image according to described original image and image sample data storehouse
In the characteristics of image of prestored digital image judge whether the characteristics of image duplicate removal in described original image；
If the characteristics of image of original image described in the existence in described image sample data storehouse, then to described former
The characteristics of image duplicate removal of beginning image；If there is not described original graph in described image sample data storehouse
The characteristics of image of picture, then extract the characteristics of image according to described original image more of described original image
New described image sample data storehouse.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610232700.2A CN105930391B (en) | 2016-04-14 | 2016-04-14 | Update method and image server of the supersolution as image sample data library in system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610232700.2A CN105930391B (en) | 2016-04-14 | 2016-04-14 | Update method and image server of the supersolution as image sample data library in system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105930391A true CN105930391A (en) | 2016-09-07 |
| CN105930391B CN105930391B (en) | 2019-06-07 |
Family
ID=56838082
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610232700.2A Active CN105930391B (en) | 2016-04-14 | 2016-04-14 | Update method and image server of the supersolution as image sample data library in system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105930391B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107798389A (en) * | 2017-11-06 | 2018-03-13 | 国网重庆市电力公司电力科学研究院 | A kind of image data set construction method, system and computer readable storage devices |
| CN107958272A (en) * | 2017-12-12 | 2018-04-24 | 北京旷视科技有限公司 | Image data set update method, device, system and computer-readable storage medium |
| CN109062988A (en) * | 2018-06-29 | 2018-12-21 | 百度在线网络技术（北京）有限公司 | The update method and device of street view image |
| CN109545113A (en) * | 2018-09-30 | 2019-03-29 | 合肥鑫晟光电科技有限公司 | A kind of method for processing abnormal data and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103914546A (en) * | 2014-04-09 | 2014-07-09 | 百度在线网络技术（北京）有限公司 | Data updating method and device thereof |
| CN104200499A (en) * | 2014-09-24 | 2014-12-10 | 深圳市英威诺科技有限公司 | Technical method for intelligently removing reduplication of information images |
| US20150348273A1 (en) * | 2009-11-11 | 2015-12-03 | Disney Enterprises, Inc. | Depth modification for display applications |
| CN105243373A (en) * | 2015-10-27 | 2016-01-13 | 北京奇虎科技有限公司 | Method for filtering facial images to prevent repeated snapshot, server, intelligent monitoring device and system |
-
2016
- 2016-04-14 CN CN201610232700.2A patent/CN105930391B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150348273A1 (en) * | 2009-11-11 | 2015-12-03 | Disney Enterprises, Inc. | Depth modification for display applications |
| CN103914546A (en) * | 2014-04-09 | 2014-07-09 | 百度在线网络技术（北京）有限公司 | Data updating method and device thereof |
| CN104200499A (en) * | 2014-09-24 | 2014-12-10 | 深圳市英威诺科技有限公司 | Technical method for intelligently removing reduplication of information images |
| CN105243373A (en) * | 2015-10-27 | 2016-01-13 | 北京奇虎科技有限公司 | Method for filtering facial images to prevent repeated snapshot, server, intelligent monitoring device and system |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107798389A (en) * | 2017-11-06 | 2018-03-13 | 国网重庆市电力公司电力科学研究院 | A kind of image data set construction method, system and computer readable storage devices |
| CN107798389B (en) * | 2017-11-06 | 2021-05-14 | 国网重庆市电力公司电力科学研究院 | Image data set construction method and system and computer readable storage device |
| CN107958272A (en) * | 2017-12-12 | 2018-04-24 | 北京旷视科技有限公司 | Image data set update method, device, system and computer-readable storage medium |
| CN109062988A (en) * | 2018-06-29 | 2018-12-21 | 百度在线网络技术（北京）有限公司 | The update method and device of street view image |
| CN109062988B (en) * | 2018-06-29 | 2019-10-25 | 百度在线网络技术（北京）有限公司 | The update method and device of street view image |
| CN109545113A (en) * | 2018-09-30 | 2019-03-29 | 合肥鑫晟光电科技有限公司 | A kind of method for processing abnormal data and device |
| US11200818B2 (en) | 2018-09-30 | 2021-12-14 | Hefei Xinsheng Optoelectronics Technology Co., Ltd. | Method and device for processing abnormal data |
| CN109545113B (en) * | 2018-09-30 | 2022-09-06 | 合肥鑫晟光电科技有限公司 | Abnormal data processing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105930391B (en) | 2019-06-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Iglovikov et al. | Ternausnet: U-net with vgg11 encoder pre-trained on imagenet for image segmentation | |
| Dai et al. | Curriculum model adaptation with synthetic and real data for semantic foggy scene understanding | |
| JP6395158B2 (en) | How to semantically label acquired images of a scene | |
| CN108921799B (en) | Remote sensing image thin cloud removing method based on multi-scale collaborative learning convolutional neural network | |
| CN105930391A (en) | Update method and image server of image sample database of super-resolution image system | |
| CN110570433B (en) | Image semantic segmentation model construction method and device based on generation countermeasure network | |
| DE202015009255U1 (en) | Automatic image organization | |
| CN110209632A (en) | A kind of electronics folder with case production, turn shelves system | |
| CN102831239B (en) | A kind of method and system building image data base | |
| CN102880726A (en) | Image filter method and image filter system | |
| CN104486585B (en) | A kind of city magnanimity monitor video management method and system based on GIS | |
| DE102022105471A1 (en) | Generating modified digital images using deep visual guided patch matching models for image restoration | |
| CN111951368B (en) | Deep learning method for point cloud, voxel and multi-view fusion | |
| Chen et al. | A new process for the segmentation of high resolution remote sensing imagery | |
| CN111967313A (en) | Unmanned aerial vehicle image annotation method assisted by deep learning target detection algorithm | |
| CN110059539A (en) | A kind of natural scene text position detection method based on image segmentation | |
| CN102855245A (en) | Image similarity determining method and image similarity determining equipment | |
| CN116597270A (en) | Road damage target detection method based on attention mechanism integrated learning network | |
| CN109543525B (en) | Table extraction method for general table image | |
| CN113610024B (en) | Multi-strategy deep learning remote sensing image small target detection method | |
| CN104992407A (en) | Image super-resolution method | |
| Karim et al. | A coverless image steganography based on robust image wavelet hashing | |
| CN111951171A (en) | HDR image generation method and device, readable storage medium and terminal equipment | |
| CN103927533B (en) | The intelligent processing method of graph text information in a kind of scanned document for earlier patents | |
| CN115457448B (en) | Intelligent extraction system for video key frames |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |