The content of the invention
An object of the present invention is to provide a kind of screen of intelligent device extracting method based on positioning image, this method energy
Enough screen of intelligent device images automatically, easily extracted within screen of intelligent device interior frame scope.Further can be with
Extract the screen of intelligent device image within screen of intelligent device outer rim scope.
According to above-mentioned purpose, the present invention proposes a kind of screen of intelligent device extracting method based on positioning image, and it is wrapped
Include step:
100:By for determine screen of intelligent device position, size, direction positioning image include in smart machine screen
On curtain;
200:Screen of intelligent device image of the collection containing positioning image simultaneously contains this smart machine screen of positioning image
Curtain image transmitting is to extracting service end;
300:Positioning image is identified the extraction service end, to determine positioning image in screen of intelligent device image
In original coordinates;
400:The extraction service end is corrected based on the original coordinates to screen of intelligent device image;
500:The extraction service end outwards finds and extracted the intelligence shown in screen of intelligent device image from positioning image
Can device screen interior frame;
600:Screen of intelligent device image within the extraction service end extraction screen of intelligent device interior frame scope.
Screen of intelligent device extracting method of the present invention based on positioning image, it passes through in screen of intelligent device
Display positioning image, and identify the positioning image in the screen of intelligent device image of collection and determine its original coordinates, due to
Original coordinates based on the positioning image can determine position, size, the direction of corresponding screen of intelligent device, so as to base
Screen of intelligent device image is corrected in the original coordinates of the positioning image, and is outwards found from positioning image and extracts smart machine
Screen interior frame, and then extract the screen of intelligent device image within screen of intelligent device interior frame scope.Wherein:
Positioning image in step 100 can preset size and be shown in the fixed position in screen of intelligent device, therefore
It is determined for position, size, the direction of screen of intelligent device.Positioning image generally has with screen of intelligent device image not
Same readily identified feature, so as to convenient identified.
The IMAQ transmission equipment with shooting image function and transmission image function, example can be passed through in step 200
Such as mobile phone, gather and transmit screen of intelligent device image.
Positioning image can be identified in step 300 by way of scanning, determines its original coordinates.If it can not find positioning
Image, then user can be prompted to can not find positioning image, and prompt user's adjustment for carrying out camera site and angle.
In some embodiments, according to original coordinates and the mapping relations pair between coordinate can it is expected in step 400
Screen of intelligent device image is corrected.The coordinate system of the expectation coordinate is generally vertical in screen of intelligent device IMAQ
The two-dimensional coordinate system of optical axis.
Because positioning image is located at the inside of screen of intelligent device interior frame in step 500 and step 600, therefore can be from
Positioning image outwards finds screen of intelligent device interior frame and extracts the smart machine within screen of intelligent device interior frame scope
Screen picture.
To sum up, the intelligence that the inventive method automatically, can be extracted easily within screen of intelligent device interior frame scope is set
Standby screen picture.
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, step 400 is also wrapped
Include:After screen of intelligent device image is corrected, it is determined that or the acquisition calibration coordinate therein for positioning image.
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, the positioning figure
Seem the positioning image of in-time generatin or the positioning image being stored in database.
In such scheme, positioning image can use the existing positioning image being stored in database, can also basis
Specific service logic in-time generatin.The positioning image being stored in database can use one-dimension code, Quick Response Code etc..
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, the positioning figure
As being image in 2 D code.
In such scheme, image in 2 D code belongs to two-dimensional localization image.If subsequently need to screen of intelligent device image
If carrying out complete extraction, then positioning the usual prioritizing selection of image has the positioning image of two dimensions, i.e. two-dimensional localization image.
Usually require to consider four points of at least quadrangle using two-dimensional localization image, can be to device screen image by this four points
Carry out complete projection correction.For Quick Response Code, because Quick Response Code has three anchor points and a hiding anchor point letter
Breath, therefore complete screen correction can be carried out by these anchor point information.
Certainly, positioning image can also select one-dimension code.Some below attention is usually required using one-dimension code positioning image:
One-dimension code is only capable of supporting the left and right correction of image, without the front and rear i.e. projection correction of enough Informational support images;
One-dimension code can not be only capable of finding screen by didactic mode by completely deriving the scope of logical calculated screen
The position of curtain, such as connected region, flooding of rivers, approximate parallel searching.
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, the extraction clothes
Business end is client or high in the clouds, and being run in the client or high in the clouds has extraction procedure.
In such scheme, the extraction service end performs correlation step by the extraction procedure.Can be according to specific
Business demand selects client or high in the clouds as extraction service end.
When the extraction service end is high in the clouds, can be handed over by high in the clouds and the smart machine of screen picture to be extracted
Mutually, so as to which the generation of positioning image be more finely controlled, these fine controls are in order to which follow-up screen picture extracts more
Simply, it is convenient, avoid using complicated back-end processing logic.Such as Quick Response Code is selected to be taken as positioning image by high in the clouds
Smart machine of the provider's passback parameter of being engaged in screen picture to be extracted, allows Quick Response Code positioning image to generate in the centre of screen,
I.e. high accounting 50%, wide accounting 50%, follow-up cloud service provider can then be carried using these parameters to screen picture
Take.
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, the correction bag
Include rotation correction and projection correction.
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, the positioning figure
As containing identity information, so that the extraction service end is sentenced when receiving the screen of intelligent device image containing positioning image
Break the screen of intelligent device image whether be the smart machine specified screen of intelligent device image.
In such scheme, the smart machine of screen picture to be extracted is audited due to being usually required that in specific business demand
Identity, therefore identity information is added in image is positioned, in order to judge whether the screen of intelligent device image is the intelligence specified
The screen of intelligent device image of energy equipment.
For Quick Response Code positions image, it usually needs the coding of information is supported, if had on screen of intelligent device image
Multiple Quick Response Codes position image, by positioning the decoding of image to Quick Response Code, can determine which belongs to screen map to be extracted
The smart machine of picture, or be not.
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, in step 500
In, size relation or positioning image edge of the extraction service end based on default positioning image and screen of intelligent device
Distance apart from screen of intelligent device interior frame reversely derives the position of screen of intelligent device interior frame, with outside from positioning image
Find and extract the screen of intelligent device interior frame shown in screen of intelligent device image.
In such scheme, positioning image can be quadrangle, such as Quick Response Code.By taking Quick Response Code as an example, usual Quick Response Code tool
There are three anchor points and a hiding anchor point information.These anchor point information can be used for being based on default Quick Response Code and intelligence
The position relationship of device screen interior frame, such as proportionate relationship or edge distance, are shown, thus can be after calibration
Intelligence in the screen of intelligent device image after correcting reversely is derived in screen of intelligent device image according to the size and location of Quick Response Code
The size and location of energy device screen interior frame, and then extract the screen of intelligent device within screen of intelligent device interior frame scope
Image.
Further, in the screen of intelligent device extracting method of the present invention based on positioning image, in the step
In 500, sentenced by regional connectivity and shown otherwise with outwards finding and extracting in screen of intelligent device image from positioning image
Screen of intelligent device interior frame.
In such scheme, calculated if described reversely derive can not be carried out according to positioning image, pass through area
The screen of intelligent device outwards found and extracted from positioning image otherwise and shown in screen of intelligent device image is sentenced in domain connection
Interior frame.It is specific to find mode, can be the mode of " flooding of rivers " or certain side based on positioning image parallel to
The property of the interior frame is found.
Further, the screen of intelligent device extracting method of the present invention based on positioning image also includes step:Institute
Extraction service end is stated to continue outwards to find and extract the intelligence shown in screen of intelligent device image from screen of intelligent device interior frame
Can device screen outer rim;And the smart machine screen within the extraction service end extraction screen of intelligent device outer rim scope
Curtain image.
It can choose whether to carry out to set intelligence in such scheme according to specific business demand or extraction fineness demand
The searching and extraction of the screen of intelligent device outer rim shown in standby screen picture, and to screen of intelligent device outer rim scope
Within screen of intelligent device image extraction.
The fineness generally includes coarseness and the class of fine granularity two.The coarseness is corresponded in screen of intelligent device image
The searching and extraction of the screen of intelligent device interior frame of display, and the intelligence within screen of intelligent device interior frame scope is set
The extraction of standby screen picture.The fine granularity corresponds to seeking for the screen of intelligent device outer rim that is shown in screen of intelligent device image
Look for and extract, and the extraction to the screen of intelligent device image within screen of intelligent device outer rim scope.
Because in a variety of smart machines, the outer rim form of expression of screen is all different, and without the rule of fixation
Rule can be followed, therefore usually require to be extracted by way of machine learning, and the mode of extraction is typically excessive in interior frame
One proportional positions goes to extract.
The business less sensitive to the outer screen between interior frame and outer rim, can omit outer screen, directly carry out thick
The screen picture extraction of granularity.
It is a further object of the present invention to provide a kind of screen of intelligent device extraction system based on positioning image, the system energy
Enough screen of intelligent device images automatically, easily extracted within screen of intelligent device interior frame scope.Further can be with
Extract the screen of intelligent device image within screen of intelligent device outer rim scope.
Based on foregoing invention purpose, present invention also offers a kind of screen of intelligent device based on positioning image to extract system
System, it performs any of the above-described screen of intelligent device extracting method based on positioning image.
Screen of intelligent device extraction system of the present invention based on positioning image, it is any of the above-described based on fixed due to performing
The screen of intelligent device extracting method of bit image, therefore equally can automatically, easily extract screen of intelligent device interior frame model
Screen of intelligent device image within enclosing.The intelligence within screen of intelligent device outer rim scope can also be further extracted to set
Standby screen picture.Have a detailed description, will not be repeated here in concrete principle the inventive method.
The system generally includes the smart machine of screen picture to be collected, IMAQ transmission equipment and extraction service
End.The smart machine of the screen picture to be collected performs step 100, and described image collection transmission equipment performs step 200, institute
State extraction service end and perform step 300- steps 600.
It is yet another object of the invention to provide a kind of purposes of screen of intelligent device image, the purposes can effectively utilize this
The screen of intelligent device image that inventive method extracts.
Based on foregoing invention purpose, present invention also offers a kind of purposes of screen of intelligent device image, will use above-mentioned
The screen of intelligent device image that any extracting method based on positioning image extracts is used for the damaged identification of screen or screen shows
Show the detection of content.
The purposes of screen of intelligent device image of the present invention, it is by using any of the above-described extraction based on positioning image
The screen of intelligent device image that method extracts is used for the damaged identification of screen or the detection of screen display content, so as to effectively sharp
The screen of intelligent device image extracted with the inventive method.
Screen of intelligent device extracting method of the present invention based on positioning image, it has advantages below and beneficial to effect
Fruit:
(1) the screen of intelligent device image within screen of intelligent device interior frame scope can automatically, be easily extracted, also
The screen of intelligent device image within screen of intelligent device outer rim scope can be extracted.
(2) greatly reduce the complexity of On-Screen Identification, while original screen letter can be efficiently conveyed for subsequent applications
Breath, the processing for subsequent applications.
(3) precision of screen picture extraction is greatly improved, interference of the unnecessary feature to model training is reduced, reduces
While process content information amount, the operational paradigm of whole automatic flow is improved.
(4) by first finding positioning image, device screen is then found according to positioning image, avoids different ambient fields
Scape increases the searching difficulty of device screen.
(5) by the correction based on positioning image, the precision that screen picture extracts further is improved so that follow-up application
Operation can be carried out efficiently.
(6) can be provided on the basis of screen picture extraction some it is corresponding prompt, such as screen shot it is imperfect, do not have
Have and find the information that positioning image etc. contributes to user to adjust.
(7) by using positioning image mode come from inside to outside searching device screen approximate range, then lead to again
The method for crossing machine learning, accurate screen position is found, influenceed so as to weed out the content outside screen, be follow-up screen
The broken screen identification of content detection, screen provides the input feature vector of training and prediction.
(8) without loss of generality, all possess screen, can connect the smart machine of internet in principle, include but is not limited to
The screen picture extracting method of tablet personal computer, notebook computer, intelligent watch etc., all it is applicable method of the present invention.
Screen of intelligent device extraction system of the present invention based on positioning image, it equally has above-mentioned advantage and had
Beneficial effect.
The purposes of screen of intelligent device image of the present invention, it equally has above-mentioned advantage and beneficial effect.
Embodiment
Come below in conjunction with Figure of description and specific embodiment to the intelligence of the present invention based on positioning image
Device screen extracting method, system and purposes are further described in detail, but the detailed description is not formed to this hair
Bright limitation.
Fig. 1 shows the screen of intelligent device extracting method of the present invention based on positioning image in a kind of embodiment
Under flow.As shown in figure 1, step should be included based on the screen of intelligent device extracting method of positioning image:
100:By for determine screen of intelligent device position, size, direction positioning image include in smart machine screen
On curtain;
200:Screen of intelligent device image of the collection containing positioning image simultaneously contains this smart machine screen of positioning image
Curtain image transmitting is to extracting service end;
300:Positioning image is identified extraction service end, to determine positioning image in screen of intelligent device image
Original coordinates;
400:Extraction service end is corrected based on original coordinates to screen of intelligent device image;
500:Extraction service end is outwards found and extracted the intelligence shown in screen of intelligent device image from positioning image and sets
Standby screen interior frame;
600:Extract the screen of intelligent device image within service end extraction screen of intelligent device interior frame scope.
In some embodiments, step 400 also includes:After screen of intelligent device image is corrected, it is determined that or obtaining
The calibration coordinate of positioning image therein.
In some embodiments, it is the positioning image of in-time generatin or the positioning figure being stored in database to position image
Picture.
In some embodiments, it is image in 2 D code to position image.
In some embodiments, extraction service end is client or high in the clouds, and being run in client or high in the clouds has extraction journey
Sequence.
In some embodiments, correction includes rotation correction and projection correction.
In some embodiments, positioning image contains identity information, so that extraction service end is being received containing positioning
During the screen of intelligent device image of image, judge the screen of intelligent device image whether be the smart machine specified smart machine
Screen picture.
In some embodiments, in step 500, extract service end and be based on default positioning image and smart machine screen
The size relation or positioning image edge of curtain reversely derive smart machine screen apart from the distance of screen of intelligent device interior frame
The position of curtain interior frame, outwards to be found from positioning image and extract the screen of intelligent device shown in screen of intelligent device image
Interior frame.
In some embodiments, in step 500, sentenced otherwise by regional connectivity outwards to be sought from positioning image
Look for and extract the screen of intelligent device interior frame shown in screen of intelligent device image.
In some embodiments, in addition to step:Extraction service end continues outwards to seek from screen of intelligent device interior frame
Look for and extract the screen of intelligent device outer rim shown in screen of intelligent device image;And extraction service end extraction smart machine
Screen of intelligent device image within screen outer rim scope.
A kind of structure of the screen of intelligent device extraction system based on positioning image of the present invention under embodiment includes
Mobile phone A to be measured as the smart machine of screen picture to be collected, shooting mobile phone B and work as IMAQ transmission equipment
Being run to extract the high in the clouds C of service end, wherein on the C of high in the clouds has extraction procedure.Image in 2 D code D as positioning image is stored in
In database, image in 2 D code D is the coding of mobile phone A identity information to be measured.
Fig. 2 shows a kind of workflow of the screen of intelligent device extraction system based on positioning image as embodiment
Journey.As shown in Fig. 2 the workflow of the screen of intelligent device extraction system based on positioning image of the embodiment includes:
Step 100:Mobile phone A to be measured reads position, size, the direction for determining mobile phone A screen to be measured from database
Image in 2 D code D, and according to the size relation of the default image in 2 D code D from high in the clouds C and mobile phone A screen to be measured
Or image in 2 D code D edge is included it with default size and location apart from the distance instruction of mobile phone A screen interior frame to be measured
On the screen of mobile phone A to be measured.
Step 200:Shooting mobile phone B gathers the mobile phone A screen picture to be measured containing image in 2 D code D and this is contained into two dimension
Code image D mobile phone A screen picture to be measured is transmitted to high in the clouds C.
Step 300:Image in 2 D code D is identified by the extraction procedure run thereon by high in the clouds C, first decoding two dimension
Code image D simultaneously compared with the identity information of mobile phone A to be measured, with judge reception screen picture whether be mobile phone A to be measured screen
Image, it is then determined that original coordinates of the image in 2 D code D in mobile phone A screen picture to be measured.
In the step, if can not find positioning image, user is prompted to can not find positioning image, and prompt user to be clapped
Act as regent and put adjustment with angle.
Step 400:High in the clouds C is entered by the extraction procedure run thereon based on original coordinates to mobile phone A screen picture to be measured
Row correction, the correction include rotation correction and projection correction.After mobile phone A screen picture to be measured is corrected, record therein
Image in 2 D code D calibration coordinate.
, can be according to the mapping relations between original coordinates and expectation coordinate to mobile phone A screen picture to be measured in the step
It is corrected.The coordinate system of the expectation coordinate is the two-dimensional coordinate system perpendicular to shooting mobile phone B IMAQ optical axises.
Step 500:High in the clouds C is outwards found from image in 2 D code D by the extraction procedure run thereon and extracts hand to be measured
The mobile phone A screen interior frame to be measured shown in machine A screen pictures.
In the step, size relations or two of the high in the clouds C based on default image in 2 D code D Yu mobile phone A screen to be measured
Dimension code image D edge reversely derives the position of mobile phone A screen interior frame to be measured apart from the distance of mobile phone A screen interior frame to be measured,
So that the mobile phone A screen interior frame to be measured shown in mobile phone A screen picture to be measured is outwards found and extracted from image in 2 D code D.Example
Such as, it is assumed that instruction requires the position of Quick Response Code generation at the center of mobile phone A screen to be measured, and with apart from mobile phone A screen to be measured
If there is preset ratio relation in upside and left side, then mobile phone A to be measured is directly found and extracted according to this proportionate relationship
Screen interior frame.
Step 600:Coarseness screen picture extracts.High in the clouds C extracts mobile phone A screen to be measured by the extraction procedure run thereon
Mobile phone A screen picture to be measured within curtain interior frame scope.
Step 700:Fine granularity screen picture extracts.High in the clouds C continues outwards to find and carry from mobile phone A screen interior frame to be measured
Take the mobile phone A screen outer rim to be measured shown in mobile phone A screen picture to be measured;High in the clouds C extracts mobile phone A screen outer rim model to be measured
Mobile phone A screen picture to be measured within enclosing.
In the step, extracted by way of machine learning in an excessive proportional positions of interior frame.
The hand to be measured extracted based on the screen of intelligent device extraction system and method that position image of above-mentioned embodiment
Machine A screen pictures can be used for the detection of the screen display content of the damaged identification of the screen of mobile phone A to be measured or mobile phone A to be measured.Tool
Body step includes:
Step 800:Mobile phone A screen picture to be measured within the mobile phone A screen outer rim scope to be measured that step 700 is extracted
Characteristic processing is carried out, such as is sized (resize), remodeling (reshape), marginalisation, the training for broken screen identification model
And prediction.
Step 900:Predicting with the parameter of broken screen identification model more for the broken screen of mobile phone A to be measured is carried out using broken screen identification model
New training.
In addition, it should also be noted that, institute in the combination of each technical characteristic and unlimited this case claim in this case
Combination described in the combination or specific embodiment of record, all technical characteristics described in this case can be to appoint
Where formula carries out independent assortment or combination, unless producing contradiction between each other.
It should be noted that listed above is only specific embodiment of the invention, it is clear that real the invention is not restricted to more than
Example is applied, the similar change for having many therewith.If those skilled in the art directly exported from present disclosure or
All deformations associated, all should belong to protection scope of the present invention.