CN108305223A - Image background blurring processing method and device - Google Patents
Image background blurring processing method and device Download PDFInfo
- Publication number
- CN108305223A CN108305223A CN201810019406.2A CN201810019406A CN108305223A CN 108305223 A CN108305223 A CN 108305223A CN 201810019406 A CN201810019406 A CN 201810019406A CN 108305223 A CN108305223 A CN 108305223A
- Authority
- CN
- China
- Prior art keywords
- image
- body profile
- camera
- background
- auxiliary
- 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
- 238000003672 processing method Methods 0.000 title claims abstract description 17
- 238000012545 processing Methods 0.000 claims abstract description 49
- 238000000034 method Methods 0.000 claims abstract description 25
- 230000000694 effects Effects 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 11
- 238000010801 machine learning Methods 0.000 claims abstract description 10
- 230000001360 synchronised effect Effects 0.000 claims description 38
- 230000009977 dual effect Effects 0.000 claims description 14
- 230000002708 enhancing effect Effects 0.000 claims description 13
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 230000008859 change Effects 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 abstract description 10
- 238000013473 artificial intelligence Methods 0.000 description 9
- 230000007704 transition Effects 0.000 description 5
- 238000010168 coupling process Methods 0.000 description 4
- 238000005859 coupling reaction Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000010365 information processing Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 210000004556 brain Anatomy 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 241001347978 Major minor Species 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000002490 cerebral effect Effects 0.000 description 1
- 238000000205 computational method Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000003708 edge detection Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 210000000653 nervous system Anatomy 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/50—Depth or shape recovery
-
- 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
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses an image background blurring processing method and device. Wherein, the method comprises the following steps: gather the image, and then according to the object recognition model, discern the main part profile of the object in this image to carry out the blurring processing to the background of main part profile in the image, wherein, the object recognition model is for using multiunit data to obtain through machine learning training, and every group data in the multiunit data all includes: an image and a subject profile of an object identified from the image. The invention solves the technical problem of poor background blurring effect in the shot image in the related technology.
Description
Technical field
The present invention relates to technical field of image processing, and processing method and dress are blurred in particular to a kind of image background
It sets.
Background technology
With being widely used for smart mobile phone, camera has replaced card camera to become big as the important module in mobile phone
Many most common portable capture apparatus.And mobile phone camera, due to the limitation of volume itself, application limitation is still more apparent,
Middle aperture is small, and focal length is short, so it is difficult to taking background blurring Deep Canvas.It is trapped in such limitation, in the related technology with soft
Part simulates the background blurring Deep Canvas that professional slr camera large aperture camera lens is shot, but there is also certain drawbacks,
For example, main body discrimination is not high, main body and background transition are not natural enough.
For above-mentioned problem, currently no effective solution has been proposed.
Invention content
An embodiment of the present invention provides a kind of image backgrounds to blur processing method and processing device, at least to solve in the related technology
Background blurring ineffective technical problem in the image of shooting.
One side according to the ... of the embodiment of the present invention provides a kind of image background virtualization processing method, including：It collects
Image；According to object identification model, the body profile of the object in described image is identified, wherein the object identification model
Multi-group data is used to show that every group of data in the multi-group data include by machine learning training：Image and from
The body profile of the object identified in described image；Virtualization processing is carried out to the background of body profile described in described image.
Optionally, the body profile for according to the object identification model, identifying the object in described image it
Before, further include：The sampled images of predetermined quantity series are obtained, and the main body wheel of object identified from the sampled images
It is wide；To the sampled images of acquisition, and the body profile of object that is identified from the sampled images, using artificial god
It is trained through network algorithm, obtains the object identification model for the predetermined quantity image series.
Optionally, carrying out virtualization processing to the background of body profile described in described image includes：Obtain described image
Depth map；According to the depth map of acquisition and the body profile identified, to the body profile in described image
Background carry out virtualization processing.
Optionally, the depth map for obtaining described image includes：Master image is intercepted and captured by main camera and passes through auxiliary camera
Intercept and capture auxiliary image；Time synchronization is carried out to the master image and the auxiliary image, obtains the main synchronous images of the main camera
With the auxiliary synchronous images of the auxiliary camera；According to the main synchronous images and the auxiliary synchronous images, described image is obtained
Depth map.
Optionally, according to the main synchronous images and the auxiliary synchronous images, the depth map for obtaining described image includes：It obtains
The image difference of the main synchronous images and the auxiliary synchronous images is taken, is obtained in the main camera and the auxiliary camera two
The distance between heart position, and obtain the picture difference of the main synchronous images and the auxiliary synchronous images；According to acquisition
Described image difference, the distance and the picture difference, obtain the depth map of described image.
Optionally, before carrying out virtualization processing to the background of body profile described in described image, further include：To described
The body profile in image carries out effect enhancing processing.
Optionally, before carrying out effect enhancing processing to the body profile in described image, further include：By adopting
The auxiliary camera collected in the dual camera of described image focuses to the corresponding main body of the body profile.
Optionally, after carrying out virtualization processing to the background of body profile described in described image, further include：By adopting
Collect the main camera in the dual camera of described image and carries out preview and output to carrying out virtualization treated image.
Another aspect according to the ... of the embodiment of the present invention additionally provides a kind of image background virtualization processing unit, including：Acquisition
Module, for collecting image；Identification module, for according to object identification model, identifying the master of the object in described image
Body profile, wherein the object identification model is trained by machine learning using multi-group data and obtained, the multi-group data
In every group of data include：The body profile of image and the object identified from described image；Blurring module, for institute
The background for stating body profile described in image carries out virtualization processing.
Another aspect according to the ... of the embodiment of the present invention, additionally provides a kind of storage medium, and the storage medium includes storage
Program, wherein equipment where controlling the storage medium when described program is run executes the figure described in above-mentioned any one
As background blurring processing method.
Another aspect according to the ... of the embodiment of the present invention additionally provides a kind of processor, and the processor is used to run program,
Wherein, image background when described program is run described in the above-mentioned any one of perform claim blurs processing method.
In embodiments of the present invention, by the way of machine learning, by the object identification model trained, knowledge has been reached
Do not go out the body profile of the objects in images of acquisition, and carry out the purpose of virtualization processing to the background of the body profile, wherein object
Body identification model show that every group of data in multi-group data include using multi-group data by machine learning training：Figure
The body profile of picture and the object identified from image.Main body and background in image are improved to which the embodiment of the present invention realizes
Discrimination so that the more natural technique effect of transition both in image, and then solve the image shot in the related technology
In background blurring ineffective technical problem.
Description of the drawings
Attached drawing described herein is used to provide further understanding of the present invention, and is constituted part of this application, this hair
Bright illustrative embodiments and their description are not constituted improper limitations of the present invention for explaining the present invention.In the accompanying drawings：
Fig. 1 is the flow chart of image background virtualization processing method according to the ... of the embodiment of the present invention；
Fig. 2 is to identify enhancing dual camera virtualization effect by AI image objects according to a kind of of the preferred embodiment of the invention
The flow chart of fruit method；
Fig. 3 is the structural schematic diagram of image background virtualization processing unit according to the ... of the embodiment of the present invention.
Specific implementation mode
In order to enable those skilled in the art to better understand the solution of the present invention, below in conjunction in the embodiment of the present invention
Attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, instead of all the embodiments.Based on the embodiments of the present invention, ordinary skill people
The every other embodiment that member is obtained without making creative work should all belong to the model that the present invention protects
It encloses.
It should be noted that term " first " in description and claims of this specification and above-mentioned attached drawing, "
Two " etc. be for distinguishing similar object, without being used to describe specific sequence or precedence.It should be appreciated that using in this way
Data can be interchanged in the appropriate case, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " comprising " and " having " and their any deformation, it is intended that cover
It includes to be not necessarily limited to for example, containing the process of series of steps or unit, method, system, product or equipment to cover non-exclusive
Those of clearly list step or unit, but may include not listing clearly or for these processes, method, product
Or the other steps or unit that equipment is intrinsic.
According to embodiments of the present invention, a kind of embodiment of the method for image background virtualization processing is provided, it should be noted that
Step shown in the flowchart of the accompanying drawings can execute in the computer system of such as a group of computer-executable instructions, and
It, in some cases, can be to execute institute different from sequence herein and although logical order is shown in flow charts
The step of showing or describing.
Fig. 1 is the flow chart of image background virtualization processing method according to the ... of the embodiment of the present invention, as shown in Figure 1, this method
Include the following steps：
Step S102, collects image；
Step S104 identifies the body profile of the object in image, wherein object identification according to object identification model
Model show that every group of data in multi-group data include using multi-group data by machine learning training：Image and from
The body profile of the object identified in image；
Step S106 carries out virtualization processing to the background of body profile in image.
Through the above steps, it may be implemented in the image background virtualization processing method of the embodiment of the present invention, using machine
The mode of study has reached the body profile for the objects in images for identifying acquisition by the object identification model trained, and
The purpose of virtualization processing is carried out to the background of the body profile, wherein object identification model is to pass through machine using multi-group data
What learning training obtained, every group of data in multi-group data include：The main body wheel of image and the object identified from image
It is wide.By above-mentioned intelligent body profile identifying processing, be effectively improved in image the recognition accuracy of main body and background with
The more natural technique effect of transition for making the two in image, relative to the dual camera merely with calibration position, using software
Simulate depth map, depth map recycled to calculate main body and background, so will be background blurring for, efficiently solve correlation
Background blurring ineffective technical problem in the image shot in technology.
For the training of object identification model, the computational methods of a variety of machine learning may be used, it is preferred that may be used
Artificial neural network algorithm is trained above-mentioned object identification model.Therefore according to object identification model, identify in image
Object body profile before, can also include：The sampled images of predetermined quantity series are obtained, and are known from sampled images
The body profile for the object not gone out；To the sampled images of acquisition, and the body profile of object that is identified from sampled images,
It is trained using artificial neural network algorithm, obtains the object identification model for predetermined quantity image series.Wherein, above-mentioned
Training data can be obtained by experiment, can also be that the equipment for widely applying image background virtualization processing method uses
Constantly acquire what accumulation reported in the process, it is equal to obtain a large amount of data by being tracked to equipment currently in use
It can be used for training.Optionally, the large number of equipment of above application image background virtualization processing method can be by pre-setting
Communication module, during real-time collected data are uploaded onto the server, train and use for machine.
It should be noted that artificial neural network can be abstracted human brain neuroid from information processing angle,
Naive model is established, different networks is formed by different connection types.Artificial neural network is a kind of non-programming, adaptation
The information processing of property, brain style, essence is to obtain a kind of parallel distributed by the transformation and dynamic behavior of network,
And in different degrees of and level upper mold apery cerebral nervous system the information processing function, there is self study, connection entropy and high speed
The advantages of finding optimization solution.Therefore artificial neural network algorithm is used to train object identification model, model can be made to have better
The ability of self-optimization and study extension, and then keep the body profile of the object identified from image more precisely and clear.
The body profile gone out according to object identification Model Identification, it is preferred that the background of body profile in image is carried out empty
Change is handled：Obtain the depth map of image；According to the depth map of acquisition and the body profile identified, in image
The background of body profile carries out virtualization processing.Wherein, depth refers to some point of the scene in image to where camera center
The distance of plane, depth map, you can reflect the depth information of each point in scene.It is combined by the physical contours identified deep
The depth information for spending each point in figure, can accurately reflect the distance relation of main body and background different location in image, into
And the virtualization that respective degrees are carried out by the distance relation is handled, to keep the transition of main body and background more natural.
In conjunction with dual camera setting in the related technology, it is preferred that the depth map for obtaining image may include：It is taken the photograph by master
Master image is intercepted and captured as head and auxiliary image is intercepted and captured by auxiliary camera；Time synchronization is carried out to master image and auxiliary image, master is obtained and takes the photograph
As the main synchronous images of head and the auxiliary synchronous images of auxiliary camera；According to main synchronous images and auxiliary synchronous images, image is obtained
Depth map.When because of image taking, light, background and main body can generate subtle variation at any time, therefore main camera is cut
It obtains master image and auxiliary image is intercepted and captured by auxiliary camera and carry out time synchronization, the maximum restored to photographed scene can be obtained
The effect of validity and consistency.Wherein, above-mentioned main camera can be adjacent with the position of auxiliary camera, between may be set to be
At a certain distance, can be on a horizontal line, can also both line and horizontal line it is angled.
For the acquisition of above-mentioned depth map, it is preferred that according to main synchronous images and auxiliary synchronous images, obtain the depth of image
Figure may include：The image difference for obtaining main synchronous images and auxiliary synchronous images obtains two center of main camera and auxiliary camera
The distance between position, and obtain the picture difference of main synchronous images and auxiliary synchronous images；According to the image difference of acquisition, away from
From and picture difference, obtain the depth map of image.Because the position of main camera and secondary camera is distinguished, and then make to collect
Image in difference position relationship and distance relation have any different, by the image difference of above-mentioned acquisition, picture difference and
The distance between two center of major-minor camera, can be with the position relationship of each point in COMPREHENSIVE CALCULATING photographed scene, and then obtains
The accurate depth map for getting the image of output, to keep the virtualization effect of background in image more natural.
To make the main body identified in image and background more degree of having any different, it is preferred that the body profile in image
Before background carries out virtualization processing, can also include：Effect enhancing processing is carried out to the body profile in image.Pass through the enhancing
Processing can make main body in image definitely.Further, effect enhancing processing is being carried out to the body profile in image
Before, can also include：The auxiliary camera in dual camera by acquiring image carries out pair the corresponding main body of body profile
It is burnt.Wherein it is possible to calculate the object edge in image using related algorithm and then complete the focusing of auxiliary camera, such as image border
Detection algorithm after being calculated the marginal information in image using the algorithm, then is matched with above-mentioned object identification models coupling, can
So that the contour of object identified is more accurate.
Preferably, after the background of body profile carries out virtualization processing in image, can also include：Schemed by acquiring
Treated to carrying out virtualization that image carries out preview and output for main camera in the dual camera of picture.Meanwhile auxiliary camera can
Depth map and focusing main body are calculated for auxiliary, carries out main body affirmed.Pass through the functional areas to main camera and auxiliary camera
Point and positioning, the effect both made definitely, and then make relative program and be arranged more targeted.
Based on above-described embodiment and preferred embodiment, in a preferred embodiment of the invention, additionally provides one kind and passing through people
The method of work intelligence AI (Artificial Intelligence) image object identification enhancing dual camera virtualization effect.Fig. 2 is
A kind of according to the preferred embodiment of the invention identifying that enhancing dual camera blurs the flow of effect method by AI image objects
Figure, wherein the dual camera position in this method in equipment is adjacent and parallel, respectively based on auxiliary camera；Wherein, main camera shooting
Head calculates depth map and AI image recognitions focusing main body for image preview and image output, auxiliary camera for assisting.
As shown in Fig. 2, this method includes mainly following detailed step：
(1) artificial neural network algorithm is used to train AI object model collection；
(2) after opening camera applications, virtualization function is opened, main camera is opened simultaneously with auxiliary camera, exports video
Stream；
(3) frame data of auxiliary camera and the frame data of main camera are intercepted and captured, and according to the time by above-mentioned two frame data
Matching synchronizes；
(4) using main camera frame data synchronous with auxiliary camera, according to the difference of two image informations and two camera shootings
The distance and picture difference of head center, calculate the depth map of this frame data；
(5) according to auxiliary camera or main camera data, edge in image is calculated using Edge-Detection Algorithm
Information is matched with AI object model collection, identifies contour of object, finds out focusing main body, and carry out for focusing body profile
Effect enhances；
(6) according to main body contour of object and depth map, the background blurring of main camera frame is calculated as a result, and carrying out figure
As preview and image export.
The preferred embodiment is cooperateed with using the identification of AI image objects with depth map, and improving mobile phone etc. can picture pick-up device pair
The discrimination of the background blurring focusing main body of camera and virtualization effect, make the virtualization effect than depth is used only in the related technology
The virtualization effect that figure calculates more has a sense of hierarchy, and definitely, profile is more clear main body.It is carried by using the identification of AI image objects
The discrimination of hi-vision main body and background, the preferred embodiment make the transition to the main body and background of focus in image more certainly
So, and then the technique effect closer to professional slr camera depth image has been reached.
According to embodiments of the present invention, a kind of device of image background virtualization processing is additionally provided.Fig. 3 is according to of the invention real
The structural schematic diagram for applying the image background virtualization processing unit of example, as shown in figure 3, the device includes：Acquisition module 32 identifies mould
Block 34, blurring module 36.Image background virtualization processing unit is illustrated below.
Acquisition module 32, for collecting image；
Identification module 34 is connected to above-mentioned acquisition module 32, for according to object identification model, identifying the object in image
The body profile of body, wherein object identification model is obtained using multi-group data by machine learning training, in multi-group data
Every group of data include：The body profile of image and the object identified from image；
Blurring module 36 is connected to above-mentioned identification module 34, is carried out at virtualization for the background to body profile in image
Reason.
Preferably, according to fig. 3 in structure, above-mentioned image background virtualization processing unit can also include：Processing module is used
According to object identification model, before the body profile for identifying the object in image, the sampling of predetermined quantity series is obtained
Image, and the body profile of object that is identified from sampled images；Module is obtained, above-mentioned processing module and identification are connected to
Module 34, for the sampled images of acquisition, and the body profile of object that is identified from sampled images, using artificial god
It is trained through network algorithm, obtains the object identification model for predetermined quantity image series.
Preferably, above-mentioned blurring module 36 may include：Acquiring unit, the depth map for obtaining image；Unit is blurred,
It is connected to above-mentioned acquiring unit, for according to the depth map of acquisition and the body profile identified, to the body profile in image
Background carry out virtualization processing.
Preferably, above-mentioned acquiring unit may include：Subelement is intercepted and captured, for intercepting and capturing master image by main camera and leading to
It crosses auxiliary camera and intercepts and captures auxiliary image；Subelement is obtained, above-mentioned intercepting and capturing subelement is connected to, for being carried out to master image and auxiliary image
Time synchronization obtains the main synchronous images of main camera and the auxiliary synchronous images of auxiliary camera；Subelement is obtained, is connected to above-mentioned
Subelement is obtained, for according to main synchronous images and auxiliary synchronous images, obtaining the depth map of image.
Preferably, above-mentioned acquisition subelement may include：First obtains time subelement, for obtaining main synchronous images and auxiliary
The image difference of synchronous images obtains the distance between two center of main camera and auxiliary camera, and obtains main synchronization
The picture difference of image and auxiliary synchronous images；Second obtains time subelement, is connected to above-mentioned first and obtains time subelement, is used for root
According to the image difference of acquisition, distance and picture difference obtain the depth map of image.
Preferably, according to fig. 3 in structure, above-mentioned image background virtualization processing unit can also include：Enhance module, even
It is connected between above-mentioned identification module 34 and blurring module 36, the background for the body profile in image carries out virtualization processing
Before, effect enhancing processing is carried out to the body profile in image.
Optionally, above-mentioned image background virtualization processing unit can also include：Focusing module is connected to above-mentioned identification module
Between 34 and enhancing module, for before carrying out effect enhancing processing to the body profile in image, passing through acquisition image
Auxiliary camera in dual camera focuses to the corresponding main body of body profile.
Optionally, above-mentioned image background virtualization processing unit can also include：Preview output module is connected to above-mentioned virtualization
Module 36, after the background for the body profile in image carries out virtualization processing, in the dual camera by acquiring image
Main camera carry out preview and output to carrying out virtualization treated image.
According to embodiments of the present invention, a kind of storage medium is additionally provided, storage medium includes the program of storage, wherein
Equipment where control storage medium executes the image background virtualization processing method of above-mentioned any one when program is run.
According to embodiments of the present invention, a kind of processor is additionally provided, processor is for running program, wherein program is run
The image background of the above-mentioned any one of Shi Zhihang blurs processing method.
The embodiments of the present invention are for illustration only, can not represent the quality of embodiment.
In the above embodiment of the present invention, all emphasizes particularly on different fields to the description of each embodiment, do not have in some embodiment
The part of detailed description may refer to the associated description of other embodiment.
In several embodiments provided herein, it should be understood that disclosed technology contents can pass through others
Mode is realized.Wherein, the apparatus embodiments described above are merely exemplary, for example, the unit division, Ke Yiwei
A kind of division of logic function, formula that in actual implementation, there may be another division manner, such as multiple units or component can combine or
Person is desirably integrated into another system, or some features can be ignored or not executed.Another point, shown or discussed is mutual
Between coupling, direct-coupling or communication connection can be INDIRECT COUPLING or communication link by some interfaces, unit or module
It connects, can be electrical or other forms.
The unit illustrated as separating component may or may not be physically separated, aobvious as unit
The component shown may or may not be physical unit, you can be located at a place, or may be distributed over multiple
On unit.Some or all of unit therein can be selected according to the actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, it can also
It is that each unit physically exists alone, it can also be during two or more units be integrated in one unit.Above-mentioned integrated list
The form that hardware had both may be used in member is realized, can also be realized in the form of SFU software functional unit.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product
When, it can be stored in a computer read/write memory medium.Based on this understanding, technical scheme of the present invention is substantially
The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words
It embodies, which is stored in a storage medium, including some instructions are used so that a computer
Equipment (can be personal computer, server or network equipment etc.) execute each embodiment the method for the present invention whole or
Part steps.And storage medium above-mentioned includes：USB flash disk, read-only memory (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), mobile hard disk, magnetic disc or CD etc. are various can to store program code
Medium.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (11)
1. a kind of image background blurs processing method, which is characterized in that including：
Collect image；
According to object identification model, the body profile of the object in described image is identified, wherein the object identification model is
It is obtained by machine learning training using multi-group data, every group of data in the multi-group data include：Image and from institute
State the body profile of the object identified in image；
Virtualization processing is carried out to the background of body profile described in described image.
2. according to the method described in claim 1, it is characterized in that, according to the object identification model, the figure is identified
Before the body profile of object as in, further include：
The sampled images of predetermined quantity series are obtained, and the body profile of object identified from the sampled images；
To the sampled images of acquisition, and the body profile of object that is identified from the sampled images, using artificial
Neural network algorithm is trained, and obtains the object identification model for the predetermined quantity image series.
3. according to the method described in claim 1, it is characterized in that, being carried out to the background of body profile described in described image empty
Change is handled：
Obtain the depth map of described image；
According to the depth map of acquisition and the body profile identified, to the back of the body of the body profile in described image
Scape carries out virtualization processing.
4. according to the method described in claim 3, it is characterized in that, the depth map for obtaining described image includes：
Master image is intercepted and captured by main camera and auxiliary image is intercepted and captured by auxiliary camera；
Time synchronization is carried out to the master image and the auxiliary image, obtains the main synchronous images of the main camera and described auxiliary
The auxiliary synchronous images of camera；
According to the main synchronous images and the auxiliary synchronous images, the depth map of described image is obtained.
5. according to the method described in claim 4, it is characterized in that, according to the main synchronous images and the auxiliary synchronous images,
Obtain described image depth map include：
The image difference of the main synchronous images and the auxiliary synchronous images is obtained, the main camera and the auxiliary camera shooting are obtained
The distance between first two center, and obtain the picture difference of the main synchronous images and the auxiliary synchronous images；
According to the described image difference of acquisition, the distance and the picture difference obtain the depth map of described image.
6. the method according to any one of claims 1 to 5, it is characterized in that, to main body wheel described in described image
Before wide background carries out virtualization processing, further include：
Effect enhancing processing is carried out to the body profile in described image.
7. according to the method described in claim 6, it is characterized in that, carrying out effect to the body profile in described image
Before enhancing processing, further include：
The auxiliary camera in dual camera by acquiring described image focuses to the corresponding main body of the body profile.
8. according to the method described in claim 6, it is characterized in that, in the background progress to body profile described in described image
After virtualization processing, further include：
The main camera in dual camera by acquiring described image carries out preview and defeated to carrying out virtualization treated image
Go out.
9. a kind of image background blurs processing unit, which is characterized in that including：
Acquisition module, for collecting image；
Identification module, for according to object identification model, identifying the body profile of the object in described image, wherein described
Object identification model show that every group of data in the multi-group data are wrapped using multi-group data by machine learning training
It includes：The body profile of image and the object identified from described image；
Blurring module carries out virtualization processing for the background to body profile described in described image.
10. a kind of storage medium, which is characterized in that the storage medium includes the program of storage, wherein run in described program
When control the storage medium where equipment perform claim require image background virtualization processing side described in any one of 1 to 8
Method.
11. a kind of processor, which is characterized in that the processor is for running program, wherein right of execution when described program is run
Profit requires the image background described in any one of 1 to 8 to blur processing method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810019406.2A CN108305223B (en) | 2018-01-09 | 2018-01-09 | Image background blurring processing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810019406.2A CN108305223B (en) | 2018-01-09 | 2018-01-09 | Image background blurring processing method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108305223A true CN108305223A (en) | 2018-07-20 |
| CN108305223B CN108305223B (en) | 2020-11-03 |
Family
ID=62868316
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810019406.2A Active CN108305223B (en) | 2018-01-09 | 2018-01-09 | Image background blurring processing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108305223B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109102460A (en) * | 2018-08-28 | 2018-12-28 | Oppo广东移动通信有限公司 | A kind of image processing method, image processing apparatus and terminal device |
| CN109785400A (en) * | 2018-11-20 | 2019-05-21 | 广东智媒云图科技股份有限公司 | A kind of sketch figure picture production method, device, electronic equipment and storage medium |
| WO2020063080A1 (en) * | 2018-09-28 | 2020-04-02 | 珠海格力电器股份有限公司 | Method, system, and cell phone for creating blurring effect during recording |
| CN112840376A (en) * | 2018-10-15 | 2021-05-25 | 华为技术有限公司 | Image processing method, device and equipment |
| CN113298845A (en) * | 2018-10-15 | 2021-08-24 | 华为技术有限公司 | Image processing method, device and equipment |
| US11122198B2 (en) | 2020-01-07 | 2021-09-14 | International Business Machines Corporation | Adjusting image capture parameters via machine learning |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080212878A1 (en) * | 2002-06-14 | 2008-09-04 | Canon Kabushiki Kaisha | Multiple image processing and synthesis using background image extraction |
| US8503801B2 (en) * | 2010-09-21 | 2013-08-06 | Adobe Systems Incorporated | System and method for classifying the blur state of digital image pixels |
| CN104751407A (en) * | 2015-03-11 | 2015-07-01 | 百度在线网络技术（北京）有限公司 | Method and device used for blurring image |
| CN105100615A (en) * | 2015-07-24 | 2015-11-25 | 青岛海信移动通信技术股份有限公司 | Image preview method, apparatus and terminal |
| CN105979165A (en) * | 2016-06-02 | 2016-09-28 | 广东欧珀移动通信有限公司 | Blurred photos generation method, blurred photos generation device and mobile terminal |
| CN106960413A (en) * | 2017-03-24 | 2017-07-18 | 深圳市金立通信设备有限公司 | A kind of method and terminal of image virtualization |
| CN107194938A (en) * | 2017-04-17 | 2017-09-22 | 上海大学 | Image outline detection method based on depth convolutional neural networks |
-
2018
- 2018-01-09 CN CN201810019406.2A patent/CN108305223B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080212878A1 (en) * | 2002-06-14 | 2008-09-04 | Canon Kabushiki Kaisha | Multiple image processing and synthesis using background image extraction |
| US8503801B2 (en) * | 2010-09-21 | 2013-08-06 | Adobe Systems Incorporated | System and method for classifying the blur state of digital image pixels |
| CN104751407A (en) * | 2015-03-11 | 2015-07-01 | 百度在线网络技术（北京）有限公司 | Method and device used for blurring image |
| CN105100615A (en) * | 2015-07-24 | 2015-11-25 | 青岛海信移动通信技术股份有限公司 | Image preview method, apparatus and terminal |
| CN105979165A (en) * | 2016-06-02 | 2016-09-28 | 广东欧珀移动通信有限公司 | Blurred photos generation method, blurred photos generation device and mobile terminal |
| CN106960413A (en) * | 2017-03-24 | 2017-07-18 | 深圳市金立通信设备有限公司 | A kind of method and terminal of image virtualization |
| CN107194938A (en) * | 2017-04-17 | 2017-09-22 | 上海大学 | Image outline detection method based on depth convolutional neural networks |
Non-Patent Citations (1)
| Title |
|---|
| 李海燕: "卷积神经网络在目标轮廓检测中的应用", 《电子世界》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109102460A (en) * | 2018-08-28 | 2018-12-28 | Oppo广东移动通信有限公司 | A kind of image processing method, image processing apparatus and terminal device |
| WO2020063080A1 (en) * | 2018-09-28 | 2020-04-02 | 珠海格力电器股份有限公司 | Method, system, and cell phone for creating blurring effect during recording |
| CN112840376A (en) * | 2018-10-15 | 2021-05-25 | 华为技术有限公司 | Image processing method, device and equipment |
| CN113298845A (en) * | 2018-10-15 | 2021-08-24 | 华为技术有限公司 | Image processing method, device and equipment |
| US12026863B2 (en) | 2018-10-15 | 2024-07-02 | Huawei Technologies Co., Ltd. | Image processing method and apparatus, and device |
| CN109785400A (en) * | 2018-11-20 | 2019-05-21 | 广东智媒云图科技股份有限公司 | A kind of sketch figure picture production method, device, electronic equipment and storage medium |
| CN109785400B (en) * | 2018-11-20 | 2023-08-04 | 广东智媒云图科技股份有限公司 | Silhouette image manufacturing method and device, electronic equipment and storage medium |
| US11122198B2 (en) | 2020-01-07 | 2021-09-14 | International Business Machines Corporation | Adjusting image capture parameters via machine learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108305223B (en) | 2020-11-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108305223A (en) | Image background blurring processing method and device | |
| Li et al. | In ictu oculi: Exposing ai created fake videos by detecting eye blinking | |
| CN110147721B (en) | Three-dimensional face recognition method, model training method and device | |
| CN109241903A (en) | Sample data cleaning method, device, computer equipment and storage medium | |
| CN111597938A (en) | Living body detection and model training method and device | |
| CN109409250A (en) | A kind of across the video camera pedestrian of no overlap ken recognition methods again based on deep learning | |
| CN110070029A (en) | A kind of gait recognition method and device | |
| CN110827312A (en) | Learning method based on cooperative visual attention neural network | |
| Jin et al. | Learning to see through reflections | |
| CN113657195A (en) | Face image recognition method, face image recognition equipment, electronic device and storage medium | |
| CN113887387A (en) | Ski field target image generation method, system and server | |
| CN113449606A (en) | Target object identification method and device, computer equipment and storage medium | |
| CN115830002A (en) | Infrared image quality evaluation method and device | |
| CN114898402A (en) | Pedestrian re-identification method, device and medium fusing human face and human body characteristics | |
| CN111860057A (en) | Face image blurring and living body detection method and device, storage medium and equipment | |
| CN105488780A (en) | Monocular vision ranging tracking device used for industrial production line, and tracking method thereof | |
| CN110188630A (en) | A kind of face identification method and camera | |
| CN115410240A (en) | Intelligent face pockmark and color spot analysis method and device and storage medium | |
| CN111931544B (en) | Living body detection method, living body detection device, computing equipment and computer storage medium | |
| CN112766065A (en) | Mobile terminal examinee identity authentication method, device, terminal and storage medium | |
| CN116506724A (en) | Photographing auxiliary method, photographing auxiliary device, medium and terminal | |
| CN111626212A (en) | Method and device for identifying object in picture, storage medium and electronic device | |
| Selim et al. | Image Quality-aware Deep Networks Ensemble for Efficient Gender Recognition in the Wild. | |
| CN113225484B (en) | Method and device for rapidly acquiring high-definition picture shielding non-target foreground | |
| CN110751034B (en) | Pedestrian behavior recognition method and terminal equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |