CN110288530A - A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding - Google Patents
A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding Download PDFInfo
- Publication number
- CN110288530A CN110288530A CN201910578592.8A CN201910578592A CN110288530A CN 110288530 A CN110288530 A CN 110288530A CN 201910578592 A CN201910578592 A CN 201910578592A CN 110288530 A CN110288530 A CN 110288530A
- Authority
- CN
- China
- Prior art keywords
- image
- super
- resolution rebuilding
- resolution
- processed
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012545 processing Methods 0.000 title claims abstract description 118
- 238000003672 processing method Methods 0.000 title claims abstract description 20
- 230000011218 segmentation Effects 0.000 claims abstract description 81
- 238000000034 method Methods 0.000 claims description 39
- 238000012549 training Methods 0.000 claims description 37
- 238000004891 communication Methods 0.000 claims description 19
- 241001465754 Metazoa Species 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 10
- 230000000007 visual effect Effects 0.000 abstract description 11
- 230000008569 process Effects 0.000 description 14
- 238000003709 image segmentation Methods 0.000 description 9
- 238000010586 diagram Methods 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000000717 retained effect Effects 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000008439 repair process Effects 0.000 description 3
- 238000005192 partition Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 239000007787 solid Substances 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000004883 computer application Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 230000001815 facial effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012360 testing method 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
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
- G06T2207/30201—Face
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The embodiment of the invention provides the processing method and processing devices that a kind of pair of image carries out super-resolution rebuilding, comprising: obtains image to be processed；Image dividing processing is carried out to image to be processed, obtains target area and background area；Wherein, target area is the region where the target in image to be processed, and background area is the region in image to be processed in addition to target area；Segmentation masking-out is generated based on target area and background area；First super-resolution rebuilding processing is carried out to image to be processed, obtains the first super-resolution rebuilding image；Second super-resolution rebuilding processing is carried out to image to be processed, obtains the second super-resolution rebuilding image；According to segmentation masking-out, the first super-resolution rebuilding image and the second super-resolution rebuilding image, the super-resolution rebuilding image of image to be processed is obtained.Using this programme, the visual effect for the high-definition picture that can be improved.
Description
Technical field
The present invention relates to computer application technologies, and the place of super-resolution rebuilding is carried out more particularly to a kind of pair of image
Manage method and device.
Background technique
When handling image, it is sometimes desirable to which target area and background area to image use different processing sides
Formula, for example, when carrying out super-resolution processing to image, in image face or the target areas such as animal hair for,
Need to retain the image detail of these target areas, and for the background area in image, then need to carry out noise repair,
To reduce the mosaic and burr in background area.
In the related technology, usual to the mode of the image progress super-resolution rebuilding processing of low resolution are as follows: firstly, low
Image in different resolution is registrated on the lattice point for the high-definition picture to be calculated, then, using interpolation algorithm to low-resolution image
It carries out interpolation and generates corresponding high-definition picture to obtain the value of the corresponding each pixel of high-definition picture.
But above-mentioned super-resolution rebuilding processing mode will lead to vision when the image to low resolution carries out interpolation
Effect is deteriorated, and generates mosaic, influences the visual effect of high-definition picture；If further to the horse in high-definition picture
The noises such as Sai Ke and burr are repaired, and may lose image detail, also will affect the visual effect of high-definition picture.
Summary of the invention
The processing method and processing device for being designed to provide a kind of pair of image and carrying out super-resolution rebuilding of the embodiment of the present invention,
Visual effect with the high-definition picture handled by super-resolution rebuilding.Specific technical solution is as follows:
The embodiment of the invention provides the processing methods that a kind of pair of image carries out super-resolution rebuilding, comprising:
Obtain image to be processed；
Image dividing processing is carried out to the image to be processed, obtains target area and background area；Wherein, the target
Region is the region where the target in the image to be processed, and the background area is that the mesh is removed in the image to be processed
Mark the region except region；
Segmentation masking-out is generated based on the target area and the background area；Wherein, the segmentation masking-out includes covering
Region and region is not covered, the cover region corresponds to the background area, and the region of not covering corresponds to the target area；
First super-resolution rebuilding processing is carried out to the image to be processed, obtains the first super-resolution rebuilding image；It is right
The image to be processed carries out the second super-resolution rebuilding processing, obtains the second super-resolution rebuilding image；
According to the segmentation masking-out, the first super-resolution rebuilding image and the second super-resolution rebuilding image,
Obtain the super-resolution rebuilding image of the image to be processed.
Optionally, described according to the segmentation masking-out, the first super-resolution rebuilding image and second super-resolution
Rate reconstruction image obtains the super-resolution rebuilding image of the image to be processed, comprising:
It is chosen from the first super-resolution rebuilding image according to the segmentation masking-out corresponding with the target area
Pixel value is as target area super-resolution rebuilding image；
It is chosen from the second super-resolution rebuilding image according to the segmentation masking-out corresponding with the background area
Pixel value is as background area super-resolution rebuilding image；
The target area super-resolution rebuilding image and the background area super-resolution rebuilding image are merged,
Obtain the super-resolution rebuilding image of the image to be processed.
It is optionally, described that segmentation masking-out is generated based on the target area and the background area, comprising:
Region is not covered using the target area as the segmentation masking-out, using the background area as the segmentation
The cover region of masking-out.
Optionally, described to be chosen from the first super-resolution rebuilding image according to the segmentation masking-out and the target
The corresponding pixel value in region is as target area super-resolution rebuilding image, comprising:
It chooses from the first super-resolution rebuilding image and is not covered described in the corresponding pixel value conduct of region with described
Target area super-resolution rebuilding image；
It is described to be chosen from the second super-resolution rebuilding image according to the segmentation masking-out and the background area pair
The pixel value answered is as background area super-resolution rebuilding image, comprising:
Pixel value corresponding with the cover region is chosen from the second super-resolution rebuilding image as the back
Scene area super-resolution rebuilding image.
Optionally, described to be chosen from the first super-resolution rebuilding image according to the segmentation masking-out and the target
The corresponding pixel value in region is as target area super-resolution rebuilding image, comprising:
Intercepted from the first super-resolution rebuilding image with it is described divide masking-out do not cover the corresponding region in region,
As the target area super-resolution rebuilding image；
It is described to be chosen from the second super-resolution rebuilding image according to the segmentation masking-out and the background area pair
The pixel value answered is as background area super-resolution rebuilding image, comprising:
Region corresponding with the segmentation cover region of masking-out is intercepted from the second super-resolution rebuilding image, is made
For the background area super-resolution rebuilding image.
Optionally, described that first super-resolution rebuilding processing is carried out to the image to be processed, obtain the first super-resolution
Reconstruction image, comprising:
The image to be processed is input in the first Super-resolution reconstruction established model that training obtains in advance, obtains the first surpassing
Resolution reconstruction image；Wherein, the first Super-resolution reconstruction established model is to be carried out based on target sample image as training set
Training obtains, and the target sample image refers to comprising the image based on target and the target.
Optionally, described that second super-resolution rebuilding processing is carried out to the image to be processed, obtain the second super-resolution
Reconstruction image, comprising:
The image to be processed is input in the second Super-resolution reconstruction established model that training obtains in advance, obtains the second surpassing
Resolution reconstruction image；Wherein, the second Super-resolution reconstruction established model be based on it is non-determine class sample image as training set into
Row training obtains, described non-to determine class sample image and refer to various different classes of images.
Optionally, the image to be processed is the image comprising face and/or hair；Wherein, the face be face and/
Or animal face, the hair are the hair and/or animal hair of people；The target area is face area and/or hair area
Domain, the background area are the region in the image to be processed in addition to the face area and the hair area.
Optionally, image dividing processing is carried out to the image to be processed, is described to the image progress first to be processed
Super-resolution rebuilding processing and described the step of carrying out the processing of the second super-resolution rebuilding to the image to be processed, carry out simultaneously.
The embodiment of the invention also provides the processing units that a kind of pair of image carries out super-resolution rebuilding, comprising:
Module is obtained, for obtaining image to be processed；
Divide module, for carrying out image dividing processing to the image to be processed, obtains target area and background area；
Wherein, the target area is the region where the target in the image to be processed, and the background area is described to be processed
Region in image in addition to the target area；
Masking-out generation module generates segmentation masking-out based on the target area and the background area；Wherein, the segmentation
Masking-out includes covering region and not covering region, and the cover region corresponds to the background area, and the region of not covering corresponds to
The target area；
First super-resolution rebuilding processing module, for being carried out at the first super-resolution rebuilding to the image to be processed
Reason, obtains the first super-resolution rebuilding image
Second super-resolution rebuilding processing module, for being carried out at the second super-resolution rebuilding to the image to be processed
Reason, obtains the second super-resolution rebuilding image；
Super-resolution rebuilding image generation module, for according to the segmentation masking-out of the image to be processed, described the first surpass
Resolution reconstruction image and the second super-resolution rebuilding image generate the super-resolution rebuilding figure of the image to be processed
Picture.
Optionally, the super-resolution rebuilding image generation module, comprising:
Target area super-resolution rebuilding image generation unit, for according to the segmentation masking-out from first super-resolution
Pixel value corresponding with the target area is chosen in rate reconstruction image as target area super-resolution rebuilding image；
Background area super-resolution rebuilding image generation unit, for according to the segmentation masking-out from second super-resolution
Pixel value corresponding with the background area is chosen in rate reconstruction image as background area super-resolution rebuilding image；
Combining unit is used for the target area super-resolution rebuilding image and the background area super-resolution rebuilding
Image merges, and obtains the super-resolution rebuilding image of the image to be processed.
Optionally, the masking-out generation module is also used to: using the target area not covering as the segmentation masking-out
Region, using the background area as the cover region of the segmentation masking-out.
Optionally, the target area super-resolution rebuilding image generation unit, is also used to: from first super-resolution
It is chosen in reconstruction image with the corresponding pixel value in region of not covering as the target area super-resolution rebuilding image；
The background area super-resolution rebuilding image generation unit is also used to from the second super-resolution rebuilding image
It is middle to choose pixel value corresponding with the cover region as the background area super-resolution rebuilding image.
Optionally, the target area super-resolution rebuilding image generation unit, is also used to: from first super-resolution
Intercepted in reconstruction image with it is described divide masking-out do not cover the corresponding region in region, as the target area Super-resolution reconstruction
Build image；
The background area super-resolution rebuilding image generation unit, is also used to: from the second super-resolution rebuilding figure
The interception region corresponding with the segmentation cover region of masking-out as in, as the background area super-resolution rebuilding image.
Optionally, the first super-resolution rebuilding processing module is also used to for the image to be processed being input in advance
In the first Super-resolution reconstruction established model that training obtains, the first super-resolution rebuilding image is obtained；Wherein, first super-resolution
Rate reconstruction model is to be trained to obtain as training set based on target sample image, and the target sample image refers to comprising mesh
Image based on mark and the target.
Optionally, the second super-resolution rebuilding processing module is also used to for the image to be processed being input in advance
In the second Super-resolution reconstruction established model that training obtains, the second super-resolution rebuilding image is obtained；Wherein, second super-resolution
Rate reconstruction model is to be determined class sample image based on non-as training set and be trained to obtain, described non-to determine class sample image and refer to respectively
The different classes of image of kind.
Optionally, the image to be processed is the image comprising face and/or hair；Wherein, the face be face and/
Or animal face, the hair are the hair and/or animal hair of people；The target area is face area and/or hair area
Domain, the background area are the region in the image to be processed in addition to the face area and the hair area.
It optionally, further include Coordination module, for the image to be processed to be input to the segmentation module, described simultaneously
First super-resolution rebuilding processing module and the second super-resolution rebuilding processing module.
The embodiment of the invention also provides a kind of electronic equipment, including processor, communication interface, memory and communication are total
Line, wherein processor, communication interface, memory complete mutual communication by communication bus；
Memory, for storing computer program；
Processor, when for executing the program stored on memory, realization is any of the above-described described to surpass image
The processing method of resolution reconstruction.
The embodiment of the invention also provides a kind of computer readable storage medium, the computer readable storage medium memory
Contain computer program, the computer program is realized when being executed by processor any of the above-described described carries out super-resolution to image
The processing method that rate is rebuild.
The processing method and processing device provided in an embodiment of the present invention that super-resolution rebuilding is carried out to image, to image to be processed
Region of different nature carried out different super-resolution processings, it will be understood that different super-resolution processings can generate phase
The different processing results answered in this way, can retain the details of target for the target area of image to be processed, and are treated
For the background area for handling image, the noises such as the mosaic generated in calculating process and burr can be repaired, so as to mention
The visual effect for the high-definition picture that height obtains.Certainly, implement any of the products of the present invention or method it is not absolutely required to same
When reach all the above advantage.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the process signal for the processing method that a kind of pair of image provided in an embodiment of the present invention carries out super-resolution rebuilding
Figure；
Fig. 2 is the scheme schematic diagram of the processing method provided in an embodiment of the present invention that super-resolution rebuilding is carried out to image；
Fig. 3 is the structural representation for the processing unit that a kind of pair of image provided in an embodiment of the present invention carries out super-resolution rebuilding
Figure；
Fig. 4 is the schematic diagram of a kind of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
When handling image, it is sometimes desirable to which target area and background area to image use different processing sides
Formula, for example, when carrying out oversubscription processing to image, in image face or the target areas such as animal hair for, need
Retain the image detail of these target areas, and for the background area in image, then need to carry out noise repair, to subtract
Mosaic and burr in few background area.
In the related technology, usual to the mode of the image progress super-resolution rebuilding processing of low resolution are as follows: firstly, low
Image in different resolution is registrated on the lattice point for the high-definition picture to be calculated, then, using interpolation algorithm to low-resolution image
It carries out interpolation and generates corresponding high-definition picture to obtain the value of the corresponding each pixel of high-definition picture.
But above-mentioned super-resolution rebuilding processing mode will lead to vision when the image to low resolution carries out interpolation
Effect is deteriorated, and generates mosaic, influences the visual effect of high-definition picture；If further to the horse in high-definition picture
The noises such as Sai Ke and burr are repaired, and may lose image detail, also will affect the visual effect of high-definition picture.
In order to solve the above-mentioned technical problem, the invention proposes the processing methods that a kind of pair of image carries out super-resolution rebuilding
And device, this method can be applied to electronic equipment, such as computer, server, video camera, specifically without limitation.
The processing method for carrying out super-resolution rebuilding to image is provided to the embodiment of the present invention on the whole below to say
It is bright.
In a kind of implementation of the invention, the above-mentioned processing method for carrying out super-resolution rebuilding to image includes:
Obtain image to be processed；
Image dividing processing is carried out to image to be processed, obtains target area and background area；Wherein, target area be to
The region where the target in image is handled, background area is the region in image to be processed in addition to target area；
Segmentation masking-out is generated based on target area and background area；Wherein, segmentation masking-out includes covering region and not covering
Region covers region and corresponds to background area, do not cover region and correspond to target area；
First super-resolution rebuilding processing is carried out to image to be processed, obtains the first super-resolution rebuilding image；Treat place
It manages image and carries out the second super-resolution rebuilding processing, obtain the second super-resolution rebuilding image；
According to segmentation masking-out, the first super-resolution rebuilding image and the second super-resolution rebuilding image, figure to be processed is obtained
The super-resolution rebuilding image of picture.
As seen from the above, the processing method provided in an embodiment of the present invention that super-resolution rebuilding is carried out to image, is obtaining
After image to be processed, different super-resolution processings has been carried out to the region of different nature of image to be processed, it will be understood that
Different super-resolution processings can generate corresponding different processing result, in this way, for the target area of image to be processed,
The details of target can be retained, and for the background area of image to be processed, the Marseille generated in calculating process can be repaired
Gram and the noises such as burr, so as to the visual effect of the high-definition picture improved.
It below will be by specific embodiment, to the place provided in an embodiment of the present invention for carrying out super-resolution rebuilding to image
Reason method is described in detail.
As shown in Figure 1, carrying out the processing method of super-resolution rebuilding for a kind of pair of image provided in an embodiment of the present invention
Flow diagram includes the following steps:
S101: image to be processed is obtained.
Image to be processed refers to the image for needing to carry out super-resolution rebuilding processing, moreover, including in image to be processed
Background locating for a certain target and the target.It is appreciated that the target and background in image to be processed is to super-resolution rebuilding
Requirement it is different, for example, in the case that target in image to be processed is face, for the face in image to be processed
For region, need to retain image detail, and for the background area in image to be processed, then it needs to carry out noise and repairs
It is multiple, to reduce the mosaic and burr in background area.
In a kind of implementation, image to be processed is the image comprising face and/or hair, wherein face can be people
Face and/or animal face, hair can be the hair and/or animal hair of people, and face that image to be processed is included and/or
Hair, the target in image as to be processed, the other content in image to be processed other than face and/or hair is to carry on the back
Scape.
Wherein, image to be processed can be individual still image, a certain frame being also possible in video or dynamic image.
S102: image dividing processing is carried out to image to be processed, obtains target area and background area.
After obtaining image to be processed, image dividing processing can be carried out to image to be processed, it will be in image to be processed
Target and background be divided into different regions, in this way, different super-resolution rebuildings is carried out to different regions convenient for subsequent
Processing.Wherein, the target area that image dividing processing obtains is the region where the target in image to be processed, and background area is
Region in image to be processed in addition to target area.
In a kind of implementation, the mode of image dividing processing is carried out to image to be processed, it can be with are as follows: firstly, detecting
The boundary of target in image to be processed, it is then possible to according to the boundary of the target in image to be processed, by image to be processed point
It is segmented into target area and background area.
For example, it can use the Image Segmentation Model that training obtains in advance and image segmentation carried out to image to be processed,
Wherein, in advance the obtained Image Segmentation Model of training can for ROI (region of interest, area-of-interest) model or
Edge parted pattern etc..
In a kind of situation, Image Segmentation Model is convolutional neural networks model, in such a case, it is possible to using following step
Suddenly, training obtains Image Segmentation Model: firstly, obtaining multiple image segmentation sample images and corresponding sample object region, so
Afterwards, multiple image segmentation sample images and corresponding sample object region are inputted into preset first training pattern, judges to export
As a result whether meet preset condition, if conditions are not met, adjustment is iterated to preset first training pattern, until output result
Meet preset condition, obtains Image Segmentation Model.
Alternatively, also can use the image partition method based on wavelet transformation, the image partition method based on genetic algorithm
Or the methods of threshold segmentation method, detect the boundary of the target in image to be processed, it is not limited in the embodiment of the present invention.
In another implementation, it also can use object detection method and image dividing processing carried out to image to be processed:
It is possible, firstly, to detect the target in image to be processed, then, the target area of image to be processed is determined according to target region
Domain, and be target area and background area by image segmentation to be processed.
Wherein, according to the target region in image to be processed, when determining the target area of image to be processed, Ke Yizhi
The corresponding target frame region of the target that will test out is connect as target area；It can also be according to default rule, to what is detected
The corresponding target frame region of target is adjusted, using target frame region adjusted as target area, for example, detecting people
After face, the boundary of the corresponding target frame region of face can be extended into preset length, using the target frame region after extension as mesh
Region is marked, in this way, the corresponding region of hair, jewellery that human face region can be made not to be detected nearby also includes into target area
Domain, so that more reasonable to the segmentation of target area and background area.
S103: segmentation masking-out is generated based on target area and background area.
Dividing masking-out is a kind of constituency, including covering region and not covering region two parts, Ke Yigen in segmentation masking-out
It is operated according to any part of selection.In this step, the cover region for dividing masking-out corresponds to the background of image to be processed
The target area that do not cover region and correspond to image to be processed of masking-out is divided in region.
It can be by target after obtaining target area and background area carrying out image dividing processing to image to be processed
Region does not cover region as segmentation masking-out, using background area as the cover region of segmentation masking-out, in this way, just obtaining wait locate
Manage the segmentation masking-out of image.
S104: the first super-resolution rebuilding processing is carried out to image to be processed, obtains the first super-resolution rebuilding image；It is right
Image to be processed carries out the second super-resolution rebuilding processing, obtains the second super-resolution rebuilding image.
Wherein, the processing of the first super-resolution rebuilding and the processing of the second super-resolution rebuilding refer to different super-resolution rebuildings
Mode for example, carrying out image super-resolution rebuilding processing using different Super-resolution reconstruction established models, or uses different parameters
Carry out super-resolution rebuilding processing, etc..
For example, it can be modeled by the way that image to be processed is input to the first Super-resolution reconstruction that training obtains in advance
Type is realized and carries out the first super-resolution rebuilding processing to image to be processed, wherein the first Super-resolution reconstruction established model is based on mesh
Mark sample image is trained as training set.
In a kind of situation, the first Super-resolution reconstruction established model is convolutional neural networks model, in such a case, it is possible to adopt
With following steps, training obtains the first Super-resolution reconstruction established model: firstly, multiple target sample images are obtained, then, to multiple
Target sample image carries out low resolution processing, obtains corresponding low-resolution image, in turn, by multiple target sample images and
Corresponding low-resolution image inputs preset second training pattern, judges to export whether result meets preset condition, if not
Meet, adjustment is iterated to preset second training pattern, until exporting result meets preset condition, obtains described the first surpassing
Resolution reconstruction model.
Wherein, multiple target sample images are the image comprising target and based on target, for example, if figure to be processed
Target as in is face, then, target sample image is the facial image based on face；If in image to be processed
Target is animal hair, then, target sample image is the image based on animal hair.
Alternatively, the first Super-resolution reconstruction established model is also possible to not limit specifically based on interpolation or based on the model of reconstruction
It is fixed.
It is similar, can by the way that image to be processed is input to the second Super-resolution reconstruction established model that training obtains in advance,
It realizes and the second super-resolution rebuilding processing is carried out to image to be processed, wherein the second Super-resolution reconstruction established model is based on non-fixed
Class sample image is trained as training set.
In a kind of situation, the second Super-resolution reconstruction established model is convolutional neural networks model, in such a case, it is possible to adopt
With following steps, training obtains the second Super-resolution reconstruction established model: firstly, obtain multiple it is non-determine class sample image, then, to more
Zhang Feiding class sample image carries out low resolution processing, obtains corresponding low-resolution image, in turn, multiple non-is determined class sample
Image and corresponding low-resolution image input preset third training pattern, judge to export whether result meets preset condition,
If conditions are not met, being iterated adjustment to preset third training pattern, until output result meets preset condition, second is obtained
Super-resolution reconstruction established model.
Wherein, non-to determine class sample image and refer to various different classes of images, that is to say, that the modeling of the second Super-resolution reconstruction
Type is a kind of more general model, poor to the effect of the super-resolution rebuilding processing of specific objective to a certain extent.
Alternatively, the second Super-resolution reconstruction established model is also possible to not limit specifically based on interpolation or based on the model of reconstruction
It is fixed.
It should be noted that in one implementation, which can carry out simultaneously with S102, that is to say, that treat
Handle image carry out image dividing processing, to image to be processed carry out the first super-resolution rebuilding processing and to image to be processed into
The step of the second super-resolution rebuilding of row processing, can carry out simultaneously.It is appreciated that the two steps between in this way, to image into
The treatment process and detection process of row super-resolution rebuilding carry out simultaneously, without successively carrying out, so as to reduce pair
The time-consuming that image is handled.
S105: it according to segmentation masking-out, the first super-resolution rebuilding image and the second super-resolution rebuilding image, obtains wait locate
Manage the super-resolution rebuilding image of image.
In a kind of implementation, it can be chosen from the first super-resolution rebuilding image and target area according to segmentation masking-out
The corresponding pixel value in domain, as target area super-resolution rebuilding image, then, according to segmentation masking-out, from the second super-resolution
Pixel value corresponding with background area is chosen in reconstruction image, as background area super-resolution rebuilding image, in turn, by target
Region super-resolution rebuilding image and background area super-resolution rebuilding image merge, and obtain the super-resolution of image to be processed
Rate reconstruction image.
In generating the step of dividing masking-out, region can not be covered using target area as segmentation masking-out, by background
Cover region of the region as segmentation masking-out, in this way, according to the cover region for dividing masking-out and region can not covered, to first
Super-resolution rebuilding image and the second super-resolution rebuilding image are handled, and the super-resolution rebuilding figure of image to be processed is obtained
Picture.
For example, in a kind of implementation, super-resolution rebuilding image can be generated according to pixel value.Firstly, from
In one super-resolution rebuilding image, pixel value corresponding with region is not covered is chosen, as target area super-resolution rebuilding figure
Then picture from the second super-resolution rebuilding image, chooses pixel value corresponding with region is covered, as background area oversubscription
Resolution reconstruction image, in this way, just having obtained super-resolution rebuilding image.
Alternatively, in another implementation, can use segmentation masking-out to target area super-resolution rebuilding image and
Background area super-resolution rebuilding image is split, and region required for then intercepting is spliced, and obtains Super-resolution reconstruction
Build image.
For example, region is not covered with segmentation masking-out it is possible, firstly, to intercept from the first super-resolution rebuilding image
Corresponding region, as target area super-resolution rebuilding image, it is then possible to be intercepted from the second super-resolution rebuilding image
Region corresponding with the segmentation cover region of masking-out, it is in turn, super to target area as background area super-resolution rebuilding image
Resolution reconstruction image and background area super-resolution rebuilding image are spliced, and super-resolution rebuilding image is obtained.
As seen from the above, the processing method provided in an embodiment of the present invention that super-resolution rebuilding is carried out to image, is obtaining
After image to be processed, different super-resolution processings has been carried out to the region of different nature of image to be processed, it will be understood that
Different super-resolution processings can generate corresponding different processing result, in this way, for the target area of image to be processed,
The details of target can be retained, and for the background area of image to be processed, the Marseille generated in calculating process can be repaired
Gram and the noises such as burr, so as to the visual effect of the high-definition picture improved.
For example, as shown in Fig. 2, for the processing side provided in an embodiment of the present invention for carrying out super-resolution rebuilding to image
The scheme schematic diagram of method.
Firstly, carrying out image dividing processing to image to be processed, target area and background area are obtained, is based on target area
The segmentation masking-out of image to be processed is generated with background area, for example, as shown in Fig. 2, carrying out using ROI model to image to be processed
Image dividing processing obtains human face region, and then the background area based on human face region and in addition to human face region, obtains wait locate
Manage the segmentation masking-out of image；
Meanwhile carrying out the first super-resolution rebuilding processing and the processing of the second super-resolution rebuilding respectively to image to be processed,
Obtain the first super-resolution rebuilding image and the second super-resolution rebuilding image, for example, as shown in Fig. 2, using faceform and
Background model carries out the first super-resolution rebuilding processing and the processing of the second super-resolution rebuilding respectively to image to be processed, obtains
The human face super-resolution reconstruction image and background super-resolution rebuilding image of image to be processed；
Then, according to the segmentation masking-out of image to be processed, the first super-resolution rebuilding image and the second super-resolution rebuilding
Image generates super-resolution rebuilding image, for example, as shown in Fig. 2, determining that face is super according to the segmentation masking-out of image to be processed
The background parts in face part and background super-resolution rebuilding image in resolution reconstruction image, to obtain super-resolution
Rate reconstruction image.
Wherein, the treatment process of super-resolution rebuilding is carried out to image and detection process carries out simultaneously, without elder generation
After carry out, so as to reduce the time-consuming handled image.
With it is above-mentioned to image carry out the processing method of super-resolution rebuilding it is corresponding, the embodiment of the invention also provides one kind
The processing unit of super-resolution rebuilding is carried out to image.
As shown in figure 3, carrying out the processing unit of super-resolution rebuilding for a kind of pair of image provided in an embodiment of the present invention
Structural schematic diagram, the device include:
Module 310 is obtained, for obtaining image to be processed；
Divide module 320, for carrying out image dividing processing to image to be processed, obtains target area and background area；
Wherein, target area is the region where the target in image to be processed, and background area is that target area is removed in image to be processed
Except region；
Masking-out generation module 330 generates segmentation masking-out based on target area and background area；Wherein, segmentation masking-out includes
It covers region and does not cover region, cover region and correspond to background area, do not cover region and correspond to target area；
First super-resolution rebuilding processing module 340, for carrying out the first super-resolution rebuilding processing to image to be processed,
Obtain the first super-resolution rebuilding image
Second super-resolution rebuilding processing module 350, for carrying out the second super-resolution rebuilding processing to image to be processed,
Obtain the second super-resolution rebuilding image；
Super-resolution rebuilding image generation module 360, for segmentation masking-out, the first super-resolution according to image to be processed
Reconstruction image and the second super-resolution rebuilding image, generate the super-resolution rebuilding image of image to be processed.
In a kind of implementation, super-resolution rebuilding image generation module 360, comprising:
Target area super-resolution rebuilding image generation unit 361, for according to segmentation masking-out from the first Super-resolution reconstruction
It builds in image and chooses pixel value corresponding with target area as target area super-resolution rebuilding image；
Background area super-resolution rebuilding image generation unit 362, for according to segmentation masking-out from the second Super-resolution reconstruction
It builds in image and chooses pixel value corresponding with background area as background area super-resolution rebuilding image；
Combining unit 363 is used for target area super-resolution rebuilding image and background area super-resolution rebuilding image
It merges, obtains the super-resolution rebuilding image of image to be processed.
In a kind of implementation, masking-out generation module 330 is also used to: using target area as the non-covered area of segmentation masking-out
Domain, using background area as the cover region of segmentation masking-out.
In a kind of implementation, target area super-resolution rebuilding image generation unit 361 is also used to: from the first oversubscription
Pixel value corresponding with region is not covered is chosen in resolution reconstruction image as target area super-resolution rebuilding image；
Background area super-resolution rebuilding image generation unit 362 is also used to select from the second super-resolution rebuilding image
Take pixel value corresponding with region is covered as background area super-resolution rebuilding image.
In a kind of implementation, target area super-resolution rebuilding image generation unit 361 is also used to: from the first oversubscription
It is intercepted in resolution reconstruction image and does not cover the corresponding region in region with divide masking-out, as target area super-resolution rebuilding figure
Picture；
Background area super-resolution rebuilding image generation unit 362, is also used to: cutting from the second super-resolution rebuilding image
Region corresponding with the segmentation cover region of masking-out is taken, as background area super-resolution rebuilding image.
In a kind of implementation, the first super-resolution rebuilding processing module 340 is also used to for image to be processed being input to pre-
In the first Super-resolution reconstruction established model that first training obtains, the first super-resolution rebuilding image is obtained；Wherein, the first super-resolution
Reconstruction model is to be trained to obtain as training set based on target sample image, and target sample image refers to comprising target and mesh
It is designated as the image of main body.
In a kind of implementation, the second super-resolution rebuilding processing module 350 is also used to for image to be processed being input to pre-
In the second Super-resolution reconstruction established model that first training obtains, the second super-resolution rebuilding image is obtained；Wherein, the second super-resolution
Reconstruction model is to be determined class sample image based on non-as training set and be trained to obtain, non-to determine class sample image and refer to various differences
The image of classification.
In a kind of implementation, image to be processed is the image comprising face and/or hair；Wherein, face be face and/
Or animal face, hair are the hair and/or animal hair of people；Target area is face area and/or hair area, background area
Domain is the region in image to be processed in addition to face area and hair area.
It further include Coordination module (not shown), for image to be processed to be input to simultaneously in a kind of implementation
Divide module, the first super-resolution rebuilding processing module and the second super-resolution rebuilding processing module.
As seen from the above, the processing unit provided in an embodiment of the present invention that super-resolution rebuilding is carried out to image, is obtaining
After image to be processed, different super-resolution processings has been carried out to the region of different nature of image to be processed, it will be understood that
Different super-resolution processings can generate corresponding different processing result, in this way, for the target area of image to be processed,
The details of target can be retained, and for the background area of image to be processed, the Marseille generated in calculating process can be repaired
Gram and the noises such as burr, so as to the visual effect of the high-definition picture improved.
The embodiment of the invention also provides a kind of electronic equipment, as shown in figure 4, include processor 401, communication interface 402,
Memory 403 and communication bus 404, wherein processor 401, communication interface 402, memory 403 are complete by communication bus 404
At mutual communication,
Memory 403, for storing computer program；
Processor 401 when for executing the program stored on memory 403, realizes following steps:
Obtain image to be processed；
Image dividing processing is carried out to image to be processed, obtains target area and background area；Wherein, target area be to
The region where the target in image is handled, background area is the region in image to be processed in addition to target area；
Segmentation masking-out is generated based on target area and background area；Wherein, segmentation masking-out includes covering region and not covering
Region covers region and corresponds to background area, do not cover region and correspond to target area；
First super-resolution rebuilding processing is carried out to image to be processed, obtains the first super-resolution rebuilding image；Treat place
It manages image and carries out the second super-resolution rebuilding processing, obtain the second super-resolution rebuilding image；
According to segmentation masking-out, the first super-resolution rebuilding image and the second super-resolution rebuilding image, figure to be processed is obtained
The super-resolution rebuilding image of picture.
The communication bus that above-mentioned electronic equipment is mentioned can be Peripheral Component Interconnect standard (Peripheral Component
Interconnect, PCI) bus or expanding the industrial standard structure (Extended Industry Standard
Architecture, EISA) bus etc..The communication bus can be divided into address bus, data/address bus, control bus etc..For just
It is only indicated with a thick line in expression, figure, it is not intended that an only bus or a type of bus.
Communication interface is for the communication between above-mentioned electronic equipment and other equipment.
Memory may include random access memory (Random Access Memory, RAM), also may include non-easy
The property lost memory (Non-Volatile Memory, NVM), for example, at least a magnetic disk storage.Optionally, memory may be used also
To be storage device that at least one is located remotely from aforementioned processor.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit,
CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal
Processing, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing
It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete
Door or transistor logic, discrete hardware components.
As seen from the above, electronic equipment provided in an embodiment of the present invention, after obtaining image to be processed, to figure to be processed
The region of different nature of picture has carried out different super-resolution processings, it will be understood that different super-resolution processings can generate
Corresponding different processing result, in this way, the details of target can be retained for the target area of image to be processed, and it is right
For the background area of image to be processed, the noises such as the mosaic generated in calculating process and burr can be repaired, so as to
Improve the visual effect of obtained high-definition picture.
In another embodiment provided by the invention, a kind of computer readable storage medium is additionally provided, which can
It reads to be stored with instruction in storage medium, when run on a computer, so that computer executes any institute in above-described embodiment
That states carries out the processing method of super-resolution rebuilding to image.
In another embodiment provided by the invention, a kind of computer program product comprising instruction is additionally provided, when it
When running on computers, so that computer executes and any in above-described embodiment described carries out super-resolution rebuilding to image
Processing method.
In the above-described embodiments, can come wholly or partly by software, hardware, firmware or any combination thereof real
It is existing.When implemented in software, it can entirely or partly realize in the form of a computer program product.The computer program
Product includes one or more computer instructions.When loading on computers and executing the computer program instructions, all or
It partly generates according to process or function described in the embodiment of the present application.The computer can be general purpose computer, dedicated meter
Calculation machine, computer network or other programmable devices.The computer instruction can store in computer readable storage medium
In, or from a computer readable storage medium to the transmission of another computer readable storage medium, for example, the computer
Instruction can pass through wired (such as coaxial cable, optical fiber, number from a web-site, computer, server or data center
User's line (DSL)) or wireless (such as infrared, wireless, microwave etc.) mode to another web-site, computer, server or
Data center is transmitted.The computer readable storage medium can be any usable medium that computer can access or
It is comprising data storage devices such as one or more usable mediums integrated server, data centers.The usable medium can be with
It is magnetic medium, (for example, floppy disk, hard disk, tape), optical medium (for example, DVD) or semiconductor medium (such as solid state hard disk
Solid State Disk (SSD)) etc..
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device reality
For applying example, electronic equipment embodiment, storage medium embodiment, since it is substantially similar to the method embodiment, so description
Fairly simple, the relevent part can refer to the partial explaination of embodiments of method.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all
Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention
It is interior.
Claims (20)
1. the processing method that a kind of pair of image carries out super-resolution rebuilding characterized by comprising
Obtain image to be processed；
Image dividing processing is carried out to the image to be processed, obtains target area and background area；Wherein, the target area
For the region where the target in the image to be processed, the background area is that the target area is removed in the image to be processed
Region except domain；
Segmentation masking-out is generated based on the target area and the background area；Wherein, the segmentation masking-out includes covering region
Region is not covered, and the cover region corresponds to the background area, and the region of not covering corresponds to the target area；
First super-resolution rebuilding processing is carried out to the image to be processed, obtains the first super-resolution rebuilding image；To described
Image to be processed carries out the second super-resolution rebuilding processing, obtains the second super-resolution rebuilding image；
According to the segmentation masking-out, the first super-resolution rebuilding image and the second super-resolution rebuilding image, obtain
The super-resolution rebuilding image of the image to be processed.
2. the method according to claim 1, wherein described according to the segmentation masking-out, first super-resolution
Rate reconstruction image and the second super-resolution rebuilding image obtain the super-resolution rebuilding image of the image to be processed, packet
It includes:
Pixel corresponding with the target area is chosen from the first super-resolution rebuilding image according to the segmentation masking-out
Value is used as target area super-resolution rebuilding image；
Pixel corresponding with the background area is chosen from the second super-resolution rebuilding image according to the segmentation masking-out
Value is used as background area super-resolution rebuilding image；
The target area super-resolution rebuilding image and the background area super-resolution rebuilding image are merged, obtained
The super-resolution rebuilding image of the image to be processed.
3. according to the method described in claim 2, it is characterized in that, described raw based on the target area and the background area
At segmentation masking-out, comprising:
Region is not covered using the target area as the segmentation masking-out, using the background area as the segmentation masking-out
Cover region.
4. according to the method described in claim 3, it is characterized in that, it is described according to the segmentation masking-out from first super-resolution
Pixel value corresponding with the target area is chosen in rate reconstruction image as target area super-resolution rebuilding image, comprising:
It is chosen from the first super-resolution rebuilding image with the corresponding pixel value in region of not covering as the target
Region super-resolution rebuilding image；
It is described chosen from the second super-resolution rebuilding image according to the segmentation masking-out it is corresponding with the background area
Pixel value is as background area super-resolution rebuilding image, comprising:
Pixel value corresponding with the cover region is chosen from the second super-resolution rebuilding image as the background area
Domain super-resolution rebuilding image.
5. according to the method described in claim 3, it is characterized in that, it is described according to the segmentation masking-out from first super-resolution
Pixel value corresponding with the target area is chosen in rate reconstruction image as target area super-resolution rebuilding image, comprising:
Intercepted from the first super-resolution rebuilding image with it is described divide masking-out do not cover the corresponding region in region, as
The target area super-resolution rebuilding image；
It is described chosen from the second super-resolution rebuilding image according to the segmentation masking-out it is corresponding with the background area
Pixel value is as background area super-resolution rebuilding image, comprising:
Region corresponding with the segmentation cover region of masking-out is intercepted from the second super-resolution rebuilding image, as institute
State background area super-resolution rebuilding image.
6. method according to claim 1 or 2, which is characterized in that described to carry out the first oversubscription to the image to be processed
Resolution reconstruction processing, obtains the first super-resolution rebuilding image, comprising:
The image to be processed is input in the first Super-resolution reconstruction established model that training obtains in advance, obtains the first super-resolution
Rate reconstruction image；Wherein, the first Super-resolution reconstruction established model is to be trained based on target sample image as training set
It obtains, the target sample image refers to comprising the image based on target and the target.
7. method according to claim 1 or 2, which is characterized in that described to carry out the second oversubscription to the image to be processed
Resolution reconstruction processing, obtains the second super-resolution rebuilding image, comprising:
The image to be processed is input in the second Super-resolution reconstruction established model that training obtains in advance, obtains the second super-resolution
Rate reconstruction image；Wherein, the second Super-resolution reconstruction established model is to be determined class sample image based on non-and instructed as training set
It gets, it is described non-to determine class sample image and refer to various different classes of images.
8. method according to claim 1 or 2, which is characterized in that the image to be processed is to include face and/or hair
Image；Wherein, the face is face and/or animal face, and the hair is the hair and/or animal hair of people；It is described
Target area is face area and/or hair area, and the background area is that the face area is removed in the image to be processed
And the region other than the hair area.
9. method according to claim 1 or 2, which is characterized in that
Image dividing processing is carried out to the image to be processed, is described to first super-resolution rebuilding of the image progress to be processed
Processing and described the step of carrying out the processing of the second super-resolution rebuilding to the image to be processed, carry out simultaneously.
10. the processing unit that a kind of pair of image carries out super-resolution rebuilding characterized by comprising
Module is obtained, for obtaining image to be processed；
Divide module, for carrying out image dividing processing to the image to be processed, obtains target area and background area；Its
In, the target area is the region where the target in the image to be processed, and the background area is the figure to be processed
Region as in addition to the target area；
Masking-out generation module generates segmentation masking-out based on the target area and the background area；Wherein, the segmentation masking-out
Including covering region and not covering region, the cover region corresponds to the background area, described not cover described in the correspondence of region
Target area；
First super-resolution rebuilding processing module is obtained for carrying out the first super-resolution rebuilding processing to the image to be processed
To the first super-resolution rebuilding image；
Second super-resolution rebuilding processing module is obtained for carrying out the second super-resolution rebuilding processing to the image to be processed
To the second super-resolution rebuilding image；
Super-resolution rebuilding image generation module, for segmentation masking-out, first super-resolution according to the image to be processed
Rate reconstruction image and the second super-resolution rebuilding image generate the super-resolution rebuilding image of the image to be processed.
11. device according to claim 10, which is characterized in that the super-resolution rebuilding image generation module, comprising:
Target area super-resolution rebuilding image generation unit, for according to the segmentation masking-out from first Super-resolution reconstruction
It builds in image and chooses pixel value corresponding with the target area as target area super-resolution rebuilding image；
Background area super-resolution rebuilding image generation unit, for according to the segmentation masking-out from second Super-resolution reconstruction
It builds in image and chooses pixel value corresponding with the background area as background area super-resolution rebuilding image；
Combining unit is used for the target area super-resolution rebuilding image and the background area super-resolution rebuilding image
It merges, obtains the super-resolution rebuilding image of the image to be processed.
12. device according to claim 11, which is characterized in that the masking-out generation module is also used to: by the target
Region does not cover region as the segmentation masking-out, using the background area as the cover region of the segmentation masking-out.
13. device according to claim 12, which is characterized in that
The target area super-resolution rebuilding image generation unit, is also used to: from the first super-resolution rebuilding image
It chooses with the corresponding pixel value in region of not covering as the target area super-resolution rebuilding image；
The background area super-resolution rebuilding image generation unit is also used to select from the second super-resolution rebuilding image
Take pixel value corresponding with the cover region as the background area super-resolution rebuilding image.
14. device according to claim 12, which is characterized in that
The target area super-resolution rebuilding image generation unit, is also used to: from the first super-resolution rebuilding image
Interception with it is described divide masking-out do not cover the corresponding region in region, as the target area super-resolution rebuilding image；
The background area super-resolution rebuilding image generation unit, is also used to: from the second super-resolution rebuilding image
Region corresponding with the segmentation cover region of masking-out is intercepted, as the background area super-resolution rebuilding image.
15. device described in 0 or 11 according to claim 1, which is characterized in that
The first super-resolution rebuilding processing module is also used to for the image to be processed being input to training in advance obtains the
In one Super-resolution reconstruction established model, the first super-resolution rebuilding image is obtained；Wherein, the first Super-resolution reconstruction established model is
It is trained to obtain as training set based on target sample image, the target sample image refers to comprising target and the target
Based on image.
16. device described in 0 or 11 according to claim 1, which is characterized in that
The second super-resolution rebuilding processing module is also used to for the image to be processed being input to training in advance obtains the
In two Super-resolution reconstruction established models, the second super-resolution rebuilding image is obtained；Wherein, the second Super-resolution reconstruction established model is
Determine class sample image based on non-as training set and be trained to obtain, it is described it is non-determine class sample image refer to it is various different classes of
Image.
17. device described in 0 or 11 according to claim 1, which is characterized in that the image to be processed be comprising face and/or
The image of hair；Wherein, the face is face and/or animal face, and the hair is the hair and/or animal hair of people；
The target area is face area and/or hair area, and the background area is that the face is removed in the image to be processed
Region other than region and the hair area.
18. device described in 0 or 11 according to claim 1, which is characterized in that further include Coordination module, for by described wait locate
Reason image is input to the segmentation module, the first super-resolution rebuilding processing module and second Super-resolution reconstruction simultaneously
Build processing module.
19. a kind of electronic equipment, which is characterized in that including processor, communication interface, memory and communication bus, wherein processing
Device, communication interface, memory complete mutual communication by communication bus；
Memory, for storing computer program；
Processor when for executing the program stored on memory, realizes any method and step of claim 1-9.
20. a kind of computer readable storage medium, which is characterized in that be stored with computer in the computer readable storage medium
Program realizes claim 1-9 any method and step when the computer program is executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910578592.8A CN110288530A (en) | 2019-06-28 | 2019-06-28 | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910578592.8A CN110288530A (en) | 2019-06-28 | 2019-06-28 | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110288530A true CN110288530A (en) | 2019-09-27 |
Family
ID=68020110
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910578592.8A Pending CN110288530A (en) | 2019-06-28 | 2019-06-28 | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110288530A (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110796600A (en) * | 2019-10-29 | 2020-02-14 | Oppo广东移动通信有限公司 | Image super-resolution reconstruction method, image super-resolution reconstruction device and electronic equipment |
| CN111145202A (en) * | 2019-12-31 | 2020-05-12 | 北京奇艺世纪科技有限公司 | Model generation method, image processing method, device, equipment and storage medium |
| CN111163265A (en) * | 2019-12-31 | 2020-05-15 | 成都旷视金智科技有限公司 | Image processing method, image processing device, mobile terminal and computer storage medium |
| CN111161369A (en) * | 2019-12-20 | 2020-05-15 | 上海联影智能医疗科技有限公司 | Image reconstruction storage method and device, computer equipment and storage medium |
| CN111553846A (en) * | 2020-05-12 | 2020-08-18 | Oppo广东移动通信有限公司 | Super-resolution processing method and device |
| CN113781309A (en) * | 2021-09-17 | 2021-12-10 | 北京金山云网络技术有限公司 | Image processing method and device and electronic equipment |
| CN114202651A (en) * | 2020-09-01 | 2022-03-18 | 武汉Tcl集团工业研究院有限公司 | Image over-segmentation method, system and medium for enhancing interested region |
| CN114626273A (en) * | 2022-03-31 | 2022-06-14 | 广州思茂信息科技有限公司 | Finite element parametric modeling method for hexahedral gridding divided bolt and nut |
| CN118590597A (en) * | 2024-07-30 | 2024-09-03 | 深圳天健电子科技有限公司 | Monitoring image generation method and device based on artificial intelligence technology |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678728A (en) * | 2016-01-19 | 2016-06-15 | 西安电子科技大学 | High-efficiency super-resolution imaging device and method with regional management |
| CN106340027A (en) * | 2016-08-26 | 2017-01-18 | 西北大学 | Calligraphy background reconstruction method based on image super resolution |
| CN108596833A (en) * | 2018-04-26 | 2018-09-28 | 广东工业大学 | Super-resolution image reconstruction method, device, equipment and readable storage medium storing program for executing |
| CN108665509A (en) * | 2018-05-10 | 2018-10-16 | 广东工业大学 | A kind of ultra-resolution ratio reconstructing method, device, equipment and readable storage medium storing program for executing |
| CN108846795A (en) * | 2018-05-30 | 2018-11-20 | 北京小米移动软件有限公司 | Image processing method and device |
| CN109064399A (en) * | 2018-07-20 | 2018-12-21 | 广州视源电子科技股份有限公司 | Image super-resolution reconstruction method and system, computer device and storage medium thereof |
| CN109389129A (en) * | 2018-09-15 | 2019-02-26 | 北京市商汤科技开发有限公司 | A kind of image processing method, electronic equipment and storage medium |
-
2019
- 2019-06-28 CN CN201910578592.8A patent/CN110288530A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105678728A (en) * | 2016-01-19 | 2016-06-15 | 西安电子科技大学 | High-efficiency super-resolution imaging device and method with regional management |
| CN106340027A (en) * | 2016-08-26 | 2017-01-18 | 西北大学 | Calligraphy background reconstruction method based on image super resolution |
| CN108596833A (en) * | 2018-04-26 | 2018-09-28 | 广东工业大学 | Super-resolution image reconstruction method, device, equipment and readable storage medium storing program for executing |
| CN108665509A (en) * | 2018-05-10 | 2018-10-16 | 广东工业大学 | A kind of ultra-resolution ratio reconstructing method, device, equipment and readable storage medium storing program for executing |
| CN108846795A (en) * | 2018-05-30 | 2018-11-20 | 北京小米移动软件有限公司 | Image processing method and device |
| CN109064399A (en) * | 2018-07-20 | 2018-12-21 | 广州视源电子科技股份有限公司 | Image super-resolution reconstruction method and system, computer device and storage medium thereof |
| CN109389129A (en) * | 2018-09-15 | 2019-02-26 | 北京市商汤科技开发有限公司 | A kind of image processing method, electronic equipment and storage medium |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110796600A (en) * | 2019-10-29 | 2020-02-14 | Oppo广东移动通信有限公司 | Image super-resolution reconstruction method, image super-resolution reconstruction device and electronic equipment |
| WO2021083059A1 (en) * | 2019-10-29 | 2021-05-06 | Oppo广东移动通信有限公司 | Image super-resolution reconstruction method, image super-resolution reconstruction apparatus, and electronic device |
| CN110796600B (en) * | 2019-10-29 | 2023-08-11 | Oppo广东移动通信有限公司 | Image super-resolution reconstruction method, image super-resolution reconstruction device and electronic equipment |
| CN111161369A (en) * | 2019-12-20 | 2020-05-15 | 上海联影智能医疗科技有限公司 | Image reconstruction storage method and device, computer equipment and storage medium |
| CN111161369B (en) * | 2019-12-20 | 2024-04-23 | 上海联影智能医疗科技有限公司 | Image reconstruction storage method, device, computer equipment and storage medium |
| CN111145202A (en) * | 2019-12-31 | 2020-05-12 | 北京奇艺世纪科技有限公司 | Model generation method, image processing method, device, equipment and storage medium |
| CN111163265A (en) * | 2019-12-31 | 2020-05-15 | 成都旷视金智科技有限公司 | Image processing method, image processing device, mobile terminal and computer storage medium |
| CN111145202B (en) * | 2019-12-31 | 2024-03-08 | 北京奇艺世纪科技有限公司 | Model generation method, image processing method, device, equipment and storage medium |
| CN111553846B (en) * | 2020-05-12 | 2023-05-26 | Oppo广东移动通信有限公司 | Super-resolution processing method and device |
| CN111553846A (en) * | 2020-05-12 | 2020-08-18 | Oppo广东移动通信有限公司 | Super-resolution processing method and device |
| CN114202651A (en) * | 2020-09-01 | 2022-03-18 | 武汉Tcl集团工业研究院有限公司 | Image over-segmentation method, system and medium for enhancing interested region |
| CN113781309A (en) * | 2021-09-17 | 2021-12-10 | 北京金山云网络技术有限公司 | Image processing method and device and electronic equipment |
| CN114626273A (en) * | 2022-03-31 | 2022-06-14 | 广州思茂信息科技有限公司 | Finite element parametric modeling method for hexahedral gridding divided bolt and nut |
| CN118590597A (en) * | 2024-07-30 | 2024-09-03 | 深圳天健电子科技有限公司 | Monitoring image generation method and device based on artificial intelligence technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110288530A (en) | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding | |
| CN110298790A (en) | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding | |
| CN108875732B (en) | Model training and instance segmentation method, device and system and storage medium | |
| US10878575B2 (en) | Foreground-aware image inpainting | |
| JP7490141B2 (en) | IMAGE DETECTION METHOD, MODEL TRAINING METHOD, IMAGE DETECTION APPARATUS, TRAINING APPARATUS, DEVICE, AND PROGRAM | |
| CN108604369B (en) | Method, device and equipment for removing image noise and convolutional neural network | |
| CN110619618A (en) | Surface defect detection method and device and electronic equipment | |
| US11080833B2 (en) | Image manipulation using deep learning techniques in a patch matching operation | |
| Fan et al. | Effect of image noise on the classification of skin lesions using deep convolutional neural networks | |
| Kanwal et al. | Region based adaptive contrast enhancement of medical X-ray images | |
| CN111862044A (en) | Ultrasonic image processing method and device, computer equipment and storage medium | |
| CN108665421B (en) | Highlight component removing device and method for face image and storage medium product | |
| WO2019167665A1 (en) | Learning device, method and program | |
| CN111696196B (en) | Three-dimensional face model reconstruction method and device | |
| CN105868678A (en) | Human face recognition model training method and device | |
| CN112001983B (en) | Method and device for generating occlusion image, computer equipment and storage medium | |
| WO2021082819A1 (en) | Image generation method and apparatus, and electronic device | |
| Omer et al. | Denoising ct images using median based filters: a review | |
| FR et al. | Segmentation of mammography by applying extreme learning machine in tumor detection | |
| Conde et al. | Lens-to-lens bokeh effect transformation. NTIRE 2023 challenge report | |
| CN113362338A (en) | Rail segmentation method, device, computer equipment and rail segmentation processing system | |
| JP6819445B2 (en) | Information processing equipment, control methods, and programs | |
| KR102126795B1 (en) | Deep learning-based image on personal information image processing system, apparatus and method therefor | |
| Wang et al. | A variational histogram equalization method for image contrast enhancement | |
| Cui et al. | Progressive dual-branch network for low-light image enhancement |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190927 |
|
| RJ01 | Rejection of invention patent application after publication |