CN107087201A - Image processing method and device - Google Patents
Image processing method and device Download PDFInfo
- Publication number
- CN107087201A CN107087201A CN201710326762.4A CN201710326762A CN107087201A CN 107087201 A CN107087201 A CN 107087201A CN 201710326762 A CN201710326762 A CN 201710326762A CN 107087201 A CN107087201 A CN 107087201A
- Authority
- CN
- China
- Prior art keywords
- image
- neutral net
- resolution
- low
- transmitting terminal
- 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 45
- 230000007935 neutral effect Effects 0.000 claims abstract description 203
- 238000000034 method Methods 0.000 claims abstract description 61
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 53
- 230000006835 compression Effects 0.000 claims abstract description 44
- 238000007906 compression Methods 0.000 claims abstract description 44
- 238000012549 training Methods 0.000 claims description 38
- 230000005540 biological transmission Effects 0.000 claims description 22
- 238000012545 processing Methods 0.000 claims description 15
- 238000005457 optimization Methods 0.000 claims description 14
- 210000005036 nerve Anatomy 0.000 claims description 4
- 238000003860 storage Methods 0.000 description 19
- 238000013528 artificial neural network Methods 0.000 description 9
- 230000008569 process Effects 0.000 description 7
- 230000009466 transformation Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 210000004218 nerve net Anatomy 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000032798 delamination Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000002592 echocardiography Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000002955 isolation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/132—Sampling, masking or truncation of coding units, e.g. adaptive resampling, frame skipping, frame interpolation or high-frequency transform coefficient masking
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/63—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using sub-band based transform, e.g. wavelets
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
- H04N21/234—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs
- H04N21/2343—Processing of video elementary streams, e.g. splicing of video streams or manipulating encoded video stream scene graphs involving reformatting operations of video signals for distribution or compliance with end-user requests or end-user device requirements
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
- H04N21/44—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream or rendering scenes according to encoded video stream scene graphs
- H04N21/4402—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream or rendering scenes according to encoded video stream scene graphs involving reformatting operations of video signals for household redistribution, storage or real-time display
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of image processing method and device.Wherein, this method includes:The data that receiving end/sending end is sent, and wavelet coefficient is determined according to the data of reception, wherein, the data that transmitting terminal is sent include the wavelet coefficient obtained to original image after wavelet decomposition;Low-resolution image is generated according to wavelet coefficient;Super-resolution rebuilding is carried out to low-resolution image according to neutral net, high-definition picture is obtained.The present invention solves method for compressing image in correlation technique can not be while meet higher compression efficiency and the technical problem of higher image restoring quality.
Description
Technical field
The present invention relates to image processing field, in particular to a kind of image processing method and device.
Background technology
In image transmitting, the limitation of bandwidth make it that the size of image and the transmission time of image can not be taken into account, and generally needs
Image is compressed.Traditional picture compression method is confined to remove redundancy, algorithm to view data from the angle of mathematics
Design framework is substantially stationary, and compression efficiency is difficult to have greatly improved again, and system is consolidated without learning functionality, generation picture quality
It is fixed, when bandwidth limitation is serious, huge is influenceed on image quality.In the prior art, can by the way of wavelet transformation come pair
Image is compressed, and super-resolution rebuilding is carried out to the low-resolution image of compression after compression of images.Wavelet transformation is near
A kind of tool of mathematical analysis being widely used over year, it overcomes defect of the short time discrete Fourier transform in single resolution ratio,
Time domain and frequency domain have the ability for characterizing signal detail information, and the window size of time and frequency is dynamically adapted, and adapt to model
Enclose extensively.In recent years, compression of images and transmission algorithm based on wavelet transformation are widely used in Image Compression.
5/3 lossless wavelet transformation is used in JPEG 2000 compression algorithm with 9/7 wavelet transformation damaged to realize picture pressure
Contracting, the image compression algorithm based on 5/3 small echo is using relatively broad.
Can not meet higher compression efficiency simultaneously for the method for compressing image in correlation technique and higher image also
The technical problem of proper mass, not yet proposes effective solution at present.
The content of the invention
The embodiments of the invention provide a kind of image processing method and device, at least to solve the image pressure in correlation technique
Compression method can not meet higher compression efficiency simultaneously and the technical problem of higher image restoring quality.
One side according to embodiments of the present invention includes there is provided a kind of image processing method, this method:Receive and send
The data sent are held, and wavelet coefficient is determined according to the data of reception, wherein, the data that transmitting terminal is sent are included to original graph
As the wavelet coefficient obtained after wavelet decomposition;Low-resolution image is generated according to wavelet coefficient;According to neutral net to low
Image in different resolution carries out super-resolution rebuilding, obtains high-definition picture.
Further, carrying out super-resolution rebuilding to low-resolution image according to neutral net includes:It is determined that current god
Through network;Undated parameter is received, wherein, undated parameter is being entered according to original image to current neutral net for transmitting terminal transmission
The parameter for the neutral net that row optimization is obtained;Current neutral net is updated according to undated parameter, the god updated
Through network;Super-resolution rebuilding is carried out to low-resolution image according to the neutral net of renewal.
Further, this method also includes:Wavelet decomposition is carried out to multiple sample images, corresponding multiple low resolutions are obtained
Rate sample image;According to multiple sample images and corresponding multiple low resolution sample image training neutral nets;After training
Neutral net as initial neutral net be stored in transmitting terminal and receiving terminal, wherein, sent for the first time according to transmitting terminal
Data carry out super-resolution rebuilding when determine current neutral net be initial neutral net.
Further, carrying out super-resolution rebuilding to low-resolution image according to neutral net includes:According to low resolution
The classification of the content selection neutral net of image;Super-resolution reconstruction is carried out to low-resolution image according to the neutral net of selection
Build.
Another aspect according to embodiments of the present invention, additionally provides a kind of image processing method, and this method includes:Transmitting terminal
Wavelet decomposition is carried out to original image, wavelet coefficient is obtained;Transmitting terminal sends wavelet coefficient to receiving terminal;Receiving terminal is according to small echo
Coefficient generates low-resolution image;Receiving terminal carries out super-resolution rebuilding according to neutral net to low-resolution image, obtains height
Image in different resolution.
Further, wavelet decomposition is carried out to original image in transmitting terminal, obtained after wavelet coefficient, this method is also wrapped
Include:Transmitting terminal generates low-resolution image according to wavelet coefficient;Transmitting terminal is according to original image and low-resolution image to sending
The current neutral net in end is optimized;After optimization in transmitting terminal to neutral net Jing Guo preset times, transmitting terminal is to connecing
Receiving end sends undated parameter, wherein, undated parameter is the parameter of the neutral net after transmitting terminal optimizes by preset times;
Receiving terminal is updated to be updated according to undated parameter after undated parameter is received to the neutral net of receiving terminal
Neutral net, and the neutral net based on renewal carries out oversubscription to the subsequently received corresponding low-resolution image of wavelet coefficient
Resolution is rebuild.
Further, wavelet decomposition is carried out to original image in transmitting terminal, obtained before wavelet coefficient, this method is also wrapped
Include:Transmitting terminal or receiving terminal carry out wavelet decomposition to multiple sample images, obtain corresponding multiple low resolution sample images;Hair
Sending end or receiving terminal train neutral net according to multiple sample images and corresponding multiple low resolution sample images;After training
Neutral net as initial neutral net be stored in transmitting terminal and receiving terminal.
Further, transmitting terminal or receiving terminal are instructed according to multiple sample images and corresponding multiple low resolution sample images
Practicing neutral net includes:Transmitting terminal or receiving terminal are carried out according to the content of low resolution sample image to low resolution sample image
Classification;The low resolution sample image of each classification is trained using corresponding neutral net, the god of multiple classifications is obtained
Through network.
Further, transmitting terminal sends wavelet coefficient to receiving terminal and included:Transmitting terminal is to wavelet coefficient according to default compression
Algorithm is encoded, and obtains code stream;Transmitting terminal sends obtained code stream to receiving terminal, wherein, receiving terminal is receiving code stream
Decoded afterwards according to default compression algorithm, obtain wavelet coefficient.
Another aspect according to embodiments of the present invention, additionally provides a kind of image processing apparatus, and the device includes:Receive single
Member, the data sent for receiving end/sending end, and wavelet coefficient is determined according to the data of reception, wherein, the number that transmitting terminal is sent
According to the wavelet coefficient for including obtaining original image after wavelet decomposition;Generation unit, for being generated according to wavelet coefficient
Low-resolution image;Reconstruction unit, for carrying out super-resolution rebuilding to low-resolution image according to neutral net, obtains high score
Resolution image.
Further, reconstruction unit includes:Determining module, for determining current neutral net;First receiving module, is used
In reception undated parameter, wherein, undated parameter is the excellent to current neutral net progress according to original image of transmitting terminal transmission
Change the parameter of obtained neutral net;Update module, for being updated according to undated parameter to current neutral net, is obtained
The neutral net of renewal;First rebuilds module, and super-resolution is carried out to low-resolution image for the neutral net according to renewal
Rebuild.
Further, reconstruction unit includes:Selecting module, for the content selection neutral net according to low-resolution image
Classification;Second rebuilds module, and super-resolution rebuilding is carried out to low-resolution image for the neutral net according to selection.
Another aspect according to embodiments of the present invention, additionally provides a kind of image processing apparatus, and the device includes:Decompose single
Member, for carrying out wavelet decomposition to original image by transmitting terminal, obtains wavelet coefficient;Transmitting element, for passing through transmitting terminal
Wavelet coefficient is sent to receiving terminal;Generation unit, for generating low-resolution image according to wavelet coefficient by receiving terminal;Rebuild
Unit, carries out super-resolution rebuilding to low-resolution image according to neutral net for receiving terminal, obtains high-definition picture.
In embodiments of the present invention, the data sent by receiving end/sending end, and determine wavelet systems according to the data of reception
Number, wherein, the data that transmitting terminal is sent include the wavelet coefficient obtained to original image after wavelet decomposition;According to small echo
Coefficient generates low-resolution image;Super-resolution rebuilding is carried out to low-resolution image according to neutral net, high-resolution is obtained
Image, solving the method for compressing image in correlation technique can not be while meets higher compression efficiency and higher image restoring
The technical problem of quality, and then realize can be while take into account higher compression efficiency and the technology of higher image restoring quality
Effect.
Brief description of the drawings
Accompanying drawing described herein is used for providing a further understanding of the present invention, constitutes the part of the application, this hair
Bright schematic description and description is used to explain the present invention, does not constitute inappropriate limitation of the present invention.In the accompanying drawings:
Fig. 1 is a kind of flow chart of optional image processing method according to embodiments of the present invention;
Fig. 2 is the flow chart of another optional image processing method according to embodiments of the present invention;
Fig. 3 is a kind of flow chart of optional wavelet decomposition according to embodiments of the present invention;
Fig. 4 is a kind of flow chart of optional training neutral net according to embodiments of the present invention;
Fig. 5 is the flow chart that a kind of optional transmitting terminal according to embodiments of the present invention carries out image processing method;
Fig. 6 is the flow chart that a kind of optional receiving terminal according to embodiments of the present invention carries out image processing method;
Fig. 7 is a kind of schematic diagram of optional image processing apparatus according to embodiments of the present invention;
Fig. 8 is the schematic diagram of another optional image processing apparatus according to embodiments of the present invention;
Fig. 9 is the schematic diagram of another optional image processing apparatus according to embodiments of the present invention.
Embodiment
In order that those skilled in the art more fully understand the present invention program, below in conjunction with the embodiment of the present invention
Accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is only
The embodiment of a part of the invention, rather than whole embodiments.Based on the embodiment in the present invention, ordinary skill people
The every other embodiment that member is obtained under the premise of creative work is not made, should all belong to the model that the present invention is protected
Enclose.
It should be noted that term " first " in description and claims of this specification and above-mentioned accompanying drawing, "
Two " etc. be for distinguishing similar object, without for describing specific order or precedence.It should be appreciated that so using
Data can exchange in the appropriate case, so as to embodiments of the invention described herein can with except illustrating herein or
Order beyond those of description is implemented.In addition, term " comprising " and " having " and their any deformation, it is intended that cover
Lid is non-exclusive to be included, for example, the process, method, system, product or the equipment that contain series of steps or unit are not necessarily limited to
Those steps or unit clearly listed, but may include not list clearly or for these processes, method, product
Or the intrinsic other steps of equipment or unit.
This application provides a kind of embodiment of image processing method.
Fig. 1 is a kind of flow chart of optional image processing method according to embodiments of the present invention, as shown in figure 1, the party
Method comprises the following steps:
Step S101, the data that receiving end/sending end is sent, and determine wavelet coefficient according to the data of reception:
Optionally, the image processing method that the embodiment is provided can be in receiving terminal or decoding end execution.Receiving
After the data that transmitting terminal is sent, wavelet coefficient is determined according to the data of receiving, wherein, the data that transmitting terminal is sent include pair
The wavelet coefficient that original image is obtained after wavelet decomposition, transmitting terminal is passing through small echo to original image (image waiting for transmission)
After the size of the mode compressed file of decomposition, wavelet coefficient is obtained, optionally, the 5/3 of one-level can be used to original image
Lossless wavelet-decomposing method, after original image is decomposed by 5/3 lossless small echo as the wavelet-decomposing method of wavelet basis, is obtained
To four subband wavelet coefficients:LL subbands wavelet coefficient, LH subbands wavelet coefficient, HL subbands wavelet coefficient, HH subband wavelet systems
Number, can choose LL subband wavelet coefficients in four subband wavelet coefficients and be transmitted.
Optionally, transmitting terminal before transmitting, can be compressed coding to wavelet coefficient, for example, transmitting terminal pass through it is pre-
If compression algorithm will be compressed to the wavelet coefficient obtained after wavelet decomposition, code stream is obtained, correspondingly, is sent receiving
Hold after transmitted stream, code stream can be decoded according to default compression algorithm, obtain wavelet coefficient.Checking can be obtained, only
When the LL subband wavelet coefficients obtained to one-level wavelet decomposition are compressed, its code stream size is only the corresponding code stream of original image
40%-50%, compression efficiency lifts more than 2 times compared with prior art.
Step S102, low-resolution image is generated according to wavelet coefficient:
After the data sent according to the transmitting terminal of reception determine wavelet coefficient, low point can be generated according to wavelet coefficient
Resolution image.Wavelet coefficient is the coefficient without dimension, and the step can use when carrying out wavelet decomposition with transmitting terminal and construct phase
Wavelet coefficient is reconstructed same small echo, and low-resolution image corresponding with wavelet coefficient is obtained after reconstruct, the low resolution
Rate image is image determined by wavelet coefficient.
Step S103, carries out super-resolution rebuilding to low-resolution image according to neutral net, obtains high-definition picture:
After low-resolution image is generated according to wavelet coefficient, oversubscription is carried out to low-resolution image according to neutral net
Resolution is rebuild, and obtains high-definition picture.Neutral net is the good neutral net of training in advance.
Alternatively, the neutral net in the embodiment can have the function of study, use can be carried out updating every time,
Wherein, update method can be that transmitting terminal optimizes in the original image and low-resolution image according to transmission as sample to continuation
Training neutral net is reached after preset times, and the undated parameter of the neutral net after renewal is sent to receiving terminal, wherein, in advance
If number of times can be multiple or once, in preset times in the case of once, original image to be sent every time corresponding
The undated parameter of the neutral net after original image is trained is sent during wavelet coefficient.Receiving terminal is receiving neutral net
Undated parameter after, current neutral net is updated according to undated parameter, the neutral net updated and basis
The neutral net of renewal carries out super-resolution rebuilding to current low-resolution image.
Specifically, carrying out super-resolution rebuilding to low-resolution image according to neutral net includes:It is determined that current nerve
Network, current neutral net is the nerve net stored in the device of image processing method of the embodiment or equipment etc. is performed
Network, when first time carrying out image reconstruction, current neutral net is initial neutral net, and initial neutral net is stored in
Transmitting terminal and receiving terminal, after undated parameter is received, wherein, undated parameter is for transmitting terminal transmission according to original image to working as
Preceding neutral net optimizes the parameter of obtained neutral net, and current neutral net is carried out more according to undated parameter
Newly, the neutral net updated, carries out super-resolution rebuilding, each according to the neutral net of renewal to low-resolution image
After the undated parameter for receiving the corresponding wavelet coefficient of original image and the neutral net optimized according to original image,
Current neutral net is updated according to undated parameter, neutral net after renewal is as when being rebuild next time
Current Situation of Neural Network.
Optionally, initial neutral net can be generated in transmitting terminal or generated in receiving terminal, in life
Into after initial neutral net, initial neutral net is stored in transmitting terminal and receiving terminal.It is in initial neutral net
In the case where receiving terminal is generated, this method also includes:Wavelet decomposition is carried out to multiple sample images, obtains corresponding multiple low
Resolution ratio sample image, wherein, multiple sample images and corresponding multiple low resolution sample image composing training Sample Storehouses, instruction
Practicing Sample Storehouse includes multiple samples pair, each sample to including a sample image and corresponding low resolution sample image,
, can be by each sample to gradually inputting neutral net when being trained to neutral net, training neutral net causes will be low
High-definition picture after resolution ratio sample image input neural network reconstruction is as identical as possible with sample image, and error is less than
Preparatory condition.After neutral net is trained according to multiple sample images and corresponding multiple low resolution sample images, it will instruct
Neutral net after white silk is stored in transmitting terminal and receiving terminal as initial neutral net, wherein, according to transmitting terminal for the first time
The data of transmission carry out determining that current neutral net is initial neutral net during super-resolution rebuilding.
Optionally, neutral net, can be divided into multiple classifications, pass through different god by the difference of the content in image
Low-resolution image is rebuild through network.Specifically, it is determined that during initial neutral net, can be according to default grader pair
Content in sample image is classified, for example, being divided into word class image and picture category image, further, picture is also
The classifications such as portrait, animation, animal can be categorized as, according to different classes of image pattern to training corresponding neutral net, are made
Preferably reconstruction effect can be reached for the image of the category by obtaining each neutral net.Stored point in transmitting terminal and receiving terminal
Not Dui Yingyu multiple classifications image multiple neutral nets after, receiving terminal is it is determined that can basis after low-resolution image
The classification of the content selection neutral net of low-resolution image, is then carried out according to the neutral net of selection to low-resolution image
Super-resolution rebuilding.
The data that the embodiment is sent by receiving end/sending end, and wavelet coefficient is determined according to the data of reception, wherein, hair
The data that sending end is sent include the wavelet coefficient obtained to original image after wavelet decomposition;Generated according to wavelet coefficient low
Image in different resolution;Super-resolution rebuilding is carried out to low-resolution image according to neutral net, high-definition picture is obtained, solves
Method for compressing image in correlation technique can not meet higher compression efficiency simultaneously and the technology of higher image restoring quality
Problem, and then realize can be while take into account higher compression efficiency and the technique effect of higher image restoring quality.
Optionally, in above-mentioned steps, perform the module for receiving data and execution is handled (example to the data received
Such as, rebuild etc.) executive agent can be identical module or different modules, different modules can be arranged on
In identical equipment, can also set on different devices, as the case may be depending on, the present invention does not make specific limit to this
It is fixed.
Present invention also provides a kind of storage medium, the storage medium includes the program of storage, wherein, when program is run
Equipment where control storage medium performs the image processing method of the above embodiment of the present invention offer.
Present invention also provides a kind of processor, the processor is used for operation program, wherein, program performs this hair when running
The image processing method that bright above-described embodiment is provided.
Present invention also provides the embodiment of another image processing method.It should be noted that what the embodiment was provided
Image processing method can be performed by the transmitting terminal for needing to send original image, and this method comprises the following steps:
Step one, wavelet decomposition is carried out to original image, obtains wavelet coefficient:
Wavelet decomposition is carried out according to the small echo of default wavelet basis to original image, multiple subband wavelet coefficients are obtained, it is optional
, one of wavelet coefficient can be chosen as data waiting for transmission.
Step 2, low-resolution image is generated according to wavelet coefficient:
The embodiment provide image processing method receiving terminal can be by neutral net to receiving wavelet coefficient
The low-resolution image of determination is rebuild.
Except wavelet coefficient is transmitted to receiving terminal, the actuating station for the image processing method that the embodiment is provided can also be by
The parameter for training the neutral net after optimization according to original image is sent to receiving terminal.
After step one determination wavelet coefficient is performed, step 2 is performed, low-resolution image is generated according to wavelet coefficient.
After step 2 is performed, step 3 is performed.
Step 3, optimizes to current neutral net according to original image and low-resolution image and obtains neutral net
Undated parameter:
After low-resolution image is generated according to wavelet coefficient, using original image and corresponding low-resolution image as
Training sample pair, is optimized to current neutral net, obtains the undated parameter of the neutral net after training optimization.
The definition of current neutral net is identical with the definition in above-described embodiment, will not be repeated here.
Correspondingly, this method can also include:Wavelet decomposition is carried out to multiple sample images, corresponding multiple low points are obtained
Resolution sample image;According to multiple sample images and corresponding multiple low resolution sample image training neutral nets;Will training
Neutral net afterwards is stored in transmitting terminal and receiving terminal as initial neutral net, wherein, to first original to be sent
Beginning image determines that current neutral net is initial neutral net when optimizing.
Optionally, included according to multiple sample images and corresponding multiple low resolution sample image training neutral nets:
Low resolution sample image is classified according to the content of low resolution sample image;To the low resolution sample of each classification
Image is trained using different neutral nets, obtains the neutral net of multiple classifications.
Step 4, wavelet coefficient and undated parameter are sent to receiving terminal:
After the undated parameter and wavelet coefficient of neutral net is determined, wavelet coefficient and undated parameter are sent to connecing
Receiving end.
, specifically, can be with it should be noted that the transmission time of wavelet coefficient and undated parameter can be nonsynchronous
It is determined that first wavelet coefficient is sent to receiving terminal after wavelet coefficient, at the same according to wavelet coefficient generate low-resolution image and
The step of training sample, then undated parameter is sent to receiving terminal after undated parameter is determined, wherein, undated parameter can
To be to send to receiving terminal or reaching in optimization training to update after preset times after every suboptimization training
Parameter is sent to receiving terminal.
Optionally, sent by wavelet coefficient and undated parameter to before receiving terminal, can be according to default compression algorithm pair
The data to be sent are compressed, specifically, and wavelet coefficient and undated parameter, which are sent to receiving terminal, to be included:To wavelet coefficient
Encoded with undated parameter according to default compression algorithm, obtain code stream;Obtained code stream is sent to receiving terminal.
The image processing method that the embodiment is provided is improved from the frame structure of image coding and decoding, incorporates nerve net
Network system, makes whole algorithm frame more flexible, and stronger to network bandwidth adaptability, the nerve net in the image processing method
Network has unsupervised self-learning function, with using, and the effect of image is generated on the premise of performance is not influenceed to become better and better,
Neutral net is introduced in decoding end, image can be carried out to thinner classification, the effect of generation image is improved.
Optionally, in above-mentioned steps, perform the module for receiving data and execution is handled (example to the data received
Such as, rebuild etc.) executive agent can be identical module or different modules, different modules can be arranged on
In identical equipment, can also set on different devices, as the case may be depending on, the present invention does not make specific limit to this
It is fixed.
Present invention also provides a kind of storage medium, the storage medium includes the program of storage, wherein, when program is run
Equipment where control storage medium performs the image processing method of the above embodiment of the present invention offer.
Present invention also provides a kind of processor, the processor is used for operation program, wherein, program performs this hair when running
The image processing method that bright above-described embodiment is provided.
Fig. 2 is the flow chart of another optional image processing method according to embodiments of the present invention, and the embodiment is provided
Image processing method can be performed by a kind of image processing system, the image processing system is carried including the embodiment of the present invention
The transmitting terminal and receiving terminal of confession, as shown in Fig. 2 this method comprises the following steps:
Step S201, transmitting terminal carries out wavelet decomposition to original image, obtains wavelet coefficient:
Transmitting terminal carries out wavelet decomposition to sent original image according to the small echo of default wavelet basis, obtains multiple subbands
Wavelet coefficient, optionally, can choose one of wavelet coefficient as data waiting for transmission.
Step S202, transmitting terminal sends wavelet coefficient to receiving terminal:
After transmitting terminal obtains wavelet coefficient, wavelet coefficient is sent to receiving terminal, optionally, transmitting terminal can sent
First data are compressed before data, the data after compression are sent to receiving terminal.
Step S203, receiving terminal generates low-resolution image according to wavelet coefficient:
Receiving terminal is after wavelet coefficient is obtained, and receiving terminal generates low-resolution image according to wavelet coefficient.Wavelet coefficient
It is the coefficient without dimension, the step, which can be used, constructs identical small echo to wavelet coefficient when carrying out wavelet decomposition with transmitting terminal
It is reconstructed, low-resolution image corresponding with wavelet coefficient is obtained after reconstruct, the low-resolution image is wavelet coefficient institute
The image of determination.
Step S204, receiving terminal carries out super-resolution rebuilding according to neutral net to low-resolution image, obtains high-resolution
Rate image:
After low-resolution image is generated according to wavelet coefficient, oversubscription is carried out to low-resolution image according to neutral net
Resolution is rebuild, and obtains high-definition picture.
Alternatively, the neutral net of transmitting terminal and receiving terminal can optimize renewal, and transmitting terminal can be according to sending every time
Original image and the corresponding low-resolution image of original image are trained as sample to the neutral net continued to transmitting terminal,
The parameter of the optimization neural network obtained after retraining preset times is sent to receiving terminal, receiving terminal can be according to acquisition
Parameter is updated to the neutral net of receiving terminal.
Specifically, wavelet decomposition is carried out to original image in transmitting terminal, after obtaining wavelet coefficient, this method can be with
Including:Transmitting terminal generates low-resolution image according to wavelet coefficient;Transmitting terminal is according to original image and low-resolution image to hair
The current neutral net of sending end is optimized;After optimization in transmitting terminal to neutral net Jing Guo preset times, transmitting terminal to
Receiving terminal sends undated parameter, wherein, undated parameter is the ginseng of the neutral net after transmitting terminal optimizes by preset times
Number;Receiving terminal is updated to obtain more according to undated parameter after undated parameter is received to the neutral net of receiving terminal
New neutral net, and the neutral net based on renewal is to the subsequently received corresponding low-resolution image progress of wavelet coefficient
Super-resolution rebuilding.
Alternatively, wavelet decomposition is carried out to original image in transmitting terminal, obtained before wavelet coefficient, this method can also be wrapped
Include:Transmitting terminal or receiving terminal carry out wavelet decomposition to multiple sample images, obtain corresponding multiple low resolution sample images;Hair
Sending end or receiving terminal train neutral net according to multiple sample images and corresponding multiple low resolution sample images;After training
Neutral net as initial neutral net be stored in transmitting terminal and receiving terminal.
Alternatively, transmitting terminal or receiving terminal are trained according to multiple sample images and corresponding multiple low resolution sample images
Neutral net can include:Transmitting terminal or receiving terminal enter according to the content of low resolution sample image to low resolution sample image
Row classification;The low resolution sample image of each classification is trained using corresponding neutral net, multiple classifications are obtained
Neutral net.
Alternatively, transmitting terminal sends wavelet coefficient to receiving terminal and can included:Transmitting terminal is to wavelet coefficient according to default pressure
Compression algorithm is encoded, and obtains code stream;Transmitting terminal sends obtained code stream to receiving terminal, wherein, receiving terminal is receiving code
Decoded after stream according to default compression algorithm, obtain wavelet coefficient.
Present invention also provides a kind of storage medium, the storage medium includes the program of storage, wherein, when program is run
Equipment where control storage medium performs the image processing method of the above embodiment of the present invention offer.
Present invention also provides a kind of processor, the processor is used for operation program, wherein, program performs this hair when running
The image processing method that bright above-described embodiment is provided.
It is of the invention by the image processing method under a kind of concrete application scene as a kind of alternative embodiment of above-described embodiment
The step of method, is described as follows:
As shown in figure 3, the process that the image processing method in the embodiment carries out wavelet decomposition to original image is:First
By whole original image x (n1, n2) decomposed using 5/3 small echo of lifting, it is low to row data acquisition by 5/3 boosting algorithm
Frequency signal and by 5/3 boosting algorithm to row data acquisition high-frequency signal, row data low frequency signal is used and adopted under 2 times of column data
Sample obtains x1(n1, n2/ 2), x is obtained using 2 times of down-samplings of column data to row data high-frequency signal2(n1, n2/ 2), to x1(n1, n2/
2) low frequency signal is obtained to column data using 5/3 boosting algorithm, obtains LL subband wavelet coefficients xLL{n1/ 2, n2/ 2 }, to x1(n1,
n2/ high-frequency signal 2) is obtained to column data using 5/3 boosting algorithm, obtain LH subband wavelet coefficients xLH{n1/ 2, n2/ 2 }, to x2
(n1, n2/ low frequency signal 2) is obtained to column data using 5/3 boosting algorithm, obtain HL subband wavelet coefficients xHL{n1/ 2, n2/ 2 },
To x2(n1, n2/ high-frequency signal 2) is obtained to column data using 5/3 boosting algorithm, obtain HH subband wavelet coefficients xHH{n1/ 2, n2/
2 }, that is, tetra- sub-band informations of LL, LH, HL and HH can be obtained after wavelet decomposition, wherein, L represents to obtain after LPF
Low frequency signal, H represents the high-frequency signal after high-pass filtering.By this isolation a sub-picture divide in order to it is multiple compared with
Small image information.Then coding is compressed to the image after wavelet transformation and is transmitted.Utilize this method for compressing image one
Aspect can realize the Lossless Compression of picture, be on the one hand adapted to different network environments and realize Delamination Transmission.
As shown in figure 4, the image processing method in the embodiment realizes the super-resolution of image by artificial neural network
The process of reconstruction is as follows:(1) the larger clearly high-resolution sample image of quantity is chosen as training sample, to each sample
The wavelet decomposition of imagery exploitation 5/3, progress down-sampling and LPF, obtain low-resolution image, obtain sample pair, each sample
To including an original image and corresponding low-resolution image, sample is to for training neural network parameter;(2) sample is utilized
To training neutral net under the conditions of unsupervised, neutral net is depth convolutional neural networks, wherein, during training neutral net
Constraints is:Corresponding high-definition picture can be obtained after ensureing low-resolution image input depth convolutional neural networks,
And the high-definition picture meets error constraints condition with original sample image similarity degree, specifically, low-resolution image
By deconvolution output image after into depth convolutional neural networks, the image of output is high-definition picture, by output
High-definition picture is contrasted with original high-resolution sample image, calculation error, corrective networks, then by next sample
This carries out training next time to input depth convolutional neural networks;Optionally, it can be divided according to the content difference of sample image
Multiple neutral nets, the picture material of one classification of each neutral net correspondence, for example, being divided sample image are not trained
Class, judgement is character image or picture, and different classes of image is trained using different neutral nets;(3) pass through
After substantial amounts of sample training, the word neutral net and picture neutral net that can be trained, can after training is completed
To choose the low resolution picture with sample same type, whether test neutral net can generate corresponding high score using the picture
Resolution picture.By above step, that is, the training of the neutral net for image super-resolution rebuilding is completed, to network parameter
Preserve, two neutral nets trained are preserved in the form of data in coding side (transmitting terminal) and decoding end (receiving terminal).
Follow-up in use, the neutral net that can train this is used as " flight data recorder ", to realize the super of low-resolution image
Resolution reconstruction.
As shown in figure 5, the process of the image processing method of coding side (transmitting terminal) is as follows:
The original image that will be sent does level of decomposition using 5/3 small echo, low-resolution image is obtained, by low resolution figure
As being compressed coding, and transmission code stream to LL low frequency sub-bands using existing image compression algorithm.
After level of decomposition, acquisition low-resolution image are being done using 5/3 small echo, always according to original image and low resolution
Image is updated to the neutral net of coding side, specifically, low-resolution image input grader is differentiated, judges class
Type, it is determined that being character image or picture, if character image, then calls the neutral net of character image to character image
Carry out image super-resolution rebuilding and optimize word neutral net according to reconstructed results, if picture, then call picture
The neutral net of image carries out image super-resolution rebuilding to picture and optimizes picture neutral net according to reconstructed results,
Optimization is reached after preset times, the parameter of the neutral net after optimization is sent to decoding end, decoding end is receiving renewal
The neutral net of decoding end storage can be updated after parameter.
As shown in fig. 6, transmitting terminal is it is determined that after 5/3 wavelet decomposition low-resolution image, be compressed coding transmission extremely
Receiving terminal, the process of the image processing method of decoding end (receiving terminal) is as follows:
Receiving terminal is decoded to data, obtains low-resolution image, and input grader is differentiated, the grader can be with
It is the grader identical grader with transmitting terminal, judges after being character image or picture, call corresponding nerve
Network carries out image super-resolution rebuilding, generates high-definition picture, output image.Wherein, the neutral net of decoding end can be with
Update, specifically, transmitting terminal is after undated parameter is sent, current neutral net is updated by decoding end.
Optionally, in the image processing method that the embodiment is provided, using 5/3 small echo as wavelet basis tectonic network,
In actual applications, also can be using other wavelet basis come tectonic network, for example, Haar small echos or 9/7 small echo etc..
Image processing method provided in an embodiment of the present invention can write program using C language or C Plus Plus, wherein, C languages
The better performances of speech program, C++ code organization is relatively good but operational efficiency is less than C programmer, and use C language preferably is entered
Row programming.
The image processing method that the embodiment is provided is by the characteristics of wavelet decomposition and deep neural network image super-resolution
It is combined, it is proposed that a kind of technical scheme of new image compressing transmission, Sample Storehouse is generated using the method for 5/3 wavelet decomposition,
Sample Storehouse preliminary classification is character image and the big classification of picture two, trains deep neural network using Sample Storehouse, is answered
Two neutral nets for character image and picture;Then, in transmitting terminal to the wavelet decomposition of imagery exploitation 5/3, by original
The LL subbands wavelet coefficient of figure carries out coding transmission;Finally, in receiving terminal according to image category, word or picture are utilized respectively
Neutral net carries out super-resolution rebuilding to picture, obtains high-definition picture and exports.
Traditional picture compression method has been difficult to have greatly improved on the basis of existing algorithm, and the embodiment is carried
This method for compressing image based on neural network image super-resolution that the image processing method of confession is proposed, is inherently provided
New compression of images thinking, can significantly lift compression efficiency.Passed using the low-resolution image of one-level wavelet decomposition
, can be more than 2 times by original picture compression improved efficiency when defeated.The program can utilize the spy of the unsupervised self study of neutral net
Point, the optimization of invisible completion neutral net, constantly lifts network performance in image encoding process, makes generation picture quality not
Disconnected lifting.
Checking can be obtained, and when the LL subband wavelet coefficients only obtained to one-level wavelet decomposition are compressed, its code stream size is only
For the 40%~50% of original bit stream, compression efficiency lifts more than 2 times compared with prior art, and the high-definition picture of decoding generation with
Original image difference less, disclosure satisfy that vision needs, in addition, the program make use of the unsupervised learning feature of neutral net, make
Picture quality can be lifted constantly during use so that the Neural Network Self-learning optimization of decoding end, system
Performance it is elongated and constantly improve with use time.
The image processing method that the embodiment is provided can be applied in computer screen picture transmission, and screen-picture transmission can
Applied to scenes such as video conference, remote desktops.These application scenarios are higher to network bandwidth requirement, and the size of bandwidth will be direct
Influence the quality of image and the smooth degree of picture.Therefore, the image processing method of the program utilizes wavelet transformation and depth god
The characteristics of through network, proposes a kind of new compression of images encoding and decoding transmission method, can be by original compression using this method
Efficiency improves more than 2 times.
It should be noted that accompanying drawing flow chart though it is shown that logical order, but in some cases, can be with
Shown or described step is performed different from order herein.
Present invention also provides a kind of embodiment of image processing apparatus.The device can be disposed on going back image
The device of former one end, can be used for performing the image processing method provided in an embodiment of the present invention for reducing image
Method.
Fig. 7 is a kind of schematic diagram of optional image processing apparatus according to embodiments of the present invention, as shown in fig. 7, the dress
Put including receiving unit 10, generation unit 20 and reconstruction unit 30, wherein, receiving unit is used for the number that receiving end/sending end is sent
According to, and wavelet coefficient is determined according to the data of reception, wherein, the data that transmitting terminal is sent include passing through small echo to original image
The wavelet coefficient obtained after decomposition;Generation unit is used to generate low-resolution image according to wavelet coefficient;Reconstruction unit is used for root
Super-resolution rebuilding is carried out to low-resolution image according to neutral net, high-definition picture is obtained.
The embodiment is by receiving unit, the data sent for receiving end/sending end, and is determined according to the data of reception small
Wave system number, wherein, the data that transmitting terminal is sent include the wavelet coefficient obtained to original image after wavelet decomposition;Generation
Unit, for generating low-resolution image according to wavelet coefficient;Reconstruction unit, for according to neutral net to low-resolution image
Super-resolution rebuilding is carried out, high-definition picture is obtained, solving the method for compressing image in correlation technique can not be while meets
The technical problem of higher compression efficiency and higher image restoring quality, and then realize can be while take into account higher compression
The technique effect of efficiency and higher image restoring quality.
Further, reconstruction unit includes:Determining module, for determining current neutral net;First receiving module, is used
In reception undated parameter, wherein, undated parameter is the excellent to current neutral net progress according to original image of transmitting terminal transmission
Change the parameter of obtained neutral net;Update module, for being updated according to undated parameter to current neutral net, is obtained
The neutral net of renewal;First rebuilds module, and super-resolution is carried out to low-resolution image for the neutral net according to renewal
Rebuild.
Further, the device also includes:Resolving cell, for carrying out wavelet decomposition to multiple sample images, is obtained pair
The multiple low resolution sample images answered;Training unit, for according to multiple sample images and corresponding multiple low resolution samples
This image trains neutral net;Memory cell, for being stored in hair using the neutral net after training as initial neutral net
Sending end and receiving terminal, wherein, determine current god when the data sent for the first time according to transmitting terminal carry out super-resolution rebuilding
It is initial neutral net through network.
Further, reconstruction unit includes:Selecting module, for the content selection neutral net according to low-resolution image
Classification;Second rebuilds module, and super-resolution rebuilding is carried out to low-resolution image for the neutral net according to selection.
Further, receiving unit includes:Second receiving module, for receiving end/sending end transmitted stream;Decoder module,
For being decoded according to default compression algorithm to code stream, wavelet coefficient is obtained, wherein, default compression algorithm is transmitting terminal to warp
Cross the compression algorithm used when the wavelet coefficient obtained after wavelet decomposition is compressed.
Fig. 8 is the schematic diagram of another optional image processing apparatus according to embodiments of the present invention, and the device can be
The device of the one end being compressed to image is arranged at, can be used for performing provided in an embodiment of the present invention be used for image progress
The image processing method of compression, as shown in figure 8, the device includes resolving cell 11, generation unit 21, optimization unit 31 and transmission
Unit 41, wherein, resolving cell is used to carry out wavelet decomposition to original image, obtains wavelet coefficient;Generation unit is used for basis
Wavelet coefficient generates low-resolution image;Optimizing unit is used for according to original image and low-resolution image to current nerve net
Network optimizes the undated parameter for obtaining neutral net;Transmitting element is used to send wavelet coefficient and undated parameter to reception
End.
Further, resolving cell is additionally operable to carry out wavelet decomposition to multiple sample images, obtains corresponding multiple low points
Resolution sample image, the device also includes:Training unit, for according to multiple sample images and corresponding multiple low resolution samples
This image trains neutral net;Memory cell, for being stored in hair using the neutral net after training as initial neutral net
Sending end and receiving terminal, wherein, determine that current neutral net is first when being optimized to first original image to be sent
The neutral net of beginning.
Further, training unit includes:Sort module, is differentiated for the content according to low resolution sample image to low
Rate sample image is classified;Training module, different nerve nets is used for the low resolution sample image to each classification
Network is trained, and obtains the neutral net of multiple classifications.
Further, transmitting element includes:Decoder module, for being calculated according to default compression wavelet coefficient and undated parameter
Method is encoded, and obtains code stream;Sending module, for obtained code stream to be sent to receiving terminal.
Fig. 9 is the schematic diagram of another optional image processing apparatus according to embodiments of the present invention, and the device can be used
In the image processing method for performing the corresponding embodiment offers of Fig. 2 of the present invention, as shown in figure 9, the device includes resolving cell 12,
Transmitting element 22, generation unit 32 and reconstruction unit 42, wherein, resolving cell is used for small to original image progress by transmitting terminal
Wave Decomposition, obtains wavelet coefficient;Transmitting element is used to send wavelet coefficient to receiving terminal by transmitting terminal;Generation unit is used to lead to
Cross receiving terminal and low-resolution image is generated according to wavelet coefficient;Reconstruction unit is used for receiving terminal according to neutral net to low resolution
Image carries out super-resolution rebuilding, obtains high-definition picture.
Above-mentioned device can include processor and memory, and said units can be stored in storage as program unit
In device, corresponding function is realized by the said procedure unit of computing device storage in memory.
Memory potentially includes the volatile memory in computer-readable medium, random access memory (RAM) and/
Or the form, such as read-only storage (ROM) or flash memory (flash RAM) such as Nonvolatile memory, memory is deposited including at least one
Store up chip.
The order of above-mentioned the embodiment of the present application does not represent the quality of embodiment.
In above-described embodiment of the application, the description to each embodiment all emphasizes particularly on different fields, and does 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 appreciated that
Arrive, disclosed technology contents can be realized by another way.
Wherein, device embodiment described above is only schematical, such as division of described unit, can be one
Kind of division of logic function, can there is other dividing mode when actually realizing, such as multiple units or component can combine or
Another system is desirably integrated into, or some features can be ignored, or do not perform.It is another, it is shown or discussed it is mutual it
Between coupling or direct-coupling or communication connection can be the INDIRECT COUPLING or communication link of unit or module by some interfaces
Connect, can be electrical or other forms.
In addition, each functional unit in the application each embodiment can be integrated in a processing unit, can also
That unit is individually physically present, can also two or more units it is integrated in a unit.Above-mentioned integrated list
Member can both be realized in the form of hardware, it would however also be possible to employ the form of SFU software functional unit is realized.
If the integrated unit is realized using in the form of SFU software functional unit and as independent production marketing or used
When, it can be stored in a computer read/write memory medium.Understood based on such, the technical scheme of the application is substantially
The part contributed in other words to prior art or all or part of the technical scheme can be in the form of software products
Embody, the computer software product is stored in a storage medium, including some instructions are to cause a computer
Equipment (can for personal computer, server or network equipment etc.) perform the application each embodiment methods described whole or
Part steps.And foregoing storage medium includes:USB flash disk, read-only storage (ROM, Read-Only Memory), arbitrary access are deposited
Reservoir (RAM, Random Access Memory), mobile hard disk, magnetic disc or CD etc. are various can be with store program codes
Medium.
Described above is only the preferred embodiment of the application, it is noted that for the ordinary skill people of the art
For member, on the premise of the application principle is not departed from, some improvements and modifications can also be made, these improvements and modifications also should
It is considered as the protection domain of the application.
Claims (13)
1. a kind of image processing method, it is characterised in that including:
The data that receiving end/sending end is sent, and wavelet coefficient is determined according to the data of reception, wherein, the number that the transmitting terminal is sent
According to the wavelet coefficient for including obtaining original image after wavelet decomposition;
Low-resolution image is generated according to the wavelet coefficient;
Super-resolution rebuilding is carried out to the low-resolution image according to neutral net, high-definition picture is obtained.
2. according to the method described in claim 1, it is characterised in that the low-resolution image is surpassed according to neutral net
Resolution reconstruction includes:
It is determined that current neutral net;
Undated parameter is received, wherein, the undated parameter is being worked as according to the original image to described for transmitting terminal transmission
Preceding neutral net optimizes the parameter of obtained neutral net;
The current neutral net is updated according to the undated parameter, the neutral net updated;
Super-resolution rebuilding is carried out to the low-resolution image according to the neutral net of the renewal.
3. method according to claim 2, it is characterised in that methods described also includes:
Wavelet decomposition is carried out to multiple sample images, corresponding multiple low resolution sample images are obtained;
According to the multiple sample image and corresponding multiple low resolution sample image training neutral nets;
The transmitting terminal and receiving terminal are stored in using the neutral net after training as initial neutral net, wherein, in basis
The data that the transmitting terminal is sent for the first time carry out determining that the current neutral net is described initial during super-resolution rebuilding
Neutral net.
4. according to the method described in claim 1, it is characterised in that the low-resolution image is surpassed according to neutral net
Resolution reconstruction includes:
According to the classification of the content selection neutral net of the low-resolution image;
Super-resolution rebuilding is carried out to the low-resolution image according to the neutral net of selection.
5. a kind of image processing method, it is characterised in that including:
Transmitting terminal carries out wavelet decomposition to original image, obtains wavelet coefficient;
The transmitting terminal sends the wavelet coefficient to receiving terminal;
The receiving terminal generates low-resolution image according to the wavelet coefficient;
The receiving terminal carries out super-resolution rebuilding according to neutral net to the low-resolution image, obtains high resolution graphics
Picture.
6. method according to claim 5, it is characterised in that wavelet decomposition is carried out to original image in transmitting terminal, obtained
After wavelet coefficient, methods described also includes:
The transmitting terminal generates low-resolution image according to the wavelet coefficient;
The transmitting terminal enters according to the original image and the low-resolution image neutral net current to the transmitting terminal
Row optimization;
After optimization in the transmitting terminal to neutral net Jing Guo preset times, the transmitting terminal is sent more to the receiving terminal
New parameter, wherein, the undated parameter is the parameter of the neutral net after the transmitting terminal optimizes by preset times;
The receiving terminal is after the undated parameter is received, according to neutral net of the undated parameter to the receiving terminal
The neutral net for being updated to be updated, and the neutral net based on the renewal is to subsequently received wavelet coefficient pair
The low-resolution image answered carries out super-resolution rebuilding.
7. method according to claim 5, it is characterised in that wavelet decomposition is carried out to original image in transmitting terminal, obtained
Before wavelet coefficient, methods described also includes:
The transmitting terminal or the receiving terminal carry out wavelet decomposition to multiple sample images, obtain corresponding multiple low resolution samples
This image;
The transmitting terminal or the receiving terminal are according to the multiple sample image and corresponding multiple low resolution sample graphs
As training neutral net;
The transmitting terminal and the receiving terminal are stored in using the neutral net after training as initial neutral net.
8. method according to claim 7, it is characterised in that the transmitting terminal or the receiving terminal are according to the multiple sample
This image and corresponding multiple low resolution sample image training neutral nets include:
The transmitting terminal or the receiving terminal are according to the content of the low resolution sample image to the low resolution sample graph
As being classified;
The low resolution sample image of each classification is trained using corresponding neutral net, the nerve of multiple classifications is obtained
Network.
9. method according to claim 5, it is characterised in that the transmitting terminal sends the wavelet coefficient bag to receiving terminal
Include:
The transmitting terminal is encoded to the wavelet coefficient according to default compression algorithm, obtains code stream;
The transmitting terminal sends the obtained code stream to the receiving terminal, wherein, the receiving terminal is receiving the code
Decoded after stream according to the default compression algorithm, obtain the wavelet coefficient.
10. a kind of image processing apparatus, it is characterised in that including:
Receiving unit, the data sent for receiving end/sending end, and wavelet coefficient is determined according to the data of reception, wherein, it is described
The data that transmitting terminal is sent include the wavelet coefficient obtained to original image after wavelet decomposition;
Generation unit, for generating low-resolution image according to the wavelet coefficient;
Reconstruction unit, for carrying out super-resolution rebuilding to the low-resolution image according to neutral net, obtains high-resolution
Image.
11. device according to claim 10, it is characterised in that the reconstruction unit includes:
Determining module, for determining current neutral net;
First receiving module, for receiving undated parameter, wherein, the undated parameter is described in the basis that the transmitting terminal is sent
Original image optimizes the parameter of obtained neutral net to the current neutral net;
Update module, for being updated according to the undated parameter to the current neutral net, the nerve updated
Network;
First rebuilds module, and Super-resolution reconstruction is carried out to the low-resolution image for the neutral net according to the renewal
Build.
12. device according to claim 10, it is characterised in that the reconstruction unit includes:
Selecting module, the classification for the content selection neutral net according to the low-resolution image;
Second rebuilds module, and super-resolution rebuilding is carried out to the low-resolution image for the neutral net according to selection.
13. a kind of image processing apparatus, it is characterised in that including:
Resolving cell, for carrying out wavelet decomposition to original image by transmitting terminal, obtains wavelet coefficient;
Transmitting element, for sending the wavelet coefficient to receiving terminal by the transmitting terminal;
Generation unit, for generating low-resolution image according to the wavelet coefficient by the receiving terminal;
Reconstruction unit, carries out super-resolution rebuilding to the low-resolution image according to neutral net for the receiving terminal, obtains
To high-definition picture.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710326762.4A CN107087201B (en) | 2017-05-10 | 2017-05-10 | Image processing method and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710326762.4A CN107087201B (en) | 2017-05-10 | 2017-05-10 | Image processing method and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107087201A true CN107087201A (en) | 2017-08-22 |
CN107087201B CN107087201B (en) | 2019-11-08 |
Family
ID=59611522
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710326762.4A Active CN107087201B (en) | 2017-05-10 | 2017-05-10 | Image processing method and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107087201B (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108197488A (en) * | 2017-12-25 | 2018-06-22 | 大国创新智能科技(东莞)有限公司 | Information hiding, extracting method and system based on big data and neural network |
CN108564546A (en) * | 2018-04-18 | 2018-09-21 | 厦门美图之家科技有限公司 | Model training method, device and photo terminal |
CN108876864A (en) * | 2017-11-03 | 2018-11-23 | 北京旷视科技有限公司 | Image coding, coding/decoding method, device, electronic equipment and computer-readable medium |
CN108960425A (en) * | 2018-07-05 | 2018-12-07 | 广东工业大学 | A kind of rending model training method, system, equipment, medium and rendering method |
CN111567056A (en) * | 2018-01-04 | 2020-08-21 | 三星电子株式会社 | Video playing device and control method thereof |
CN112399120A (en) * | 2019-08-14 | 2021-02-23 | 三星电子株式会社 | Electronic device and control method thereof |
WO2023174144A1 (en) * | 2022-03-17 | 2023-09-21 | 阿里巴巴(中国)有限公司 | Super-resolution reconstruction method and apparatus for cloud desktop image, device, and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106067161A (en) * | 2016-05-24 | 2016-11-02 | 深圳市未来媒体技术研究院 | A kind of method that image is carried out super-resolution |
CN106204447A (en) * | 2016-06-30 | 2016-12-07 | 北京大学 | The super resolution ratio reconstruction method with convolutional neural networks is divided based on total variance |
CN106228512A (en) * | 2016-07-19 | 2016-12-14 | 北京工业大学 | Based on learning rate adaptive convolutional neural networks image super-resolution rebuilding method |
US20170003368A1 (en) * | 2015-07-02 | 2017-01-05 | The General Hospital Corporation | System and Method For High Resolution Diffusion Imaging |
-
2017
- 2017-05-10 CN CN201710326762.4A patent/CN107087201B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20170003368A1 (en) * | 2015-07-02 | 2017-01-05 | The General Hospital Corporation | System and Method For High Resolution Diffusion Imaging |
CN106067161A (en) * | 2016-05-24 | 2016-11-02 | 深圳市未来媒体技术研究院 | A kind of method that image is carried out super-resolution |
CN106204447A (en) * | 2016-06-30 | 2016-12-07 | 北京大学 | The super resolution ratio reconstruction method with convolutional neural networks is divided based on total variance |
CN106228512A (en) * | 2016-07-19 | 2016-12-14 | 北京工业大学 | Based on learning rate adaptive convolutional neural networks image super-resolution rebuilding method |
Non-Patent Citations (2)
Title |
---|
胥妍: ""图像超分辨率重建算法研究"", 《中国博士学位论文全文数据库》 * |
郭丙华 等: ""小波去噪和神经网络相融合的超分辨率图像重建"", 《激光杂志》 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108876864A (en) * | 2017-11-03 | 2018-11-23 | 北京旷视科技有限公司 | Image coding, coding/decoding method, device, electronic equipment and computer-readable medium |
CN108876864B (en) * | 2017-11-03 | 2022-03-08 | 北京旷视科技有限公司 | Image encoding method, image decoding method, image encoding device, image decoding device, electronic equipment and computer readable medium |
CN108197488B (en) * | 2017-12-25 | 2020-04-14 | 大国创新智能科技(东莞)有限公司 | Information hiding and extracting method and system based on big data and neural network |
CN108197488A (en) * | 2017-12-25 | 2018-06-22 | 大国创新智能科技(东莞)有限公司 | Information hiding, extracting method and system based on big data and neural network |
US11457273B2 (en) | 2018-01-04 | 2022-09-27 | Samsung Electronics Co., Ltd. | Video playback device and control method thereof |
CN111567056A (en) * | 2018-01-04 | 2020-08-21 | 三星电子株式会社 | Video playing device and control method thereof |
US11831948B2 (en) | 2018-01-04 | 2023-11-28 | Samsung Electronics Co., Ltd. | Video playback device and control method thereof |
CN111567056B (en) * | 2018-01-04 | 2022-10-14 | 三星电子株式会社 | Video playing device and control method thereof |
CN108564546B (en) * | 2018-04-18 | 2020-08-04 | 厦门美图之家科技有限公司 | Model training method and device and photographing terminal |
CN108564546A (en) * | 2018-04-18 | 2018-09-21 | 厦门美图之家科技有限公司 | Model training method, device and photo terminal |
CN108960425B (en) * | 2018-07-05 | 2022-04-19 | 广东工业大学 | Rendering model training method, system, equipment, medium and rendering method |
CN108960425A (en) * | 2018-07-05 | 2018-12-07 | 广东工业大学 | A kind of rending model training method, system, equipment, medium and rendering method |
US11574385B2 (en) | 2019-08-14 | 2023-02-07 | Samsung Electronics Co., Ltd. | Electronic apparatus and control method for updating parameters of neural networks while generating high-resolution images |
CN112399120B (en) * | 2019-08-14 | 2023-09-19 | 三星电子株式会社 | Electronic device and control method thereof |
CN112399120A (en) * | 2019-08-14 | 2021-02-23 | 三星电子株式会社 | Electronic device and control method thereof |
WO2023174144A1 (en) * | 2022-03-17 | 2023-09-21 | 阿里巴巴(中国)有限公司 | Super-resolution reconstruction method and apparatus for cloud desktop image, device, and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN107087201B (en) | 2019-11-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107087201A (en) | Image processing method and device | |
CN107689034B (en) | Denoising method and denoising device | |
US10834415B2 (en) | Devices for compression/decompression, system, chip, and electronic device | |
Toderici et al. | Variable rate image compression with recurrent neural networks | |
US11861809B2 (en) | Electronic apparatus and image processing method thereof | |
CN110751597B (en) | Video super-resolution method based on coding damage repair | |
CN107172428B (en) | The transmission method of image, device and system | |
CN108038546A (en) | Method and apparatus for compressing neutral net | |
WO2020238439A1 (en) | Video quality-of-service enhancement method under restricted bandwidth of wireless ad hoc network | |
CN112053408A (en) | Face image compression method and device based on deep learning | |
CN115409716B (en) | Video processing method, device, storage medium and equipment | |
CN110880194A (en) | Image compression method based on convolutional neural network | |
CN113947136A (en) | Image compression and classification method and device and electronic equipment | |
CN110189385B (en) | Model training and picture compression and decompression method and device, medium and electronic equipment | |
CN114979672A (en) | Video encoding method, decoding method, electronic device, and storage medium | |
CN113225590B (en) | Video super-resolution enhancement method and device, computer equipment and storage medium | |
CN108171325A (en) | Sequential integrated network, code device and the decoding apparatus that a kind of multiple dimensioned face restores | |
CN115941966B (en) | Video compression method and electronic equipment | |
CN108230253A (en) | Image recovery method, device, electronic equipment and computer storage media | |
CN115690238A (en) | Image generation and model training method, device, equipment and storage medium | |
CN114170082A (en) | Video playing method, image processing method, model training method, device and electronic equipment | |
US20220164934A1 (en) | Image processing method and apparatus, device, video processing method and storage medium | |
CN113592723B (en) | Video enhancement method and device, electronic equipment and storage medium | |
Xiong et al. | Deep feature compression with collaborative coding of image texture | |
Mahajan et al. | PixelPyramids: Exact Inference Models from Lossless Image Pyramids |
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 |