CN109064399A - Image super-resolution rebuilding method and system, computer equipment and its storage medium - Google Patents
Image super-resolution rebuilding method and system, computer equipment and its storage medium Download PDFInfo
- Publication number
- CN109064399A CN109064399A CN201810803433.9A CN201810803433A CN109064399A CN 109064399 A CN109064399 A CN 109064399A CN 201810803433 A CN201810803433 A CN 201810803433A CN 109064399 A CN109064399 A CN 109064399A
- Authority
- CN
- China
- Prior art keywords
- image
- label value
- super
- split
- image data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformation in the plane of the image
- G06T3/40—Scaling the whole image or part thereof
- G06T3/4053—Super resolution, i.e. output image resolution higher than sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
Abstract
The present invention relates to a kind of image super-resolution rebuilding methods and system, computer equipment and its storage medium, belong to technical field of image processing.Described image super resolution ratio reconstruction method, comprising: low resolution image to be reconstructed is split according to scene content, and the label value of the image data in each split sence region is respectively set；The image data in each split sence region and its corresponding label value are subjected to split and obtain the input data of deep learning network；The input data is input to deep learning network and carries out parameter learning, and super-resolution image is reconstructed according to the network parameter that study obtains.The technical solution solves the problems, such as the relevant details for being difficult to accurately recover scene content different in image in the prior art, can accurately recover the relevant details of scene content different in image, improve the picture quality of reconstruction.
Description
Technical field
The present invention relates to technical field of image processing, more particularly to a kind of image super-resolution rebuilding method and system,
Computer equipment and its storage medium.
Background technique
Image super-resolution rebuilding (or reconstruct), refers to from the image or video sequence of a width low resolution and recovers high score
The image of rate.Conventional solution mainly uses sparse coding to realize the reconstruction of super-resolution, respectively from low-resolution image
Collection and high resolution graphics image set learn two sparse dictionary collection and its reflect relationship, and low-resolution image is mapped to low resolution
In dictionary, corresponding high-resolution dictionary coefficient is obtained by dictionary mapping relations, and then reconstruct high-resolution image；Separately
Outside, low resolution is learnt by depth convolutional network using depth convolution kernel as dictionary with the appearance of deep learning technology
Image reconstructs high-resolution image to the mapping between high-definition picture, achieves higher than sparse coding method
Effect.
But in realizing process of the present invention, inventor has found in technology as described above that at least there are the following problems: in needle
When to image super-resolution reconstruct, after different scene contents carries out super-resolution, the details for needing to restore is not identical.For example,
Smooth metope and the details that mixed and disorderly meadow needs to embody in Super-resolution Reconstruction are also not exactly the same, and the prior art is difficult to
The relevant details for accurately recovering scene content different in image affect the picture quality of reconstruction.
Summary of the invention
Based on this, it is necessary to for being difficult to accurately recover asking for the relevant details of scene content different in image
Topic provides a kind of image super-resolution rebuilding method and system.
A kind of image super-resolution rebuilding method, includes the following steps:
Low resolution image to be reconstructed is split according to scene content, and each split sence region is respectively set
The label value of image data；
The image data in each split sence region and its corresponding label value are subjected to split and obtain deep learning network
Input data；
The input data is input to deep learning network and carries out parameter learning, and the network parameter obtained according to study
Reconstruct super-resolution image.
Above-mentioned image super-resolution rebuilding method is first split low resolution image to be reconstructed according to scene content
And label value is set, it is then added to the scene information in image as priori in super-resolution deep learning network, in turn
Reconstruct super-resolution image；The technical solution institute learning parameter can adapt to different scene contents, can be preferably to figure
As carrying out super-resolution rebuilding, the relevant details of scene content different in image are accurately recovered, the figure of reconstruction is improved
Image quality amount.
In one embodiment, the step of label value of the image data that each split sence region is respectively set packet
It includes:
For the type of various scene contents, the corresponding label value of various types institute is set；
Corresponding label value is obtained according to the type of the image data in each split sence region；
The label value got is written in the image data in the split sence region.
In one embodiment, described to spell the image data in each split sence region and its corresponding label value
Closing the step of obtaining the input data of deep learning network includes:
The RGB triple channel image data in each split sence region obtained；
The label value L in split sence region where obtaining the RGB triple channel image data；
The image data RGBL that the RGB triple channel image data and label value L are pieced together to four-way, as depth
Practise the input data of network.
In one embodiment, the deep learning network is convolutional neural networks.
In one embodiment, described that the input data is input to deep learning network progress parameter learning, and root
The step of reconstructing super-resolution image according to the obtained network parameter of study include:
The image data RGBL input convolutional neural networks of four-way are subjected to parameter learning and obtain convolution kernel, according to described
Convolution kernel rebuilds each split sence region of low resolution image to be reconstructed, exports super-resolution image.
A kind of image super-resolution rebuilding system, comprising:
Divide module, for being split according to scene content to low resolution image to be reconstructed, and is respectively set each
The label value of the image data in split sence region；
Die section is obtained for the image data in each split sence region and its corresponding label value to be carried out split
The input data of deep learning network；
Module is rebuild, carries out parameter learning for the input data to be input to deep learning network, and according to study
Obtained network parameter reconstructs super-resolution image.
Above-mentioned image super-resolution rebuilding system is realized according to scene content by segmentation module to low resolution to be reconstructed
Image is split and is arranged label value, and the scene information in image is added to super-resolution depth by die section
In learning network, and then super-resolution image is gone out by reconstruction remodelling；The technical solution institute learning parameter can adapt to difference
Scene content, can preferably to image carry out super-resolution rebuilding, accurately recover scene content different in image
Relevant details, improve the picture quality of reconstruction.
In one embodiment, the segmentation module is further used for the type for various scene contents, is arranged various
The corresponding label value of type institute；Corresponding label value is obtained according to the type of the image data in each split sence region；By institute
The label value got is stated to be written in the image data in the split sence region.
In one embodiment, the die section is further used for the RGB threeway in each split sence region obtained
Road image data；The label value L in split sence region where obtaining the RGB triple channel image data；By the RGB triple channel
Image data and label value L piece together the image data RGBL of four-way, the input data as deep learning network.
In addition, there is a need asking for the relevant details for being difficult to accurately recover scene content different in image
Topic provides a kind of computer equipment and its storage medium.
A kind of computer equipment, including memory, processor and be stored on the memory and can be in the processing
The computer program run on device, the processor realize such as above-mentioned image super-resolution rebuilding when executing the computer program
Method.
Above-mentioned computer equipment can accurately recover image by the computer program run on the processor
The relevant details of middle different scene content, improve the picture quality of reconstruction.
A kind of computer storage medium, is stored thereon with computer program, realizes when which is executed by processor as above
State image super-resolution rebuilding method.
Above-mentioned computer storage medium can be recovered accurately different in image by the computer program that it is stored
Scene content relevant details, improve the picture quality of reconstruction.
Detailed description of the invention
Fig. 1 is the image super-resolution rebuilding method flow chart of one embodiment；
Fig. 2 is the image super-resolution rebuilding system structure diagram of one embodiment；
Fig. 3 is the module map for being able to achieve a computer system of the embodiment of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Refering to what is shown in Fig. 1, Fig. 1 is the image super-resolution rebuilding method flow chart of one embodiment, include the following steps:
S10 is split low resolution image to be reconstructed according to scene content, and each split sence area is respectively set
The label value of the image data in domain.
In above-mentioned steps, in conjunction with the scene content in low resolution image to be reconstructed, image is first split, herein
It when being split, can be finely divided according to actual needs, subdivision degree is higher, and when subsequent reconstruction image corresponds to details can also be with
It is finer, for the label value being arranged after segmentation, it is contemplated that the factor of image procossing can be described using digital form.
For example, mainly including the low resolution image to be reconstructed of sky and two kinds of meadow scene content to a width, can adopt
With corresponding algorithm, two split sence regions are divided the image into according to sky and meadow, then according to the common of setting
Class label, such as 0 indicates sky, and 1 indicates meadow etc..Mark is written according to the content in region in image in the region divided the image into
Label, i.e., label 1 is written in the image data in meadow region, and label 0 is written in the image data of sky areas；Certainly, if divided
Three classifications are segmented into, then unknown classification individually sets label value 2 etc..
In one embodiment, the label value of the image data in each split sence region is set in the step S10
Process may include steps of:
Firstly, it is directed to the type of various scene contents, the corresponding label value of setting various types institute；Specifically, can root
According to all types of scene content, each type is first set and corresponds to a kind of label value.
Then, corresponding label value is obtained according to the type of the image data in each split sence region；Specifically, when needing
When being split to a low resolution image to be reconstructed, the type of each image data is found according to specific splitting scheme
Label value.
Finally, the label value got is written in the image data in the split sence region；Specifically, will search
To type corresponding label value be written in the image data in each split sence region.
The scheme of above-described embodiment is combined in use by the way that the corresponding label value of various types scene content is first arranged
Splitting scheme finds label value and is written in image data, is convenient for image dividing processing, improves image real time transfer efficiency.
The image data in each split sence region and its corresponding label value are carried out split and obtain deep learning by S20
The input data of network.
In above-mentioned steps, based on the application of deep learning network, by by the image data in split sence region and its
Corresponding label value generates input data, for during deep learning.
The technical solution by having increased label value newly in input data, using the scene areas information in image as priori
It is added in super-resolution deep learning network, so that the input data of deep learning network increases a dimension, so that institute
Learning parameter can adapt to different scene contents.
In one embodiment, the split process of step S20, may include steps of:
S201, the RGB triple channel image data in each split sence region of acquisition；
S202, the label value L in split sence region where obtaining the RGB triple channel image data；
The RGB triple channel image data and label value L are pieced together the image data RGBL of four-way by S203, as
The input data of deep learning network.
The scheme of above-described embodiment is the processing scheme for being directed to rgb format image, by RGB triple channel image data and
The label value L in split sence region pieces together the data of RGBL four-way, and input deep learning network is learnt, so that
Different scene contents is enough adapted in learning process.
Certainly, other than above-described embodiment mode, the image of the multichannels such as YUV and YCbCr can also be directed to.
As embodiment, the deep learning network can use convolutional neural networks (Convolutional Neural
Network, CNN), CNN network can use super-resolution algorithms commonly full convolutional network, can export as high resolution graphics
Picture.
The input data is input to deep learning network and carries out parameter learning, and the network obtained according to study by S30
Reconstruction goes out super-resolution image.
In above-mentioned steps, by being added to super-resolution deep learning net for the scene areas information in image as priori
In network, learned parameter can reflect the difference of scene content, so that the content of reconstruction image is more careful.
By taking the above-mentioned low resolution image to be reconstructed including sky and two kinds of meadow scene content as an example, relative to tradition side
The picture quality that case reconstructs is more average, and uses the technical solution of the embodiment of the present invention, the image data of sky areas
The image recovered can be more smooth；The image data in meadow region can be finer and smoother in the image recovered.
In one embodiment, the process for reconstructing super-resolution image of step S30, may include steps of:
The image data RGBL input convolutional neural networks of four-way are subjected to parameter learning and obtain convolution kernel, according to described
Convolution kernel rebuilds each split sence region of low resolution image to be reconstructed, exports super-resolution image.
The scheme of above-described embodiment utilizes the image data in each split sence region in rgb format image reconstruction
RGBL learns to obtain convolution kernel, reconstructs super-resolution image.Scene content different in image can accurately be recovered
Relevant details improve the picture quality of reconstruction.
In addition, entire low resolution image to be reconstructed can also be protruded wherein as a label as embodiment
Certain a kind of details.For example, it is assumed that thering is personage to have meadow again in low resolution image to be reconstructed, in order to protrude the oversubscription of personage
Resolution effect can set entire segmentation tag value to the label value of personage, be sent in deep learning network and tested,
Without each split sence region for segmentation, different label values are set.
In conclusion the scheme of the embodiment of the present invention, the image super-resolution method based on deep learning, and fully consider
To the scene content of image, the information that low-resolution image is divided is added in deep learning network and carries out parameter learning.By
In the label value that image segmentation is added, enable depth network according to picture material learning network parameter, so that is generated is super
Image in different resolution effect is more preferable.
It is the image super-resolution rebuilding system structure diagram of one embodiment with reference to Fig. 2, Fig. 2, comprising:
Divide module 10, for being split according to scene content to low resolution image to be reconstructed, and is respectively set each
The label value of the image data in a split sence region；
Die section 20 is obtained for the image data in each split sence region and its corresponding label value to be carried out split
To the input data of deep learning network；
Module 30 is rebuild, carries out parameter learning for the input data to be input to deep learning network, and according to
Acquistion to network parameter reconstruct super-resolution image.
In one embodiment, the segmentation module is further used for the type for various scene contents, is arranged various
The corresponding label value of type institute；Corresponding label value is obtained according to the type of the image data in each split sence region；By institute
The label value got is stated to be written in the image data in the split sence region.
In one embodiment, the die section is further used for the RGB threeway in each split sence region obtained
Road image data；The label value L in split sence region where obtaining the RGB triple channel image data；By the RGB triple channel
Image data and label value L piece together the image data RGBL of four-way, the input data as deep learning network.
In one embodiment, the deep learning network can be convolutional neural networks.
The reconstruction module 30 is further used for joining the image data RGBL input convolutional neural networks of four-way
Mathematics acquistion carries out weight to convolution kernel, according to each split sence region of the convolution kernel to low resolution image to be reconstructed
It builds, exports super-resolution image.
Image super-resolution rebuilding system of the invention and image super-resolution rebuilding method of the invention correspond,
The technical characteristic and its advantages that the embodiment of above-mentioned image super-resolution rebuilding method illustrates are suitable for Image Super-resolution
In the embodiment of rate reconstructing system, hereby give notice that.
Based on example as described above, a kind of computer equipment is also provided in one embodiment, the computer equipment packet
The computer program that includes memory, processor and storage on a memory and can run on a processor, wherein processor executes
It realizes when described program such as any one image super-resolution rebuilding method in the various embodiments described above.
Above-mentioned computer equipment can accurately recover image by the computer program run on the processor
The relevant details of middle different scene content, improve the picture quality of reconstruction.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, it is non-volatile computer-readable that the program can be stored in one
It takes in storage medium, in the embodiment of the present invention, which be can be stored in the storage medium of computer system, and by the calculating
At least one processor in machine system executes, and includes the process such as the embodiment of above-mentioned each sleep householder method with realization.Its
In, the storage medium can be magnetic disk, CD, read-only memory (Read-Only Memory, ROM) or random storage
Memory body (Random Access Memory, RAM) etc..
Accordingly, a kind of storage medium is also provided in one embodiment, is stored thereon with computer program, wherein the journey
It realizes when sequence is executed by processor such as any one image super-resolution rebuilding in the various embodiments described above.
Above-mentioned computer storage medium can be recovered accurately different in image by the computer program that it is stored
Scene content relevant details, improve the picture quality of reconstruction.
Fig. 3 is the module map for being able to achieve a computer system of the embodiment of the present invention.The computer system is one
It is suitable for the invention the example of computer environment, is not construed as proposing any restrictions to use scope of the invention.
Computer system can not be construed to need to rely on or one or more of the illustrative computer system with diagram
The combination of component.
Computer system shown in Fig. 3 is the example for being suitable for computer system of the invention.With difference
Other frameworks of subsystem configuration also can be used.Such as have big well known desktop computer, notebook, personal digital assistant,
The similar devices such as smart phone, tablet computer, portable media player, set-top box can be adapted for some implementations of the invention
Example.But it is not limited to equipment enumerated above.
As shown in figure 3, the computer system includes processor 310, memory 320 and system bus 322.Including memory
320 and processor 310 including various system components be connected on system bus 322.Processor 310 is one and is used to pass through meter
Basic arithmetic sum logical operation executes the hardware of computer program instructions in calculation machine system.Memory 320 is one and is used for
Temporarily or permanently store the physical equipment of calculation procedure or data (for example, program state information).System bus 320 can be
Any one in the bus structures of following several types, including memory bus or storage control, peripheral bus and part
Bus.Processor 310 and memory 320 can carry out data communication by system bus 322.Wherein memory 320 includes only
It reads memory (ROM) or flash memory (being all not shown in figure) and random access memory (RAM), RAM typically refers to be loaded with behaviour
Make the main memory of system and application program.
Computer system further includes display interface 330 (for example, graphics processing unit), display equipment 340 (for example, liquid crystal
Display), audio interface 350 (for example, sound card) and audio frequency apparatus 360 (for example, loudspeaker).Show equipment 340 and audio
Equipment 360 is the media device for experiencing multimedia content.
The computer system generally comprises a storage equipment 370.Storing equipment 370 can be from a variety of computer-readable Jie
It is selected in matter, computer-readable medium refers to any available medium that can be accessed by computer system 300, including moves
Dynamic and fixed two media.For example, computer-readable medium includes but is not limited to, and flash memory (miniature SD card), CD-
ROM, digital versatile disc (DVD) or other optical disc storages, cassette, tape, disk storage or other magnetic storage apparatus, or
It can be used for storing information needed and can be by any other medium of computer system accesses.
The computer system further includes input unit 380 and input interface 390 (for example, I/O controller).User can lead to
Input unit 380 is crossed, such as the touch panel equipment in keyboard, mouse, display device 340, input instruction and information to computer
In system.Input unit 380 is usually connected on system bus 322 by input interface 390, but can also be by other
Interface or bus structures are connected, such as universal serial bus (USB).
The computer system can carry out logical connection with one or more network equipment in a network environment.The network equipment
It can be PC, server, router, smart phone, tablet computer or other common network nodes.Computer system
300 are connected by local area network (LAN) interface 400 or mobile comm unit 410 with the network equipment.Local area network (LAN) refers to
In finite region, such as family, school, computer laboratory or the office building using the network media, interconnect the meter of composition
Calculation machine network.WiFi and twisted pair wiring Ethernet are two kinds of technologies of most common building local area network.WiFi is a kind of to make
Computer system swapping data or the technology that wireless network is connected to by radio wave.Mobile comm unit 410 can be one
It answers and makes a phone call by radio communication diagram while movement in a wide geographic area.Other than call, move
Dynamic communication unit 410 is also supported to carry out internet access in 2G, 3G or the 4G cellular communication system for providing mobile data service.
It should be pointed out that other includes that the computer system of subsystems more more or fewer than computer system can also fit
For inventing.
As detailed above, image super-resolution rebuilding method can be executed by being suitable for the invention computer system
Process.Computer system executes these by way of the software instruction that processor 310 is run in computer-readable medium
Operation.These software instructions can be read into memory from another equipment from storage equipment 370 or by lan interfaces 400
In 320.The software instruction being stored in memory 320 makes processor 310 execute above-mentioned image super-resolution rebuilding method.
In addition, also can equally realize the present invention by hardware circuit or hardware circuit combination software instruction.Therefore, the present invention is realized simultaneously
It is not limited to the combination of any specific hardware circuit and software.
Each technical characteristic of embodiment described above can be combined arbitrarily, for simplicity of description, not to above-mentioned reality
It applies all possible combination of each technical characteristic in example to be all described, as long as however, the combination of these technical characteristics is not deposited
In contradiction, all should be considered as described in this specification.
The embodiments described above only express several embodiments of the present invention, and the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to protection of the invention
Range.Therefore, the scope of protection of the patent of the invention shall be subject to the appended claims.
Claims (10)
1. a kind of image super-resolution rebuilding method, which comprises the steps of:
Low resolution image to be reconstructed is split according to scene content, and the image in each split sence region is respectively set
The label value of data；
The image data in each split sence region and its corresponding label value are subjected to split and obtain the defeated of deep learning network
Enter data；
The input data is input to deep learning network and carries out parameter learning, and is rebuild according to the network parameter that study obtains
Super-resolution image out.
2. image super-resolution rebuilding method according to claim 1, which is characterized in that described that each segmentation is respectively set
The step of label value of the image data of scene areas includes:
For the type of various scene contents, the corresponding label value of various types institute is set；
Corresponding label value is obtained according to the type of the image data in each split sence region；
The label value got is written in the image data in the split sence region.
3. image super-resolution rebuilding method according to claim 1, which is characterized in that described by each split sence area
The image data in domain and its corresponding label value carry out the step of split obtains the input data of deep learning network
The RGB triple channel image data in each split sence region obtained；
The label value L in split sence region where obtaining the RGB triple channel image data；
The image data RGBL that the RGB triple channel image data and label value L are pieced together to four-way, as deep learning net
The input data of network.
4. image super-resolution rebuilding method according to claim 3, which is characterized in that the deep learning network is volume
Product neural network.
5. image super-resolution rebuilding method according to claim 4, which is characterized in that described that the input data is defeated
Enter to deep learning network and carries out parameter learning, and the step of super-resolution image is reconstructed according to the network parameter that study obtains
Include:
The image data RGBL input convolutional neural networks of four-way are subjected to parameter learning and obtain convolution kernel, according to the convolution
The each split sence region for checking low resolution image to be reconstructed is rebuild, and super-resolution image is exported.
6. a kind of image super-resolution rebuilding system characterized by comprising
Divide module, for being split according to scene content to low resolution image to be reconstructed, and each segmentation is respectively set
The label value of the image data of scene areas；
Die section obtains depth for the image data in each split sence region and its corresponding label value to be carried out split
The input data of learning network；
Module is rebuild, carries out parameter learning for the input data to be input to deep learning network, and obtain according to study
Network parameter reconstruct super-resolution image.
7. image super-resolution rebuilding system according to claim 6, which is characterized in that the segmentation module, further
For being directed to the type of various scene contents, the corresponding label value of setting various types institute；According to each split sence region
The type of image data obtains corresponding label value；The label value got is written to the picture number in the split sence region
In.
8. image super-resolution rebuilding system according to claim 6, which is characterized in that the die section, further
The RGB triple channel image data in each split sence region for acquisition；Obtain the RGB triple channel image data place point
Cut the label value L of scene areas；The RGB triple channel image data and label value L are pieced together to the image data of four-way
RGBL, the input data as deep learning network.
9. a kind of computer equipment, including memory, processor and it is stored on the memory and can be in the processor
The computer program of upper operation, which is characterized in that the processor realized when executing the computer program as claim 1 to
Image super-resolution rebuilding method described in 5 any one.
10. a kind of computer storage medium, is stored thereon with computer program, which is characterized in that the program is executed by processor
Image super-resolution rebuilding method of the Shi Shixian as described in claim 1 to 5 any one.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810803433.9A CN109064399B (en) | 2018-07-20 | 2018-07-20 | Image super-resolution reconstruction method and system, computer device and storage medium thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810803433.9A CN109064399B (en) | 2018-07-20 | 2018-07-20 | Image super-resolution reconstruction method and system, computer device and storage medium thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109064399A true CN109064399A (en) | 2018-12-21 |
| CN109064399B CN109064399B (en) | 2023-01-24 |
Family
ID=64817698
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810803433.9A Active CN109064399B (en) | 2018-07-20 | 2018-07-20 | Image super-resolution reconstruction method and system, computer device and storage medium thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109064399B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110136062A (en) * | 2019-05-10 | 2019-08-16 | 武汉大学 | A kind of super resolution ratio reconstruction method of combination semantic segmentation |
| CN110288530A (en) * | 2019-06-28 | 2019-09-27 | 北京金山云网络技术有限公司 | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding |
| CN110298790A (en) * | 2019-06-28 | 2019-10-01 | 北京金山云网络技术有限公司 | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding |
| CN110619603A (en) * | 2019-08-29 | 2019-12-27 | 浙江师范大学 | Single image super-resolution method for optimizing sparse coefficient |
| CN110958469A (en) * | 2019-12-13 | 2020-04-03 | 联想(北京)有限公司 | Video processing method and device, electronic equipment and storage medium |
| CN111598776A (en) * | 2020-04-29 | 2020-08-28 | Oppo广东移动通信有限公司 | Image processing method, image processing apparatus, storage medium, and electronic device |
| CN111640067A (en) * | 2020-06-10 | 2020-09-08 | 华侨大学 | Single image super-resolution reconstruction method based on three-channel convolutional neural network |
| CN111800630A (en) * | 2019-04-09 | 2020-10-20 | Tcl集团股份有限公司 | Method and system for reconstructing video super-resolution and electronic equipment |
| CN112734647A (en) * | 2021-01-20 | 2021-04-30 | 支付宝(杭州)信息技术有限公司 | Image processing method and device |
| 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 |
| US20210272239A1 (en) * | 2019-04-29 | 2021-09-02 | Intel Corporation | Method and apparatus for person super resolution from low resolution image |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013026659A (en) * | 2011-07-15 | 2013-02-04 | Univ Of Tsukuba | Super-resolution image processing device and dictionary creating device for super-resolution image processing |
| US20130156297A1 (en) * | 2011-12-15 | 2013-06-20 | Microsoft Corporation | Learning Image Processing Tasks from Scene Reconstructions |
| CN103489173A (en) * | 2013-09-23 | 2014-01-01 | 百年金海科技有限公司 | Video image super-resolution reconstruction method |
| US20170148139A1 (en) * | 2015-11-25 | 2017-05-25 | Heptagon Micro Optics Pte. Ltd. | Super-resolution image reconstruction using high-frequency band extraction |
| CN107578377A (en) * | 2017-08-31 | 2018-01-12 | 北京飞搜科技有限公司 | A kind of super-resolution image reconstruction method and system based on deep learning |
| CN107977929A (en) * | 2016-10-21 | 2018-05-01 | 中国电信股份有限公司 | Image Super Resolution Processing method and apparatus |
-
2018
- 2018-07-20 CN CN201810803433.9A patent/CN109064399B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2013026659A (en) * | 2011-07-15 | 2013-02-04 | Univ Of Tsukuba | Super-resolution image processing device and dictionary creating device for super-resolution image processing |
| US20130156297A1 (en) * | 2011-12-15 | 2013-06-20 | Microsoft Corporation | Learning Image Processing Tasks from Scene Reconstructions |
| CN103489173A (en) * | 2013-09-23 | 2014-01-01 | 百年金海科技有限公司 | Video image super-resolution reconstruction method |
| US20170148139A1 (en) * | 2015-11-25 | 2017-05-25 | Heptagon Micro Optics Pte. Ltd. | Super-resolution image reconstruction using high-frequency band extraction |
| CN107977929A (en) * | 2016-10-21 | 2018-05-01 | 中国电信股份有限公司 | Image Super Resolution Processing method and apparatus |
| CN107578377A (en) * | 2017-08-31 | 2018-01-12 | 北京飞搜科技有限公司 | A kind of super-resolution image reconstruction method and system based on deep learning |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111800630A (en) * | 2019-04-09 | 2020-10-20 | Tcl集团股份有限公司 | Method and system for reconstructing video super-resolution and electronic equipment |
| US11734798B2 (en) * | 2019-04-29 | 2023-08-22 | Intel Corporation | Method and apparatus for person super resolution from low resolution image |
| US20210272239A1 (en) * | 2019-04-29 | 2021-09-02 | Intel Corporation | Method and apparatus for person super resolution from low resolution image |
| CN110136062A (en) * | 2019-05-10 | 2019-08-16 | 武汉大学 | A kind of super resolution ratio reconstruction method of combination semantic segmentation |
| CN110136062B (en) * | 2019-05-10 | 2020-11-03 | 武汉大学 | Super-resolution reconstruction method combining semantic segmentation |
| CN110288530A (en) * | 2019-06-28 | 2019-09-27 | 北京金山云网络技术有限公司 | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding |
| CN110298790A (en) * | 2019-06-28 | 2019-10-01 | 北京金山云网络技术有限公司 | A kind of pair of image carries out the processing method and processing device of super-resolution rebuilding |
| CN110619603A (en) * | 2019-08-29 | 2019-12-27 | 浙江师范大学 | Single image super-resolution method for optimizing sparse coefficient |
| CN110619603B (en) * | 2019-08-29 | 2023-11-10 | 浙江师范大学 | Single image super-resolution method for optimizing sparse coefficient |
| 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 |
| CN110958469A (en) * | 2019-12-13 | 2020-04-03 | 联想(北京)有限公司 | Video processing method and device, electronic equipment and storage medium |
| CN111598776B (en) * | 2020-04-29 | 2023-06-30 | Oppo广东移动通信有限公司 | Image processing method, image processing device, storage medium and electronic apparatus |
| CN111598776A (en) * | 2020-04-29 | 2020-08-28 | Oppo广东移动通信有限公司 | Image processing method, image processing apparatus, storage medium, and electronic device |
| CN111640067B (en) * | 2020-06-10 | 2022-08-09 | 华侨大学 | Single image super-resolution reconstruction method based on three-channel convolutional neural network |
| CN111640067A (en) * | 2020-06-10 | 2020-09-08 | 华侨大学 | Single image super-resolution reconstruction method based on three-channel convolutional neural network |
| CN112734647A (en) * | 2021-01-20 | 2021-04-30 | 支付宝(杭州)信息技术有限公司 | Image processing method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109064399B (en) | 2023-01-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109064399A (en) | Image super-resolution rebuilding method and system, computer equipment and its storage medium | |
| CN108022212B (en) | High-resolution picture generation method, generation device and storage medium | |
| CN110189246B (en) | Image stylization generation method and device and electronic equipment | |
| US11244430B2 (en) | Digital image fill | |
| US9626586B2 (en) | Optimized fast hessian matrix computation architecture | |
| CN110363753B (en) | Image quality evaluation method and device and electronic equipment | |
| CN110378410B (en) | Multi-label scene classification method and device and electronic equipment | |
| CN104851117A (en) | Method for fusing image with two-dimensional barcode and device thereof | |
| WO2020107625A1 (en) | Video classification method and apparatus, electronic device, and computer readable storage medium | |
| CN110210532B (en) | Background color generation method and device and electronic equipment | |
| CN103997687A (en) | Techniques for adding interactive features to videos | |
| CN111105375A (en) | Image generation method, model training method and device thereof, and electronic equipment | |
| CN112399249A (en) | Multimedia file generation method and device, electronic equipment and storage medium | |
| WO2021147276A1 (en) | Data processing method and apparatus, and chip, electronic device and storage medium | |
| CN112785493A (en) | Model training method, style migration method, device, equipment and storage medium | |
| US11328454B2 (en) | Enabling secure video sharing by exploiting data sparsity | |
| US20150242988A1 (en) | Methods of eliminating redundant rendering of frames | |
| CN110197459B (en) | Image stylization generation method and device and electronic equipment | |
| CN112492382A (en) | Video frame extraction method and device, electronic equipment and storage medium | |
| CN110728732A (en) | Image reconstruction method, device, equipment and medium | |
| Noguera et al. | Interaction and visualization of 3D virtual environments on mobile devices | |
| EP2903288A1 (en) | Method and apparatus for constructing an epitome from an image | |
| CN111696041B (en) | Image processing method and device and electronic equipment | |
| CN111696010A (en) | Scene-based training method, server, terminal device and storage medium | |
| CN112204585A (en) | Processor, electronic device and control method thereof |
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 |