CN109859107A - Remote Sensing Image Super Resolution method, apparatus, equipment and readable storage medium storing program for executing - Google Patents
Remote Sensing Image Super Resolution method, apparatus, equipment and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN109859107A CN109859107A CN201910111151.7A CN201910111151A CN109859107A CN 109859107 A CN109859107 A CN 109859107A CN 201910111151 A CN201910111151 A CN 201910111151A CN 109859107 A CN109859107 A CN 109859107A
- Authority
- CN
- China
- Prior art keywords
- remote sensing
- super resolution
- image
- generator
- sensing image
- 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
Abstract
The embodiment of the invention discloses a kind of Remote Sensing Image Super Resolution method, apparatus, equipment and computer readable storage mediums.Wherein, method includes constructing super resolution image in advance based on SRGAN algorithm to generate model, it will be input in super-resolution image generation model wait mention high-resolution remote sensing images, to realize Remote Sensing Image Super Resolution, it includes generator, arbiter and network layer that super resolution image, which generates model,；The loss function of generator and arbiter is based on obtained by earth mover's distance；Generator is using high-resolution remote sensing image as input picture, obtained by training residual error convolutional network.The application, which realizes, fast and efficiently carries out Remote Sensing Image Super Resolution, and earth mover's distance is defined as loss function, loses index as reflection physical training condition, not only reduces training difficulty, also effectively improve model training stability, precision and accuracy.
Description
Technical field
The present embodiments relate to super resolution image processing technology fields, more particularly to a kind of Remote Sensing Image Super Resolution
Method, apparatus, equipment and computer readable storage medium.
Background technique
For image as information most direct way is obtained, the promotion to its carried information content is always that the unremitting of the mankind chases after
It asks, has some idea of from the continuous renewal of high-definition picture in recent years development.But in many fields such as remote sensing, medical treatment, security protection
Deng, the promotion of image resolution ratio suffers from the limitation of the hardware cost, manufacturing process and information transmission conditions of imaging sensor,
Especially to the satellite remote sensing field of volume, power consumption, weight demands harshness, the brought volume and weight of Optical System Design is increased
Increase can not ignore, therefore how under the premise of not increasing Satellite Camera volume and weight, utilize low point of multidate
Resolution image obtains more high-frequency informations, that is to say, that Remote Sensing Image Super Resolution reconstruct becomes those skilled in the art urgently
Problem to be solved.
For a long time, the practical application of super-resolution technique also has been limited to the acquisition of Multitemporal Remote Sensing Images, effectively
Load be the low rail remote sensing satellite that linear array push sweeps camera to return to the period longer, such as US Terrestrial observation satellite series is general
Repetition period is 16 days, if the multiple image for obtaining Same Scene needs long time, and during this, atural object
Scene can most probably change very much, thus being based on the biggish multi-temporal data realization super-resolution reconstruction of time span can not
Capable.
With geostationary orbit remote sensing technology, optical remote sensing constellation networking technology and based on single star face battle array CMOS camera skill
Art flourishes, and obtaining multi-temporal data in second grade or minute grade becomes possibility, thus super-resolution reconstruction technology also will
New opportunity to develop is welcome in China quickly.
SRGAN (Super-Resolution Using a Generative Adversarial Network) be based on
Method is trained GAN (Generative Adversarial Networks, production fight network), including a life
It grows up to be a useful person and an arbiter, computing resource consumption is also reduced while promoting resolution ratio.
But not only there is the losses of training difficulty, generator and arbiter can not indicate training process, life for this method
The problems such as at sample lack of diversity；And the input and output object of SRGAN is common RGB picture, is had with remote sensing images
It is distinguished, the super-resolution reconstruction of remote sensing images can not be suitable for.
Summary of the invention
The embodiment of the present disclosure provides a kind of Remote Sensing Image Super Resolution method, apparatus, equipment and computer-readable storage
Medium realizes and fast and efficiently carries out Remote Sensing Image Super Resolution.
In order to solve the above technical problems, the embodiment of the present invention the following technical schemes are provided:
On the one hand the embodiment of the present invention provides a kind of Remote Sensing Image Super Resolution method, comprising:
Obtain remote sensing images to be processed；
The remote sensing images to be processed are input to the super resolution image constructed in advance to generate in model, are obtained described wait locate
Manage the super resolution image of remote sensing images；
Wherein, the super resolution image generates model and is based on SRGAN model, including generator, arbiter and network layer；Institute
The loss function for stating generator and the arbiter is based on obtained by earth mover's distance；The generator is with high-definition remote sensing
Image is input picture, obtained by training residual error convolutional network.
Optionally, the loss function of the generator isThe loss function of the arbiter is
Wherein, x is the input picture that the super resolution image generates model, fwFor arbiter network function, E is expectation
Value, PrTo input the sample distribution that the super resolution image generates model, PgThe sample distribution generated for the generator.
Optionally, the last layer of the arbiter does not include sigmoid activation primitive, and the arbiter is updated
The absolute value of parameter differences value in journey is within a preset range.
Optionally, the activation primitive of the last layer of the network layer is band leakage rectification function, and the network layer is not
Include batch normalization layer.
Optionally, it is root mean square pass-algorithm that the super resolution image, which generates the optimization algorithm of model,.
Optionally, it further includes visualization learning layer that the super resolution image, which generates model, for monitoring under loss function
Drop state.
Optionally, the convolution kernel of the last layer of the generator is 9*9.
On the other hand the embodiment of the present invention provides a kind of Remote Sensing Image Super Resolution device, comprising:
Model prebuild module generates model, the super-resolution for constructing super resolution image in advance based on SRGAN algorithm
It includes generator, arbiter and network layer that image, which generates model,；The loss function of the generator and the arbiter be based on
Obtained by earth mover's distance；The generator is using high-resolution remote sensing image as input picture, obtained by training residual error convolutional network；
Remote sensing images obtain module, for obtaining remote sensing images to be processed；
Super-resolution remote sensing images generation module, for the remote sensing images to be processed to be input to the super-resolution constructed in advance
Image generates in model, obtains the super resolution image of the remote sensing images to be processed.
The embodiment of the invention also provides a kind of Remote Sensing Image Super Resolution equipment, including processor, the processor is used
The step of realizing the Remote Sensing Image Super Resolution method as described in preceding any one when executing the computer program stored in memory.
The embodiment of the present invention finally additionally provides a kind of computer readable storage medium, the computer readable storage medium
On be stored with Remote Sensing Image Super Resolution program, realize when the Remote Sensing Image Super Resolution program is executed by processor as former
The step of one Remote Sensing Image Super Resolution method.
The advantages of technical solution provided by the present application, is, earth mover's distance is defined as loss function, instructs as reflection
Practice state and lose index, can produce significant gradient to update generator, so that generating distribution is pulled to true distribution, reduces
Training difficulty solves the problems, such as that generation sample is single, it is ensured that the diversity for generating sample effectively improves model training and stablizes
Property, precision and accuracy；In addition, training super resolution image to generate model in advance using remote sensing images, realize fast and efficiently
Carry out Remote Sensing Image Super Resolution.
In addition, the embodiment of the present invention provides corresponding realization device, equipment also directed to Remote Sensing Image Super Resolution method
And computer readable storage medium, further such that the method has more practicability, described device, equipment and computer-readable
Storage medium has the advantages that corresponding.
It should be understood that the above general description and the following detailed description are merely exemplary, this can not be limited
It is open.
Detailed description of the invention
It, below will be to embodiment or correlation for the clearer technical solution for illustrating the embodiment of the present invention or the relevant technologies
Attached drawing needed in technical description is briefly described, it should be apparent that, the accompanying drawings in the following description is only this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is a kind of flow diagram of Remote Sensing Image Super Resolution method provided in an embodiment of the present invention；
Fig. 2 is a kind of disclosure structural framing schematic diagram of generator shown according to an exemplary embodiment；
Fig. 3 is a kind of disclosure structural framing schematic diagram of arbiter shown according to an exemplary embodiment；
Fig. 4 is that a kind of disclosure super resolution image shown according to an exemplary embodiment generates the structural framing of model and shows
It is intended to；
Fig. 5 is a kind of specific embodiment structure chart of Remote Sensing Image Super Resolution device provided in an embodiment of the present invention.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, with reference to the accompanying drawings and detailed description
The present invention is described in further detail.Obviously, described embodiments are only a part of the embodiments of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
The description and claims of this application and term " first ", " second ", " third " " in above-mentioned attached drawing
Four " etc. be for distinguishing different objects, rather than for describing specific sequence.Furthermore term " includes " and " having " and
Their any deformations, it is intended that cover and non-exclusive include.Such as contain a series of steps or units process, method,
System, product or equipment are not limited to listed step or unit, but may include the step of not listing or unit.
After describing the technical solution of the embodiment of the present invention, the various non-limiting realities of detailed description below the application
Apply mode.
Referring first to Fig. 1, Fig. 1 is a kind of process signal of Remote Sensing Image Super Resolution method provided in an embodiment of the present invention
Figure, the embodiment of the present invention may include the following contents:
S101: super resolution image is constructed based on SRGAN algorithm in advance and generates model.
S102: remote sensing images to be processed are obtained.
S103: remote sensing images to be processed are input to the super resolution image constructed in advance and are generated in model, are obtained to be processed
The super resolution image of remote sensing images.
It includes generator, arbiter and network layer that super resolution image, which generates model,.Super resolution image generate model be
Gained after modifying on the basis of SRGAN model, so that super resolution image generates model and can not only surpass to remote sensing images
Resolution processing, also raising model training stability, precision and accuracy.
Model training have a problem that i.e. when the deconditioning generator and arbiter, if over training is sentenced
Other device, generator just go down without calligraphy learning, and it is poor to will lead to modelling effect if too early deconditioning.How reasonable balance is needed
The training degree of generator and arbiter guarantees model training stability, is very important, that is to say, that if settable one
A loss function index reflects training, and the difficulty of model training just will be greatly reduced.
Earth mover's distance, i.e. Wasserstein distance are called Earth-Mover (EM) distance, are defined as follows:
Π(Pr,Pg) it is PrAnd PgThe set of all possible Joint Distribution to combine.It is possible for each
For closing distribution, can therefrom sample to obtain an authentic specimen x and a generation sample y, and calculate this distance to sample |
| x-y | |, it is possible to calculate the desired value E that sample is adjusted the distance under Joint Distribution γ(x, y)~γ[||x-y||].In all possibility
Joint Distribution in can be to the lower bound that this desired value is gotJust it is defined as
Wasserstein distance.
Since Wasserstein distance has such superior property, can define earth mover's distance is generator and arbiter
Loss function.Significant gradient can be generated in this way to update generator, so that generating distribution is pulled to true distribution.
But because in Wasserstein distance definitionCannot direct solution, formula 1 can be transformed to as
Lower form:
It is exactly to apply a limitation to a continuous function f, it is desirable that there are one here with Lipschitzian continuity has been arrived
Constant K >=0 makes any two element x of domain1And x2All meet | f (x1)-f(x2)|≤Kx1-x2|, at this time just claim letter
The lipschitz constant of number f is K.
Lipschitz constant of the formula (2) in f to be found a function | | f | |LUnder conditions of K, to all possible satisfactions
The f of condition is gotThe upper bound, then again divided by K.Particularly, can be determined with one group of parameter w
A series of possible function f of justicew, at this time solution formula 2 can approximation become to solve following form:
Then f is indicated with a neural network with parameter w, since the capability of fitting of neural network is powerful enough,
Function fwIt is enough the sup that height is required close to formula (2) | | f ‖L≤ K.
Then a containing parameter w can be constructed, the last layer is not the arbiter network f of non-linear active coatingw, in limitation w
Under conditions of a certain range, so thatMaximum value is got as far as possible, at this time L
It will be similar to really be distributed and generate the Wasserstein distance between distribution.
Following generator will approximatively minimize Wasserstein distance, can be minimized L, due to Wasserstein
The advantageous property of distance needs not worry about the problem of generator gradient disappears.Consider further that the first item of L is unrelated with generator,
Two loss functions that super resolution image generates model are just obtained.
The loss function of generator are as follows:
The loss function of arbiter are as follows:
Wherein, x is the input picture that super resolution image generates model, fwFor arbiter network function, E is desired value, PrFor
Input the sample distribution that the super resolution image generates model, PgThe sample distribution generated for the generator.
Generator is using high-resolution remote sensing image as input picture, obtained by training residual error convolutional network.The knot of generator
Structure is seen shown in Fig. 2, and the design of residual error convolutional network can be used in generator, to restrain faster, each residual error modular structure phase
With (totally 5 residual blocks in figure), each residual error module may include two convolutional layers (Conv), two ReLU layers of (Rectified
Linear Unit, line rectification function) and by element operation summation layer (Elementwise Sum).It is another in addition to residual block in figure
One module without marking each layer name is identical with its previous modular structure, i.e., the module comprising convolutional layer, ReLU layers and
Pixelshuffle*2 layers.In addition, the convolution kernel of generator the last layer be 9*9 (the last layer of traditional SRGAN model
Convolution kernel is 1*1).The sub-pix that the amplification of generator and reconstruct part can be used in efficient sub-pix convolutional neural networks is set
Meter allows picture just to increase resolution ratio in rearmost network layer, and computing resource consumption is reduced while promoting resolution ratio.Remote sensing is super
The task of resolution ratio is the remote sensing images by inputting a low resolution, exports the remote sensing images of a ultrahigh resolution.?
In the training generator stage, high-resolution remote sensing image I can be usedHRAs input, it is distant that low resolution is then obtained by down-sampling
Feel image ILR, the down-sampling factor is r.The remote sensing images for being C to a channel, can be by IHRSize be described as W × H × C, it is right
The I answeredLRSize is rW × rH × C.
The structure of arbiter is seen shown in Fig. 3, all the same without the modular structure for marking each layer name, includes convolution
Layer and with leakage rectify function layer.What is done due to the arbiter of original GAN is true and false two classification task, so the last layer is
Sigmoid activation primitive, but arbiter f nowwWhat is done is approximate fits Wasserstein distance, belongs to recurrence task, can
The sigmoid activation of the last layer is taken away, and the absolute value of the parameter differences value of arbiter at no point in the update process is in default model
Their absolute value is truncated to no more than one fixed constant in enclosing, namely every time after the parameter of update arbiter.
VGG19, such as super resolution image shown in Fig. 4 can be used to generate the structure of model, VGG16 mould for the main body of arbiter
Type will extract generator and full resolution pricture to generate the feature of super resolution image, and using the feature of extraction as the defeated of arbiter
Enter, VGG16 is in this as a kind of feature extractor.
Since the remote sensing image that super resolution image generates the input and output of the generator of model has much in spatial distribution
Similar characters of ground object, and batch normalization layer of tradition SRGAN model destroys original sky completely when standardized feature
Between characters of ground object.Therefore part layer and parameter in network is needed to go to restore characters of ground object when rebuilding, identical
Under parameter, there are the partial parameters that also to take out of batch normalization layer to go to repair characters of ground object, therefore effect can decline, therefore remove and criticize
Normalization layer.Network layer does not include batch normalization layer, lets out in addition, the activation primitive of the last layer of network layer may also be configured to band
Dew rectification function, i.e., be changed to Leaky-ReLU (Leaky- for the activation primitive Sigmoid of the last layer of network layer
Rectified Linear Unit, band leakage rectification function).
In order to improve model computational efficiency and precision, root mean square biography is can be used in the optimization algorithm that super resolution image generates model
Pass algorithm.
In order to further enhance model training stability, model training difficulty is reduced, super resolution image generates model and may be used also
Including visualizing learning layer, for monitoring the decline state of loss function.Any visual chemistry can be used in visualization study
Algorithm is practised, the application does not do any restriction to this.
In technical solution provided in an embodiment of the present invention, earth mover's distance is defined as loss function, is instructed as reflection
Practice state and lose index, can produce significant gradient to update generator, so that generating distribution is pulled to true distribution, reduces
Training difficulty solves the problems, such as that generation sample is single, it is ensured that the diversity for generating sample effectively improves model training and stablizes
Property, precision and accuracy；In addition, training super resolution image to generate model in advance using remote sensing images, realize fast and efficiently
Carry out Remote Sensing Image Super Resolution.
The embodiment of the present invention provides corresponding realization device also directed to Remote Sensing Image Super Resolution method, further such that
The method has more practicability.Remote Sensing Image Super Resolution device provided in an embodiment of the present invention is introduced below, under
The Remote Sensing Image Super Resolution device of text description can correspond to each other reference with above-described Remote Sensing Image Super Resolution method.
Referring to Fig. 5, Fig. 5 is Remote Sensing Image Super Resolution device provided in an embodiment of the present invention in a kind of specific embodiment
Under structure chart, the device can include:
Model prebuild module 501 generates model, super-resolution for constructing super resolution image in advance based on SRGAN algorithm
It includes generator, arbiter and network layer that image, which generates model,；The loss function of generator and arbiter be based on bull-dozer away from
From gained；Generator is using high-resolution remote sensing image as input picture, obtained by training residual error convolutional network.
Remote sensing images obtain module 502, for obtaining remote sensing images to be processed.
Super-resolution remote sensing images generation module 503, for remote sensing images to be processed to be input to the super-resolution constructed in advance
Image generates in model, obtains the super resolution image of remote sensing images to be processed.
Optionally, in some embodiments of the present embodiment, the loss function of generator isDifferentiate
The loss function of device is
Wherein, x is the input picture that super resolution image generates model, fwFor arbiter network function, E is desired value, PrFor
Input the sample distribution that the super resolution image generates model, PgThe sample distribution generated for the generator.
Optionally, the last layer of arbiter does not include sigmoid activation primitive, and the ginseng of arbiter at no point in the update process
The absolute value of number difference value is within a preset range.
Optionally, the activation primitive of the last layer of network layer is band leakage rectification function, and network layer does not include batch mark
Standardization layer.
Optionally, it is root mean square pass-algorithm that super resolution image, which generates the optimization algorithm of model,.
Optionally, it further includes visualization learning layer that super resolution image, which generates model, for monitoring the decline shape of loss function
State.
Optionally, the convolution kernel of the last layer of generator is 9*9.
The function of each functional module of Remote Sensing Image Super Resolution device described in the embodiment of the present invention can be according to the above method
Method specific implementation in embodiment, specific implementation process are referred to the associated description of above method embodiment, herein not
It repeats again.
From the foregoing, it will be observed that the embodiment of the present invention, which realizes, fast and efficiently carries out Remote Sensing Image Super Resolution.
The embodiment of the invention also provides a kind of Remote Sensing Image Super Resolution equipment, specifically can include:
Memory, for storing computer program；
Processor realizes Remote Sensing Image Super Resolution side described in any one embodiment as above for executing computer program
The step of method.
The function of each functional module of Remote Sensing Image Super Resolution equipment described in the embodiment of the present invention can be according to the above method
Method specific implementation in embodiment, specific implementation process are referred to the associated description of above method embodiment, herein not
It repeats again.
From the foregoing, it will be observed that the embodiment of the present invention, which realizes, fast and efficiently carries out Remote Sensing Image Super Resolution.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored with Remote Sensing Image Super Resolution journey
Sequence, Remote Sensing Image Super Resolution described in any one embodiment as above when the Remote Sensing Image Super Resolution program is executed by processor
The step of method.
The function of each functional module of computer readable storage medium described in the embodiment of the present invention can be according to above method reality
The method specific implementation in example is applied, specific implementation process is referred to the associated description of above method embodiment, herein no longer
It repeats.
From the foregoing, it will be observed that the embodiment of the present invention, which realizes, fast and efficiently carries out Remote Sensing Image Super Resolution.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with it is other
The difference of embodiment, same or similar part may refer to each other between each embodiment.For being filled disclosed in embodiment
For setting, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is referring to method part
Explanation.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure
And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
The step of method described in conjunction with the examples disclosed in this document or algorithm, can directly be held with hardware, processor
The combination of capable software module or the two is implemented.Software module can be placed in random access memory (RAM), memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In any other form of storage medium well known in field.
It to a kind of Remote Sensing Image Super Resolution method, apparatus provided by the present invention, equipment and computer-readable deposits above
Storage media is described in detail.It is used herein that a specific example illustrates the principle and implementation of the invention,
The above description of the embodiment is only used to help understand the method for the present invention and its core ideas.It should be pointed out that for this technology
For the those of ordinary skill in field, without departing from the principle of the present invention, several improvement can also be carried out to the present invention
And modification, these improvements and modifications also fall within the scope of protection of the claims of the present invention.
Claims (10)
1. a kind of Remote Sensing Image Super Resolution method characterized by comprising
Obtain remote sensing images to be processed；
The remote sensing images to be processed are input to the super resolution image constructed in advance to generate in model, are obtained described to be processed distant
Feel the super resolution image of image；
Wherein, the super resolution image generates model and is based on SRGAN model, including generator, arbiter and network layer；The life
Growing up to be a useful person with the loss function of the arbiter is based on obtained by earth mover's distance；The generator is with high-resolution remote sensing image
For input picture, training residual error convolutional network gained.
2. Remote Sensing Image Super Resolution method according to claim 1, which is characterized in that the loss function of the generator
ForThe loss function of the arbiter is
Wherein, x is the input picture that the super resolution image generates model, fwFor arbiter network function, E is desired value, PrFor
Input the sample distribution that the super resolution image generates model, PgThe sample distribution generated for the generator.
3. Remote Sensing Image Super Resolution method according to claim 2, which is characterized in that the last layer of the arbiter
Not comprising sigmoid activation primitive, and the absolute value of the parameter differences value of the arbiter at no point in the update process is in preset range
It is interior.
4. Remote Sensing Image Super Resolution method according to claim 2, which is characterized in that the last layer of the network layer
Activation primitive be band leakage rectification function, and the network layer does not include batch normalization layer.
5. Remote Sensing Image Super Resolution method according to claim 2, which is characterized in that the super resolution image generates mould
The optimization algorithm of type is root mean square pass-algorithm.
6. according to claim 1 to Remote Sensing Image Super Resolution method described in 5 any one, which is characterized in that the oversubscription
Distinguishing that image generates model further includes visualization learning layer, for monitoring the decline state of loss function.
7. Remote Sensing Image Super Resolution method according to claim 6, which is characterized in that the last layer of the generator
Convolution kernel be 9*9.
8. a kind of Remote Sensing Image Super Resolution device characterized by comprising
Model prebuild module generates model, the super resolution image for constructing super resolution image in advance based on SRGAN algorithm
Generating model includes generator, arbiter and network layer；The loss function of the generator and the arbiter is based on soil-shifting
Machine is apart from gained；The generator is using high-resolution remote sensing image as input picture, obtained by training residual error convolutional network；
Remote sensing images obtain module, for obtaining remote sensing images to be processed；
Super-resolution remote sensing images generation module, for the remote sensing images to be processed to be input to the super resolution image constructed in advance
It generates in model, obtains the super resolution image of the remote sensing images to be processed.
9. a kind of Remote Sensing Image Super Resolution equipment, which is characterized in that including processor, the processor is for executing memory
It is realized when the computer program of middle storage as described in any one of claim 1 to 7 the step of Remote Sensing Image Super Resolution method.
10. a kind of computer readable storage medium, which is characterized in that be stored with remote sensing figure on the computer readable storage medium
As super-resolution program, realize when the Remote Sensing Image Super Resolution program is executed by processor such as any one of claim 1 to 7
The step of Remote Sensing Image Super Resolution method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910111151.7A CN109859107B (en) | 2019-02-12 | 2019-02-12 | Remote sensing image super-resolution method, device, equipment and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910111151.7A CN109859107B (en) | 2019-02-12 | 2019-02-12 | Remote sensing image super-resolution method, device, equipment and readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109859107A true CN109859107A (en) | 2019-06-07 |
| CN109859107B CN109859107B (en) | 2023-04-07 |
Family
ID=66897767
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910111151.7A Active CN109859107B (en) | 2019-02-12 | 2019-02-12 | Remote sensing image super-resolution method, device, equipment and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109859107B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110363792A (en) * | 2019-07-19 | 2019-10-22 | 广东工业大学 | A kind of method for detecting change of remote sensing image based on illumination invariant feature extraction |
| CN111179172A (en) * | 2019-12-24 | 2020-05-19 | 浙江大学 | Remote sensing satellite super-resolution implementation method and device based on unmanned aerial vehicle aerial data, electronic equipment and storage medium |
| CN111667431A (en) * | 2020-06-09 | 2020-09-15 | 云南电网有限责任公司电力科学研究院 | Method and device for manufacturing cloud and fog removing training set based on image conversion |
| CN111882046A (en) * | 2020-09-27 | 2020-11-03 | 北京声智科技有限公司 | Multimedia data identification method, device, equipment and computer storage medium |
| CN112365553A (en) * | 2019-07-24 | 2021-02-12 | 北京新唐思创教育科技有限公司 | Human body image generation model training, human body image generation method and related device |
| CN112734638A (en) * | 2020-12-24 | 2021-04-30 | 桂林理工大学 | Remote sensing image super-resolution reconstruction method and device and storage medium |
| CN113643183A (en) * | 2021-10-14 | 2021-11-12 | 湖南大学 | Non-matching remote sensing image weak supervised learning super-resolution reconstruction method and system |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180075581A1 (en) * | 2016-09-15 | 2018-03-15 | Twitter, Inc. | Super resolution using a generative adversarial network |
| CN108022213A (en) * | 2017-11-29 | 2018-05-11 | 天津大学 | Video super-resolution algorithm for reconstructing based on generation confrontation network |
| CN109087243A (en) * | 2018-06-29 | 2018-12-25 | 中山大学 | A kind of video super-resolution generation method generating confrontation network based on depth convolution |
-
2019
- 2019-02-12 CN CN201910111151.7A patent/CN109859107B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180075581A1 (en) * | 2016-09-15 | 2018-03-15 | Twitter, Inc. | Super resolution using a generative adversarial network |
| CN108022213A (en) * | 2017-11-29 | 2018-05-11 | 天津大学 | Video super-resolution algorithm for reconstructing based on generation confrontation network |
| CN109087243A (en) * | 2018-06-29 | 2018-12-25 | 中山大学 | A kind of video super-resolution generation method generating confrontation network based on depth convolution |
Non-Patent Citations (1)
| Title |
|---|
| 杜彦璞: "基于生成对抗网络的遥感图像超分辨率方法研究", 《中国优秀硕士学位论文全文数据库信息科技辑2019》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110363792A (en) * | 2019-07-19 | 2019-10-22 | 广东工业大学 | A kind of method for detecting change of remote sensing image based on illumination invariant feature extraction |
| CN112365553A (en) * | 2019-07-24 | 2021-02-12 | 北京新唐思创教育科技有限公司 | Human body image generation model training, human body image generation method and related device |
| CN111179172A (en) * | 2019-12-24 | 2020-05-19 | 浙江大学 | Remote sensing satellite super-resolution implementation method and device based on unmanned aerial vehicle aerial data, electronic equipment and storage medium |
| CN111179172B (en) * | 2019-12-24 | 2021-11-02 | 浙江大学 | Remote sensing satellite super-resolution implementation method and device based on unmanned aerial vehicle aerial data, electronic equipment and storage medium |
| CN111667431A (en) * | 2020-06-09 | 2020-09-15 | 云南电网有限责任公司电力科学研究院 | Method and device for manufacturing cloud and fog removing training set based on image conversion |
| CN111667431B (en) * | 2020-06-09 | 2023-04-14 | 云南电网有限责任公司电力科学研究院 | Method and device for manufacturing cloud and fog removing training set based on image conversion |
| CN111882046A (en) * | 2020-09-27 | 2020-11-03 | 北京声智科技有限公司 | Multimedia data identification method, device, equipment and computer storage medium |
| CN112734638A (en) * | 2020-12-24 | 2021-04-30 | 桂林理工大学 | Remote sensing image super-resolution reconstruction method and device and storage medium |
| CN112734638B (en) * | 2020-12-24 | 2022-08-05 | 桂林理工大学 | Remote sensing image super-resolution reconstruction method and device and storage medium |
| CN113643183A (en) * | 2021-10-14 | 2021-11-12 | 湖南大学 | Non-matching remote sensing image weak supervised learning super-resolution reconstruction method and system |
| CN113643183B (en) * | 2021-10-14 | 2021-12-21 | 湖南大学 | Non-matching remote sensing image weak supervised learning super-resolution reconstruction method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109859107B (en) | 2023-04-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109859107A (en) | Remote Sensing Image Super Resolution method, apparatus, equipment and readable storage medium storing program for executing | |
| CN114549731B (en) | Method and device for generating visual angle image, electronic equipment and storage medium | |
| DE102005008824B4 (en) | A method of registering two images using a graphics processor and the computer readable program storage device therefor | |
| CN109584178A (en) | Image repair method, device and storage medium | |
| CN104268934B (en) | Method for reconstructing three-dimensional curve face through point cloud | |
| CN109493347A (en) | The method and system that the object of sparse distribution is split in the picture | |
| CN108229497A (en) | Image processing method, device, storage medium, computer program and electronic equipment | |
| US8743119B2 (en) | Model-based face image super-resolution | |
| CN103810729B (en) | A kind of based on isocontour raster image vector quantized method | |
| Du et al. | Accelerated super-resolution MR image reconstruction via a 3D densely connected deep convolutional neural network | |
| CN104899835B (en) | Image Super-resolution processing method based on blind blur estimation and anchoring space mapping | |
| CN109657483A (en) | A kind of image encryption method and system | |
| CN107977930A (en) | A kind of image super-resolution method and its system | |
| CN101937555B (en) | Parallel generation method of pulse compression reference matrix based on GPU (Graphic Processing Unit) core platform | |
| CN109559278B (en) | Super resolution image reconstruction method and system based on multiple features study | |
| CN110335196A (en) | A kind of super-resolution image reconstruction method and system based on fractal decoding | |
| Tang et al. | Master: Meta Style Transformer for Controllable Zero-Shot and Few-Shot Artistic Style Transfer | |
| CN106157248A (en) | A kind of joint line network based on grid generates method | |
| CN110276711A (en) | Graphics process | |
| Zhang et al. | Lightweight transformer backbone for medical object detection | |
| CN110706167B (en) | Fine completion processing method and device for remote sensing image to-be-repaired area | |
| Wang et al. | Adaptive patch-based POCS approach for super resolution reconstruction of 4D-CT lung data | |
| Yang | Super resolution using dual path connections | |
| Ye et al. | Cascade Multiscale Swin-Conv Network for Fast MRI Reconstruction | |
| Kang et al. | Snow accumulation effects in screen space for real-time terrain rendering |
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 |