CN110211036A - A kind of image super-resolution rebuilding method and device based on intelligent priori - Google Patents
A kind of image super-resolution rebuilding method and device based on intelligent priori Download PDFInfo
- Publication number
- CN110211036A CN110211036A CN201910338546.0A CN201910338546A CN110211036A CN 110211036 A CN110211036 A CN 110211036A CN 201910338546 A CN201910338546 A CN 201910338546A CN 110211036 A CN110211036 A CN 110211036A
- Authority
- CN
- China
- Prior art keywords
- resolution
- super
- image
- priori
- supplement
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 40
- 239000013589 supplement Substances 0.000 claims abstract description 27
- 238000012360 testing method Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 description 13
- 230000008901 benefit Effects 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000012545 processing 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
- 238000013135 deep learning Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000001737 promoting effect Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
A kind of image super-resolution rebuilding method and device based on intelligent priori, this method comprises: using the single-frame images super-resolution method based on study to low-resolution image YkCarry out super-resolution rebuilding, the high resolution sequence image X after being rebuildk；By the high-definition picture XkFor to sequence image YkInformation merges the regular conditions supplement of super-resolution rebuilding, realizes sequence image super-resolution reconstruction.The present invention passes through the priori characteristic to single frames study acquisition special scenes, and it is used in the oversubscription algorithm for reconstructing merged based on multiple image information, so that newly-generated image resolution ratio can keep picture characteristics as much as possible, the accuracy of image super-resolution rebuilding is improved.
Description
Technical field
The present invention relates to technical field of image processing more particularly to a kind of image super-resolution rebuildings based on intelligent priori
Method and device.
Background technique
For this problem of sequence image super-resolution, from the point of view of mathematical angle, it belongs to an inverse problem, i.e., according to portion
Divide the anti-unique solution for pushing away primal problem of observed result.The maximum difficult point of inverse problem is its ill-posedness: from known sight
The solution of problem calculated by survey condition is not unique.This allows for constrain solution from the essence of problem.
For sequence image super-resolution problem, its solution is to infer a panel height image in different resolution, therefore, it is necessary to consider high score
The characteristics of resolution image.It is usually solved in the prior art with model below, its foundation is energy functional minimum:
Wherein, YkFor low resolution observed image, image size M × N, X are original high-resolution image, image size qM
× qN, q are interpolation factor, and q > 1, D are lack sampling of the high-resolution to low-resolution image, SkExpression degrades model, K=
1 ... ... N.
In this model, first item data itemIndicate reconstruction error item, Section 2It is
Regularization term is defined based on priori knowledge and image statistics.Regularization coefficient λ is for balancing fidelity term and canonical
?.It is the key that algorithm design, is that sequence image super-resolution algorithm can effectively ensure that.
Currently, various effective super resolution algorithms are all to propose suitable regularization according to the characteristics of high-definition picture
Come to solution X constrain, the solution for enabling it to acquire meets physical process as far as possible, and is of practical significance, it is different just
Then item constitutes different sequence image super-resolution algorithms.Traditional regular terms design is usually by artificially summarizing the experience, such as
Smoothness properties, image low-rank characteristic etc..Such regular terms can assist error fidelity term to obtain unique solution, still, equally bring
To the loss of high-frequency signal.Undoubtedly too extensive moreover, no matter what picture material is all constrained with the same canonical, do not have
The otherness for embodying content does not follow the methodology of particular problem concrete analysis, necessarily causes super-resolution rebuilding not smart enough
Carefully, lack using specific aim.
Summary of the invention
The object of the present invention is to provide a kind of images based on intelligent priori for promoting image super-resolution rebuilding accuracy
Super resolution ratio reconstruction method and device.
To achieve the above object, there is provided a kind of image oversubscription based on intelligent priori for the first technical solution of the invention
Resolution method for reconstructing, comprising:
Using the single frames super-resolution method based on study to low-resolution image YkSuper-resolution rebuilding is carried out, weight is obtained
High-definition picture X after buildingk；
By the high-definition picture XkSupplement priori item as the Regularization function for image super-resolution rebuilding
Part；
Image super-resolution rebuilding is realized according to the Regularization function after supplement priori conditions.
In the present embodiment, because of XkIt is middle necessarily to contain the characteristic of part original high-resolution image X, then can incite somebody to action | |
Xk- X | | as a priori, for avoid two images because tonal gradation difference caused by influence, may further useWherein ΩkFor XkAll pixel average gray values, Ω be X all pixel average gray values.
When two images above carry out operation, approximate same place pixel is subtracted each other.Wherein X image size and XkImage
Size is identical.
Further, in order to balance image individualized feature as supplement priori item and as the ratio for counting general priori item
It is heavy, further include the steps that weight coefficient is arranged in present invention implementation.
Objective function may finally be formed:
Analyzed in above formula image particular content carrying characteristic, more targetedly, than in the past only with general statistical just
Then the method for item is advantageously.
Wherein, λ1、λ2The respectively weight coefficient of supplement priori item and the general priori item of statistics, needs test adjustment to generate,Common statistics regular terms can be used, it is ensured that smoothly wait fundamental characteristics, it is ensured that be not in singular value.
To achieve the above object, there is provided a kind of image oversubscription based on intelligent priori for the second technical solution of the invention
Resolution reconstructing device, comprising:
First rebuilds module, for utilizing the single-frame images super-resolution method based on study to low-resolution image YkInto
Row super-resolution rebuilding, the high-definition picture X after being rebuildk；
Configuration module is used for the high-definition picture XkAs the Regularization function for image super-resolution rebuilding
Supplement priori conditions；
Second rebuilds module, for realizing image super-resolution rebuilding according to the Regularization function after supplement priori conditions.
Image super-resolution rebuilding method and device provided by the invention based on intelligent priori, by low resolution figure
As YkCarry out super-resolution rebuilding, the high-definition picture X after being rebuildk, and the high-definition picture X that will be obtainedkAs with
In the supplement priori conditions of the Regularization function of image super-resolution rebuilding, so that newly-generated image resolution ratio can be use up
Picture characteristics is possibly kept, the accuracy of image super-resolution rebuilding is promoted.
The above description is only an overview of the technical scheme of the present invention, in order to better understand the technical means of the present invention,
And it can be implemented in accordance with the contents of the specification, and in order to allow above and other objects of the present invention, feature and advantage can
It is clearer and more comprehensible, the followings are specific embodiments of the present invention.
Detailed description of the invention
By reading the following detailed description of the preferred embodiment, various other advantages and benefits are common for this field
Technical staff will become clear.The drawings are only for the purpose of illustrating a preferred embodiment, and is not considered as to the present invention
Limitation.And throughout the drawings, the same reference numbers will be used to refer to the same parts.In the accompanying drawings:
Fig. 1 is a kind of flow chart of image super-resolution rebuilding method based on intelligent priori of the embodiment of the present invention；
Fig. 2 is a kind of structural block diagram of image super-resolution rebuilding device based on intelligent priori of the embodiment of the present invention.
Specific embodiment
Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings, it should be understood that preferred reality described herein
Apply example only for the purpose of illustrating and explaining the present invention and is not intended to limit the present invention.
Embodiment 1
Exemplary embodiments of the present disclosure are described in more detail below with reference to accompanying drawings.Although showing the disclosure in attached drawing
Exemplary embodiment, it being understood, however, that may be realized in various forms the disclosure without should be by embodiments set forth here
It is limited.On the contrary, these embodiments are provided to facilitate a more thoroughly understanding of the present invention, and can be by the scope of the present disclosure
It is fully disclosed to those skilled in the art.
Those skilled in the art of the present technique are appreciated that unless otherwise defined, all terms used herein (including technology art
Language and scientific term), there is meaning identical with the general understanding of those of ordinary skill in fields of the present invention.Should also
Understand, those terms such as defined in the general dictionary, it should be understood that have in the context of the prior art
The consistent meaning of meaning, and unless otherwise will not be explained in an idealized or overly formal meaning by specific definitions.
Fig. 1 is a kind of flow chart of image super-resolution rebuilding method based on intelligent priori of the embodiment of the present invention.Ginseng
According to Fig. 1, the image super-resolution rebuilding method provided in an embodiment of the present invention based on intelligent priori, specifically includes the following steps:
S11, to low-resolution image YkCarry out super-resolution rebuilding, the high-definition picture X after being rebuildk.Especially
Ground, to low-resolution image YkCarrying out super-resolution rebuilding is realized using the single frames super-resolution method based on study.
S12, by the high-definition picture XkSupplement as the Regularization function for image super-resolution rebuilding is first
Test condition；
S13, image super-resolution rebuilding is realized according to the Regularization function after supplement priori conditions.
Image super-resolution rebuilding method provided in an embodiment of the present invention based on intelligent priori, by low resolution figure
As YkCarry out super-resolution rebuilding, the high-definition picture X after being rebuildk, and the high-definition picture X that will be obtainedkAs with
In the supplement priori conditions of the Regularization function of image super-resolution rebuilding, so that newly-generated image resolution ratio can be use up
Picture characteristics is possibly kept, the accuracy of image super-resolution rebuilding is promoted.
In the present embodiment, because of XkIt is middle necessarily to contain the characteristic of part original high-resolution image X, then can incite somebody to action | |
Xk- X | | as a priori, for avoid two images because tonal gradation difference caused by influence, may further useWherein ΩkFor XkAll pixel average gray values, Ω be X all pixel average gray values.
When two images above carry out operation, approximate same place pixel is subtracted each other.Wherein X image size and XkImage
Size is identical.
In above-described embodiment, single frames super-resolution method refers mainly to the method based on machine learning, mainly includes at present
The methods of deep learning, joint regression study, such method is not only believed with more preferably peak value compared to traditional interpolation method
It makes an uproar and compares, and information delta is more prominent, i.e., with true high-definition picture apart from closer.In practical applications, single frames is super
Point method enables to every secondary low-resolution image to generate a secondary high-definition picture by training, still, due to being study
The knowledge come thinks to have carried out oversubscription to information with it more accurately say it is to image when carrying out oversubscription to low partial image
Characteristic is enhanced, or think using study method by low-resolution image generate high-definition picture when, by low resolution
The picture characteristics that rate image corresponds to high-definition picture is added, i.e., personalized elder generation is obtained by the method for machine learning
It tests.Therefore, we will learn to realize the high-definition picture obtained after super-resolution processing as a Xiang Chong based on single-frame images
The characteristic priori wanted.It is expressed with formula:
Yk→Xk
The method that above formula indicates single frames generates high resolution sequence.It is considered that XkIn contain the characteristic of part X, then
It can incite somebody to action
||Xk- X | | as a priori, for avoid two images because tonal gradation difference caused by influence, can make
With
Wherein ΩkFor XkAll pixel average gray values, Ω be X all pixel average gray
Value.
When two images above carry out operation, it is desirable that be that the pixel being registrated is subtracted each other, to reach approximate same place picture
The target that element subtracts each other.Wherein X image size and XkImage size is identical.So just there is condition:
(1) fidelity term
(2) machine learning priori item
(3) general priori item is counted
According to Optimum Theory, it is as follows to form objective function:
Theoretically, the characteristic of image particular content carrying is analyzed in above formula, more targetedly, is led to than previous
Advantageously with statistics regular terms, it can be solved by numerical optimizations such as convex optimizations.
Wherein, λ1、λ2Test adjustment is needed to generate,Common statistics regular terms can be used, it is ensured that smooth to wait substantially
Characteristic, it is ensured that be not in singular value.
For embodiment of the method, for simple description, therefore, it is stated as a series of action combinations, but this field
Technical staff should be aware of, and embodiment of that present invention are not limited by the describe sequence of actions, because implementing according to the present invention
Example, some steps may be performed in other sequences or simultaneously.Secondly, those skilled in the art should also know that, specification
Described in embodiment belong to preferred embodiment, the actions involved are not necessarily necessary for embodiments of the present invention.
Fig. 2 diagrammatically illustrates the structure of the image super-resolution rebuilding device based on intelligent priori of the embodiment of the present invention
Block diagram.Referring to Fig. 2, the image super-resolution rebuilding device based on intelligent priori of the embodiment of the present invention includes:
First rebuilds module 201, is used for using the super-resolution method based on study to low-resolution image YkSurpassed
Resolution reconstruction, the high-definition picture X after being rebuildk；
Configuration module 202 is used for the high-definition picture XkAs the regularization for image super-resolution rebuilding
The supplement priori conditions of function；
Second rebuilds module 203, for realizing image super-resolution weight according to the Regularization function after supplement priori conditions
It builds.
Image super-resolution rebuilding method and device provided in an embodiment of the present invention based on intelligent priori, by low point
Resolution image YkCarry out super-resolution rebuilding, the high-definition picture X after being rebuildk, and the high-definition picture X that will be obtainedk
As the supplement priori conditions of the Regularization function for image super-resolution rebuilding, so that newly-generated image resolution ratio
Picture characteristics can be kept as much as possible, promote the accuracy of image super-resolution rebuilding.
It will be appreciated by those of skill in the art that although some embodiments in this include included in other embodiments
Certain features rather than other feature, but the combination of the feature of different embodiments means to be within the scope of the present invention simultaneously
And form different embodiments.For example, in the following claims, the one of any of embodiment claimed all may be used
Come in a manner of in any combination using.
Finally, it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
Present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: it still may be used
To modify the technical solutions described in the foregoing embodiments or equivalent replacement of some of the technical features；
And these are modified or replaceed, technical solution of various embodiments of the present invention that it does not separate the essence of the corresponding technical solution spirit and
Range.
Claims (5)
1. a kind of image super-resolution rebuilding method based on intelligent priori characterized by comprising
Using the single frames super-resolution method based on study to low-resolution image YkSuper-resolution rebuilding is carried out, after obtaining reconstruction
High-definition picture Xk；
By the high-definition picture XkFor to sequence image YkInformation merges the regular conditions supplement of super-resolution rebuilding,
Realize sequence image super-resolution reconstruction.
2. a kind of image super-resolution rebuilding method based on intelligent priori according to claim 1, which is characterized in that institute
It states the high-definition picture XkThe step of supplement priori conditions as the Regularization function for image super-resolution rebuilding
Suddenly it specifically includes: according to the high-definition picture XkBuilding supplement prior model, the supplement prior model are as follows: | | Xk-X||；
Wherein, X is original high-resolution image, XkFor the high-definition picture after reconstruction.
3. a kind of image super-resolution rebuilding method based on intelligent priori according to claim 1, which is characterized in that institute
It states the high-definition picture XkThe step of supplement priori conditions as the Regularization function for image super-resolution rebuilding
Suddenly it specifically includes: according to the high-definition picture XkBuilding supplement prior model, the supplement prior model are as follows:Wherein ΩkFor XkAll pixel average gray values, Ω be X all pixel average gray values.
4. a kind of image super-resolution rebuilding method based on intelligent priori according to claim 3, which is characterized in that institute
State method further include: the weight coefficient that setting supplement priori item and regular terms are made；Correspondingly, after the priori conditions according to supplement
Regularization function realize image super-resolution rebuilding, comprising: use following objective function, according to supplement priori conditions after just
Then change function and realize that image super-resolution rebuilding, objective function are as follows:
Wherein, λ1、λ2The respectively weight coefficient of supplement priori item and the general priori item of statistics, needs test adjustment to generate,Common statistics regular terms can be used, it is ensured that smoothly wait fundamental characteristics, it is ensured that be not in singular value.
5. a kind of image super-resolution rebuilding device based on intelligent priori characterized by comprising
First rebuilds module, for utilizing the single frames super-resolution method based on study to low-resolution image YkCarry out super-resolution
Rate is rebuild, the high-definition picture X after being rebuildk；
Configuration module is used for the high-definition picture XkSuper-resolution rebuilding is being merged just as sequence image information
Then change the supplement priori conditions of function；
Second rebuilds module, for realizing image super-resolution rebuilding according to the Regularization function after supplement priori conditions.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910338546.0A CN110211036A (en) | 2019-04-25 | 2019-04-25 | A kind of image super-resolution rebuilding method and device based on intelligent priori |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910338546.0A CN110211036A (en) | 2019-04-25 | 2019-04-25 | A kind of image super-resolution rebuilding method and device based on intelligent priori |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110211036A true CN110211036A (en) | 2019-09-06 |
Family
ID=67786330
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910338546.0A Pending CN110211036A (en) | 2019-04-25 | 2019-04-25 | A kind of image super-resolution rebuilding method and device based on intelligent priori |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110211036A (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104408697A (en) * | 2014-10-23 | 2015-03-11 | 西安电子科技大学 | Image super-resolution reconstruction method based on genetic algorithm and regular prior model |
| EP2927864A1 (en) * | 2012-11-29 | 2015-10-07 | NEC Corporation | Image processing device and image processing method |
-
2019
- 2019-04-25 CN CN201910338546.0A patent/CN110211036A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2927864A1 (en) * | 2012-11-29 | 2015-10-07 | NEC Corporation | Image processing device and image processing method |
| CN104408697A (en) * | 2014-10-23 | 2015-03-11 | 西安电子科技大学 | Image super-resolution reconstruction method based on genetic algorithm and regular prior model |
Non-Patent Citations (4)
| Title |
|---|
| 吴忠 等: "基于学习的多帧图像超分辨率重建技术研究", 《福建电脑》 * |
| 孟庆玉等: "基于学习的多帧图像超分辨率重建技术探究", 《数字技术与应用》 * |
| 黄全亮等: "融合学习算法的单帧图像超分辨率复原", 《计算机工程与应用》 * |
| 黄吉庆等: "基于多种正则化的改进超分辨率重建算法", 《计算机工程与应用》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Huang et al. | Bidirectional recurrent convolutional networks for multi-frame super-resolution | |
| CN111754403A (en) | Image super-resolution reconstruction method based on residual learning | |
| CN110675321A (en) | Super-resolution image reconstruction method based on progressive depth residual error network | |
| CN110136060B (en) | Image super-resolution reconstruction method based on shallow dense connection network | |
| CN109191392B (en) | Image super-resolution reconstruction method driven by semantic segmentation | |
| Chen et al. | Single image super-resolution using deep CNN with dense skip connections and inception-resnet | |
| CN102915527A (en) | Face image super-resolution reconstruction method based on morphological component analysis | |
| CN105631807A (en) | Single-frame image super resolution reconstruction method based on sparse domain selection | |
| CN112801904B (en) | Hybrid degraded image enhancement method based on convolutional neural network | |
| Guan et al. | Srdgan: learning the noise prior for super resolution with dual generative adversarial networks | |
| CN114170088A (en) | Relational reinforcement learning system and method based on graph structure data | |
| CN113139904A (en) | Image blind super-resolution method and system | |
| CN107392855B (en) | Image super-resolution reconstruction method based on sparse self-coding network and extremely fast learning | |
| Zhou et al. | MSAR‐DefogNet: Lightweight cloud removal network for high resolution remote sensing images based on multi scale convolution | |
| Pham et al. | CSIE-M: compressive sensing image enhancement using multiple reconstructed signals for internet of things surveillance systems | |
| Zhang et al. | Crafting training degradation distribution for the accuracy-generalization trade-off in real-world super-resolution | |
| CN112184552A (en) | Sub-pixel convolution image super-resolution method based on high-frequency feature learning | |
| CN110211036A (en) | A kind of image super-resolution rebuilding method and device based on intelligent priori | |
| Moeller et al. | Image denoising—old and new | |
| Wu et al. | Compressed sensing via a deep convolutional auto-encoder | |
| Zhang et al. | Temporal compressive imaging reconstruction based on a 3D-CNN network | |
| CN111260552B (en) | Progressive learning-based image super-resolution method | |
| CN113205005B (en) | Low-illumination low-resolution face image reconstruction method | |
| CN111008930B (en) | Fabric image super-resolution reconstruction method | |
| Zhang et al. | Deep external and internal learning for noisy compressive sensing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190906 |
|
| RJ01 | Rejection of invention patent application after publication |