CN109767389B - Self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local combined prior - Google Patents
Self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local combined prior Download PDFInfo
- Publication number
- CN109767389B CN109767389B CN201910035555.2A CN201910035555A CN109767389B CN 109767389 B CN109767389 B CN 109767389B CN 201910035555 A CN201910035555 A CN 201910035555A CN 109767389 B CN109767389 B CN 109767389B
- Authority
- CN
- China
- Prior art keywords
- image
- resolution
- reconstruction
- local
- estimated
- 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.)
- Active
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Image Processing (AREA)
Abstract
The invention discloses a self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local joint priori. Mainly comprises the following steps: using an adaptive weighted double-norm prior in a blur kernel estimation sub-process to obtain an estimated blur kernel and an initial high resolution image; taking the estimated fuzzy core and the initial high-resolution image as inputs of a non-blind reconstruction sub-process; in the non-blind reconstruction sub-process, the estimated fuzzy core and the initial high-resolution image are used as known conditions, and the high-resolution image is estimated by using a local and non-local combined prior and maximum posterior probability reconstruction model. And taking the reconstruction result as a new input high-resolution image, repeatedly executing the two steps until the maximum reconstruction times are reached, and finally outputting the final reconstruction result. The method can reconstruct high-quality high-resolution images from low-resolution images under the condition of unknown fuzzy kernels, and the reconstructed remote sensing images can be applied to the fields of military, agriculture, civilian life and the like.
Description
Technical Field
The invention relates to an image resolution improvement technology, in particular to a self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local combined prior, and belongs to the field of image processing.
Background
The high-resolution remote sensing image is widely applied to the fields of military, agriculture, civilian life and the like. However, the acquisition of the remote sensing image is susceptible to vibrations and movements, causing blurring degradation, and thus the quality of the remote sensing image may be unsatisfactory. Typically, image degradation includes blurring and downsampling. Therefore, in practical applications, it is necessary to purposefully enhance the quality of the acquired remote sensing image by using an image processing technology. In order to solve these problems, extensive studies have been made on image deblurring and resolution improvement. This problem is studied herein as blind super-resolution reconstruction. The single image super-resolution reconstruction technology is one of methods for improving the resolution of images, has the characteristics of easiness in implementation, low cost, strong applicability and the like, and estimates corresponding high-resolution images from observed single low-resolution images. The single image super-resolution reconstruction problem has serious morbidity because the same low resolution image may correspond to a plurality of different high resolution images. There is therefore a need for a stable, reliable high resolution estimate through image prior constraints. The current super-resolution reconstruction method of a single image can be roughly divided into three types: a super-resolution method based on interpolation, a super-resolution method based on reconstruction, and a super-resolution method based on learning. These three types of methods have different characteristics, such as interpolation-based methods generally acquire an interpolated image only according to an interpolation kernel, consider less the influence of blur and noise, and have a relatively limited application range. The super-resolution method based on reconstruction fully mines the information of the degraded image, and can generally well inhibit the artifacts. Learning-based methods tend to have a faster execution speed and can recover image detail well. In addition, performing super-resolution reconstruction under an unknown blur kernel is a greater challenge for image restoration, namely blind super-resolution reconstruction. Blind super-resolution reconstruction is generally divided into two sub-processes, fuzzy kernel estimation and non-blind super-resolution reconstruction.
Disclosure of Invention
The invention aims to organically combine the two sub-processes of fuzzy kernel estimation and non-blind super-resolution, so as to construct a high-efficiency and high-performance remote sensing image super-resolution blind reconstruction method.
The invention provides a self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local joint priori, which mainly comprises the following operation steps:
(1) In the fuzzy core estimation sub-process, self-adaptive weighted double-norm prior is used to obtain an estimated fuzzy core and an initial high-resolution image;
(2) Taking the estimated fuzzy core and the initial high-resolution image as inputs of a non-blind reconstruction sub-process;
(3) In the non-blind reconstruction sub-process, the estimated fuzzy core and the initial high-resolution image are used as known conditions, and the high-resolution image is estimated by using a local and non-local combined prior and maximum posterior probability reconstruction model.
(4) And (3) taking the reconstruction result of the step (3) as a new input high-resolution image, repeatedly executing the step (3) and the step (4) until the maximum reconstruction times are reached, and finally outputting the final reconstruction result.
Drawings
FIG. 1 is a schematic block diagram of a self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local joint prior
FIG. 2 is a graph comparing an initial high resolution image and a bicubic reconstructed image: wherein (a) (c) is a double reconstruction of the original high resolution image obtained by the fuzzy nucleonic process and (b) (d) is a result of using bicubic double reconstruction
FIG. 3 is a graph comparing the fuzzy core estimation results of the present invention with other methods: wherein (a) (e) (i) is a true blur kernel image, (b) (f) (j) is a blur kernel estimated by the present invention, (c) (g) (k) is a blur kernel estimated by the deblurring contrast method 1, (d) (h) (l) is a blur kernel estimated by the deblurring contrast method 2
Fig. 4 is a graph comparing the results of the double reconstruction of the test image "mobilehomepark" by the six methods of the present invention: wherein, (a) is an input low resolution image, (i) is an original high resolution image, (b) (c) (d) (e) (f) (g) (h) is a contrast method 1, a contrast method 2, a contrast method 3, a contrast method 4, a contrast method 5, a Bicubic and a reconstruction result of the invention respectively
Detailed Description
The invention is further described below with reference to the accompanying drawings:
in fig. 1, the self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local joint prior can be specifically divided into the following steps:
(1) In the fuzzy core estimation sub-process, self-adaptive weighted double-norm prior is used to obtain an estimated fuzzy core and an initial high-resolution image;
(2) Taking the estimated fuzzy core and the initial high-resolution image as inputs of a non-blind reconstruction sub-process;
(3) In the non-blind reconstruction sub-process, the estimated fuzzy core and the initial high-resolution image are used as known conditions, and the high-resolution image is estimated by using a local and non-local combined prior and maximum posterior probability reconstruction model.
(4) And (3) taking the reconstruction result of the step (3) as a new input high-resolution image, repeatedly executing the step (3) and the step (4) until the maximum reconstruction times are reached, and finally outputting the final reconstruction result.
Specifically, in the step (1), we first input a low-resolution blurred image, construct a reconstruction framework based on the maximum posterior probability, and use an adaptive weighted double-norm prior and a convolution consistency prior as constraint conditions in the framework, wherein the adaptive weighted double-norm prior respectively constrains an estimated blur kernel and an estimated high-resolution image, and the convolution consistency prior constrains the estimated high-resolution image. As shown in formula (1):
λ represents a parameter of the convolved output of the first term deblurred image; h is a matrix representation of the fuzzy kernel k; d is a downsampling matrix that reduces the resolution of the original high-resolution image; x is the original high resolution image and y is the observed low resolution blurred image; alpha x ，β x ，α k ，β k Is a regularization parameter; η is a convolution consistency constraint parameter;
is a convolution consistency constraint for reducing morbidity, high resolution image +.>
The method can be obtained by reconstruction of the existing super-resolution algorithm; from l p Norms and l 2 Image prior term of norm construction>
And fuzzy kernel a priori term->
Together, an adaptive dual-norm weighting prior is formed, wherein the weighting matrix W is used to adaptively determine whether the current region of the image is a non-edge or edge region, and emphasize/according to the weight 2 Smoothing and noise suppression of norms to non-edge regions of an image, and l p Sharpening of edge regions by norms, wherein each term W in W i Defined as formula (2):
it represents the local non-smoothness of a 5*5 image block centered on the ith pixel. X is X i Is the center pixel, Ω, of the image block i Is the index set of all pixels in the image block, X ij Is X i Neighboring pixels at the j-position.
Through step (1), an estimated blur kernel k and an initial high resolution image x can be obtained.
In the step (2), we reconstruct the high resolution image through the super resolution non-blind reconstruction sub-process. Wherein the estimated blur kernel k and the initial high resolution image x obtained in step (1) are used as known inputs, thereby combining the two sub-processes, as in fig. 1.
In the step (3), we construct a reconstruction framework based on the maximum posterior probability, and use the combined local and non-local priors as image priors, see formula (3):
wherein J AHNLTV Non-local image prior, J AGD Is a local image prior, and ζ and θ are regularization parameters that balance the two prior. In this step, the blur kernel is known and the initial high resolution image is used as the iteration start, thereby reconstructing a high resolution image.
In the step (4), the result of the step (3) is taken as a new initial high-resolution image, and the step (3) and the step (4) are repeatedly executed. Until the set maximum reconstruction times are reached, the reconstruction result is the final output.
To better illustrate the effectiveness of the present invention, an initial high resolution image comparison experiment, a blur kernel estimation comparison experiment, and a final high resolution image reconstruction comparison experiment were performed on a commonly used test image, "mobilehomepark", respectively.
The initial high resolution image comparison experiment is shown in fig. 2. Fig. 2 (a) and 2 (c) are initial high resolution images reconstructed 2 times from the blur kernel estimation sub-process of the present invention, and fig. 2 (b) and 2 (d) are high resolution images reconstructed 2 times bicubic for the originally observed low resolution.
The fuzzy core estimation comparison experiment is shown in fig. 3. (a) (e) (i) is a true blur kernel image, (b) (f) (j) is a blur kernel estimated by the present invention, (c) (g) (k) is a blur kernel estimated by the deblurring contrast method 1, (d) (h) (l) is a blur kernel estimated by the deblurring contrast method 2. Two comparative algorithms are:
deblurring contrast method 1: xu et al, reference "Xu L, zheng S, jia J." Unnatural L0 sparse representation for natural image deblurring, "Proceedings of the IEEE conference on computer vision and pattern recogntion.2013:1107-1114.
Deblurring contrast method 2: the method proposed by Shao et al, reference "Shao W Z, li H B, elad M," Bi-l0-l2-norm regularization for blind motion deblurring, "Journal ofVisual Communication and Image Representation,2015,33:42-59.
The final high resolution image reconstruction comparison experiment is shown in fig. 4. (a) For inputting a low resolution image, (i) is an original high resolution image, (b) (c) (d) (e) (f) (g) (h) is a comparison method 1, a comparison method 2, a comparison method 3, a comparison method 4, a comparison method 5, bicubic, and the reconstruction result of the present invention, respectively.
Comparison method 1: the method proposed by Shao et al, reference "Shao W Z, elad M," Simple, accurate, and robust nonparametric blind super-resolution, "International Conference on Image and graphics, springer, cham, 2015:333-348".
Comparison method 2: the fuzzy core estimation sub-process adopts a defuzzification comparison method 2, and the non-blind reconstruction sub-process: the method proposed by Buades et al, reference "Buades A, coll B, morel J M," Image enhancement by non-local reverse heat equation, "Preprint CMLA,2006, 22:2006".
Comparison method 3: the fuzzy core estimation sub-process adopts a defuzzification comparison method 2, and the non-blind reconstruction sub-process: the method proposed by Ren et al, reference "Ren C, he X, nguyen T Q," Single image super-resolution via adaptive high-dimensional non-local total variation and adaptive geometric feature, "IEEE Transactions on Image Processing,2017,26 (1): 90-106".
Comparison method 4: the fuzzy core estimation sub-process adopts a defuzzification comparison method 1, and the non-blind reconstruction sub-process: the method proposed by Dong et al, reference "Dong W, zhang L, shi G, et al," Nonlocally centralized sparse representation for image restoration, "IEEE Transactions on Image Processing,2013,22 (4): 1620-1630.
Comparison method 5: the fuzzy core estimation sub-process adopts a defuzzification comparison method 1, and the non-blind reconstruction sub-process: the method proposed by Buades et al, reference "Buades A, coll B, morel J M," Image enhancement by non-local reverse heat equation, "Preprint CMLA,2006, 22:2006".
The content of the reconstruction contrast experiment of the final high resolution image is as follows:
the low resolution blurred images generated by the remote sensing test image library "UCMerced" simulation were reconstructed 2-fold using Bicubic, method 1, method 2, method 3, method 4, method 5 and the present invention, respectively. The blurring degradation of the low resolution image is realized by eight blurring kernels. The objective evaluation parameters of the super-resolution reconstruction result are shown in table one. The larger the objective evaluation parameters PSNR (Peak Signal to Noise Ratio) and SSIM (Structure Similarity Index) are, the better the image quality is represented. Algorithm test platform: processor Inter Core i5CPU (3.3 GHz) and memory 16G.
List one
From the objective parameters shown in the table one, the invention obtains the highest PSNR and SSIM values for eight different fuzzy cores on the remote sensing image test library, which represents that the reconstruction result of the invention has better quality.
In conclusion, compared with a comparison method, the reconstruction result has certain advantages in subjective and objective evaluation. Therefore, the invention is a high-performance single image super-resolution reconstruction method.
Claims (1)
1. The self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local joint priori is characterized by comprising the following steps of:
step one: in the fuzzy core estimation sub-process, self-adaptive weighted double-norm prior is used to obtain an estimated fuzzy core and an initial high-resolution image; firstly, inputting a low-resolution blurred image, constructing a reconstruction frame based on maximum posterior probability, wherein the frame adopts self-adaptive weighted double-norm prior and convolution consistency prior as constraint conditions, wherein the self-adaptive weighted double-norm prior respectively constrains an estimated blurred kernel and an estimated high-resolution image, and the convolution consistency prior constrains the estimated high-resolution image, as shown in a formula (1):
λ represents a parameter of the convolved output of the first term deblurred image; h is a matrix representation of the fuzzy kernel k; d is a downsampling matrix that reduces the resolution of the original high-resolution image; x is the original high resolution image and y is the observed low resolution blurred image; alpha x ，β x ，α k ，β k Is a regularization parameter; η is a convolution consistency constraint parameter;
is a convolution consistency constraint for reducing morbidity, high resolution image +.>
The method can be obtained by reconstruction of the existing super-resolution algorithm; from l p Norms and l 2 Image prior term of norm construction>
And fuzzy kernel a priori term->
Together, an adaptive dual-norm weighting prior is formed, wherein the weighting matrix W is used to adaptively determine whether the current region of the image is a non-edge or edge region, and emphasize/according to the weight 2 Smoothing and noise suppression of norms to non-edge regions of an image, and l p Sharpening of edge regions by norms, wherein each term W in W i Defined as formula (2):
it represents the local non-smoothness of a 5*5 image block centered on the ith pixel; x is X i Is the center pixel, Ω, of the image block i Is the index set of all pixels in the image block, X ij Is X i A neighbor image at the j position;
step two: taking the estimated fuzzy core and the initial high-resolution image as inputs of a non-blind reconstruction sub-process;
step three: in the non-blind reconstruction sub-process, the estimated fuzzy core and the initial high-resolution image are used as known conditions, and the high-resolution image is estimated by using a local and non-local combined prior and maximum posterior probability reconstruction model;
step four: and (3) taking the reconstruction result of the step (III) as a new input high-resolution image, repeatedly executing the step (III) and the step (IV) until the maximum reconstruction times are reached, and finally outputting the final reconstruction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910035555.2A CN109767389B (en) | 2019-01-15 | 2019-01-15 | Self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local combined prior |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910035555.2A CN109767389B (en) | 2019-01-15 | 2019-01-15 | Self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local combined prior |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109767389A CN109767389A (en) | 2019-05-17 |
| CN109767389B true CN109767389B (en) | 2023-06-20 |
Family
ID=66454040
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910035555.2A Active CN109767389B (en) | 2019-01-15 | 2019-01-15 | Self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local combined prior |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109767389B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113096013B (en) * | 2021-03-31 | 2021-11-26 | 南京理工大学 | Blind image super-resolution reconstruction method and system based on imaging modeling and knowledge distillation |
| CN113627357B (en) * | 2021-08-13 | 2022-07-08 | 哈尔滨工业大学 | High-spatial-high-spectral-resolution intrinsic decomposition method and system for remote sensing image |
| WO2023050258A1 (en) * | 2021-09-30 | 2023-04-06 | Baidu.Com Times Technology (Beijing) Co., Ltd. | Robust and efficient blind super-resolution using variational kernel autoencoder |
| CN114202459B (en) * | 2021-10-26 | 2024-03-22 | 北京工业大学 | Blind image super-resolution method based on depth priori |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005055136A2 (en) * | 2003-11-26 | 2005-06-16 | Board Of Regents, The University Of Texas System | Method and apparatus for phase-sensitive magnetic resonance imaging |
| JP2008154237A (en) * | 1998-04-06 | 2008-07-03 | Seiko Epson Corp | Image data interpolation apparatus, computer for image data interpolation processing, image data interpolation method, and medium recorded with image data interpolation program |
| WO2010050914A1 (en) * | 2008-10-31 | 2010-05-06 | Hewlett-Packard Development Company, L.P. | Method and system for enhancing image signals to alter the perception of depth by human viewers |
| CN104217405A (en) * | 2014-09-23 | 2014-12-17 | 闽江学院 | Salt-pepper noise filter method for image fusing local information and global information |
| WO2015084966A2 (en) * | 2013-12-04 | 2015-06-11 | Razzor Technologies Inc. | Adaptive sharpening in image processing and display |
| WO2015180053A1 (en) * | 2014-05-28 | 2015-12-03 | 北京大学深圳研究生院 | Method and apparatus for rapidly reconstructing super-resolution image |
| US9237257B1 (en) * | 2013-06-14 | 2016-01-12 | Xilinx, Inc. | Circuits for and methods of generating a digital image |
| CN107025632A (en) * | 2017-04-13 | 2017-08-08 | 首都师范大学 | A kind of image super-resolution rebuilding method and system |
| CN109064402A (en) * | 2018-07-27 | 2018-12-21 | 四川大学 | Based on the single image super resolution ratio reconstruction method for enhancing non local total variation model priori |
Family Cites Families (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8269727B2 (en) * | 2007-01-03 | 2012-09-18 | Apple Inc. | Irregular input identification |
| US8422788B2 (en) * | 2008-08-26 | 2013-04-16 | Microsoft Corporation | Automatic image straightening |
| US8330827B2 (en) * | 2009-11-19 | 2012-12-11 | Eastman Kodak Company | Increasing image resolution using combined differential image |
| US20110122224A1 (en) * | 2009-11-20 | 2011-05-26 | Wang-He Lou | Adaptive compression of background image (acbi) based on segmentation of three dimentional objects |
| US8594464B2 (en) * | 2011-05-26 | 2013-11-26 | Microsoft Corporation | Adaptive super resolution for video enhancement |
| US9692939B2 (en) * | 2013-05-29 | 2017-06-27 | Yeda Research And Development Co. Ltd. | Device, system, and method of blind deblurring and blind super-resolution utilizing internal patch recurrence |
| JP6236259B2 (en) * | 2013-09-06 | 2017-11-22 | 株式会社東芝 | Image processing apparatus, image processing method, and image processing program |
| CN105205783B (en) * | 2015-09-14 | 2016-11-16 | 河海大学 | A kind of SAR image method for rebuilding blind super-resolution of associating prior estimate |
| CN105550989B (en) * | 2015-12-09 | 2018-11-30 | 西安电子科技大学 | The image super-resolution method returned based on non local Gaussian process |
| CN105913392A (en) * | 2016-04-08 | 2016-08-31 | 西安电子科技大学昆山创新研究院 | Degraded image overall quality improving method in complex environment |
| CN106251297A (en) * | 2016-07-19 | 2016-12-21 | 四川大学 | A kind of estimation based on multiple image fuzzy core the rebuilding blind super-resolution algorithm of improvement |
| CN107680040A (en) * | 2017-09-26 | 2018-02-09 | 北京空间飞行器总体设计部 | A kind of blind convolution super-resolution reconstruction method of multiframe based on bayesian criterion and device |
-
2019
- 2019-01-15 CN CN201910035555.2A patent/CN109767389B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008154237A (en) * | 1998-04-06 | 2008-07-03 | Seiko Epson Corp | Image data interpolation apparatus, computer for image data interpolation processing, image data interpolation method, and medium recorded with image data interpolation program |
| WO2005055136A2 (en) * | 2003-11-26 | 2005-06-16 | Board Of Regents, The University Of Texas System | Method and apparatus for phase-sensitive magnetic resonance imaging |
| WO2010050914A1 (en) * | 2008-10-31 | 2010-05-06 | Hewlett-Packard Development Company, L.P. | Method and system for enhancing image signals to alter the perception of depth by human viewers |
| US9237257B1 (en) * | 2013-06-14 | 2016-01-12 | Xilinx, Inc. | Circuits for and methods of generating a digital image |
| WO2015084966A2 (en) * | 2013-12-04 | 2015-06-11 | Razzor Technologies Inc. | Adaptive sharpening in image processing and display |
| WO2015180053A1 (en) * | 2014-05-28 | 2015-12-03 | 北京大学深圳研究生院 | Method and apparatus for rapidly reconstructing super-resolution image |
| CN104217405A (en) * | 2014-09-23 | 2014-12-17 | 闽江学院 | Salt-pepper noise filter method for image fusing local information and global information |
| CN107025632A (en) * | 2017-04-13 | 2017-08-08 | 首都师范大学 | A kind of image super-resolution rebuilding method and system |
| CN109064402A (en) * | 2018-07-27 | 2018-12-21 | 四川大学 | Based on the single image super resolution ratio reconstruction method for enhancing non local total variation model priori |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109767389A (en) | 2019-05-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109767389B (en) | Self-adaptive weighted double-norm remote sensing image blind super-resolution reconstruction method based on local and non-local combined prior | |
| Wang et al. | Real-esrgan: Training real-world blind super-resolution with pure synthetic data | |
| Siu et al. | Review of image interpolation and super-resolution | |
| Huang et al. | Robust single-image super-resolution based on adaptive edge-preserving smoothing regularization | |
| CN112164011B (en) | Motion image deblurring method based on self-adaptive residual error and recursive cross attention | |
| Min et al. | Blind deblurring via a novel recursive deep CNN improved by wavelet transform | |
| CN108460723B (en) | Bilateral total variation image super-resolution reconstruction method based on neighborhood similarity | |
| CN108648162A (en) | A kind of gradient correlation TV factor graph picture denoising deblurring methods based on noise level | |
| CN107085826B (en) | Image super-resolution reconstruction method based on weighted overlap non-local regression prior | |
| Wang et al. | An improved image blind deblurring based on dark channel prior | |
| Zhao et al. | Deep pyramid generative adversarial network with local and nonlocal similarity features for natural motion image deblurring | |
| CN115345791A (en) | Infrared image deblurring algorithm based on attention mechanism residual error network model | |
| Laroche et al. | Deep model-based super-resolution with non-uniform blur | |
| López-Tapia et al. | Fast and robust cascade model for multiple degradation single image super-resolution | |
| Zhao et al. | Single depth image super-resolution with multiple residual dictionary learning and refinement | |
| CN110852947B (en) | Infrared image super-resolution method based on edge sharpening | |
| Zin et al. | Local image denoising using RAISR | |
| Li et al. | Guided iterative back-projection scheme for single-image super-resolution | |
| CN115760638A (en) | End-to-end deblurring super-resolution method based on deep learning | |
| Panda et al. | An Improved DCT interpolation using Bilateral filter | |
| CN106846286A (en) | Video super-resolution algorithm for reconstructing based on various complementary priori | |
| Gan et al. | Adaptive joint nonlocal means denoising back projection for image super resolution | |
| CN108648143B (en) | Image resolution enhancement method using sequence image | |
| CN112348745A (en) | Video super-resolution reconstruction method based on residual convolutional network | |
| Panda et al. | A novel image upscaling method using high order error sharpening |
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 |