CN107563965A - Jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks - Google Patents
Jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks Download PDFInfo
- Publication number
- CN107563965A CN107563965A CN201710783918.1A CN201710783918A CN107563965A CN 107563965 A CN107563965 A CN 107563965A CN 201710783918 A CN201710783918 A CN 201710783918A CN 107563965 A CN107563965 A CN 107563965A
- Authority
- CN
- China
- Prior art keywords
- convolutional neural
- neural networks
- image
- compressed image
- jpeg compressed
- 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
Abstract
The invention discloses a kind of jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks.Mainly include the following steps that：For jpeg compressed image, the Super-resolution reconstruction established model based on convolutional neural networks is built；Using training image, the convolutional neural networks of structure are trained；The low-resolution image through JPEG compression is rebuild using the convolutional neural networks model trained.The convolutional neural networks framework built in the present invention can be optimized training end to end by going pinch effect network, increase resolution network and quality enhancing network to form to it.The method of the invention can reduce the compression noise in jpeg compressed image and improve its resolution ratio.The inventive method can be applied to the field such as image and video compress, Digital multimedia communications.
Description
Technical field
The present invention relates to the increased quality technology of jpeg compressed image, and in particular to a kind of based on convolutional neural networks
Jpeg compressed image super resolution ratio reconstruction method, belongs to image processing field.
Background technology
In fields such as military and medical treatment, the image and video of high quality help to obtain more abundant and more accurate letter
Breath, the resolution ratio of wherein image and video is an important index.With the improvement of people ' s living standards and high definition is shown
The popularization of equipment, in daily life, requirement of the people to image and the resolution ratio of video also more and more higher.In recent years, each neck
The imaging device that domain uses also has very big development, can obtain that resolution ratio is higher, image and video of better quality.But
In some cases, restricted by many factors such as economy, environment, the image or video of acquisition still can not reach real sometimes
The demand of border application.Super-resolution rebuilding technology can carry out increase resolution to the image and video gathered, have very strong
Practicality.In in the past few decades, scholars conduct in-depth research to super-resolution technique, especially single image
Super-resolution rebuilding, it is proposed that many effective methods.But most of image super-resolution rebuilding method assumes that acquisition
Image not through overcompression or in the absence of compression noise.However, the objective condition such as transmission bandwidth and memory capacity are limited to, it is existing
The image in living that grows directly from seeds often all have passed through compression processing.Therefore, directly using traditional super-resolution method in actual life
Compression image rebuild, it is difficult to obtain and gratifying rebuild effect.
The content of the invention
The purpose of the present invention is to be directed to the jpeg compressed image being widely present in daily life to propose a kind of effective oversubscription
Resolution method for reconstructing.
Jpeg compressed image super resolution ratio reconstruction method proposed by the present invention based on convolutional neural networks, it is main include with
Lower operating procedure：
(1) jpeg compressed image is directed to, builds the Super-resolution reconstruction established model based on convolutional neural networks；
(2) training image, the convolutional neural networks built in training step (1) are utilized；
(3) the convolutional neural networks model trained in step (2) is utilized to enter the low-resolution image through JPEG compression
Row is rebuild.
Brief description of the drawings
Fig. 1 is the block diagram of the jpeg compressed image super resolution ratio reconstruction method of the invention based on convolutional neural networks.
Fig. 2 is comparison diagram (super-resolution of the present invention with four kinds of methods to the reconstructed results of test image " Butterfly "
It is 2 to rebuild the factor, 10) JPEG compression quality factor is：Wherein, (a) is test image, and (b) (c) (d) (e) (f) is respectively double
Cubic interpolation, control methods 1, control methods 2, control methods 3 and reconstructed results of the invention
Fig. 3 is comparison diagram (super-resolution rebuilding of the present invention with four kinds of methods to the reconstructed results of test image " Ppt3 "
The factor is 2,20) JPEG compression quality factor is：Wherein, (a) is test image, and (b) (c) (d) (e) (f) is respectively bicubic
Interpolation, control methods 1, control methods 2, control methods 3 and reconstructed results of the invention
Fig. 4 is comparison diagram (super-resolution rebuilding of the present invention with four kinds of methods to the reconstructed results of test image " House "
The factor is 2,30) JPEG compression quality factor is：Wherein, (a) is test image, and (b) (c) (d) (e) (f) is respectively bicubic
Interpolation, control methods 1, control methods 2, control methods 3 and reconstructed results of the invention
Embodiment
The invention will be further described below in conjunction with the accompanying drawings：
In Fig. 1, the jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks, it can specifically be divided into following
Three steps：
(1) jpeg compressed image is directed to, builds the Super-resolution reconstruction established model based on convolutional neural networks；
(2) training image, the convolutional neural networks built in training step (1) are utilized；
(3) the convolutional neural networks model trained in step (2) is utilized to enter the low-resolution image through JPEG compression
Row is rebuild.
Specifically, in the step (1), we build the reconstruction model as shown in Figure 1 based on convolutional neural networks.Tool
Body, constructed model is by going pinch effect network, increase resolution network and quality enhancing network to form.Wherein, go to compress
Effect network mainly realizes the removal of JPEG compression noise, and its number of plies is K1Layer (in the present invention, K1=20)；Increase resolution net
The major function of network be enlarged drawing to expected resolution ratio, its number of plies is K2Layer (in the present invention, K2=10)；Quality strengthens net
The purpose of network is that the quality of reconstruction image is further lifted in target resolution space, and its number of plies is K3Layer (in the present invention, K3
=10).In the reconstruction model based on convolutional neural networks that the present invention is built, the preceding K of pinch effect network is removed1- 1 convolution
The number of filter that layer uses is 64, and convolution kernel size is 3 × 3；The number of filter of its last convolutional layer is 1, convolution
Core size is 3 × 3.The preceding K of increase resolution network2The number of filter that -1 convolutional layer uses is 64, and convolution kernel size is 3
×3；Its last layer is warp lamination, and number of filter 1, convolution kernel size is 9 × 9.Quality strengthens the preceding K of network1-1
The number of filter that individual convolutional layer uses is 64, and convolution kernel size is 3 × 3；The number of filter of its last convolutional layer is
1, convolution kernel size is 3 × 3.The reconstruction model based on convolutional neural networks that the present invention is built can be optimized end to end
Training.
In the step (2), we are first by the high-definition picture for trainingSampled, obtainedSo
Afterwards, with JPEG under the different compression quality factors to the image after samplingIt is compressed, the low resolution observation simulated
ImageFinally, utilize WithCarry out the reconstruction model based on convolutional neural networks of structure in training step (1).
In training process, first individually pinch effect network, increase resolution network and quality enhancing network, Ran Houyong are removed in training stage by stage
The parameter initialization whole network model learnt, and then carry out optimization training end to end.
In the step (3), the input of convolutional neural networks is original image to be reconstructed, i.e., low after JPEG compression
Image in different resolution.The output of convolutional neural networks is the final result rebuild.
In order to verify the validity of the inventive method, with standard testing image " Butterfly ", " Ppt3 " and " House "
Tested.Original " Butterfly ", " Ppt3 " and " House " image is respectively such as Fig. 2 (a), 3 (a), shown in 4 (a).Mould
The generating mode for the low-resolution image through JPEG compression intended：High-resolution test chart picture is carried out with bicubic interpolation method
2 times of down-samplings, then the image after sampling is compressed under the different compression quality factors with JPEG, the image after compression is
For image to be reconstructed.Choose bicubic interpolation and three kinds of single image super-resolution rebuilding algorithms method as a comparison.Wherein, side
The model of method 1 and method 2 is all trained according to the process that degrades in the present invention.The super-resolution rebuilding algorithm of three kinds of contrasts
For：
Method 1：The method that Timofte et al. is proposed, bibliography " R.Timofte, V.De Smet, L.Van Gool, A
+:Adjusted anchored neighborhood regression for fast super-resolution,in:
Proceedings of the Asian Conference on Computer Vision(ACCV),2014,pp.111-
126.”。
Method 2：The method that Dong et al. is proposed, bibliography " C.Dong, C.C.Loy, X.Tang, Accelerating
the super-resolution convolutional neural network,in:Proceedings of the
European Conference on Computer Vision(ECCV),2016,pp.391-407.”。
Method 3：The method that Kang et al. is proposed, bibliography " L.W.Kang, C.C.Hsu, B.Zhuang, C.W.Lin,
C.H.Yeh,Learning-based joint super-resolution and deblocking for a highly
compressed image,IEEE Trans.Multimedia.17(7)(2015)921-934.”。
The content of contrast experiment is as follows：
Experiment 1, respectively with bicubic interpolation, method 1, method 2, method 3 and the inventive method are to by " Butterfly "
The low-resolution image of test image simulation generation carries out 2 times of reconstructions.In this experiment, the quality factor of JPEG compression is set to 10.
The reconstructed results of " Butterfly " original image and each method are respectively such as Fig. 2 (a), Fig. 2 (b), Fig. 2 (c), Fig. 2 (d), Fig. 2 (e)
And shown in Fig. 2 (f).Be in the experiment of one, table each method reconstructed results PSNR (Peak Signal to Noise Ratio) and
SSIM (Structure Similarity Index) parameter.
Table one
Experiment 2, respectively with bicubic interpolation, method 1, method 2, method 3 and the inventive method to being tested by " Ppt3 "
The low-resolution image of image simulation generation carries out 2 times of reconstructions.In this experiment, the quality factor of JPEG compression is set to 20.
The reconstructed results of " Ppt3 " original image and each method are respectively such as Fig. 3 (a), Fig. 3 (b), Fig. 3 (c), Fig. 3 (d), Fig. 3 (e) and Fig. 3
(f) shown in.Table two is PSNR the and SSIM parameters of each method reconstructed results in this experiment.
Table two
Experiment 3, respectively with bicubic interpolation, method 1, method 2, method 3 and the inventive method to being tested by " House "
The low-resolution image of image simulation generation carries out 2 times of reconstructions.In this experiment, the quality factor of JPEG compression is set to 30.
The reconstructed results of " House " original image and each method are respectively such as Fig. 4 (a), Fig. 4 (b), Fig. 4 (c), Fig. 4 (d), Fig. 4 (e) and figure
Shown in 4 (f).Table three is PSNR the and SSIM parameters of each method reconstructed results in this experiment.
Table three
From the experimental result shown in Fig. 2, Fig. 3 and Fig. 4 can be seen that in the result of bicubic interpolation and method 3 contain than
Obvious compression noise, image visual effect are poor；Method 1 and method 2 can remove Partial shrinkage noise, but image it is overall some
It is fuzzy；Without obvious compression noise in the result of the present invention, and image, than more visible, edge keeps more preferable, and visual effect is more preferably.
In PSNR the and SSIM parameters given from table one, table two and table three, the present invention achieves on two indices
Highest value, and lifted obvious.Therefore, the quality of reconstructed results of the invention is higher.
The subjective vision effect and objective parameter of Integrated comparative each method reconstructed results, it can be seen that the inventive method pair
The reconstruction effect of jpeg compressed image is more preferable, and suitable for the compression image under the different compression quality factors.Therefore, for
Jpeg compressed image, the present invention are a kind of effective super resolution ratio reconstruction methods.
Claims (5)
1. the jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks, it is characterised in that comprise the following steps：
Step 1：For jpeg compressed image, the Super-resolution reconstruction established model based on convolutional neural networks is built；
Step 2：Using training image, the convolutional neural networks that are built in training step one；
Step 3：The low-resolution image through JPEG compression is carried out using the convolutional neural networks model trained in step 2
Rebuild.
2. the jpeg compressed image super resolution ratio reconstruction method according to claim 1 based on convolutional neural networks, it is special
It is by removing pinch effect network, increase resolution network and quality enhancing network to levy in constructed reconstruction model in step 1
Form.
3. the jpeg compressed image super resolution ratio reconstruction method according to claim 1 based on convolutional neural networks, it is special
Training can be optimized end to end in constructed model in step 1 by levying.
4. the jpeg compressed image super resolution ratio reconstruction method according to claim 1 based on convolutional neural networks, it is special
Sign is the network training method used in step 2, i.e., first pinch effect network, increase resolution net are removed in training stage by stage
Network and quality enhancing network, are then optimized end to end to whole model again.
5. the jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks according to claim 1-4, its
The super resolution ratio reconstruction method for jpeg compressed image design is characterised as, but this method can be extended to other images
Or video compression standard.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710783918.1A CN107563965A (en) | 2017-09-04 | 2017-09-04 | Jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710783918.1A CN107563965A (en) | 2017-09-04 | 2017-09-04 | Jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107563965A true CN107563965A (en) | 2018-01-09 |
Family
ID=60978900
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710783918.1A Pending CN107563965A (en) | 2017-09-04 | 2017-09-04 | Jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107563965A (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108416748A (en) * | 2018-02-26 | 2018-08-17 | 阿博茨德（北京）科技有限公司 | The image pre-processing method and device of JPEG compression document |
| CN108629737A (en) * | 2018-05-09 | 2018-10-09 | 复旦大学 | A method of improving jpeg format image spatial resolution |
| CN108765290A (en) * | 2018-05-29 | 2018-11-06 | 天津大学 | A kind of super resolution ratio reconstruction method based on improved dense convolutional neural networks |
| CN109118428A (en) * | 2018-06-07 | 2019-01-01 | 西安电子科技大学 | A kind of image super-resolution rebuilding method based on feature enhancing |
| CN109361919A (en) * | 2018-10-09 | 2019-02-19 | 四川大学 | A kind of image coding efficiency method for improving combined super-resolution and remove pinch effect |
| CN109584161A (en) * | 2018-11-29 | 2019-04-05 | 四川大学 | The Remote sensed image super-resolution reconstruction method of convolutional neural networks based on channel attention |
| CN109996085A (en) * | 2019-04-30 | 2019-07-09 | 北京金山云网络技术有限公司 | Model training method, image processing method, device and electronic equipment |
| CN110062125A (en) * | 2019-04-18 | 2019-07-26 | 西安万像电子科技有限公司 | Image transfer method and device |
| WO2020048354A1 (en) * | 2018-09-04 | 2020-03-12 | 杭州海康威视数字技术股份有限公司 | Neural network model compression method and apparatus, and computer device |
| WO2020057492A1 (en) * | 2018-09-19 | 2020-03-26 | 北京金山办公软件股份有限公司 | Image compression and decompression method and apparatus, electronic device, and storage medium |
| CN110930333A (en) * | 2019-11-22 | 2020-03-27 | 北京金山云网络技术有限公司 | Image restoration method and device, electronic equipment and computer-readable storage medium |
| KR102098375B1 (en) * | 2018-11-15 | 2020-04-08 | 충남대학교산학협력단 | System for increasing resolution of jpeg compressed image and method thereof |
| CN111726638A (en) * | 2019-03-18 | 2020-09-29 | 四川大学 | HEVC (high efficiency video coding) optimization method combining decompression effect and super-resolution |
| CN112150356A (en) * | 2019-06-28 | 2020-12-29 | 四川大学 | Single compressed image super-resolution reconstruction method based on cascade framework |
| CN112188217A (en) * | 2019-07-01 | 2021-01-05 | 四川大学 | JPEG compressed image decompression effect removing method combining DCT domain and pixel domain learning |
| CN112509071A (en) * | 2021-01-29 | 2021-03-16 | 电子科技大学 | Chroma information compression and reconstruction method assisted by luminance information |
| CN113225590A (en) * | 2021-05-06 | 2021-08-06 | 深圳思谋信息科技有限公司 | Video super-resolution enhancement method and device, computer equipment and storage medium |
| WO2022011571A1 (en) * | 2020-07-14 | 2022-01-20 | Oppo广东移动通信有限公司 | Video processing method and apparatus, and device, decoder, system and storage medium |
| CN114913072A (en) * | 2022-05-16 | 2022-08-16 | 中国第一汽车股份有限公司 | Image processing method and device, storage medium and processor |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069825A (en) * | 2015-08-14 | 2015-11-18 | 厦门大学 | Image super resolution reconstruction method based on deep belief network |
| CN105611303A (en) * | 2016-03-07 | 2016-05-25 | 京东方科技集团股份有限公司 | Image compression system, decompression system, training method and device, and display device |
| CN106709875A (en) * | 2016-12-30 | 2017-05-24 | 北京工业大学 | Compressed low-resolution image restoration method based on combined deep network |
| CN106791927A (en) * | 2016-12-23 | 2017-05-31 | 福建帝视信息科技有限公司 | A kind of video source modeling and transmission method based on deep learning |
| CN106952229A (en) * | 2017-03-15 | 2017-07-14 | 桂林电子科技大学 | Image super-resolution rebuilding method based on the enhanced modified convolutional network of data |
-
2017
- 2017-09-04 CN CN201710783918.1A patent/CN107563965A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069825A (en) * | 2015-08-14 | 2015-11-18 | 厦门大学 | Image super resolution reconstruction method based on deep belief network |
| CN105611303A (en) * | 2016-03-07 | 2016-05-25 | 京东方科技集团股份有限公司 | Image compression system, decompression system, training method and device, and display device |
| CN106791927A (en) * | 2016-12-23 | 2017-05-31 | 福建帝视信息科技有限公司 | A kind of video source modeling and transmission method based on deep learning |
| CN106709875A (en) * | 2016-12-30 | 2017-05-24 | 北京工业大学 | Compressed low-resolution image restoration method based on combined deep network |
| CN106952229A (en) * | 2017-03-15 | 2017-07-14 | 桂林电子科技大学 | Image super-resolution rebuilding method based on the enhanced modified convolutional network of data |
Cited By (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108416748A (en) * | 2018-02-26 | 2018-08-17 | 阿博茨德（北京）科技有限公司 | The image pre-processing method and device of JPEG compression document |
| CN108629737A (en) * | 2018-05-09 | 2018-10-09 | 复旦大学 | A method of improving jpeg format image spatial resolution |
| CN108629737B (en) * | 2018-05-09 | 2022-11-18 | 复旦大学 | Method for improving JPEG format image space resolution |
| CN108765290A (en) * | 2018-05-29 | 2018-11-06 | 天津大学 | A kind of super resolution ratio reconstruction method based on improved dense convolutional neural networks |
| CN109118428A (en) * | 2018-06-07 | 2019-01-01 | 西安电子科技大学 | A kind of image super-resolution rebuilding method based on feature enhancing |
| WO2020048354A1 (en) * | 2018-09-04 | 2020-03-12 | 杭州海康威视数字技术股份有限公司 | Neural network model compression method and apparatus, and computer device |
| WO2020057492A1 (en) * | 2018-09-19 | 2020-03-26 | 北京金山办公软件股份有限公司 | Image compression and decompression method and apparatus, electronic device, and storage medium |
| CN110933432A (en) * | 2018-09-19 | 2020-03-27 | 珠海金山办公软件有限公司 | Image compression method, image decompression method, image compression device, image decompression device, electronic equipment and storage medium |
| CN109361919A (en) * | 2018-10-09 | 2019-02-19 | 四川大学 | A kind of image coding efficiency method for improving combined super-resolution and remove pinch effect |
| KR102098375B1 (en) * | 2018-11-15 | 2020-04-08 | 충남대학교산학협력단 | System for increasing resolution of jpeg compressed image and method thereof |
| CN109584161A (en) * | 2018-11-29 | 2019-04-05 | 四川大学 | The Remote sensed image super-resolution reconstruction method of convolutional neural networks based on channel attention |
| CN111726638A (en) * | 2019-03-18 | 2020-09-29 | 四川大学 | HEVC (high efficiency video coding) optimization method combining decompression effect and super-resolution |
| CN110062125A (en) * | 2019-04-18 | 2019-07-26 | 西安万像电子科技有限公司 | Image transfer method and device |
| CN109996085A (en) * | 2019-04-30 | 2019-07-09 | 北京金山云网络技术有限公司 | Model training method, image processing method, device and electronic equipment |
| CN109996085B (en) * | 2019-04-30 | 2021-05-14 | 北京金山云网络技术有限公司 | Model training method, image processing method and device and electronic equipment |
| CN112150356A (en) * | 2019-06-28 | 2020-12-29 | 四川大学 | Single compressed image super-resolution reconstruction method based on cascade framework |
| CN112188217A (en) * | 2019-07-01 | 2021-01-05 | 四川大学 | JPEG compressed image decompression effect removing method combining DCT domain and pixel domain learning |
| CN112188217B (en) * | 2019-07-01 | 2022-03-04 | 四川大学 | JPEG compressed image decompression effect removing method combining DCT domain and pixel domain learning |
| CN110930333A (en) * | 2019-11-22 | 2020-03-27 | 北京金山云网络技术有限公司 | Image restoration method and device, electronic equipment and computer-readable storage medium |
| WO2022011571A1 (en) * | 2020-07-14 | 2022-01-20 | Oppo广东移动通信有限公司 | Video processing method and apparatus, and device, decoder, system and storage medium |
| CN112509071A (en) * | 2021-01-29 | 2021-03-16 | 电子科技大学 | Chroma information compression and reconstruction method assisted by luminance information |
| CN112509071B (en) * | 2021-01-29 | 2021-04-30 | 电子科技大学 | Chroma information compression and reconstruction method assisted by luminance information |
| CN113225590A (en) * | 2021-05-06 | 2021-08-06 | 深圳思谋信息科技有限公司 | Video super-resolution enhancement method and device, computer equipment and storage medium |
| CN114913072A (en) * | 2022-05-16 | 2022-08-16 | 中国第一汽车股份有限公司 | Image processing method and device, storage medium and processor |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107563965A (en) | Jpeg compressed image super resolution ratio reconstruction method based on convolutional neural networks | |
| CN108537733B (en) | Super-resolution reconstruction method based on multi-path deep convolutional neural network | |
| Wang et al. | Esrgan: Enhanced super-resolution generative adversarial networks | |
| CN109118432B (en) | Image super-resolution reconstruction method based on rapid cyclic convolution network | |
| CN108647775B (en) | Super-resolution image reconstruction method based on full convolution neural network single image | |
| CN110276721A (en) | Image super-resolution rebuilding method based on cascade residual error convolutional neural networks | |
| CN112734646B (en) | Image super-resolution reconstruction method based on feature channel division | |
| CN110189253A (en) | A kind of image super-resolution rebuilding method generating confrontation network based on improvement | |
| CN111932461B (en) | Self-learning image super-resolution reconstruction method and system based on convolutional neural network | |
| CN109035146A (en) | A kind of low-quality image oversubscription method based on deep learning | |
| CN107590775B (en) | Image super-resolution amplification method using regression tree field | |
| Luo et al. | Lattice network for lightweight image restoration | |
| Yang et al. | Image super-resolution based on deep neural network of multiple attention mechanism | |
| CN107833182A (en) | The infrared image super resolution ratio reconstruction method of feature based extraction | |
| CN112001843A (en) | Infrared image super-resolution reconstruction method based on deep learning | |
| CN112017116B (en) | Image super-resolution reconstruction network based on asymmetric convolution and construction method thereof | |
| CN113516601A (en) | Image restoration technology based on deep convolutional neural network and compressed sensing | |
| CN112288630A (en) | Super-resolution image reconstruction method and system based on improved wide-depth neural network | |
| CN112950480A (en) | Super-resolution reconstruction method integrating multiple receptive fields and dense residual attention | |
| CN110288529B (en) | Single image super-resolution reconstruction method based on recursive local synthesis network | |
| CN112150356A (en) | Single compressed image super-resolution reconstruction method based on cascade framework | |
| CN108492252A (en) | Face image super-resolution reconstruction method based on secondary reconstruction | |
| CN106709874B (en) | Compressed low-resolution face image restoration method based on face structure correlation | |
| CN110047038B (en) | Single-image super-resolution reconstruction method based on hierarchical progressive network | |
| CN111461978A (en) | Attention mechanism-based resolution-by-resolution enhanced image super-resolution restoration method |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180109 |