CN109255755B - Image super-resolution reconstruction method based on multi-column convolutional neural network - Google Patents
Image super-resolution reconstruction method based on multi-column convolutional neural network Download PDFInfo
- Publication number
- CN109255755B CN109255755B CN201811241002.4A CN201811241002A CN109255755B CN 109255755 B CN109255755 B CN 109255755B CN 201811241002 A CN201811241002 A CN 201811241002A CN 109255755 B CN109255755 B CN 109255755B
- Authority
- CN
- China
- Prior art keywords
- image
- resolution
- training
- model
- convolutional neural
- 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
- 238000013527 convolutional neural network Methods 0.000 title claims abstract description 39
- 238000000034 method Methods 0.000 title claims abstract description 33
- 238000012549 training Methods 0.000 claims abstract description 51
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 23
- 238000000605 extraction Methods 0.000 claims abstract description 9
- 238000013135 deep learning Methods 0.000 claims abstract description 7
- 230000003416 augmentation Effects 0.000 claims description 16
- 238000005457 optimization Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 abstract description 6
- 101100365548 Caenorhabditis elegans set-14 gene Proteins 0.000 abstract description 3
- 230000000694 effects Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 241000556720 Manga Species 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000003709 image segmentation Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an image super-resolution reconstruction method based on a multi-column convolutional neural network. First, a multi-column convolutional neural network model is designed according to a deep learning algorithm, including a feature extraction portion and an image reconstruction portion. The original image is then cut into small blocks and these high resolution small blocks are downsampled to obtain low resolution small blocks, which are used to build the training set. And finally, training the model by using a random gradient descent algorithm to obtain a model for reconstructing the low-resolution image into the high-resolution image, and recovering the input low-resolution image reconstruction into the corresponding high-resolution image. The method of the invention tests on five general image databases of Set5, set14, BSDS100, urban100 and Manga109, and has higher robustness and accuracy.
Description
Technical Field
The invention relates to an image super-resolution reconstruction method, in particular to a super-resolution reconstruction method based on a multi-column convolutional neural network image, belonging to the utilization of image processing and reconstruction technologies.
Background
With the development of information technology, images have been widely used in various scenes as a main information transmission medium therein. In many fields, there is a high demand for the quality of images, so it has been difficult for the information age of high-speed development for low-quality images to meet the demands of specific scenes. Image resolution is an important indicator for measuring image quality, and the higher the image resolution, the more detailed information the image contains. Image Super-Resolution (SR) reconstruction belongs to an image processing technology, and a High-Resolution (HR) image is reconstructed from a Low-Resolution (LR) image. Super-resolution reconstruction of images has wide applications such as face recognition, medical imaging and remote sensing techniques.
At present, convolutional neural networks (Convolutional Neural Networks, CNN) have made significant progress in computer vision tasks such as target detection, human behavior recognition, and image segmentation. In particular, the super-resolution method based on the convolutional neural network has better reconstruction effect than the traditional methods such as dictionary learning, local linear regression, random forest and the like. In 2014, dong et al have achieved Image super-resolution reconstruction (Super Resolution Convolutional Neural Network, SRCNN) using convolutional neural networks, see references Dong, chao, et al, "Image super-resolution using deep convolutional networks," IEEE transactions on pattern analysis and machine intelligence 38.2.2 (2016): 295-307. The low-resolution image after bicubic interpolation pretreatment is input into an end-to-end deep convolutional neural network, and the mapping relation between the low-resolution image and the high-resolution image is gradually subjected to learning treatment. Because the end-to-end training mode in the deep learning is adopted, compared with the traditional method, the method remarkably improves the image super-resolution reconstruction effect.
Although the proposed image super-resolution algorithm based on the convolutional neural network solves the problems of poor robustness, complex calculation and the like of the traditional image super-resolution reconstruction algorithm, the existing image super-resolution method based on the convolutional neural network firstly uses a bicubic interpolation (Bicubic Interpolation) method to amplify a low-resolution image to the size of a high-resolution image to be reconstructed before extracting the low-resolution image features, extracts features from the bicubic interpolation image, introduces a lot of redundant information into the bicubic interpolation image, and is not helpful for feature extraction. Therefore, the existing method also has the problems of poor reconstruction capability, poor visual effect and the like for images with rich details.
Disclosure of Invention
The invention aims to reconstruct a low-resolution image with higher quality, and provides an image super-resolution method based on a multi-column convolutional neural network. The method can effectively improve the peak signal-to-noise ratio and the structural similarity of the super-resolution reconstructed image, and has better effect on subjective vision. In addition, the method has important reference significance for the application of the convolutional neural network in the super-resolution of the image.
To achieve the above object, the present invention is conceived as follows:
first, a multi-column convolutional neural network model is designed according to a deep learning algorithm, including a feature extraction portion and an image reconstruction portion. The original image is then cut into small blocks and these high resolution small blocks are downsampled to obtain low resolution small blocks, which are used to build the training set. And finally, training the model by using a random gradient descent algorithm to obtain a model for reconstructing the low-resolution image to the high-resolution image, namely the image super-resolution reconstruction model of the multi-column convolutional neural network.
According to the conception, the invention adopts the following technical scheme:
an image super-resolution method based on a multi-column convolutional neural network comprises the following steps:
step 2, image augmentation (Image Augmentation): the large-scale data set is a precondition of successfully using the depth network, and the image augmentation is to generate similar training samples but different training samples by making a series of random changes on the training images, so that the scale of the training data set is enlarged; the scale of the training set is increased through image augmentation, and the dependence of the model on certain attributes is reduced, so that the generalization capability of the model is improved, and the used image augmentation method comprises rotation, scaling and mirroring;
step 3, training set establishment: cutting an original image into small blocks on the training set with the increased scale obtained according to the step 2, and downsampling the small blocks with high resolution so as to obtain small blocks with low resolution, and establishing the training set by using the small block pairs with low resolution and high resolution;
step 4, training a multi-column convolutional neural network model: training an image super-resolution reconstruction model on the training set obtained in the step 3, and obtaining a model for reconstructing a low-resolution image to a high-resolution image after training by using a random gradient descent algorithm through an optimization algorithm;
step 5, reconstructing the super-resolution of the image: the model trained in step 4 reconstructs the input low resolution image back into a corresponding high resolution image.
The method mainly considers the multi-scale characteristics of the image, so that the multi-scale characteristics in the image can be effectively extracted and fused by means of a multi-column convolution neural network model. Features are directly extracted from the low-resolution image, so that the calculated amount is reduced, and the reconstruction speed of the image is improved. In order to accelerate the convergence rate of the image super-resolution reconstruction model, the extracted multi-scale features are used for reconstructing interpolation images of the high-resolution image and the bicubic interpolation image instead of directly reconstructing the high-resolution image from the features, so that the training difficulty of a network is reduced, and meanwhile, the super-resolution reconstruction quality of the image is improved.
Compared with the prior art, the invention has the following obvious prominent substantive features and obvious advantages:
1. the method fully considers the multi-scale characteristics of the image, namely, the condition that objects in the image have different scales. An image super-resolution reconstruction model based on a multi-column convolutional neural network is provided.
2. The method directly extracts the features from the low-resolution image which is not preprocessed, so that the calculated amount of the features is reduced, and the reconstruction speed of the model is improved.
3. The method utilizes the extracted multi-scale features to reconstruct the interpolation images of the high-resolution image and the bicubic interpolation image instead of directly reconstructing the high-resolution image from the features, reduces the training difficulty of the model and improves the super-resolution reconstruction quality of the image.
Drawings
Fig. 1 is a network structure block diagram of an image super-resolution reconstruction method based on a multi-column convolutional neural network.
FIG. 2 is a graph showing a comparison of the effect of super-resolution reconstruction when the magnification of "button" in the Set5 test Set is 2.
Fig. 3 is a comparison of super-resolution reconstruction effects at a magnification of 3 in the BSDS100 test set of "21077".
Fig. 4 is a comparison of the super-resolution reconstruction effect at an "img023" magnification of 4 in the Urban100 test set.
FIG. 5 is a comparison of the effect of super-resolution reconstruction at a magnification of 4 in the Manga109 test set.
Detailed Description
Preferred embodiments of the present invention are described in detail below with reference to the attached drawing figures:
the structure of the multi-column convolutional neural network of this embodiment is shown in fig. 1. The method is realized by programming simulation in a Ubuntu 16.04, pyTorch environment. First, a multi-column convolutional neural network model is designed according to a deep learning algorithm, including a feature extraction portion and an image reconstruction portion. The original image is then cut into small blocks and these high resolution small blocks are downsampled to obtain low resolution small blocks, which are used to build the training set. And finally, training the model by using a random gradient descent algorithm to obtain a model for reconstructing the low-resolution image to the high-resolution image, namely the image super-resolution reconstruction model of the multi-column convolutional neural network.
The method specifically comprises the following steps:
step 2, image augmentation (Image Augmentation): the large-scale data set is a precondition of successfully using the depth network, and the image augmentation is to generate similar training samples but different training samples by making a series of random changes on the training images, so that the scale of the training data set is enlarged; the scale of the training set is increased through image augmentation, and the dependence of the model on certain attributes is reduced, so that the generalization capability of the model is improved, and the used image augmentation method comprises rotation, scaling and mirroring;
step 3, training set establishment: cutting an original image into small blocks on the training set with the increased scale obtained according to the step 2, and downsampling the small blocks with high resolution so as to obtain small blocks with low resolution, and establishing the training set by using the small block pairs with low resolution and high resolution;
step 4, training a multi-column convolutional neural network model: training an image super-resolution reconstruction model on the training set obtained in the step 3, and obtaining a model for reconstructing a low-resolution image to a high-resolution image after training by using a random gradient descent algorithm through an optimization algorithm;
step 5, reconstructing the super-resolution of the image: the model trained in step 4 reconstructs the input low resolution image back into a corresponding high resolution image.
In the step 1, a cascade multi-column convolutional neural network is proposed to extract multi-scale features from the low-resolution image, and then reconstruct a corresponding high-resolution image, and the network structure is shown in fig. 1. The proposed network framework uses a number of Multi-Column modules (Multi-Column blocks), each consisting of three columns of convolution layers of different convolution kernel sizes. The proposed model predicts an interpolated image between the bicubic interpolated image and the target high resolution image from the input low resolution image. The proposed model is divided into two parts, a feature extraction part and an image reconstruction part.
In the feature extraction section, first, a convolution layer having 64 convolution kernels of 3×3 is used to extract coarse features. Three cascaded multi-column modules are then used to extract multi-scale features. In this model, no bias is used, so the calculation formula for the convolution layer is as follows:
in the above formula, W l And x represents the input of the learnable weights and convolution layers, respectively. Sigma represents the activation function, in which model a linear unit (Leaky Rectified Linear Unit) with leakage correction is used.
Finally, the extracted features are upsampled using a deconvolution layer, followed by a deconvolution layer using a 3 x 3 convolution layer to obtain the residual image. The calculation formula of the output image size of the deconvolution layer is as follows:
X out ＝(X in -1)×λ-2×ρ+κ, (2)
in the above formula, X in And X out Respectively the input and output of the deconvolution layer, λ represents the step size of the deconvolution, ρ represents the number of rows added 0 on each side of the input, and κ represents the size of the deconvolution kernel. Obviously, λ needs to be set to be the same as the magnification. Table 1 gives the parameter settings of the deconvolution layer at different magnifications.
TABLE 1
In the proposed model, convolution kernels of different sizes are used in each column to extract features. The detailed structure is shown in fig. 1. The calculation formula of the receptive field gamma of the convolution layer is as follows:
γ＝κ+(κ-1)×(n-1), (3)
in the above formula, κ represents the size of the convolution kernel, and n represents the number of convolution layers in each column. According to the above formula, a convolution layer with a convolution kernel size of 3×3, a convolution layer with a convolution kernel size of 5×5, and a convolution layer with a convolution kernel size of 7×7 are used in a multi-column module, so that a receptive field with the same size can be obtained.
In order to extract more reliable features, features extracted from different columns need to be fused on the same receptive field. The feature fusion adopts the method that a1 multiplied by 1 convolution layer is added to the last layer of each column, and then the feature images of the columns are added as elements to enter the fusion. The benefit of adding a1 x 1 convolution layer is that there can be more complex combinations of multi-scale features. In general, more multi-column modules may have better performance, and this embodiment uses three multi-column modules for performance and efficiency tradeoffs.
At the image reconstruction module, a 3 x 3 convolution layer is used to predict the residual image of the high resolution image and the bicubic interpolated image. And adding the residual image predicted by the network and the bicubic interpolation image through elements, so that a corresponding high-resolution image can be reconstructed. The calculation formula of the output image is as follows:
in the above formula, x and
representing the input low resolution image and the output high resolution image of the model, respectively. />
Represents bicubic interpolation,/>
Representing the proposed model.
In the step 2, the training set image used is composed of 91 pictures of Yang and 200 pictures of BSDS. See reference Yang, jiamao, et al, "Image super-resolution via sparse presentation," IEEE transactions on Image processing 19.11.11 (2010): 2861-2873. See also references Martin, david, et al, "A database ofhuman segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statics," Computer Vision,2001.ICCV2001.Proceedings.Eighth IEEE International Conference on.Vol.2.IEEE,2001. The image augmentation mode mainly comprises scaling, rotation and mirroring. Wherein the scaling factor is 1-fold, 0.7-fold and 0.5-fold; the rotation angles are 0 °, 90 °, 180 ° and 270 °; the mirror image is a horizontal mirror image or remains original. After image augmentation, 23 additional versions are obtained in addition to the original.
In said step 3, the original image is cut into small blocks on the scaled-up training set obtained according to step 2, and these small blocks of high resolution are downsampled, resulting in small blocks of low resolution, which are used to build the training set. When the magnification is 2, the image block size is 82×82, the step size is 64, and downsampling is the inverse of the magnification, i.e., 1/2 times. Similarly, when the magnification is 3, the image block size is 123×123, the step size is 48, and the downsampling magnification is 1/3; when the magnification is 4, the image block size is 164×164, the step size is 32, and the downsampling magnification is 1/4. The input low resolution image blocks are all 41 x 41 in size.
In the step 4, training the image super-resolution reconstruction model on the training set obtained in the step 3, the optimization algorithm uses a random gradient descent algorithm (Stochastic Gradient Descent), the batch size is set to 64, the momentum parameter is set to 0.9, and the weight attenuation is set to 10 -4 The learning rate was set to 0.1 and dropped 10 times after every 20 iteration cycles. Since the initial learning rate is relatively high, gradient slices are used to prevent gradient explosion, the gradient slices are set to 0.4, and a model for reconstructing a low-resolution image into a high-resolution image can be obtained after training is completed.
In the step 5, the model trained in the step 4 restores the input low-resolution image reconstruction to the corresponding high-resolution image.
The image super-resolution reconstruction method based on the multi-column convolutional neural network provided by the invention is evaluated by performing experiments on five image databases of Set5, set14, BSDS100, urban100 and Manga 109. Set5, set14, and BSDS100 comprise natural images; urban100 contains an image of a city scene; manga 109. The environment of the experiment is a PyTorch platform under Ubuntu 16.04 operating system, the memory is 16GB, and the GPU is GeForce1070. The peak signal-to-noise ratio (Peak Signal to Noise Ratio, PSNR) and the structural similarity coefficient (Structural Similarity Index, SSIM) are used as evaluation indexes of the super-resolution reconstruction model, the higher the PSNR is, the closer the SSIM is to 1, the higher the coincidence degree between the representation model and the original image is, and the higher the accuracy is, and the results are shown in Table 2. Figures 2-5 compare the effect of different algorithms on the reconstruction of these test sets.
TABLE 2
In table 2, see the following references:
4.Timofte,Radu,Vincent De Smet,and Luc Van Gool."A+:Adjusted anchored neighborhood regression for fast super-resolution."Asian Conference on Computer Vision.Springer,Cham,2014.
5.Huang,Jia-Bin,Abhishek Singh,and Narendra Ahuja."Single image super-resolution from transformed self-exemplars."Proceedings oftheIEEE Conference on Computer Vision andPatternRecognition.2015.
6.Schulter,Samuel,Christian Leistner,and Horst Bischof."Fast and accurate image upscaling with super-resolution forests."Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2015.
7.Dong,Chao,Chen Change Loy,and Xiaoou Tang."Accelerating the super-resolution convolutional neural network."European Conference on Computer Vision.Springer,Cham,2016.
8.Kim,Jiwon,Jung Kwon Lee,and Kyoung Mu Lee."Accurate image super-resolution using very deep convolutional networks."Proceedings oftheIEEE conference on computervision andpattern recognition.2016.
9.Kim,Jiwon,Jung Kwon Lee,and Kyoung Mu Lee."Deeply-recursive convolutional network for image super-resolution."Proceedings oftheIEEE conference on computervision andpattern recognition.2016.
10.Lai,Wei-Sheng,et al."Deep laplacian pyramid networks for fast and accurate superresolution."IEEE Conference on Computer Vision andPatternRecognition.Vol.2.No.3.2017.
wherein the best algorithm for the experimental results is indicated by bold font and the second best algorithm is indicated by underline. From the table it can be seen that the method of the invention has better robustness and accuracy over five databases. The experiment shows that the method has better robustness and accuracy in the super-resolution reconstruction of the image, has low calculation complexity and can be better suitable for real-time video quality monitoring.
Claims (1)
1. The image super-resolution method based on the multi-column convolutional neural network is characterized by comprising the following steps of:
step 1, building a multi-column convolutional neural network model: designing a multi-column convolutional neural network model according to a deep learning algorithm, wherein the multi-column convolutional neural network model comprises a feature extraction part and an image reconstruction part;
step 2, image augmentation: the large-scale data set is a precondition of successfully using the depth network, and the image augmentation is to generate similar training samples but different training samples by making a series of random changes on the training images, so that the scale of the training data set is enlarged; the scale of the training set is increased through image augmentation, and the dependence of the model on certain attributes is reduced, so that the generalization capability of the model is improved, and the used image augmentation method comprises rotation, scaling and mirroring;
step 3, training set establishment: cutting an original image into small blocks on the training set with the increased scale obtained according to the step 2, and downsampling the small blocks with high resolution so as to obtain small blocks with low resolution, and establishing the training set by using the small block pairs with low resolution and high resolution;
step 4, training a multi-column convolutional neural network model: training an image super-resolution reconstruction model on the training set obtained in the step 3, and obtaining a model for reconstructing a low-resolution image to a high-resolution image after training by using a random gradient descent algorithm through an optimization algorithm;
step 5, reconstructing the super-resolution of the image: the model trained in step 4 reconstructs the input low resolution image back into a corresponding high resolution image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811241002.4A CN109255755B (en) | 2018-10-24 | 2018-10-24 | Image super-resolution reconstruction method based on multi-column convolutional neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811241002.4A CN109255755B (en) | 2018-10-24 | 2018-10-24 | Image super-resolution reconstruction method based on multi-column convolutional neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109255755A CN109255755A (en) | 2019-01-22 |
| CN109255755B true CN109255755B (en) | 2023-05-23 |
Family
ID=65046172
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811241002.4A Active CN109255755B (en) | 2018-10-24 | 2018-10-24 | Image super-resolution reconstruction method based on multi-column convolutional neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109255755B (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109949224B (en) * | 2019-02-26 | 2023-06-30 | 北京悦图遥感科技发展有限公司 | Deep learning-based cascade super-resolution reconstruction method and device |
| CN110060208B (en) * | 2019-04-22 | 2022-10-28 | 中国科学技术大学 | Method for improving reconstruction performance of super-resolution algorithm |
| CN110246083B (en) * | 2019-05-10 | 2023-02-24 | 杭州电子科技大学 | Fluorescence microscopic image super-resolution imaging method |
| CN110991213A (en) * | 2019-05-14 | 2020-04-10 | 杨春燕 | Automatic big data distribution platform |
| CN112446826A (en) * | 2019-09-03 | 2021-03-05 | 联咏科技股份有限公司 | Method and device for image super-resolution, image enhancement and model training |
| CN110689509B (en) * | 2019-09-10 | 2024-04-19 | 上海大学 | Video super-resolution reconstruction method based on cyclic multi-column 3D convolution network |
| CN110647934B (en) * | 2019-09-20 | 2022-04-08 | 北京百度网讯科技有限公司 | Training method and device for video super-resolution reconstruction model and electronic equipment |
| CN110728219B (en) * | 2019-09-29 | 2023-09-26 | 天津大学 | 3D face generation method based on multi-column multi-scale graph convolution neural network |
| CN111709900A (en) * | 2019-10-21 | 2020-09-25 | 上海大学 | High dynamic range image reconstruction method based on global feature guidance |
| CN111127317B (en) * | 2019-12-02 | 2023-07-25 | 深圳供电局有限公司 | Image super-resolution reconstruction method, device, storage medium and computer equipment |
| CN111091515B (en) * | 2019-12-24 | 2022-08-09 | 展讯通信(天津)有限公司 | Image restoration method and device, and computer-readable storage medium |
| CN111681168B (en) * | 2020-06-05 | 2023-03-21 | 杭州电子科技大学 | Low-resolution cell super-resolution reconstruction method based on parallel residual error network |
| EP3937120B1 (en) * | 2020-07-08 | 2023-12-20 | Sartorius Stedim Data Analytics AB | Computer-implemented method, computer program product and system for processing images |
| CN112419150B (en) * | 2020-11-06 | 2024-04-02 | 中国科学技术大学 | Image super-resolution reconstruction method of arbitrary multiple based on bilateral upsampling network |
| CN113139899A (en) * | 2021-03-31 | 2021-07-20 | 桂林电子科技大学 | Design method of high-quality light-weight super-resolution reconstruction network model |
| CN113409195A (en) * | 2021-07-06 | 2021-09-17 | 中国标准化研究院 | Image super-resolution reconstruction method based on improved deep convolutional neural network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101719266A (en) * | 2009-12-25 | 2010-06-02 | 西安交通大学 | Affine transformation-based frontal face image super-resolution reconstruction method |
| CN104463324A (en) * | 2014-11-21 | 2015-03-25 | 长沙马沙电子科技有限公司 | Convolution neural network parallel processing method based on large-scale high-performance cluster |
| CN106650786A (en) * | 2016-11-14 | 2017-05-10 | 沈阳工业大学 | Image recognition method based on multi-column convolutional neural network fuzzy evaluation |
| CN107358576A (en) * | 2017-06-24 | 2017-11-17 | 天津大学 | Depth map super resolution ratio reconstruction method based on convolutional neural networks |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI624804B (en) * | 2016-11-07 | 2018-05-21 | 盾心科技股份有限公司 | A method and system for providing high resolution image through super-resolution reconstrucion |
-
2018
- 2018-10-24 CN CN201811241002.4A patent/CN109255755B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101719266A (en) * | 2009-12-25 | 2010-06-02 | 西安交通大学 | Affine transformation-based frontal face image super-resolution reconstruction method |
| CN104463324A (en) * | 2014-11-21 | 2015-03-25 | 长沙马沙电子科技有限公司 | Convolution neural network parallel processing method based on large-scale high-performance cluster |
| CN106650786A (en) * | 2016-11-14 | 2017-05-10 | 沈阳工业大学 | Image recognition method based on multi-column convolutional neural network fuzzy evaluation |
| CN107358576A (en) * | 2017-06-24 | 2017-11-17 | 天津大学 | Depth map super resolution ratio reconstruction method based on convolutional neural networks |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109255755A (en) | 2019-01-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109255755B (en) | Image super-resolution reconstruction method based on multi-column convolutional neural network | |
| CN111369440B (en) | Model training and image super-resolution processing method, device, terminal and storage medium | |
| CN107610194B (en) | Magnetic resonance image super-resolution reconstruction method based on multi-scale fusion CNN | |
| Ma et al. | Structure-preserving image super-resolution | |
| CN112232134B (en) | Human body posture estimation method based on hourglass network and attention mechanism | |
| CN111709882B (en) | Super-resolution fusion calculation method based on sub-pixel convolution and feature segmentation | |
| CN112819910A (en) | Hyperspectral image reconstruction method based on double-ghost attention machine mechanism network | |
| Guo et al. | MambaIR: A Simple Baseline for Image Restoration with State-Space Model | |
| CN110288524B (en) | Deep learning super-resolution method based on enhanced upsampling and discrimination fusion mechanism | |
| CN116012581A (en) | Image segmentation method based on dual attention fusion | |
| Yang et al. | Image super-resolution reconstruction based on improved Dirac residual network | |
| Shi et al. | Exploiting multi-scale parallel self-attention and local variation via dual-branch transformer-cnn structure for face super-resolution | |
| Gendy et al. | Balanced spatial feature distillation and pyramid attention network for lightweight image super-resolution | |
| Deng et al. | Multiple frame splicing and degradation learning for hyperspectral imagery super-resolution | |
| Xie et al. | Adaptive densely connected single image super-resolution | |
| CN117097876B (en) | Event camera image reconstruction method based on neural network | |
| CN117351363A (en) | Remote sensing image building extraction method based on transducer | |
| CN116433654A (en) | Improved U-Net network spine integral segmentation method | |
| CN116708807A (en) | Compression reconstruction method and compression reconstruction device for monitoring video | |
| CN115187463A (en) | Landslide remote sensing image set super-resolution reconstruction method and system | |
| CN115909088A (en) | Optical remote sensing image target detection method based on super-resolution feature aggregation | |
| CN114332278A (en) | OCTA image motion correction method based on deep learning | |
| Zhang et al. | Single infrared remote sensing image super-resolution via supervised deep learning | |
| CN110569790A (en) | Residential area element extraction method based on texture enhancement convolutional network | |
| Wang et al. | Infrared small target detection based on the combination of single image super-resolution reconstruction and YOLOX |
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 |