Video super-resolution - Knowledge (XXG)

1158: 3194: 911:(the multi-resolution mixture network) consists of three modules: bottleneck, exchange, and residual. Bottleneck unit extract features that have the same resolution as input frames. Exchange module exchange features between neighboring frames and enlarges feature maps. Residual module extract features after exchange one 2878:. Models are ranked by BSQ-rate over subjective score. The resolution of ground-truth frames is 1920×1080. The tested scale factor is 4. 17 models were tested. 5 video codecs were used to compress ground-truth videos. Top combinations of Super-Resolution methods and video codecs are performed in the table: 531:

to estimate transformation from low-resolution frame to high-resolution one. To improve the final result these methods consider temporal correlation among low-resolution sequences. Some approaches also consider temporal correlation among high-resolution sequence. To approximate Kalman filter a common

430:

Video super-resolution approaches tend to have more components than the image counterparts as they need to exploit the additional temporal dimension. Complex designs are not uncommon. Some most essential components for VSR are guided by four basic functionalities: Propagation, Alignment, Aggregation,

3221:

There are situations where hand motion is simply not present because the device is stabilized (e.g. placed on a tripod). There is a way to simulate natural hand motion by intentionally slightly moving the camera. The movements are extremely small so they don't interfere with regular photos. You can

1070:

invertible spatio-temporal network) consists of spatial, temporal and reconstruction module. Spatial module composed of residual invertible blocks (RIB), which extract spatial features effectively. The output of the spatial module is processed by the temporal module, which extracts spatio-temporal

2085:

The Youku-VESR Challenge was organized to check models' ability to cope with degradation and noise, which are real for Youku online video-watching application. The proposed dataset consists of 1000 videos, each length is 4–6 seconds. The resolution of ground-truth frames is 1920×1080. The tested

3217:

When we capture a lot of sequential photos with a smartphone or handheld camera, there is always some movement present between the frames because of the hand motion. We can take advantage of this hand tremor by combining the information on those images. We choose a single image as the "base" or

2569:

The MSU Video Super-Resolution Benchmark was organized by MSU and proposed three types of motion, two ways to lower resolution, and eight types of content in the dataset. The resolution of ground-truth frames is 1920×1280. The tested scale factor is 4. 14 models were tested. To evaluate models'

1765:

Dataset REDS was collected for this challenge. It consists of 30 videos of 100 frames each. The resolution of ground-truth frames is 1280×720. The tested scale factor is 4. To evaluate models' performance PSNR and SSIM were used. The best participants' results are performed in the table:

457:

methods could be used too, generating high-resolution frames independently from their neighbours, but it's less effective and introduces temporal instability. There are a few traditional methods, which consider the video super-resolution task as an optimization problem. Last years

3209:

is used to reconstruct the photos from partial color information. A single frame doesn't give us enough data to fill in the missing colors, however, we can receive some of the missing information from multiple images taken one after the other. This process is known as

680:. The motion compensation transformer (MCT) is used for motion estimation. The sub-pixel motion compensation layer (SPMC) compensates motion. Fusion step uses encoder-decoder architecture and ConvLSTM module to unit information from both spatial and temporal dimensions 809:

Another way to align neighboring frames with target one is deformable convolution. While usual convolution has fixed kernel, deformable convolution on the first step estimate shifts for kernel and then do convolution. Examples of such methods:

3125:

In many areas, working with video, we deal with different types of video degradation, including downscaling. The resolution of video can be degraded because of imperfections of measuring devices, such as optical degradations and limited

3130:. Bad light and weather conditions add noise to video. Object and camera motion also decrease video quality. Super Resolution techniques help to restore the original video. It's useful in a wide range of applications, such as 1122:

non-local method) extract spatio-temporal features by non-local residual blocks, then fuse them by progressive fusion residual block (PFRB). The result of these blocks is a residual image. The final result is gained by adding

3204:

Reconstructing details on digital photographs is a difficult task since these photographs are already incomplete: the camera sensor elements measure only the intensity of the light, not directly its color. A process called

4469:

Caballero, Jose; Ledig, Christian; Aitken, Andrew; Acosta, Alejandro; Totz, Johannes; Wang, Zehan; Shi, Wenzhe (2016-11-16). "Real-Time Video Super-Resolution with Spatio-Temporal Networks and Motion Compensation".

905:(frame and feature-context video super-resolution) takes unaligned low-resolution frames and output high-resolution previous frames to simultaneously restore high-frequency details and maintain temporal consistency 2869:

The MSU Super-Resolution for Video Compression Benchmark was organized by MSU. This benchmark tests models' ability to work with compressed videos. The dataset consists of 9 videos, compressed with different

4928:

Bao, Wenbo; Lai, Wei-Sheng; Zhang, Xiaoyun; Gao, Zhiyong; Yang, Ming-Hsuan (2021-03-01). "MEMC-Net: Motion Estimation and Motion Compensation Driven Neural Network for Video Interpolation and Enhancement".

1042:(the spatio-temporal convolutional network) extract features in the spatial module, pass them through the recurrent temporal module and final reconstruction module. Temporal consistency is maintained by 1513:

A few benchmarks in video super-resolution were organized by companies and conferences. The purposes of such challenges are to compare diverse algorithms and to find the state-of-the-art for the task.

1077:(the residual recurrent convolutional network) is a bidirectional recurrent network, which calculates a residual image. Then the final result is gained by adding a bicubically upsampled input frame 134: 1763:

and proposed two tracks for Video Super-Resolution: clean (only bicubic degradation) and blur (blur added firstly). Each track had more than 100 participants and 14 final results were submitted.

1134:(the novel video super‐resolution network) aligns frames with target one by temporal‐spatial non‐local operation. To integrate information from aligned frames an attention‐based mechanism is used 837:(The temporally deformable alignment network) consists of an alignment module and a reconstruction module. Alignment performed by deformable convolution based on feature extraction and alignment 764:

back-projection network). The input of each recurrent projection module features from the previous frame, features from the consequence of frames, and optical flow between neighboring frames

716:

in coarse-to-fine manner. Then the low-resolution optical flow is estimated by a space-to-depth transformation. The final super-resolution result is gained from aligned low-resolution frames

554:

estimate motion between frames, upscale a reference frame, and warp neighboring frames to the high-resolution reference one. To construct result, these upscaled frames are fused together by

6083:

Yi, Peng; Wang, Zhongyuan; Jiang, Kui; Jiang, Junjun; Lu, Tao; Ma, Jiayi (2020). "A Progressive Fusion Generative Adversarial Network for Realistic and Consistent Video Super-Resolution".

5315:

Isobe, Takashi; Li, Songjiang; Jia, Xu; Yuan, Shanxin; Slabaugh, Gregory; Xu, Chunjing; Li, Ya-Li; Wang, Shengjin; Tian, Qi (2020). "Video Super-Resolution With Temporal Group Attention".

2161:). Dataset consists of 328 video sequences of 120 frames each. The resolution of ground-truth frames is 1920×1080. The tested scale factor is 16. Top methods are performed in the table: 606:

is often used along with MAP and helps to preserve similarity in neighboring patches. Huber MRFs are used to preserve sharp edges. Gaussian MRF can smooth some edges, but remove noise.

6159:

Zvezdakova, A. V.; Kulikov, D. L.; Zvezdakov, S. V.; Vatolin, D. S. (2020). "BSQ-rate: a new approach for video-codec performance comparison and drawbacks of current solutions".

1760: 1544: 744:(task-oriented flow) is a combination of optical flow network and reconstruction network. Estimated optical flow is suitable for a particular task, such as video super resolution 395: 336: 5587:

Jo, Younghyun; Oh, Seoung Wug; Kang, Jaeyeon; Kim, Seon Joo (2018). "Deep Video Super-Resolution Network Using Dynamic Upsampling Filters Without Explicit Motion Compensation".

4049:

Huhle, Benjamin; Schairer, Timo; Jenke, Philipp; Straßer, Wolfgang (2010). "Fusion of range and color images for denoising and resolution enhancement with a non-local filter".

243: 6278: 827:(The deformable non-local network) has alignment module, based on deformable convolution with the hierarchical feature fusion module (HFFB) for better quality) and non-local 662:

uses a spatial motion compensation transformer module (MCT), which estimates and compensates motion. Then a series of convolutions performed to extract feature and fuse them

817:(The enhanced deformable video restoration) can be divided into two main modules: the pyramid, cascading and deformable (PCD) module for alignment and the temporal-spatial 5723:

Zhang, Dongyang; Shao, Jie; Liang, Zhenwen; Liu, Xueliang; Shen, Heng Tao (2020). "Multi-branch Networks for Video Super-Resolution with Dynamic Reconstruction Strategy".

5069:

Chan, Kelvin C. K.; Wang, Xintao; Yu, Ke; Dong, Chao; Loy, Chen Change (2020-12-03). "BasicVSR: The Search for Essential Components in Video Super-Resolution and Beyond".

4534:

Liu, Ding; Wang, Zhaowen; Fan, Yuchen; Liu, Xianming; Wang, Zhangyang; Chang, Shiyu; Huang, Thomas (2017). "Robust Video Super-Resolution with Learned Temporal Dynamics".

1248:

Currently, there aren't so many objective metrics to verify video super-resolution method's ability to restore real details. Research is currently underway in this area.

5358:

Lucas, Alice; Lopez-Tapia, Santiago; Molina, Rafael; Katsaggelos, Aggelos K. (2019). "Generative Adversarial Networks and Perceptual Losses for Video Super-Resolution".

2570:

performance PSNR and SSIM were used with shift compensation. Also proposed a few new metrics: ERQAv1.0, QRCRv1.0, and CRRMv1.0. Top methods are performed in the table:

6308: 6230: 4720:

Chu, Mengyu; Xie, You; Mayer, Jonas; Leal-Taixé, Laura; Thuerey, Nils (2020-07-08). "Learning temporal coherence via self-supervision for GAN-based video generation".

3327:

Bose, N.K.; Kim, H.C.; Zhou, B. (1994). "Performance analysis of the TLS algorithm for image reconstruction from a sequence of undersampled noisy and blurred frames".

843:

for Video Super-Resolution uses the multi-scale dilated deformable convolution for frame alignment and the Modulative Feature Fusion Branch to integrate aligned frames

686:(robust video super-resolution) have two branches: one for spatial alignment and another for temporal adaptation. The final frame is a weighted sum of branches' output 4826:

Wang, Zhongyuan; Yi, Peng; Jiang, Kui; Jiang, Junjun; Han, Zhen; Lu, Tao; Ma, Jiayi (2019). "Multi-Memory Convolutional Neural Network for Video Super-Resolution".

1275:

for evaluation. It's important to verify models' ability to restore small details, text, and objects with complicated structure, to cope with big motion and noise.

421: 362: 299: 271: 166: 6602: 6303: 3360:

Tekalp, A.M.; Ozkan, M.K.; Sezan, M.I. (1992). "High-resolution image reconstruction from lower-resolution image sequences and space-varying image restoration".

5133:

Wang, Xintao; Chan, Kelvin C. K.; Yu, Ke; Dong, Chao; Loy, Chen Change (2019-05-07). "EDVR: Video Restoration with Enhanced Deformable Convolutional Networks".

210: 188: 6021:

Isobe, Takashi; Jia, Xu; Gu, Shuhang; Li, Songjiang; Wang, Shengjin; Tian, Qi (2020-08-02). "Video Super-Resolution with Recurrent Structure-Detail Network".

3644:

Costa, Guilherme Holsbach; Bermudez, Jos Carlos Moreira (2007). "Statistical Analysis of the LMS Algorithm Applied to Super-Resolution Image Reconstruction".

3222:

observe these motions on Google Pixel 3 phone by holding it perfectly still (e.g. pressing it against the window) and maximally pinch-zooming the viewfinder.

1023:(the dynamic multiple branch network) has three branches to exploit information from multiple resolutions. Finally, information from branches fuse dynamically 6535: 6268: 5566:

Li, Wenbo; Tao, Xin; Guo, Taian; Qi, Lu; Lu, Jiangbo; Jia, Jiaya (2020-07-23). "MuCAN: Multi-Correspondence Aggregation Network for Video Super-Resolution".

5256:

Song, Huihui; Xu, Wenjie; Liu, Dong; Liua, Bo; Liub, Qingshan; Metaxas, Dimitris N. (2021). "Multi-Stage Feature Fusion Network for Video Super-Resolution".

4194:

Farsiu, Sina; Robinson, Dirk; Elad, Michael; Milanfar, Peyman (2003-11-20). "Robust shift and add approach to superresolution". In Tescher, Andrew G. (ed.).

867:) divide input frames to N groups dependent on time difference and extract information from each group independently. Fast Spatial Alignment module based on 5620:

Li, Sheng; He, Fengxiang; Du, Bo; Zhang, Lefei; Xu, Yonghao; Tao, Dacheng (2019-04-05). "Fast Spatio-Temporal Residual Network for Video Super-Resolution".

750:(the multi-memory convolutional neural network) aligns frames with target one and then generates the final HR-result through the feature extraction, detail 6263: 4006:

Cheng, Ming-Hui; Chen, Hsuan-Ying; Leou, Jin-Jang (2011). "Video super-resolution reconstruction using a mobile search strategy and adaptive patch size".

3881:

Nasir, Haidawati; Stankovic, Vladimir; Marshall, Stephen (2011). "Singular value decomposition based fusion for super-resolution image reconstruction".

572:

join motion estimation and frames fusion to one step. It is performed by consideration of patches similarities. Weights for fusion can be calculated by

517:

assume some function between low-resolution and high-resolution frames and try to improve their guessed function in each step of an iterative process.

6298: 3588:

Farsiu, Sina; Elad, Michael; Milanfar, Peyman (2006-01-15). "A practical approach to superresolution". In Apostolopoulos, John G.; Said, Amir (eds.).

1099:(recurrent latent state propagation) fully convolutional network cell with highly efficient propagation of temporal information through a hidden state 580:

or adaptive patch size. Calculating intra-frame similarity help to preserve small details and edges. Parameters for fusion also can be calculated by

6366: 1089:(the bidirectional temporal-recurrent propagation network) use bidirectional recurrent scheme. Final-result combined from two branches with channel 5090:

Naoto Chiche, Benjamin; Frontera-Pons, Joana; Woiselle, Arnaud; Starck, Jean-Luc (2020-11-09). "Deep Unrolled Network for Video Super-Resolution".

770:(the motion estimation and motion compensation network) uses both motion estimation network and kernel estimation network to warp frames adaptively 738:

between consecutive frames and from this approximate HR optical flow to yield output frame. The discriminator assesses the quality of the generator

3973:

Zhuo, Yue; Liu, Jiaying; Ren, Jie; Guo, Zongming (2012). "Nonlocal based Super Resolution with rotation invariance and search window relocation".

4651:

Wang, Longguang; Guo, Yulan; Liu, Li; Lin, Zaiping; Deng, Xinpu; An, Wei (2020). "Deep Video Super-Resolution Using HR Optical Flow Estimation".

490:, which helps to extend the spectrum of captured signal and though increase resolution. There are different approaches for these methods: using 27: 6383: 5641:

Kim, Soo Ye; Lim, Jeongyeon; Na, Taeyoung; Kim, Munchurl (2019). "Video Super-Resolution Based on 3D-CNNS with Consideration of Scene Change".

4610:

Kim, Tae Hyun; Sajjadi, Mehdi S. M.; Hirsch, Michael; Schölkopf, Bernhard (2018). "Spatio-Temporal Transformer Network for Video Restoration".

2146: 1633: 1601: 6403: 5658: 5604: 5342: 5240: 5117: 5050: 4912: 4627: 4594: 4551: 4518: 4410: 3990: 3898: 3865: 3832: 4217:

Chantas, G.K.; Galatsanos, N.P.; Woods, N.A. (2007). "Super-Resolution Based on Fast Registration and Maximum a Posteriori Reconstruction".

3201:

Video super-resolution finds its practical use in some modern smartphones and cameras, where it is used to reconstruct digital photographs.

6313: 6283: 6273: 6253: 6223: 4143:

Elad, M.; Feuer, A. (1997). "Restoration of a single superresolution image from several blurred, noisy, and undersampled measured images".

486:. The high-resolution frame is estimated in this domain. Finally, this result frame is transformed to the spatial domain. Some methods use 4426:

Kappeler, Armin; Yoo, Seunghwan; Dai, Qiqin; Katsaggelos, Aggelos K. (2016). "Video Super-Resolution With Convolutional Neural Networks".

1105:(the recurrent structure-detail network) divide input frame into structure and detail components and process them in two parallel streams 1745: 1589: 1209: 503: 5213:

Tian, Yapeng; Zhang, Yulun; Fu, Yun; Xu, Chenliang (2020). "TDAN: Temporally-Deformable Alignment Network for Video Super-Resolution".

899:. Generator upsamples input frames, extracts features and fuses them. Discriminator assess the quality of result high-resolution frames 6644: 6468: 6293: 6000:

Fuoli, Dario; Gu, Shuhang; Timofte, Radu (2019-09-17). "Efficient Video Super-Resolution through Recurrent Latent Space Propagation".

470:

There are several traditional methods for video upscaling. These methods try to use some natural preferences and effectively estimate

1071:

information and then fuses important features. The final result is calculated in the reconstruction module by deconvolution operation

6323: 5699: 4773:

Xue, Tianfan; Chen, Baian; Wu, Jiajun; Wei, Donglai; Freeman, William T. (2019-02-12). "Video Enhancement with Task-Oriented Flow".

4326: 4293: 3799: 3712: 3628: 3572: 3539: 3377: 3344: 3311: 782:(the multi-stage multi-reference bootstrapping method) aligns frames and then have two-stage of SR-reconstruction to improve quality 5515:

Zhu, Xiaobin; Li, Zhuangzi; Lou, Jungang; Shen, Qing (2021). "Video super-resolution based on a spatio-temporal matching network".

4990:

Bare, Bahetiyaer; Yan, Bo; Ma, Chenxi; Li, Ke (2019). "Real-time video super-resolution via motion convolution kernel estimation".

4885:

Haris, Muhammad; Shakhnarovich, Gregory; Ukita, Norimichi (2019). "Recurrent Back-Projection Network for Video Super-Resolution".

1228:

LPIPS (Learned Perceptual Image Patch Similarity) compares the perceptual similarity of frames based on high-order image structure

6473: 5879:

Li, Dingyi; Liu, Yu; Wang, Zengfu (2019). "Video Super-Resolution Using Non-Simultaneous Fully Recurrent Convolutional Network".

5033:

Kalarot, Ratheesh; Porikli, Fatih (2019). "MultiBoot Vsr: Multi-Stage Multi-Reference Bootstrapping for Video Super-Resolution".

4342:

Joshi, M.V.; Chaudhuri, S.; Panuganti, R. (2005). "A Learning-Based Method for Image Super-Resolution From Zoomed Observations".

3251: 1140:

also incorporates multi-scale structure and hybrid convolutions to extract wide-range dependencies. To avoid some artifacts like

985:(The fast spatio-temporal residual network) includes a few modules: LR video shallow feature extraction net (LFENet), LR feature 5682:

Luo, Jianping; Huang, Shaofei; Yuan, Yuan (2020). "Video Super-Resolution using Multi-scale Pyramid 3D Convolutional Networks".

3522:

Cohen, B.; Avrin, V.; Dinstein, I. (2000). "Polyphase back-projection filtering for resolution enhancement of image sequences".

6458: 6216: 4276:

Rajan, D.; Chaudhuri, S. (2001). "Generation of super-resolution images from blurred observations using Markov random fields".

646:(deep draft-ensemble learning) generates a series of SR feature maps and then process them together to estimate the final frame 595: 3914:

Protter, M.; Elad, M.; Takeda, H.; Milanfar, P. (2009). "Generalizing the Nonlocal-Means to Super-Resolution Reconstruction".

4393:

Liao, Renjie; Tao, Xin; Li, Ruiyu; Ma, Ziyang; Jia, Jiaya (2015). "Video Super-Resolution via Deep Draft-Ensemble Learning".

3294:

Kim, S. P.; Bose, N. K.; Valenzuela, H. M. (1989). "Reconstruction of high resolution image from noise undersampled frames".

888: 723: 545: 5938:

Isobe, Takashi; Zhu, Fang; Jia, Xu; Wang, Shengjin (2020-08-13). "Revisiting Temporal Modeling for Video Super-resolution".

63: 45:, the main goal is not only to restore more fine details while saving coarse ones, but also to preserve motion consistency. 1013:

to extract spatial and temporal features simultaneously, which then passed through reconstruction module with 3D sub-pixel

6567: 6498: 6349: 3180: 1271:

While deep learning approaches of video super-resolution outperform traditional ones, it's crucial to form a high-quality

591: 3555:

Katsaggelos, A.K. (1997). "An iterative weighted regularized algorithm for improving the resolution of video sequences".

1083:(the recurrent residual network) uses a recurrent sequence of residual blocks to extract spatial and temporal information 619:

In approaches with alignment, neighboring frames are firstly aligned with target one. One can align frames by performing

5787:

Huang, Yan; Wang, Wei; Wang, Liang (2018). "Video Super-Resolution via Bidirectional Recurrent Convolutional Networks".

4309:

Zibetti, Marcelo Victor Wust; Mayer, Joceli (2006). "Outlier Robust and Edge-Preserving Simultaneous Super-Resolution".

2149:

and had two tracks on video extreme super-resolution: first track checks the fidelity with reference frame (measured by

2086:

scale factor is 4. PSNR and VMAF metrics were used for performance evaluation. Top methods are performed in the table:

1090: 1006: 864: 828: 818: 563: 518: 6318: 4084:

Takeda, Hiroyuki; Farsiu, Sina; Milanfar, Peyman (2007). "Kernel Regression for Image Processing and Reconstruction".

1215: 1197: 4491:

Tao, Xin; Gao, Hongyun; Liao, Renjie; Wang, Jue; Jia, Jiaya (2017). "Detail-Revealing Deep Video Super-Resolution".

1114:

Non-local methods extract both spatial and temporal information. The key idea is to use all possible positions as a

999:

to extract spatio-temporal information. Model also has a special approach for frames, where scene change is detected

6639: 6463: 2150: 1737: 1707: 1674: 1645: 1613: 1585: 1558: 1177: 1162: 918: 537: 474:

between frames. The high-resolution frame is reconstructed based on both natural preferences and estimated motion.

3393:

Goldberg, N.; Feuer, A.; Goodwin, G.C. (2003). "Super-resolution reconstruction using spatio-temporal filtering".

6649: 6515: 6428: 4614:. Lecture Notes in Computer Science. Vol. 11207. Cham: Springer International Publishing. pp. 111–127. 3611:

Jing Tian; Kai-Kuang Ma (2005). "A new state-space approach for super-resolution image sequence reconstruction".

3176: 639:

is a warping operation, which aligns one frame to another based on motion information. Examples of such methods:

533: 6134: 427:

kernel, downscaling operation and additive noise should be estimated for given input to achieve better results.

6619: 6493: 6194: 4278:

2001 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.01CH37221)

3524:

2000 IEEE International Conference on Acoustics, Speech, and Signal Processing. Proceedings (Cat. No.00CH37100)

3362:[Proceedings] ICASSP-92: 1992 IEEE International Conference on Acoustics, Speech, and Signal Processing 3261: 3231: 3211: 1252: 1032: 800:(unrolled network for video super-resolution) adapted unrolled optimization algorithms to solve the VSR problem 761: 653: 454: 42: 41:) is the process of generating high-resolution video frames from the given low-resolution video frames. Unlike 4567:

Sajjadi, Mehdi S. M.; Vemulapalli, Raviteja; Brown, Matthew (2018). "Frame-Recurrent Video Super-Resolution".

3487:

Bose, N.K.; Lertrattanapanich, S.; Chappalli, M.B. (2004). "Superresolution with second generation wavelets".

3280:

Chan, Kelvin CK, et al. "BasicVSR: The search for essential components in video super-resolution and beyond."

3731:

Elad, M.; Feuer, A. (1999). "Superresolution restoration of an image sequence: adaptive filtering approach".

6555: 6545: 6288: 3848:

Simonyan, K.; Grishin, S.; Vatolin, D.; Popov, D. (2008). "Fast video super-resolution via classification".

1724: 1694: 1067: 599: 6592: 6560: 6339: 1043: 696:, upsample it to high-resolution and warp previous output frame by using this high-resolution optical flow 491: 20: 3183:

recognition (as preprocessing step). The interest to super-resolution is growing with the development of

367: 308: 6597: 6408: 6354: 3782:

Pickering, M.; Frater, M.; Arnold, J. (2005). "Arobust approach to super-resolution sprite generation".

3241: 3184: 3082: 2154: 1741: 1711: 1678: 1649: 1617: 1562: 1191: 1145: 1124: 3815:

Nasonov, Andrey V.; Krylov, Andrey S. (2010). "Fast Super-Resolution Using Weighted Median Filtering".

1569: 217: 48:

There are many approaches for this task, but this problem still remains to be popular and challenging.

6520: 6503: 6483: 6453: 5888: 5524: 5461: 5427:

Yan, Bo; Lin, Chuming; Tan, Weimin (2019-09-28). "Frame and Feature-Context Video Super-Resolution".

5377: 5265: 5177: 4835: 4670: 4226: 4152: 4093: 3923: 3740: 3653: 3437: 1396: 896: 731: 6525: 6388: 3127: 2321:

Challenge's conditions are the same as AIM 2019 Challenge. Top methods are performed in the table:

1463: 976: 961: 673: 636: 624: 603: 495: 880:

Methods without alignment do not perform alignment as a first step and just process input frames.

30:

VSR and SISR methods' outputs comparison. VSR restores more details by using temporal information.

6540: 6448: 6433: 6393: 6176: 6116: 6022: 6001: 5939: 5920: 5820: 5767: 5748: 5705: 5664: 5621: 5567: 5548: 5428: 5409: 5367: 5320: 5297: 5218: 5167: 5134: 5095: 5070: 5015: 4972: 4938: 4890: 4867: 4808: 4782: 4755: 4729: 4702: 4660: 4572: 4496: 4471: 4451: 4375: 4258: 4125: 4031: 3955: 3677: 3469: 3246: 2158: 1682: 1621: 1256: 1251:

Another way to assess the performance of the video super-resolution algorithm is to organize the

1181: 936: 577: 19:

This article is about video frame restoration technique. For video upscaling tool by Nvidia, see

943:

temporal features and cross-scale nonlocal-correspondence to extract self-similarities in frames

788:

aligns frames with optical flow and then fuse their features in a recurrent bidirectional scheme

5766:

Aksan, Emre; Hilliges, Otmar (2019-02-18). "STCN: Stochastic Temporal Convolutional Networks".

1035:

convolutional neural networks perform video super-resolution by storing temporal dependencies.

6478: 6423: 6375: 6108: 6100: 6065: 5982: 5912: 5904: 5861: 5812: 5804: 5740: 5695: 5654: 5600: 5540: 5497: 5479: 5401: 5393: 5338: 5289: 5281: 5236: 5195: 5113: 5046: 5007: 4964: 4956: 4908: 4859: 4851: 4800: 4747: 4694: 4686: 4633: 4623: 4590: 4547: 4514: 4443: 4406: 4367: 4359: 4322: 4289: 4250: 4242: 4176: 4168: 4117: 4109: 4066: 4023: 3986: 3947: 3939: 3894: 3861: 3828: 3795: 3764: 3756: 3708: 3669: 3624: 3568: 3535: 3504: 3461: 3453: 3410: 3373: 3340: 3307: 1441: 1237:

tLP calculates how LPIPS changes from frame to frame in comparison with the reference sequence

669: 656:), but takes multiple frames as input. Input frames are first aligned by the Druleas algorithm 628: 620: 581: 499: 487: 471: 444:

Upsampling describes the method to transform the aggregated features to the final output image

5092:

2020 Tenth International Conference on Image Processing Theory, Tools and Applications (IPTA)

3695:

Elad, M.; Feuer, A. (1999). "Super-resolution reconstruction of continuous image sequences".

6587: 6550: 6398: 6258: 6168: 6092: 6055: 5972: 5896: 5851: 5796: 5732: 5687: 5646: 5592: 5532: 5487: 5469: 5385: 5330: 5273: 5228: 5185: 5105: 5038: 4999: 4948: 4900: 4843: 4792: 4739: 4678: 4615: 4582: 4539: 4506: 4435: 4398: 4351: 4314: 4281: 4234: 4199: 4160: 4101: 4058: 4015: 3978: 3931: 3886: 3853: 3820: 3787: 3748: 3700: 3661: 3616: 3593: 3560: 3527: 3496: 3445: 3402: 3365: 3332: 3299: 3236: 3187: 3172: 3146: 1241: 1219: 1141: 892: 727: 483: 3214:

and can be used to restore a single image of good quality from multiple sequential frames.

1689: 1304: 400: 341: 278: 250: 145: 6582: 6530: 6239: 5154:

Wang, Hua; Su, Dewei; Liu, Chuangchuang; Jin, Longcun; Sun, Xianfang; Peng, Xinyi (2019).

3143:(to discover better some organs or tissues for clinical analysis and medical intervention) 3140: 1260: 1206:

integrates explicit motion information by estimating distortions along motion trajectories

1115: 926: 706:

by U-style network based on Unet and compensate motion by a trilinear interpolation method

573: 453:

When working with video, temporal information could be used to improve upscaling quality.

5838:

Zhu, Xiaobin; Li, Zhuangzi; Zhang, Xiao-Yu; Li, Changsheng; Liu, Yaqi; Xue, Ziyu (2019).

5892: 5528: 5465: 5381: 5269: 5181: 4839: 4674: 4230: 4156: 4097: 3927: 3744: 3657: 3441: 26: 6607: 6577: 6488: 6413: 6344: 5492: 5449: 3883:

2011 IEEE International Conference on Signal and Image Processing Applications (ICSIPA)

3158: 1719: 521:(POCS), that defines a specific cost function, also can be used for iterative methods. 305:

Super-resolution is an inverse operation, so its problem is to estimate frame sequence

195: 173: 5448:

Tian, Zhiqiang; Wang, Yudiao; Du, Shaoyi; Lan, Xuguang (2020-07-10). Yang, You (ed.).

3975:

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

3406: 1052:(the bidirectional recurrent convolutional network) has two subnetworks: with forward 438:

Alignment concerns on the spatial transformation applied to misaligned images/features

6633: 6572: 6180: 6120: 5752: 5709: 5668: 5552: 5450:"A multiresolution mixture generative adversarial network for video super-resolution" 5301: 5035:

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

5019: 4759: 4706: 1223: 712:(super-resolution optical flow for video super-resolution) calculate high-resolution 555: 541: 528: 459: 5924: 5413: 4976: 4871: 4812: 4129: 4035: 3681: 989:

and up-sampling module (LSRNet) and two residual modules: spatio-temporal and global

776:(real-time video super-resolution) aligns frames with estimated convolutional kernel 6510: 5109: 4455: 4379: 4262: 3959: 3256: 3134: 2748: 1550: 1418: 1232: 1212:

predicts subjective video quality based on a reference and distorted video sequence

1119: 1057: 1053: 986: 940: 922: 751: 735: 713: 703: 693: 677: 5961:"Bidirectional Temporal-Recurrent Propagation Networks for Video Super-Resolution" 5824: 3473: 1244:

as the primary feature to measure the similarity between two corresponding frames.

1157: 5856: 5839: 5536: 5474: 5334: 5232: 5003: 4019: 3282:

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

498:

algorithm, space-varying or spatio-temporal varying filtering. Other methods use

6418: 5190: 5155: 4619: 3982: 3890: 3500: 3206: 2871: 2827: 1348: 1203: 1118:

sum. This strategy may be more effective than local approaches (the progressive

1014: 1010: 996: 972: 957: 953: 424: 6096: 5977: 5960: 5800: 5736: 4952: 4796: 4285: 4062: 3531: 3369: 3137:(to improve video captured from the camera and recognize car numbers and faces) 3089: 2913: 6172: 5650: 4355: 3857: 3791: 3697:

Proceedings 1999 International Conference on Image Processing (Cat. 99CH36348)

3620: 3193: 3164: 2856: 2647: 1539: 868: 853: 6104: 6069: 5986: 5908: 5900: 5865: 5808: 5744: 5544: 5483: 5397: 5389: 5285: 5277: 5199: 5042: 5011: 4960: 4855: 4847: 4804: 4751: 4690: 4682: 4637: 4447: 4439: 4363: 4318: 4246: 4172: 4113: 4070: 4027: 3943: 3935: 3760: 3704: 3673: 3564: 3508: 3457: 3449: 3414: 3336: 1889: 1657: 1596: 668:(detail-revealing deep video super-resolution) consists of three main steps: 576:. To strength searching for similar patches, one can use rotation invariance 5691: 5596: 4904: 4743: 4586: 4238: 4105: 3665: 3152: 2000: 1628: 1326: 1170: 925:

temporal features. Non-local matching block integrates super-resolution and

794:

is a refined version of BasicVSR with a recurrent coupled propagation scheme

6112: 5916: 5816: 5501: 5405: 5293: 5166:. Institute of Electrical and Electronics Engineers (IEEE): 177734–177744. 4968: 4863: 4698: 4371: 4254: 4180: 4151:(12). Institute of Electrical and Electronics Engineers (IEEE): 1646–1658. 4121: 3951: 3768: 3465: 3436:(11). Institute of Electrical and Electronics Engineers (IEEE): 2889–2900. 5366:(7). Institute of Electrical and Electronics Engineers (IEEE): 3312–3327. 5317:

2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

5215:

2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

4887:

2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

4834:(5). Institute of Electrical and Electronics Engineers (IEEE): 2530–2544. 4543: 4510: 4225:(7). Institute of Electrical and Electronics Engineers (IEEE): 1821–1830. 3824: 3652:(5). Institute of Electrical and Electronics Engineers (IEEE): 2084–2095. 2755: 4402: 3052: 3022: 2992: 2875: 1663: 1272: 559: 435:

Propagation refers to the way in which features are propagated temporally

5840:"Residual Invertible Spatio-Temporal Network for Video Super-Resolution" 4937:(3). Institute of Electrical and Electronics Engineers (IEEE): 933–948. 4434:(2). Institute of Electrical and Electronics Engineers (IEEE): 109–122. 4350:(3). Institute of Electrical and Electronics Engineers (IEEE): 527–537. 4344:

IEEE Transactions on Systems, Man, and Cybernetics - Part B: Cybernetics

4092:(2). Institute of Electrical and Electronics Engineers (IEEE): 349–366. 3739:(3). Institute of Electrical and Electronics Engineers (IEEE): 387–395. 3112: 2936: 1231:

tOF measures pixel-wise motion similarity with reference frame based on

6060: 6043: 5264:. Institute of Electrical and Electronics Engineers (IEEE): 2923–2934. 4659:. Institute of Electrical and Electronics Engineers (IEEE): 4323–4336. 3303: 3298:. Vol. 129. Berlin/Heidelberg: Springer-Verlag. pp. 315–326. 4203: 4164: 3922:(1). Institute of Electrical and Electronics Engineers (IEEE): 36–51. 3752: 3597: 1255:. People are asked to compare the corresponding frames, and the final 16:

Generating high-resolution video frames from given low-resolution ones

6027: 5944: 5772: 5626: 5572: 5433: 5139: 5075: 4476: 2791: 2712: 1963: 1371: 1194:

measures the similarity of structure between two corresponding frames

1169:

The common way to estimate the performance of video super-resolution

502:, which helps to find similarities in neighboring local areas. Later 3428:

Mallat, S (2010). "Super-Resolution With Sparse Mixing Estimators".

2784: 2037: 1180:

calculates the difference between two corresponding frames based on

1060:. The result of the network is a composition of two branches' output 6006: 5589:

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition

5372: 5325: 5223: 5172: 5100: 4943: 4895: 4787: 4734: 4665: 4577: 4569:

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition

4501: 3059: 3029: 2999: 2969: 2719: 2611: 6208: 5684:

Proceedings of the 28th ACM International Conference on Multimedia

3192: 2820: 2683: 2676: 1852: 1156: 929:. At the final step, SR-result is got on the global wavelet domain 632: 2640: 692:(frame recurrent video super-resolution) estimate low-resolution 598:

estimation. Regularization parameter for MAP can be estimated by

6212: 3329:

Proceedings of 1st International Conference on Image Processing

462:

based methods for video upscaling outperform traditional ones.

6085:

IEEE Transactions on Pattern Analysis and Machine Intelligence

5789:

IEEE Transactions on Pattern Analysis and Machine Intelligence

5725:

IEEE Transactions on Circuits and Systems for Video Technology

4931:

IEEE Transactions on Pattern Analysis and Machine Intelligence

3726: 3724: 3197:

Simulating the natural hand movements by "jiggling" the camera

5844:

Proceedings of the AAAI Conference on Artificial Intelligence

5643:

2019 IEEE International Conference on Image Processing (ICIP)

2157:). The second track checks the perceptual quality of videos ( 57:

Most research considers the degradation process of frames as

4536:

2017 IEEE International Conference on Computer Vision (ICCV)

4493:

2017 IEEE International Conference on Computer Vision (ICCV)

4395:

2015 IEEE International Conference on Computer Vision (ICCV)

3218:

reference frame and align every other frame relative to it.

3149:(to help in the investigation during the criminal procedure) 1218:

is a full-reference image quality assessment index based on

6195:"See Better and Further with Super Res Zoom on the Pixel 3" 3850:

2008 15th IEEE International Conference on Image Processing

3557:

Proceedings of International Conference on Image Processing

594:

estimate more probable image. Another group of methods use

3331:. Vol. 3. IEEE Comput. Soc. Press. pp. 571–574. 6044:"Video super-resolution with non-local alignment network" 6042:

Zhou, Chao; Chen, Can; Ding, Fei; Zhang, Dengyin (2021).

5156:"Deformable Non-Local Network for Video Super-Resolution" 5064: 5062: 4781:(8). Springer Science and Business Media LLC: 1106–1125. 3817:

2010 20th International Conference on Pattern Recognition

1720:

MSU Super-Resolution for Video Compression Benchmark 2022

1574: 558:, weighted median filter, adaptive normalized averaging, 441:

Aggregation defines the steps to combine aligned features

3395:

Journal of Visual Communication and Image Representation

1485: 960:

use both spatial and temporal information. They perform

939:

network) uses temporal multi-correspondence strategy to

1240:

FSIM (Feature Similarity Index for Image Quality) uses

979:. The model estimates kernels for specific input frames 482:

Firstly the low-resolution frame is transformed to the

3784:

IEEE International Conference on Image Processing 2005

3613:

IEEE International Conference on Image Processing 2005

129:{\displaystyle \{y\}=(\{x\}*k)\downarrow {_{s}}+\{n\}} 1662:

ICIP (International Conference of Image Processing),

1222:

and the notion of image information extracted by the

1204:

MOVIE (Motion-based Video Integrity Evaluation index)

1200:

shows information similarity with the reference frame

403: 370: 344: 311: 281: 253: 220: 198: 176: 148: 66: 2865:

MSU Super-Resolution for Video Compression Benchmark

971:(the dynamic upsampling filters) uses deformable 3D 615:

Aligned by motion estimation and motion compensation

6374: 6365: 6332: 6246: 702:(the spatio-temporal transformer network) estimate 3155:(to improve quality of video of stars and planets) 415: 389: 356: 330: 293: 265: 237: 204: 182: 160: 128: 5959:Han, Lei; Fan, Cien; Yang, Ye; Zou, Lian (2020). 5460:(7). Public Library of Science (PLoS): e0235352. 4311:2006 International Conference on Image Processing 3296:Lecture Notes in Control and Information Sciences 4728:(4). Association for Computing Machinery (ACM). 627:(MEMC) or by using Deformable convolution (DC). 3590:Visual Communications and Image Processing 2006 6224: 4196:Applications of Digital Image Processing XXVI 8: 1714:with shift compensation, QRCRv1.0, CRRMv1.0 1165:visualization of the output of a VSR method. 410: 404: 384: 371: 351: 345: 325: 312: 288: 282: 260: 254: 155: 149: 123: 117: 88: 82: 73: 67: 917:(the spatio-temporal matching network) use 6371: 6231: 6217: 6209: 4428:IEEE Transactions on Computational Imaging 1759:The NTIRE 2019 Challenge was organized by 1210:VMAF (Video Multimethod Assessment Fusion) 1148:, they use generative adversarial training 995:(The 3D super-resolution network) uses 3D 590:use statistical theory to solve the task. 168:— original high-resolution frame sequence, 6059: 6026: 6005: 5976: 5943: 5855: 5771: 5625: 5571: 5491: 5473: 5432: 5371: 5324: 5222: 5189: 5171: 5138: 5099: 5074: 4942: 4894: 4786: 4733: 4664: 4576: 4500: 4475: 4280:. Vol. 3. IEEE. pp. 1837–1840. 3526:. Vol. 4. IEEE. pp. 2171–2174. 1690:MSU Video Super-Resolution Benchmark 2021 1547:(Computer Vision and pattern recognition) 402: 374: 369: 343: 315: 310: 280: 252: 228: 224: 219: 197: 175: 147: 107: 103: 65: 4775:International Journal of Computer Vision 3592:. Vol. 6077. SPIE. p. 607703. 2880: 2572: 2323: 2163: 2088: 1768: 1636:(European Conference on Computer Vision) 1604:(European Conference on Computer Vision) 1515: 1433:A lot of fast, difficult, diverse motion 1388:A lot of fast, difficult, diverse motion 1363:A lot of fast, difficult, diverse motion 1277: 852:Some methods align frames by calculated 25: 4051:Computer Vision and Image Understanding 3699:. Vol. 3. IEEE. pp. 459–463. 3273: 3161:(to alleviate observation of an object) 525:Iterative adaptive filtering algorithms 6384:3D reconstruction from multiple images 3646:IEEE Transactions on Signal Processing 3559:. IEEE Comput. Soc. pp. 474–477. 3489:Signal Processing: Image Communication 631:gives information about the motion of 6404:Simultaneous localization and mapping 5881:IEEE Transactions on Image Processing 5360:IEEE Transactions on Image Processing 5258:IEEE Transactions on Image Processing 4828:IEEE Transactions on Image Processing 4653:IEEE Transactions on Image Processing 4219:IEEE Transactions on Image Processing 4198:. Vol. 5203. SPIE. p. 121. 4145:IEEE Transactions on Image Processing 4086:IEEE Transactions on Image Processing 3916:IEEE Transactions on Image Processing 3733:IEEE Transactions on Image Processing 3430:IEEE Transactions on Image Processing 1391:Few details, text in a few sequences 1366:Few details, text in a few sequences 506:was used for video super resolution. 7: 2565:MSU Video Super-Resolution Benchmark 1198:IFC (Information Fidelity Criterion) 43:single-image super-resolution (SISR) 504:second-generation wavelet transform 390:{\displaystyle \{{\overline {x}}\}} 331:{\displaystyle \{{\overline {x}}\}} 6469:Automatic number-plate recognition 3167:(to strength microscopes' ability) 1658:Mobile Video Restoration Challenge 1192:SSIM (Structural similarity index) 964:and maintain temporal consistency 841:Multi-Stage Feature Fusion Network 754:and feature reconstruction modules 14: 6161:Programming and Computer Software 1321:Some small details, without text 1216:VIF (Visual Information Fidelity) 1163:PSNR (Peak signal-to-noise ratio) 805:Aligned by deformable convolution 515:Iterative back-projection methods 238:{\displaystyle \downarrow {_{s}}} 6474:Automated species identification 3364:. IEEE. pp. 169–172 vol.3. 3252:Ultra-high-definition television 1161:Top: original sequence. Bottom: 301:— low-resolution frame sequence. 6459:Audio-visual speech recognition 6135:"MSU VSR Benchmark Methodology" 3171:It also helps to solve task of 1458:Without small details and text 592:maximum likelihood (ML) methods 6304:Recognition and categorization 5110:10.1109/ipta50016.2020.9286636 4057:(12). Elsevier BV: 1336–1345. 1178:PSNR (Peak signal-noise ratio) 221: 100: 97: 79: 1: 6568:Optical character recognition 6499:Content-based image retrieval 4014:(5). Elsevier BV: 1284–1297. 3407:10.1016/s1047-3203(03)00042-7 654:single image super resolution 652:is based on SRCNN (model for 546:recursive least squares (RLS) 492:weighted least squares theory 455:Single image super-resolution 5857:10.1609/aaai.v33i01.33015981 5591:. IEEE. pp. 3224–3232. 5537:10.1016/j.patcog.2020.107619 5475:10.1371/journal.pone.0235352 5335:10.1109/cvpr42600.2020.00803 5319:. IEEE. pp. 8005–8014. 5233:10.1109/cvpr42600.2020.00342 5217:. IEEE. pp. 3357–3366. 5037:. IEEE. pp. 2060–2069. 5004:10.1016/j.neucom.2019.07.089 4889:. IEEE. pp. 3892–3901. 4722:ACM Transactions on Graphics 4571:. IEEE. pp. 6626–6634. 4538:. IEEE. pp. 2526–2534. 4495:. IEEE. pp. 4482–4490. 4313:. IEEE. pp. 1741–1744. 4020:10.1016/j.sigpro.2010.12.016 3819:. IEEE. pp. 2230–2233. 1005:(the multi-scale pyramid 3D 519:Projections onto convex sets 379: 320: 5191:10.1109/access.2019.2958030 4620:10.1007/978-3-030-01219-9_7 4612:Computer Vision – ECCV 2018 3983:10.1109/icassp.2012.6288018 3891:10.1109/icsipa.2011.6144138 3501:10.1016/j.image.2004.02.001 3495:(5). Elsevier BV: 387–391. 3401:(4). Elsevier BV: 508–525. 2890:BSQ-rate (Subjective score) 956:work on spatial domain, 3D 610:Deep learning based methods 6666: 6464:Automatic image annotation 6299:Noise reduction techniques 6097:10.1109/TPAMI.2020.3042298 5978:10.3390/electronics9122085 5801:10.1109/TPAMI.2017.2701380 5737:10.1109/TCSVT.2020.3044451 4953:10.1109/tpami.2019.2941941 4797:10.1007/s11263-018-01144-2 4397:. IEEE. pp. 531–539. 4286:10.1109/icassp.2001.941300 4063:10.1016/j.cviu.2009.11.004 3977:. IEEE. pp. 853–856. 3885:. IEEE. pp. 393–398. 3852:. IEEE. pp. 349–352. 3532:10.1109/icassp.2000.859267 3370:10.1109/icassp.1992.226249 2145:The challenge was held by 1436:Few details, without text 1413:Few details, without text 935:(the multi-correspondence 919:discrete wavelet transform 604:Markov random fields (MRF) 596:maximum a posteriori (MAP) 18: 6645:Film and video technology 6616: 6429:Free viewpoint television 6173:10.1134/S0361768820030111 5651:10.1109/ICIP.2019.8803297 4356:10.1109/tsmcb.2005.846647 3858:10.1109/icip.2008.4711763 3792:10.1109/icip.2005.1529896 3621:10.1109/icip.2005.1529892 2081:Youku-VESR Challenge 2019 1806:Runtime per image in sec 1801:Runtime per image in sec 1727:(Moscow State University) 1697:(Moscow State University) 1570:Youku-VESR Challenge 2019 1517:Comparison of benchmarks 1173:is to use a few metrics: 1028:Recurrent neural networks 734:. Generator estimates LR 722:(the temporally coherent 570:Non-parametric algorithms 496:total least squares (TLS) 6494:Computer-aided diagnosis 5901:10.1109/TIP.2018.2877334 5390:10.1109/tip.2019.2895768 5278:10.1109/tip.2021.3056868 5043:10.1109/cvprw.2019.00258 4998:. Elsevier BV: 236–245. 4848:10.1109/tip.2018.2887017 4683:10.1109/tip.2020.2967596 4440:10.1109/tci.2016.2532323 4319:10.1109/icip.2006.312718 3936:10.1109/tip.2008.2008067 3786:. IEEE. pp. I-897. 3705:10.1109/icip.1999.817156 3615:. IEEE. pp. I-881. 3565:10.1109/icip.1997.638811 3450:10.1109/tip.2010.2049927 3337:10.1109/icip.1994.413741 3262:High-dynamic-range video 3232:Super-resolution imaging 2874:standards and different 2603:Runtime per image in sec 2345:Runtime per image in sec 2185:Runtime per image in sec 1257:mean opinion score (MOS) 1220:natural scene statistics 534:least mean squares (LMS) 245:— downscaling operation, 212:— convolution operation, 53:Mathematical explanation 6556:Moving object detection 6546:Medical image computing 6309:Research infrastructure 6279:Image sensor technology 5692:10.1145/3394171.3413587 5597:10.1109/cvpr.2018.00340 4905:10.1109/cvpr.2019.00402 4744:10.1145/3386569.3392457 4587:10.1109/cvpr.2018.00693 4239:10.1109/tip.2007.896664 4106:10.1109/tip.2006.888330 3666:10.1109/tsp.2007.892704 1343:A lot of small details 1293:Ground-truth resolution 1279:Comparison of datasets 821:(TSA) module for fusion 600:Tikhonov regularization 6593:Video content analysis 6561:Small object detection 6340:Computer stereo vision 5686:. pp. 1882–1890. 5645:. pp. 2831–2835. 5094:. IEEE. pp. 1–6. 3198: 3128:size of camera sensors 1397:Ultra Video Dataset 4K 1166: 1127:upsampled input frame 1044:long short-term memory 574:nonlocal-means filters 417: 391: 358: 332: 295: 267: 239: 206: 184: 162: 130: 35:Video super-resolution 31: 21:Video Super Resolution 6598:Video motion analysis 6409:Structure from motion 6355:3D object recognition 4544:10.1109/iccv.2017.274 4511:10.1109/iccv.2017.479 3825:10.1109/icpr.2010.546 3242:High definition video 3196: 1259:is calculated as the 1253:subjective evaluation 1160: 1007:convolutional network 848:Aligned by homography 588:Probabilistic methods 418: 416:{\displaystyle \{x\}} 397:is close to original 392: 359: 357:{\displaystyle \{y\}} 333: 296: 294:{\displaystyle \{y\}} 268: 266:{\displaystyle \{n\}} 240: 207: 185: 163: 161:{\displaystyle \{x\}} 131: 29: 6521:Foreground detection 6504:Reverse image search 6484:Bioimage informatics 6454:Activity recognition 6048:IET Image Processing 4403:10.1109/iccv.2015.68 1755:NTIRE 2019 Challenge 1540:NTIRE 2019 Challenge 871:used to align frames 401: 368: 342: 338:from frame sequence 309: 279: 251: 218: 196: 174: 146: 64: 6588:Autonomous vehicles 6526:Gesture recognition 6389:2D to 3D conversion 5893:2019ITIP...28.1342L 5529:2021PatRe.11007619Z 5517:Pattern Recognition 5466:2020PLoSO..1535352T 5382:2019ITIP...28.3312L 5270:2021ITIP...30.2923S 5182:2019IEEEA...7q7734W 4840:2019ITIP...28.2530W 4675:2020ITIP...29.4323W 4231:2007ITIP...16.1821C 4157:1997ITIP....6.1646E 4098:2007ITIP...16..349T 3928:2009ITIP...18...36P 3745:1999ITIP....8..387E 3658:2007ITSP...55.2084C 3442:2010ITIP...19.2889M 2893:BSQ-rate (ERQAv2.0) 2883: 2575: 2326: 2166: 2091: 1771: 1518: 1318:Without fast motion 1280: 1224:human visual system 977:motion compensation 962:motion compensation 876:Spatial non-aligned 674:motion compensation 637:motion compensation 625:motion compensation 560:AdaBoost classifier 536:. One can also use 466:Traditional methods 6603:Video surveillance 6541:Landmark detection 6449:3D pose estimation 6434:Volumetric capture 6394:Gaussian splatting 6350:Object recognition 6264:Commercial systems 6061:10.1049/ipr2.12134 3304:10.1007/bfb0042742 3247:Display resolution 3199: 3135:video surveillance 3000:Real-ESRGAN + x264 2902:BSQ-rate (MS-SSIM) 2881: 2573: 2324: 2317:AIM 2020 Challenge 2164: 2141:AIM 2019 Challenge 2089: 1933:CyberverseSanDiego 1769: 1629:AIM 2020 Challenge 1597:AIM 2019 Challenge 1516: 1278: 1182:mean squared error 1167: 578:similarity measure 413: 387: 354: 328: 291: 263: 235: 202: 180: 158: 126: 32: 6640:Signal processing 6627: 6626: 6536:Image restoration 6479:Augmented reality 6444: 6443: 6424:4D reconstruction 6376:3D reconstruction 6269:Feature detection 5731:(10): 3954–3966. 5660:978-1-5386-6249-6 5606:978-1-5386-6420-9 5344:978-1-7281-7168-5 5242:978-1-7281-7168-5 5119:978-1-7281-8750-1 5052:978-1-7281-2506-0 4914:978-1-7281-3293-8 4629:978-3-030-01218-2 4596:978-1-5386-6420-9 4553:978-1-5386-1032-9 4520:978-1-5386-1032-9 4412:978-1-4673-8391-2 4204:10.1117/12.507194 4165:10.1109/83.650118 4008:Signal Processing 3992:978-1-4673-0046-9 3900:978-1-4577-0242-6 3867:978-1-4244-1765-0 3834:978-1-4244-7542-1 3753:10.1109/83.748893 3598:10.1117/12.644391 3212:burst photography 3188:computer displays 3118: 3117: 2862: 2861: 2562: 2561: 2314: 2313: 2138: 2137: 2106:Avengers Assemble 2078: 2077: 1752: 1751: 1506: 1505: 1290:Mean video length 1263:overall ratings. 670:motion estimation 629:Motion estimation 621:motion estimation 582:kernel regression 500:wavelet transform 488:Fourier transform 382: 323: 273:— additive noise, 205:{\displaystyle *} 183:{\displaystyle k} 6657: 6650:Image processing 6551:Object detection 6516:Face recognition 6399:Shape from focus 6372: 6259:Digital geometry 6233: 6226: 6219: 6210: 6203: 6202: 6191: 6185: 6184: 6156: 6150: 6149: 6147: 6146: 6139:Video Processing 6131: 6125: 6124: 6091:(5): 2264–2280. 6080: 6074: 6073: 6063: 6054:(8): 1655–1667. 6039: 6033: 6032: 6030: 6018: 6012: 6011: 6009: 5997: 5991: 5990: 5980: 5956: 5950: 5949: 5947: 5935: 5929: 5928: 5887:(3): 1342–1355. 5876: 5870: 5869: 5859: 5835: 5829: 5828: 5795:(4): 1015–1028. 5784: 5778: 5777: 5775: 5763: 5757: 5756: 5720: 5714: 5713: 5679: 5673: 5672: 5638: 5632: 5631: 5629: 5617: 5611: 5610: 5584: 5578: 5577: 5575: 5563: 5557: 5556: 5512: 5506: 5505: 5495: 5477: 5445: 5439: 5438: 5436: 5424: 5418: 5417: 5375: 5355: 5349: 5348: 5328: 5312: 5306: 5305: 5253: 5247: 5246: 5226: 5210: 5204: 5203: 5193: 5175: 5151: 5145: 5144: 5142: 5130: 5124: 5123: 5103: 5087: 5081: 5080: 5078: 5066: 5057: 5056: 5030: 5024: 5023: 4987: 4981: 4980: 4946: 4925: 4919: 4918: 4898: 4882: 4876: 4875: 4823: 4817: 4816: 4790: 4770: 4764: 4763: 4737: 4717: 4711: 4710: 4668: 4648: 4642: 4641: 4607: 4601: 4600: 4580: 4564: 4558: 4557: 4531: 4525: 4524: 4504: 4488: 4482: 4481: 4479: 4466: 4460: 4459: 4423: 4417: 4416: 4390: 4384: 4383: 4339: 4333: 4332: 4306: 4300: 4299: 4273: 4267: 4266: 4214: 4208: 4207: 4191: 4185: 4184: 4140: 4134: 4133: 4081: 4075: 4074: 4046: 4040: 4039: 4003: 3997: 3996: 3970: 3964: 3963: 3911: 3905: 3904: 3878: 3872: 3871: 3845: 3839: 3838: 3812: 3806: 3805: 3779: 3773: 3772: 3728: 3719: 3718: 3692: 3686: 3685: 3641: 3635: 3634: 3608: 3602: 3601: 3585: 3579: 3578: 3552: 3546: 3545: 3519: 3513: 3512: 3484: 3478: 3477: 3425: 3419: 3418: 3390: 3384: 3383: 3357: 3351: 3350: 3324: 3318: 3317: 3291: 3285: 3278: 3237:Image resolution 3173:object detection 3147:forensic science 2905:BSQ-rate (LPIPS) 2884: 2576: 2515:based on STARnet 2327: 2167: 2092: 1899:ensemble of RDN, 1772: 1519: 1296:Motion in frames 1281: 1242:phase congruency 1046:(LSTM) mechanism 863:(Temporal Group 856:between frames. 635:between frames. 538:steepest descent 484:frequency domain 478:Frequency domain 431:and Upsampling. 422: 420: 419: 414: 396: 394: 393: 388: 383: 375: 363: 361: 360: 355: 337: 335: 334: 329: 324: 316: 300: 298: 297: 292: 272: 270: 269: 264: 244: 242: 241: 236: 234: 233: 232: 211: 209: 208: 203: 189: 187: 186: 181: 167: 165: 164: 159: 135: 133: 132: 127: 113: 112: 111: 6665: 6664: 6660: 6659: 6658: 6656: 6655: 6654: 6630: 6629: 6628: 6623: 6612: 6583:Robotic mapping 6531:Image denoising 6440: 6361: 6328: 6294:Motion analysis 6242: 6240:Computer vision 6237: 6207: 6206: 6193: 6192: 6188: 6158: 6157: 6153: 6144: 6142: 6133: 6132: 6128: 6082: 6081: 6077: 6041: 6040: 6036: 6020: 6019: 6015: 5999: 5998: 5994: 5958: 5957: 5953: 5937: 5936: 5932: 5878: 5877: 5873: 5837: 5836: 5832: 5786: 5785: 5781: 5765: 5764: 5760: 5722: 5721: 5717: 5702: 5681: 5680: 5676: 5661: 5640: 5639: 5635: 5619: 5618: 5614: 5607: 5586: 5585: 5581: 5565: 5564: 5560: 5514: 5513: 5509: 5447: 5446: 5442: 5426: 5425: 5421: 5357: 5356: 5352: 5345: 5314: 5313: 5309: 5255: 5254: 5250: 5243: 5212: 5211: 5207: 5153: 5152: 5148: 5132: 5131: 5127: 5120: 5089: 5088: 5084: 5068: 5067: 5060: 5053: 5032: 5031: 5027: 4989: 4988: 4984: 4927: 4926: 4922: 4915: 4884: 4883: 4879: 4825: 4824: 4820: 4772: 4771: 4767: 4719: 4718: 4714: 4650: 4649: 4645: 4630: 4609: 4608: 4604: 4597: 4566: 4565: 4561: 4554: 4533: 4532: 4528: 4521: 4490: 4489: 4485: 4468: 4467: 4463: 4425: 4424: 4420: 4413: 4392: 4391: 4387: 4341: 4340: 4336: 4329: 4308: 4307: 4303: 4296: 4275: 4274: 4270: 4216: 4215: 4211: 4193: 4192: 4188: 4142: 4141: 4137: 4083: 4082: 4078: 4048: 4047: 4043: 4005: 4004: 4000: 3993: 3972: 3971: 3967: 3913: 3912: 3908: 3901: 3880: 3879: 3875: 3868: 3847: 3846: 3842: 3835: 3814: 3813: 3809: 3802: 3781: 3780: 3776: 3730: 3729: 3722: 3715: 3694: 3693: 3689: 3643: 3642: 3638: 3631: 3610: 3609: 3605: 3587: 3586: 3582: 3575: 3554: 3553: 3549: 3542: 3521: 3520: 3516: 3486: 3485: 3481: 3427: 3426: 3422: 3392: 3391: 3387: 3380: 3359: 3358: 3354: 3347: 3326: 3325: 3321: 3314: 3293: 3292: 3288: 3279: 3275: 3270: 3228: 3185:high definition 3141:medical imaging 3123: 2899:BSQ-rate (PSNR) 2896:BSQ-rate (VMAF) 2867: 2567: 2319: 2143: 2083: 1900: 1807: 1802: 1797: 1792: 1787: 1782: 1764: 1757: 1511: 1384: 1374:(complete sets) 1269: 1261:arithmetic mean 1155: 1112: 1030: 950: 948:3D convolutions 878: 850: 807: 617: 612: 566:based filters. 512: 480: 468: 451: 399: 398: 366: 365: 340: 339: 307: 306: 277: 276: 249: 248: 225: 216: 215: 194: 193: 172: 171: 144: 143: 104: 62: 61: 55: 50: 24: 17: 12: 11: 5: 6663: 6661: 6653: 6652: 6647: 6642: 6632: 6631: 6625: 6624: 6617: 6614: 6613: 6611: 6610: 6608:Video tracking 6605: 6600: 6595: 6590: 6585: 6580: 6578:Remote sensing 6575: 6570: 6565: 6564: 6563: 6558: 6548: 6543: 6538: 6533: 6528: 6523: 6518: 6513: 6508: 6507: 6506: 6496: 6491: 6489:Blob detection 6486: 6481: 6476: 6471: 6466: 6461: 6456: 6451: 6445: 6442: 6441: 6439: 6438: 6437: 6436: 6431: 6421: 6416: 6414:View synthesis 6411: 6406: 6401: 6396: 6391: 6386: 6380: 6378: 6369: 6363: 6362: 6360: 6359: 6358: 6357: 6347: 6345:Motion capture 6342: 6336: 6334: 6330: 6329: 6327: 6326: 6321: 6316: 6311: 6306: 6301: 6296: 6291: 6286: 6281: 6276: 6271: 6266: 6261: 6256: 6250: 6248: 6244: 6243: 6238: 6236: 6235: 6228: 6221: 6213: 6205: 6204: 6199:Google AI Blog 6186: 6167:(3): 183–194. 6151: 6126: 6075: 6034: 6013: 5992: 5951: 5930: 5871: 5830: 5779: 5758: 5715: 5700: 5674: 5659: 5633: 5612: 5605: 5579: 5558: 5507: 5440: 5419: 5350: 5343: 5307: 5248: 5241: 5205: 5146: 5125: 5118: 5082: 5058: 5051: 5025: 4992:Neurocomputing 4982: 4920: 4913: 4877: 4818: 4765: 4712: 4643: 4628: 4602: 4595: 4559: 4552: 4526: 4519: 4483: 4461: 4418: 4411: 4385: 4334: 4327: 4301: 4294: 4268: 4209: 4186: 4135: 4076: 4041: 3998: 3991: 3965: 3906: 3899: 3873: 3866: 3840: 3833: 3807: 3800: 3774: 3720: 3713: 3687: 3636: 3629: 3603: 3580: 3573: 3547: 3540: 3514: 3479: 3420: 3385: 3378: 3352: 3345: 3319: 3312: 3286: 3272: 3271: 3269: 3266: 3265: 3264: 3259: 3254: 3249: 3244: 3239: 3234: 3227: 3224: 3169: 3168: 3162: 3159:remote sensing 3156: 3150: 3144: 3138: 3122: 3119: 3116: 3115: 3110: 3107: 3104: 3101: 3098: 3095: 3092: 3086: 3085: 3080: 3077: 3074: 3071: 3068: 3065: 3062: 3056: 3055: 3050: 3047: 3044: 3041: 3038: 3035: 3032: 3026: 3025: 3020: 3017: 3014: 3011: 3008: 3005: 3002: 2996: 2995: 2990: 2987: 2984: 2981: 2978: 2975: 2972: 2966: 2965: 2962: 2959: 2956: 2953: 2950: 2947: 2944: 2940: 2939: 2934: 2931: 2928: 2925: 2922: 2919: 2916: 2910: 2909: 2906: 2903: 2900: 2897: 2894: 2891: 2888: 2866: 2863: 2860: 2859: 2854: 2851: 2848: 2845: 2842: 2839: 2836: 2833: 2830: 2824: 2823: 2818: 2815: 2812: 2809: 2806: 2803: 2800: 2797: 2794: 2788: 2787: 2782: 2779: 2776: 2773: 2770: 2767: 2764: 2761: 2758: 2752: 2751: 2746: 2743: 2740: 2737: 2734: 2731: 2728: 2725: 2722: 2716: 2715: 2710: 2707: 2704: 2701: 2698: 2695: 2692: 2689: 2686: 2680: 2679: 2674: 2671: 2668: 2665: 2662: 2659: 2656: 2653: 2650: 2644: 2643: 2638: 2635: 2632: 2629: 2626: 2623: 2620: 2617: 2614: 2608: 2607: 2604: 2601: 2598: 2595: 2592: 2589: 2586: 2583: 2580: 2566: 2563: 2560: 2559: 2556: 2553: 2550: 2547: 2544: 2541: 2539: 2535: 2534: 2531: 2528: 2525: 2522: 2519: 2516: 2513: 2509: 2508: 2505: 2502: 2499: 2496: 2493: 2490: 2487: 2483: 2482: 2479: 2476: 2473: 2470: 2467: 2464: 2461: 2457: 2456: 2453: 2450: 2447: 2444: 2441: 2438: 2435: 2431: 2430: 2427: 2424: 2421: 2418: 2415: 2412: 2409: 2405: 2404: 2401: 2398: 2395: 2392: 2389: 2386: 2383: 2379: 2378: 2375: 2372: 2369: 2366: 2363: 2360: 2357: 2353: 2352: 2349: 2346: 2343: 2340: 2337: 2334: 2331: 2318: 2315: 2312: 2311: 2308: 2305: 2302: 2299: 2296: 2293: 2290: 2287: 2283: 2282: 2279: 2276: 2273: 2270: 2267: 2264: 2261: 2258: 2254: 2253: 2250: 2247: 2244: 2241: 2238: 2235: 2232: 2229: 2225: 2224: 2221: 2218: 2215: 2212: 2209: 2206: 2203: 2200: 2196: 2195: 2192: 2189: 2186: 2183: 2180: 2177: 2174: 2171: 2142: 2139: 2136: 2135: 2132: 2129: 2125: 2124: 2121: 2118: 2114: 2113: 2110: 2107: 2103: 2102: 2099: 2096: 2082: 2079: 2076: 2075: 2072: 2069: 2066: 2063: 2060: 2057: 2054: 2051: 2048: 2045: 2041: 2040: 2035: 2032: 2029: 2026: 2023: 2020: 2017: 2014: 2011: 2008: 2004: 2003: 1998: 1995: 1992: 1989: 1986: 1983: 1980: 1977: 1974: 1971: 1967: 1966: 1961: 1958: 1955: 1952: 1949: 1946: 1943: 1940: 1937: 1934: 1930: 1929: 1926: 1923: 1920: 1917: 1914: 1911: 1908: 1905: 1902: 1897: 1893: 1892: 1887: 1884: 1881: 1878: 1875: 1872: 1869: 1866: 1863: 1860: 1856: 1855: 1850: 1847: 1844: 1841: 1838: 1835: 1832: 1829: 1826: 1823: 1819: 1818: 1815: 1812: 1809: 1804: 1799: 1794: 1789: 1784: 1779: 1776: 1756: 1753: 1750: 1749: 1734: 1731: 1728: 1722: 1716: 1715: 1704: 1701: 1698: 1692: 1686: 1685: 1672: 1669: 1666: 1660: 1654: 1653: 1643: 1640: 1637: 1631: 1625: 1624: 1611: 1608: 1605: 1599: 1593: 1592: 1583: 1580: 1577: 1572: 1566: 1565: 1556: 1553: 1548: 1542: 1536: 1535: 1532: 1531:Upscale factor 1529: 1526: 1523: 1510: 1507: 1504: 1503: 1500: 1497: 1494: 1491: 1488: 1482: 1481: 1478: 1475: 1472: 1469: 1466: 1460: 1459: 1456: 1455:Diverse motion 1453: 1450: 1447: 1444: 1438: 1437: 1434: 1431: 1428: 1425: 1422: 1415: 1414: 1411: 1410:Diverse motion 1408: 1405: 1402: 1399: 1393: 1392: 1389: 1386: 1381: 1378: 1375: 1368: 1367: 1364: 1361: 1358: 1355: 1352: 1345: 1344: 1341: 1338: 1335: 1332: 1329: 1323: 1322: 1319: 1316: 1313: 1310: 1307: 1301: 1300: 1297: 1294: 1291: 1288: 1285: 1268: 1265: 1246: 1245: 1238: 1235: 1229: 1226: 1213: 1207: 1201: 1195: 1189: 1154: 1151: 1150: 1149: 1135: 1111: 1108: 1107: 1106: 1100: 1094: 1084: 1078: 1072: 1061: 1047: 1029: 1026: 1025: 1024: 1018: 1017:for upsampling 1000: 990: 980: 949: 946: 945: 944: 930: 912: 906: 900: 877: 874: 873: 872: 849: 846: 845: 844: 838: 832: 822: 806: 803: 802: 801: 795: 789: 783: 777: 771: 765: 755: 745: 739: 726:) consists of 717: 707: 697: 687: 681: 663: 657: 647: 616: 613: 611: 608: 552:Direct methods 532:way is to use 511: 510:Spatial domain 508: 479: 476: 467: 464: 450: 447: 446: 445: 442: 439: 436: 412: 409: 406: 386: 381: 378: 373: 353: 350: 347: 327: 322: 319: 314: 303: 302: 290: 287: 284: 274: 262: 259: 256: 246: 231: 227: 223: 213: 201: 191: 190:— blur kernel, 179: 169: 157: 154: 151: 137: 136: 125: 122: 119: 116: 110: 106: 102: 99: 96: 93: 90: 87: 84: 81: 78: 75: 72: 69: 54: 51: 15: 13: 10: 9: 6: 4: 3: 2: 6662: 6651: 6648: 6646: 6643: 6641: 6638: 6637: 6635: 6622: 6621: 6620:Main category 6615: 6609: 6606: 6604: 6601: 6599: 6596: 6594: 6591: 6589: 6586: 6584: 6581: 6579: 6576: 6574: 6573:Pose tracking 6571: 6569: 6566: 6562: 6559: 6557: 6554: 6553: 6552: 6549: 6547: 6544: 6542: 6539: 6537: 6534: 6532: 6529: 6527: 6524: 6522: 6519: 6517: 6514: 6512: 6509: 6505: 6502: 6501: 6500: 6497: 6495: 6492: 6490: 6487: 6485: 6482: 6480: 6477: 6475: 6472: 6470: 6467: 6465: 6462: 6460: 6457: 6455: 6452: 6450: 6447: 6446: 6435: 6432: 6430: 6427: 6426: 6425: 6422: 6420: 6417: 6415: 6412: 6410: 6407: 6405: 6402: 6400: 6397: 6395: 6392: 6390: 6387: 6385: 6382: 6381: 6379: 6377: 6373: 6370: 6368: 6364: 6356: 6353: 6352: 6351: 6348: 6346: 6343: 6341: 6338: 6337: 6335: 6331: 6325: 6322: 6320: 6317: 6315: 6312: 6310: 6307: 6305: 6302: 6300: 6297: 6295: 6292: 6290: 6287: 6285: 6282: 6280: 6277: 6275: 6272: 6270: 6267: 6265: 6262: 6260: 6257: 6255: 6252: 6251: 6249: 6245: 6241: 6234: 6229: 6227: 6222: 6220: 6215: 6214: 6211: 6201:. 2018-10-15. 6200: 6196: 6190: 6187: 6182: 6178: 6174: 6170: 6166: 6162: 6155: 6152: 6140: 6136: 6130: 6127: 6122: 6118: 6114: 6110: 6106: 6102: 6098: 6094: 6090: 6086: 6079: 6076: 6071: 6067: 6062: 6057: 6053: 6049: 6045: 6038: 6035: 6029: 6024: 6017: 6014: 6008: 6003: 5996: 5993: 5988: 5984: 5979: 5974: 5970: 5966: 5962: 5955: 5952: 5946: 5941: 5934: 5931: 5926: 5922: 5918: 5914: 5910: 5906: 5902: 5898: 5894: 5890: 5886: 5882: 5875: 5872: 5867: 5863: 5858: 5853: 5850:: 5981–5988. 5849: 5845: 5841: 5834: 5831: 5826: 5822: 5818: 5814: 5810: 5806: 5802: 5798: 5794: 5790: 5783: 5780: 5774: 5769: 5762: 5759: 5754: 5750: 5746: 5742: 5738: 5734: 5730: 5726: 5719: 5716: 5711: 5707: 5703: 5701:9781450379885 5697: 5693: 5689: 5685: 5678: 5675: 5670: 5666: 5662: 5656: 5652: 5648: 5644: 5637: 5634: 5628: 5623: 5616: 5613: 5608: 5602: 5598: 5594: 5590: 5583: 5580: 5574: 5569: 5562: 5559: 5554: 5550: 5546: 5542: 5538: 5534: 5530: 5526: 5522: 5518: 5511: 5508: 5503: 5499: 5494: 5489: 5485: 5481: 5476: 5471: 5467: 5463: 5459: 5455: 5451: 5444: 5441: 5435: 5430: 5423: 5420: 5415: 5411: 5407: 5403: 5399: 5395: 5391: 5387: 5383: 5379: 5374: 5369: 5365: 5361: 5354: 5351: 5346: 5340: 5336: 5332: 5327: 5322: 5318: 5311: 5308: 5303: 5299: 5295: 5291: 5287: 5283: 5279: 5275: 5271: 5267: 5263: 5259: 5252: 5249: 5244: 5238: 5234: 5230: 5225: 5220: 5216: 5209: 5206: 5201: 5197: 5192: 5187: 5183: 5179: 5174: 5169: 5165: 5161: 5157: 5150: 5147: 5141: 5136: 5129: 5126: 5121: 5115: 5111: 5107: 5102: 5097: 5093: 5086: 5083: 5077: 5072: 5065: 5063: 5059: 5054: 5048: 5044: 5040: 5036: 5029: 5026: 5021: 5017: 5013: 5009: 5005: 5001: 4997: 4993: 4986: 4983: 4978: 4974: 4970: 4966: 4962: 4958: 4954: 4950: 4945: 4940: 4936: 4932: 4924: 4921: 4916: 4910: 4906: 4902: 4897: 4892: 4888: 4881: 4878: 4873: 4869: 4865: 4861: 4857: 4853: 4849: 4845: 4841: 4837: 4833: 4829: 4822: 4819: 4814: 4810: 4806: 4802: 4798: 4794: 4789: 4784: 4780: 4776: 4769: 4766: 4761: 4757: 4753: 4749: 4745: 4741: 4736: 4731: 4727: 4723: 4716: 4713: 4708: 4704: 4700: 4696: 4692: 4688: 4684: 4680: 4676: 4672: 4667: 4662: 4658: 4654: 4647: 4644: 4639: 4635: 4631: 4625: 4621: 4617: 4613: 4606: 4603: 4598: 4592: 4588: 4584: 4579: 4574: 4570: 4563: 4560: 4555: 4549: 4545: 4541: 4537: 4530: 4527: 4522: 4516: 4512: 4508: 4503: 4498: 4494: 4487: 4484: 4478: 4473: 4465: 4462: 4457: 4453: 4449: 4445: 4441: 4437: 4433: 4429: 4422: 4419: 4414: 4408: 4404: 4400: 4396: 4389: 4386: 4381: 4377: 4373: 4369: 4365: 4361: 4357: 4353: 4349: 4345: 4338: 4335: 4330: 4328:1-4244-0480-0 4324: 4320: 4316: 4312: 4305: 4302: 4297: 4295:0-7803-7041-4 4291: 4287: 4283: 4279: 4272: 4269: 4264: 4260: 4256: 4252: 4248: 4244: 4240: 4236: 4232: 4228: 4224: 4220: 4213: 4210: 4205: 4201: 4197: 4190: 4187: 4182: 4178: 4174: 4170: 4166: 4162: 4158: 4154: 4150: 4146: 4139: 4136: 4131: 4127: 4123: 4119: 4115: 4111: 4107: 4103: 4099: 4095: 4091: 4087: 4080: 4077: 4072: 4068: 4064: 4060: 4056: 4052: 4045: 4042: 4037: 4033: 4029: 4025: 4021: 4017: 4013: 4009: 4002: 3999: 3994: 3988: 3984: 3980: 3976: 3969: 3966: 3961: 3957: 3953: 3949: 3945: 3941: 3937: 3933: 3929: 3925: 3921: 3917: 3910: 3907: 3902: 3896: 3892: 3888: 3884: 3877: 3874: 3869: 3863: 3859: 3855: 3851: 3844: 3841: 3836: 3830: 3826: 3822: 3818: 3811: 3808: 3803: 3801:0-7803-9134-9 3797: 3793: 3789: 3785: 3778: 3775: 3770: 3766: 3762: 3758: 3754: 3750: 3746: 3742: 3738: 3734: 3727: 3725: 3721: 3716: 3714:0-7803-5467-2 3710: 3706: 3702: 3698: 3691: 3688: 3683: 3679: 3675: 3671: 3667: 3663: 3659: 3655: 3651: 3647: 3640: 3637: 3632: 3630:0-7803-9134-9 3626: 3622: 3618: 3614: 3607: 3604: 3599: 3595: 3591: 3584: 3581: 3576: 3574:0-8186-8183-7 3570: 3566: 3562: 3558: 3551: 3548: 3543: 3541:0-7803-6293-4 3537: 3533: 3529: 3525: 3518: 3515: 3510: 3506: 3502: 3498: 3494: 3490: 3483: 3480: 3475: 3471: 3467: 3463: 3459: 3455: 3451: 3447: 3443: 3439: 3435: 3431: 3424: 3421: 3416: 3412: 3408: 3404: 3400: 3396: 3389: 3386: 3381: 3379:0-7803-0532-9 3375: 3371: 3367: 3363: 3356: 3353: 3348: 3346:0-8186-6952-7 3342: 3338: 3334: 3330: 3323: 3320: 3315: 3313:3-540-51424-4 3309: 3305: 3301: 3297: 3290: 3287: 3283: 3277: 3274: 3267: 3263: 3260: 3258: 3255: 3253: 3250: 3248: 3245: 3243: 3240: 3238: 3235: 3233: 3230: 3229: 3225: 3223: 3219: 3215: 3213: 3208: 3202: 3195: 3191: 3189: 3186: 3182: 3178: 3174: 3166: 3163: 3160: 3157: 3154: 3151: 3148: 3145: 3142: 3139: 3136: 3133: 3132: 3131: 3129: 3120: 3114: 3111: 3108: 3105: 3102: 3099: 3096: 3093: 3091: 3090:RealSR + x265 3088: 3087: 3084: 3081: 3078: 3075: 3072: 3069: 3066: 3063: 3061: 3060:COMISR + x264 3058: 3057: 3054: 3051: 3048: 3045: 3042: 3039: 3036: 3033: 3031: 3030:SwinIR + x265 3028: 3027: 3024: 3021: 3018: 3015: 3012: 3009: 3006: 3003: 3001: 2998: 2997: 2994: 2991: 2988: 2985: 2982: 2979: 2976: 2973: 2971: 2970:SwinIR + x264 2968: 2967: 2963: 2960: 2957: 2954: 2951: 2948: 2945: 2943:ahq-11 + x264 2942: 2941: 2938: 2935: 2932: 2929: 2926: 2923: 2920: 2917: 2915: 2914:RealSR + x264 2912: 2911: 2907: 2904: 2901: 2898: 2895: 2892: 2889: 2886: 2885: 2879: 2877: 2873: 2864: 2858: 2855: 2852: 2849: 2846: 2843: 2840: 2837: 2834: 2831: 2829: 2826: 2825: 2822: 2819: 2816: 2813: 2810: 2807: 2804: 2801: 2798: 2795: 2793: 2790: 2789: 2786: 2783: 2780: 2777: 2774: 2771: 2768: 2765: 2762: 2759: 2757: 2754: 2753: 2750: 2747: 2744: 2741: 2738: 2735: 2732: 2729: 2726: 2723: 2721: 2718: 2717: 2714: 2711: 2708: 2705: 2702: 2699: 2696: 2693: 2690: 2687: 2685: 2682: 2681: 2678: 2675: 2672: 2669: 2666: 2663: 2660: 2657: 2654: 2651: 2649: 2646: 2645: 2642: 2639: 2636: 2633: 2630: 2627: 2624: 2621: 2618: 2615: 2613: 2610: 2609: 2605: 2602: 2599: 2596: 2593: 2590: 2587: 2584: 2581: 2578: 2577: 2571: 2564: 2557: 2554: 2551: 2548: 2545: 2542: 2540: 2537: 2536: 2532: 2529: 2526: 2523: 2520: 2517: 2514: 2511: 2510: 2506: 2503: 2500: 2497: 2494: 2491: 2488: 2485: 2484: 2480: 2477: 2474: 2471: 2468: 2465: 2462: 2459: 2458: 2454: 2451: 2448: 2445: 2442: 2439: 2437:based on EDVR 2436: 2433: 2432: 2428: 2425: 2422: 2419: 2416: 2413: 2410: 2407: 2406: 2402: 2399: 2396: 2393: 2390: 2387: 2384: 2381: 2380: 2376: 2374:1 × 2080 Ti 6 2373: 2370: 2367: 2364: 2361: 2358: 2355: 2354: 2350: 2347: 2344: 2341: 2338: 2336:Params number 2335: 2332: 2329: 2328: 2322: 2316: 2309: 2306: 2303: 2300: 2298:second result 2297: 2294: 2291: 2289:based on EDSR 2288: 2285: 2284: 2280: 2277: 2274: 2271: 2268: 2265: 2262: 2259: 2256: 2255: 2251: 2248: 2245: 2242: 2239: 2236: 2233: 2230: 2227: 2226: 2222: 2219: 2216: 2213: 2210: 2207: 2204: 2202:based on EDVR 2201: 2198: 2197: 2193: 2190: 2187: 2184: 2181: 2178: 2175: 2172: 2169: 2168: 2162: 2160: 2156: 2152: 2148: 2140: 2133: 2130: 2127: 2126: 2122: 2119: 2116: 2115: 2111: 2108: 2105: 2104: 2100: 2097: 2094: 2093: 2087: 2080: 2073: 2070: 2067: 2064: 2061: 2058: 2055: 2052: 2049: 2046: 2043: 2042: 2039: 2036: 2033: 2030: 2027: 2024: 2021: 2018: 2015: 2012: 2009: 2006: 2005: 2002: 1999: 1996: 1993: 1990: 1987: 1984: 1981: 1978: 1975: 1972: 1969: 1968: 1965: 1962: 1959: 1956: 1953: 1950: 1947: 1944: 1941: 1938: 1935: 1932: 1931: 1927: 1924: 1921: 1918: 1915: 1912: 1909: 1906: 1903: 1898: 1895: 1894: 1891: 1888: 1885: 1882: 1879: 1876: 1873: 1870: 1867: 1864: 1861: 1858: 1857: 1854: 1851: 1848: 1845: 1842: 1839: 1836: 1833: 1830: 1827: 1824: 1821: 1820: 1816: 1813: 1810: 1805: 1803:(clean track) 1800: 1795: 1790: 1788:(clean track) 1785: 1783:(clean track) 1780: 1777: 1774: 1773: 1767: 1762: 1754: 1747: 1743: 1739: 1735: 1732: 1729: 1726: 1723: 1721: 1718: 1717: 1713: 1709: 1705: 1702: 1699: 1696: 1693: 1691: 1688: 1687: 1684: 1680: 1676: 1673: 1670: 1667: 1665: 1661: 1659: 1656: 1655: 1651: 1647: 1644: 1641: 1638: 1635: 1632: 1630: 1627: 1626: 1623: 1619: 1615: 1612: 1609: 1606: 1603: 1600: 1598: 1595: 1594: 1591: 1587: 1584: 1581: 1578: 1576: 1573: 1571: 1568: 1567: 1564: 1560: 1557: 1554: 1552: 1549: 1546: 1543: 1541: 1538: 1537: 1533: 1530: 1527: 1524: 1521: 1520: 1514: 1508: 1501: 1498: 1495: 1492: 1489: 1487: 1484: 1483: 1479: 1476: 1473: 1470: 1467: 1465: 1462: 1461: 1457: 1454: 1451: 1448: 1445: 1443: 1442:Space-Time SR 1440: 1439: 1435: 1432: 1429: 1426: 1423: 1420: 1417: 1416: 1412: 1409: 1406: 1403: 1400: 1398: 1395: 1394: 1390: 1387: 1383:from 640×360 1382: 1379: 1376: 1373: 1370: 1369: 1365: 1362: 1359: 1356: 1353: 1351:(test SR set) 1350: 1347: 1346: 1342: 1339: 1336: 1333: 1330: 1328: 1325: 1324: 1320: 1317: 1314: 1311: 1308: 1306: 1303: 1302: 1299:Fine details 1298: 1295: 1292: 1289: 1286: 1283: 1282: 1276: 1274: 1266: 1264: 1262: 1258: 1254: 1249: 1243: 1239: 1236: 1234: 1230: 1227: 1225: 1221: 1217: 1214: 1211: 1208: 1205: 1202: 1199: 1196: 1193: 1190: 1187: 1183: 1179: 1176: 1175: 1174: 1172: 1164: 1159: 1152: 1147: 1143: 1139: 1136: 1133: 1130: 1129: 1128: 1126: 1121: 1117: 1109: 1104: 1101: 1098: 1095: 1092: 1088: 1085: 1082: 1079: 1076: 1073: 1069: 1065: 1062: 1059: 1056:and backward 1055: 1051: 1048: 1045: 1041: 1038: 1037: 1036: 1034: 1027: 1022: 1019: 1016: 1012: 1008: 1004: 1001: 998: 994: 991: 988: 984: 981: 978: 974: 970: 967: 966: 965: 963: 959: 955: 947: 942: 938: 934: 931: 928: 924: 920: 916: 913: 910: 907: 904: 901: 898: 897:discriminator 894: 890: 886: 883: 882: 881: 875: 870: 866: 862: 859: 858: 857: 855: 847: 842: 839: 836: 833: 830: 826: 823: 820: 816: 813: 812: 811: 804: 799: 796: 793: 790: 787: 784: 781: 780:MultiBoot VSR 778: 775: 772: 769: 766: 763: 759: 756: 753: 749: 746: 743: 740: 737: 733: 732:discriminator 729: 725: 721: 718: 715: 711: 708: 705: 701: 698: 695: 691: 688: 685: 682: 679: 675: 671: 667: 664: 661: 658: 655: 651: 648: 645: 642: 641: 640: 638: 634: 630: 626: 622: 614: 609: 607: 605: 601: 597: 593: 589: 585: 583: 579: 575: 571: 567: 565: 561: 557: 556:median filter 553: 549: 547: 543: 542:least squares 539: 535: 530: 529:Kalman filter 526: 522: 520: 516: 509: 507: 505: 501: 497: 493: 489: 485: 477: 475: 473: 465: 463: 461: 460:deep learning 456: 448: 443: 440: 437: 434: 433: 432: 428: 426: 407: 376: 348: 317: 285: 275: 257: 247: 229: 226: 214: 199: 192: 177: 170: 152: 142: 141: 140: 120: 114: 108: 105: 94: 91: 85: 76: 70: 60: 59: 58: 52: 49: 46: 44: 40: 36: 28: 22: 6618: 6511:Eye tracking 6367:Applications 6333:Technologies 6319:Segmentation 6198: 6189: 6164: 6160: 6154: 6143:. Retrieved 6141:. 2021-04-26 6138: 6129: 6088: 6084: 6078: 6051: 6047: 6037: 6028:2008.00455v1 6016: 5995: 5971:(12): 2085. 5968: 5964: 5954: 5945:2008.05765v2 5933: 5884: 5880: 5874: 5847: 5843: 5833: 5792: 5788: 5782: 5773:1902.06568v1 5761: 5728: 5724: 5718: 5683: 5677: 5642: 5636: 5627:1904.02870v1 5615: 5588: 5582: 5573:2007.11803v1 5561: 5520: 5516: 5510: 5457: 5453: 5443: 5434:1909.13057v1 5422: 5363: 5359: 5353: 5316: 5310: 5261: 5257: 5251: 5214: 5208: 5163: 5159: 5149: 5140:1905.02716v1 5128: 5091: 5085: 5076:2012.02181v1 5034: 5028: 4995: 4991: 4985: 4934: 4930: 4923: 4886: 4880: 4831: 4827: 4821: 4778: 4774: 4768: 4725: 4721: 4715: 4656: 4652: 4646: 4611: 4605: 4568: 4562: 4535: 4529: 4492: 4486: 4477:1611.05250v2 4464: 4431: 4427: 4421: 4394: 4388: 4347: 4343: 4337: 4310: 4304: 4277: 4271: 4222: 4218: 4212: 4195: 4189: 4148: 4144: 4138: 4089: 4085: 4079: 4054: 4050: 4044: 4011: 4007: 4001: 3974: 3968: 3919: 3915: 3909: 3882: 3876: 3849: 3843: 3816: 3810: 3783: 3777: 3736: 3732: 3696: 3690: 3649: 3645: 3639: 3612: 3606: 3589: 3583: 3556: 3550: 3523: 3517: 3492: 3488: 3482: 3433: 3429: 3423: 3398: 3394: 3388: 3361: 3355: 3328: 3322: 3295: 3289: 3281: 3276: 3257:Oversampling 3220: 3216: 3203: 3200: 3170: 3124: 2908:Open source 2882:Top methods 2868: 2606:Open source 2574:Top methods 2568: 2408:BOE-IOT-AIBD 2400:1 × Titan Xp 2351:Open source 2320: 2211:first result 2194:Open source 2144: 2084: 1817:Open source 1808:(blur track) 1798:(blur track) 1793:(blur track) 1758: 1512: 1385:to 4096×2160 1270: 1250: 1247: 1233:optical flow 1185: 1168: 1137: 1131: 1113: 1102: 1096: 1086: 1080: 1074: 1063: 1049: 1039: 1031: 1020: 1002: 997:convolutions 992: 982: 968: 958:convolutions 954:convolutions 951: 932: 914: 908: 902: 891:consists of 884: 879: 860: 851: 840: 834: 824: 814: 808: 797: 791: 785: 779: 773: 767: 757: 747: 741: 736:optical flow 719: 714:optical flow 709: 704:optical flow 699: 694:optical flow 689: 683: 665: 659: 649: 643: 618: 587: 586: 569: 568: 551: 550: 524: 523: 514: 513: 481: 469: 452: 429: 304: 138: 56: 47: 38: 34: 33: 6419:Visual hull 6314:Researchers 5965:Electronics 5160:IEEE Access 3207:demosaicing 3121:Application 2872:Video codec 2582:Multi-frame 2478:1 × 1080 Ti 2199:fenglinglwb 2071:GTX 1080 Ti 2034:GTX 1080 Ti 1960:RTX 2080 Ti 1340:SLow motion 1125:bicubically 1015:convolution 1011:convolution 973:convolution 937:aggregation 6634:Categories 6289:Morphology 6247:Categories 6145:2021-05-12 6007:1909.08080 5523:: 107619. 5373:1806.05764 5326:2007.10595 5224:1812.02898 5173:1909.10692 5101:2102.11720 4944:1810.08768 4896:1903.10128 4788:1711.09078 4735:1811.09393 4666:2001.02129 4578:1801.04590 4502:1704.02738 3268:References 3165:microscopy 2887:Model name 2585:Subjective 2579:Model name 2333:Model name 2325:Top teams 2275:TensorFlow 2249:2× 1080 Ti 2220:4× Titan X 2173:Model name 2165:Top teams 2128:ALONG_NTES 2090:Top teams 1957:TensorFlow 1925:Tesla V100 1886:Tesla V100 1778:Model name 1770:Top teams 1736:ERQAv2.0, 1706:ERQAv1.0, 1579:Youku-VESR 1509:Benchmarks 1449:100 frames 1427:100 frames 1404:10 seconds 1171:algorithms 1142:flickering 1009:) uses 3D 869:homography 854:homography 6181:219157416 6121:227282569 6105:0162-8828 6070:1751-9659 5987:2079-9292 5909:1057-7149 5866:2374-3468 5809:0162-8828 5753:235057646 5745:1051-8215 5710:222278621 5669:202763112 5553:225285804 5545:0031-3203 5484:1932-6203 5398:1057-7149 5302:231864067 5286:1057-7149 5200:2169-3536 5020:201264266 5012:0925-2312 4961:0162-8828 4856:1057-7149 4805:0920-5691 4760:209460786 4752:0730-0301 4707:210023539 4691:1057-7149 4638:0302-9743 4448:2333-9403 4364:1083-4419 4247:1057-7149 4173:1057-7149 4114:1057-7149 4071:1077-3142 4028:0165-1684 3944:1057-7149 3761:1057-7149 3674:1053-587X 3509:0923-5965 3458:1057-7149 3415:1047-3203 3190:and TVs. 3181:character 3153:astronomy 2684:DynaVSR-R 2538:CET CVLab 2044:XJTU-IAIR 1901:RCAN, DUF 1896:SuperRior 1525:Organizer 1522:Benchmark 1496:1920×1080 1474:4096×2160 1421:(test SR) 1407:4096×2160 1380:2 seconds 1349:Vimeo-90K 1334:31 frames 1312:43 frames 1093:mechanism 1091:attention 1033:Recurrent 952:While 2D 927:denoising 893:generator 885:VSRResNet 865:Attention 829:attention 819:attention 762:recurrent 728:generator 380:¯ 321:¯ 222:↓ 200:∗ 101:↓ 92:∗ 6324:Software 6284:Learning 6274:Geometry 6254:Datasets 6113:33270559 5925:53044490 5917:30346282 5817:28489532 5502:32649694 5454:PLOS ONE 5414:73415655 5406:30714918 5294:33560986 4977:53046739 4969:31722471 4872:58595890 4864:30571634 4813:40412298 4699:31995491 4372:15971920 4255:17605380 4181:18285235 4130:12116009 4122:17269630 4036:17920263 3952:19095517 3769:18262881 3682:52857681 3466:20457549 3226:See also 2876:bitrates 2600:CRRMv1.0 2597:QRCRv1.0 2588:ERQAv1.0 2555:1 × P100 2452:1 × V100 2426:1 × 1080 2382:Team-WVU 2359:EVESRNet 2286:HIT-XLab 2278:Titan Xp 2257:baseline 2188:Platform 1859:UIUC-IFP 1849:TITAN Xp 1822:HelloVSR 1811:Platform 1748:, LPIPS 1652:, LPIPS 1534:Metrics 1464:Harmonic 1452:1280×720 1430:1280×720 1357:7 frames 1267:Datasets 1146:ghosting 1116:weighted 1068:residual 786:BasicVSR 768:MEMC-Net 364:so that 5889:Bibcode 5525:Bibcode 5493:7351143 5462:Bibcode 5378:Bibcode 5266:Bibcode 5178:Bibcode 4836:Bibcode 4671:Bibcode 4456:9356783 4380:3162908 4263:1811280 4227:Bibcode 4153:Bibcode 4094:Bibcode 3960:2142115 3924:Bibcode 3741:Bibcode 3654:Bibcode 3438:Bibcode 3284:. 2021. 2792:RRN-10L 2756:DUF-28L 2527:0.249 s 2489:FineNet 2411:3D-MGBP 2356:KirinUK 2348:GPU/CPU 2304:PyTorch 2246:PyTorch 2217:PyTorch 2191:GPU/CPU 2134:40.405 2123:41.227 2112:41.617 2068:PyTorch 2031:PyTorch 1997:TITAN X 1994:PyTorch 1922:PyTorch 1916:120.000 1883:PyTorch 1846:PyTorch 1742:MS-SSIM 1528:Dataset 1372:Xiph HD 1360:448×256 1337:960×540 1315:720×480 1284:Dataset 1273:dataset 1153:Metrics 1138:MSHPFNL 993:3DSRnet 792:IconVSR 720:TecoGAN 710:SOF-VSR 644:Deep-DE 449:Methods 139:where: 6179: 6119: 6111: 6103: 6068: 5985: 5923: 5915: 5907: 5864: 5825:136582 5823: 5815: 5807: 5751: 5743: 5708: 5698: 5667: 5657: 5603: 5551: 5543: 5500: 5490: 5482: 5412: 5404: 5396: 5341: 5300: 5292: 5284: 5239: 5198: 5116: 5049: 5018: 5010: 4975: 4967: 4959: 4911: 4870: 4862: 4854: 4811: 4803: 4758: 4750: 4705: 4697: 4689: 4636: 4626: 4593: 4550: 4517: 4454: 4446: 4409: 4378: 4370: 4362: 4325: 4292: 4261: 4253: 4245: 4179: 4171: 4128: 4120: 4112: 4069: 4034: 4026: 3989: 3958: 3950: 3942: 3897: 3864: 3831: 3798: 3767: 3759: 3711: 3680: 3672: 3627: 3571: 3538: 3507: 3474:856101 3472: 3464: 3456: 3413: 3376: 3343: 3310: 2841:25.989 2828:RealSR 2805:24.252 2769:25.852 2733:30.244 2697:28.377 2661:31.291 2625:31.071 2552:0.04 s 2549:0.6112 2524:0.6165 2498:0.6256 2472:0.6321 2466:31.14M 2446:0.6353 2434:sr xxx 2423:4.83 s 2420:0.6304 2394:0.6378 2388:29.51M 2368:0.6450 2362:45.29M 2228:NERCMS 2131:37.632 2120:37.681 2117:NJU_L1 2109:37.851 2065:13.000 2059:0.8301 2022:0.8307 2016:0.8782 2007:NERCMS 1985:0.8333 1979:0.8804 1948:0.8067 1942:0.8822 1936:RecNet 1907:0.8811 1874:0.8430 1868:0.8748 1837:0.8647 1831:0.8962 1639:Vid3oC 1607:Vid3oC 1287:Videos 1120:fusion 1110:Videos 1058:fusion 1054:fusion 987:fusion 909:MRMNet 903:FFCVSR 831:module 752:fusion 742:TOFlow 678:fusion 660:VESPCN 650:VSRnet 633:pixels 544:(LS), 472:motion 6177:S2CID 6117:S2CID 6023:arXiv 6002:arXiv 5940:arXiv 5921:S2CID 5821:S2CID 5768:arXiv 5749:S2CID 5706:S2CID 5665:S2CID 5622:arXiv 5568:arXiv 5549:S2CID 5429:arXiv 5410:S2CID 5368:arXiv 5321:arXiv 5298:S2CID 5219:arXiv 5168:arXiv 5135:arXiv 5096:arXiv 5071:arXiv 5016:S2CID 4973:S2CID 4939:arXiv 4891:arXiv 4868:S2CID 4809:S2CID 4783:arXiv 4756:S2CID 4730:arXiv 4703:S2CID 4661:arXiv 4573:arXiv 4497:arXiv 4472:arXiv 4452:S2CID 4376:S2CID 4259:S2CID 4126:S2CID 4032:S2CID 3956:S2CID 3678:S2CID 3470:S2CID 3109:1.206 3106:1.033 3103:1.064 3100:1.617 3097:1.622 3094:0.502 3079:1.118 3076:0.672 3073:6.081 3070:1.302 3067:0.969 3064:0.367 3049:1.474 3046:4.641 3043:8.130 3040:1.304 3037:1.575 3034:0.346 3019:0.733 3016:0.881 3013:7.874 3010:0.698 3007:5.580 3004:0.335 2989:0.559 2986:0.736 2983:6.268 2980:0.642 2977:0.760 2974:0.304 2961:0.656 2958:0.719 2955:0.873 2952:0.753 2949:0.883 2946:0.271 2933:0.591 2930:0.487 2927:0.675 2924:0.775 2921:0.770 2918:0.196 2850:0.886 2847:0.000 2844:0.767 2838:0.690 2835:3.749 2817:0.390 2814:0.989 2811:0.557 2808:0.790 2802:0.627 2799:3.887 2781:2.392 2778:0.993 2775:0.549 2772:0.830 2766:0.645 2763:3.910 2742:0.994 2739:0.557 2736:0.883 2730:0.706 2727:4.036 2709:5.664 2706:0.997 2703:0.557 2700:0.865 2694:0.709 2691:4.751 2673:1.499 2670:0.996 2667:0.629 2664:0.898 2658:0.740 2655:5.040 2634:0.992 2631:0.629 2628:0.894 2622:0.737 2619:5.561 2612:DBVSR 2546:21.77 2521:21.91 2495:22.08 2469:22.28 2443:22.43 2417:22.48 2397:4.9 s 2391:22.48 2371:6.1 s 2365:22.83 2301:60.00 2292:21.45 2263:21.75 2234:22.35 2205:22.53 2101:VMAF 2056:28.86 2047:FSTDN 2028:6.020 2025:6.020 2019:28.98 2013:30.91 1991:1.390 1988:1.390 1982:28.92 1976:30.97 1954:3.000 1951:3.000 1945:27.71 1939:31.00 1904:31.13 1880:0.980 1877:0.980 1871:29.46 1865:30.81 1843:3.562 1840:2.788 1834:30.17 1828:31.79 1796:SSIM 1791:PSNR 1786:SSIM 1781:PSNR 1575:Youku 1327:SPMCS 1132:NLVSR 1087:BTRPN 1066:(the 1064:RISTN 983:FSTRN 933:MuCAN 887:like 774:RTVSR 760:(the 748:MMCNN 690:FRVSR 666:DRVSR 6109:PMID 6101:ISSN 6066:ISSN 5983:ISSN 5913:PMID 5905:ISSN 5862:ISSN 5813:PMID 5805:ISSN 5741:ISSN 5696:ISBN 5655:ISBN 5601:ISBN 5541:ISSN 5498:PMID 5480:ISSN 5402:PMID 5394:ISSN 5339:ISBN 5290:PMID 5282:ISSN 5237:ISBN 5196:ISSN 5114:ISBN 5047:ISBN 5008:ISSN 4965:PMID 4957:ISSN 4909:ISBN 4860:PMID 4852:ISSN 4801:ISSN 4748:ISSN 4695:PMID 4687:ISSN 4634:ISSN 4624:ISBN 4591:ISBN 4548:ISBN 4515:ISBN 4444:ISSN 4407:ISBN 4368:PMID 4360:ISSN 4323:ISBN 4290:ISBN 4251:PMID 4243:ISSN 4177:PMID 4169:ISSN 4118:PMID 4110:ISSN 4067:ISSN 4024:ISSN 3987:ISBN 3948:PMID 3940:ISSN 3895:ISBN 3862:ISBN 3829:ISBN 3796:ISBN 3765:PMID 3757:ISSN 3709:ISBN 3670:ISSN 3625:ISBN 3569:ISBN 3536:ISBN 3505:ISSN 3462:PMID 3454:ISSN 3411:ISSN 3374:ISBN 3341:ISBN 3308:ISBN 3179:and 3177:face 2720:TDAN 2648:LGFN 2594:SSIM 2591:PSNR 2501:13 s 2463:MAHA 2342:SSIM 2339:PSNR 2330:Team 2307:V100 2295:0.60 2272:0.09 2266:0.60 2260:RLSP 2243:0.51 2237:0.63 2231:PFNL 2214:0.35 2208:0.64 2179:SSIM 2176:PSNR 2170:Team 2155:SSIM 2153:and 2151:PSNR 2147:ECCV 2098:PSNR 2095:Team 2010:PFNL 1973:RBPN 1862:WDVR 1825:EDVR 1775:Team 1761:CVPR 1746:VMAF 1738:PSNR 1712:SSIM 1710:and 1708:PSNR 1679:SSIM 1675:PSNR 1664:Kwai 1650:SSIM 1646:PSNR 1634:ECCV 1618:SSIM 1614:PSNR 1602:ECCV 1590:VMAF 1586:PSNR 1563:SSIM 1559:PSNR 1551:REDS 1545:CVPR 1486:CDVL 1419:REDS 1354:7824 1305:Vid4 1103:RSDN 1097:RLSP 1075:RRCN 1050:BRCN 1040:STCN 1021:DMBN 1003:MP3D 975:for 941:fuse 923:fuse 915:STMN 895:and 835:TDAN 825:DNLN 815:EDVR 798:UVSR 758:RBPN 730:and 700:STTN 684:RVSR 676:and 623:and 527:use 425:Blur 6169:doi 6093:doi 6056:doi 5973:doi 5897:doi 5852:doi 5797:doi 5733:doi 5688:doi 5647:doi 5593:doi 5533:doi 5521:110 5488:PMC 5470:doi 5386:doi 5331:doi 5274:doi 5229:doi 5186:doi 5106:doi 5039:doi 5000:doi 4996:367 4949:doi 4901:doi 4844:doi 4793:doi 4779:127 4740:doi 4679:doi 4616:doi 4583:doi 4540:doi 4507:doi 4436:doi 4399:doi 4352:doi 4315:doi 4282:doi 4235:doi 4200:doi 4161:doi 4102:doi 4059:doi 4055:114 4016:doi 3979:doi 3932:doi 3887:doi 3854:doi 3821:doi 3788:doi 3749:doi 3701:doi 3662:doi 3617:doi 3594:doi 3561:doi 3528:doi 3497:doi 3446:doi 3403:doi 3366:doi 3333:doi 3300:doi 3113:YES 3083:YES 3053:YES 3023:YES 2993:YES 2964:NO 2937:YES 2857:YES 2821:YES 2796:YES 2785:YES 2760:YES 2749:YES 2724:YES 2713:YES 2688:YES 2677:YES 2652:YES 2641:YES 2616:YES 2558:NO 2533:NO 2512:TTI 2507:NO 2486:lyl 2481:NO 2475:4 s 2460:ZZX 2455:NO 2449:4 s 2429:NO 2414:53M 2403:NO 2377:NO 2310:NO 2281:NO 2252:NO 2223:NO 2182:MOS 2159:MOS 2074:NO 2038:YES 2001:YES 1970:TTI 1964:YES 1928:NO 1890:YES 1853:YES 1814:GPU 1725:MSU 1695:MSU 1683:MOS 1622:MOS 1186:MSE 1144:or 1081:RRN 969:DUF 921:to 889:GAN 861:TGA 724:GAN 564:SVD 562:or 39:VSR 6636:: 6197:. 6175:. 6165:46 6163:. 6137:. 6115:. 6107:. 6099:. 6089:PP 6087:. 6064:. 6052:15 6050:. 6046:. 5981:. 5967:. 5963:. 5919:. 5911:. 5903:. 5895:. 5885:28 5883:. 5860:. 5848:33 5846:. 5842:. 5819:. 5811:. 5803:. 5793:40 5791:. 5747:. 5739:. 5729:31 5727:. 5704:. 5694:. 5663:. 5653:. 5599:. 5547:. 5539:. 5531:. 5519:. 5496:. 5486:. 5478:. 5468:. 5458:15 5456:. 5452:. 5408:. 5400:. 5392:. 5384:. 5376:. 5364:28 5362:. 5337:. 5329:. 5296:. 5288:. 5280:. 5272:. 5262:30 5260:. 5235:. 5227:. 5194:. 5184:. 5176:. 5162:. 5158:. 5112:. 5104:. 5061:^ 5045:. 5014:. 5006:. 4994:. 4971:. 4963:. 4955:. 4947:. 4935:43 4933:. 4907:. 4899:. 4866:. 4858:. 4850:. 4842:. 4832:28 4830:. 4807:. 4799:. 4791:. 4777:. 4754:. 4746:. 4738:. 4726:39 4724:. 4701:. 4693:. 4685:. 4677:. 4669:. 4657:29 4655:. 4632:. 4622:. 4589:. 4581:. 4546:. 4513:. 4505:. 4450:. 4442:. 4430:. 4405:. 4374:. 4366:. 4358:. 4348:35 4346:. 4321:. 4288:. 4257:. 4249:. 4241:. 4233:. 4223:16 4221:. 4175:. 4167:. 4159:. 4147:. 4124:. 4116:. 4108:. 4100:. 4090:16 4088:. 4065:. 4053:. 4030:. 4022:. 4012:91 4010:. 3985:. 3954:. 3946:. 3938:. 3930:. 3920:18 3918:. 3893:. 3860:. 3827:. 3794:. 3763:. 3755:. 3747:. 3735:. 3723:^ 3707:. 3676:. 3668:. 3660:. 3650:55 3648:. 3623:. 3567:. 3534:. 3503:. 3493:19 3491:. 3468:. 3460:. 3452:. 3444:. 3434:19 3432:. 3409:. 3399:14 3397:. 3372:. 3339:. 3306:. 3175:, 2832:NO 1744:, 1740:, 1681:, 1677:, 1648:, 1642:16 1620:, 1616:, 1610:16 1588:, 1561:, 1502:— 1480:— 1424:30 1401:16 1377:70 1331:30 672:, 602:. 584:. 548:. 540:, 494:, 423:. 6232:e 6225:t 6218:v 6183:. 6171:: 6148:. 6123:. 6095:: 6072:. 6058:: 6031:. 6025:: 6010:. 6004:: 5989:. 5975:: 5969:9 5948:. 5942:: 5927:. 5899:: 5891:: 5868:. 5854:: 5827:. 5799:: 5776:. 5770:: 5755:. 5735:: 5712:. 5690:: 5671:. 5649:: 5630:. 5624:: 5609:. 5595:: 5576:. 5570:: 5555:. 5535:: 5527:: 5504:. 5472:: 5464:: 5437:. 5431:: 5416:. 5388:: 5380:: 5370:: 5347:. 5333:: 5323:: 5304:. 5276:: 5268:: 5245:. 5231:: 5221:: 5202:. 5188:: 5180:: 5170:: 5164:7 5143:. 5137:: 5122:. 5108:: 5098:: 5079:. 5073:: 5055:. 5041:: 5022:. 5002:: 4979:. 4951:: 4941:: 4917:. 4903:: 4893:: 4874:. 4846:: 4838:: 4815:. 4795:: 4785:: 4762:. 4742:: 4732:: 4709:. 4681:: 4673:: 4663:: 4640:. 4618:: 4599:. 4585:: 4575:: 4556:. 4542:: 4523:. 4509:: 4499:: 4480:. 4474:: 4458:. 4438:: 4432:2 4415:. 4401:: 4382:. 4354:: 4331:. 4317:: 4298:. 4284:: 4265:. 4237:: 4229:: 4206:. 4202:: 4183:. 4163:: 4155:: 4149:6 4132:. 4104:: 4096:: 4073:. 4061:: 4038:. 4018:: 3995:. 3981:: 3962:. 3934:: 3926:: 3903:. 3889:: 3870:. 3856:: 3837:. 3823:: 3804:. 3790:: 3771:. 3751:: 3743:: 3737:8 3717:. 3703:: 3684:. 3664:: 3656:: 3633:. 3619:: 3600:. 3596:: 3577:. 3563:: 3544:. 3530:: 3511:. 3499:: 3476:. 3448:: 3440:: 3417:. 3405:: 3382:. 3368:: 3349:. 3335:: 3316:. 3302:: 2853:— 2745:— 2637:— 2543:— 2530:— 2518:— 2504:— 2492:— 2440:— 2385:— 2269:— 2240:— 2062:— 2053:— 2050:— 1919:— 1913:— 1910:— 1733:4 1730:— 1703:4 1700:— 1671:— 1668:— 1582:4 1555:4 1499:— 1493:— 1490:— 1477:— 1471:— 1468:— 1446:5 1309:4 1188:) 1184:( 411:} 408:x 405:{ 385:} 377:x 372:{ 352:} 349:y 346:{ 326:} 318:x 313:{ 289:} 286:y 283:{ 261:} 258:n 255:{ 230:s 178:k 156:} 153:x 150:{ 124:} 121:n 118:{ 115:+ 109:s 98:) 95:k 89:} 86:x 83:{ 80:( 77:= 74:} 71:y 68:{ 37:( 23:.

Text is available under the Creative Commons Attribution-ShareAlike License. Additional terms may apply.