DRVI: Dual Refinement for Video Interpolation
The quality of a video clip is considered to be poor if the resolution or the frame rate is low. Video interpolation is thus introduced to enhance video quality and provide a better viewing experience to users. However, there are still some challenges, like the blur caused by motion changes. In this...
Main Authors: | , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2021-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/9513293/ |
id |
doaj-97d2e373ee2f4666965fa8d547dea819 |
---|---|
record_format |
Article |
spelling |
doaj-97d2e373ee2f4666965fa8d547dea8192021-08-19T23:00:26ZengIEEEIEEE Access2169-35362021-01-01911356611357610.1109/ACCESS.2021.31045269513293DRVI: Dual Refinement for Video InterpolationXuanyi Wu0https://orcid.org/0000-0001-9817-070XZhenkun Zhou1Anup Basu2https://orcid.org/0000-0002-7695-4148Department of Computing Science, University of Alberta, Edmonton, AB, CanadaHuawei Fields Laboratory, Hangzhou, ChinaDepartment of Computing Science, University of Alberta, Edmonton, AB, CanadaThe quality of a video clip is considered to be poor if the resolution or the frame rate is low. Video interpolation is thus introduced to enhance video quality and provide a better viewing experience to users. However, there are still some challenges, like the blur caused by motion changes. In this paper, we introduce a dual refinement technique for video interpolation (DRVI). It has three main steps, namely flow refinement, frame synthesis, and Haar refinement. The flow refinement can generate accurate bi-directional flows, which are more suitable for frame interpolation tasks. The Haar refinement uses the Discrete Wavelet Transform (DWT). It can preserve information in different frequency domains and also speed up the learning process. We also add an arbitrary time approximation module to allow multi-frame generation. The number of learnable parameters in our model is much less than existing methods; still, it has excellent performance. Our method is trained on Vimeo90K (Xue <italic>et al.</italic>, 2019) and tested on three well-known datasets to demonstrate its effectiveness.https://ieeexplore.ieee.org/document/9513293/Video interpolationframe interpolationoptical flowvideo quality |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Xuanyi Wu Zhenkun Zhou Anup Basu |
spellingShingle |
Xuanyi Wu Zhenkun Zhou Anup Basu DRVI: Dual Refinement for Video Interpolation IEEE Access Video interpolation frame interpolation optical flow video quality |
author_facet |
Xuanyi Wu Zhenkun Zhou Anup Basu |
author_sort |
Xuanyi Wu |
title |
DRVI: Dual Refinement for Video Interpolation |
title_short |
DRVI: Dual Refinement for Video Interpolation |
title_full |
DRVI: Dual Refinement for Video Interpolation |
title_fullStr |
DRVI: Dual Refinement for Video Interpolation |
title_full_unstemmed |
DRVI: Dual Refinement for Video Interpolation |
title_sort |
drvi: dual refinement for video interpolation |
publisher |
IEEE |
series |
IEEE Access |
issn |
2169-3536 |
publishDate |
2021-01-01 |
description |
The quality of a video clip is considered to be poor if the resolution or the frame rate is low. Video interpolation is thus introduced to enhance video quality and provide a better viewing experience to users. However, there are still some challenges, like the blur caused by motion changes. In this paper, we introduce a dual refinement technique for video interpolation (DRVI). It has three main steps, namely flow refinement, frame synthesis, and Haar refinement. The flow refinement can generate accurate bi-directional flows, which are more suitable for frame interpolation tasks. The Haar refinement uses the Discrete Wavelet Transform (DWT). It can preserve information in different frequency domains and also speed up the learning process. We also add an arbitrary time approximation module to allow multi-frame generation. The number of learnable parameters in our model is much less than existing methods; still, it has excellent performance. Our method is trained on Vimeo90K (Xue <italic>et al.</italic>, 2019) and tested on three well-known datasets to demonstrate its effectiveness. |
topic |
Video interpolation frame interpolation optical flow video quality |
url |
https://ieeexplore.ieee.org/document/9513293/ |
work_keys_str_mv |
AT xuanyiwu drvidualrefinementforvideointerpolation AT zhenkunzhou drvidualrefinementforvideointerpolation AT anupbasu drvidualrefinementforvideointerpolation |
_version_ |
1721201892771495936 |