Abstract
Due to the limitation of shooting conditions in the real world, there exist many insufficient-resolution videos. To be transmitted under limited bandwidth conditions, low-resolution videos often have to be compressed further, which introduces more compression artifacts and results in severer damage to the video quality. Accordingly, video super-resolution in practical applications must repair compressed damage and down-sampling damage simultaneously, which can be intuitively solved by two subtasks, including compressed video quality enhancement (VQE) and video super-resolution (VSR). Therefore, we proposed a novel model, the Feature Multiplexing Video Super-Resolution for Compressed Video (FM-VSR) to effectively handle such multi-tasks. Firstly, we use a complicated Deformable Regression Pyramid (DRP) module to align the reference frame and each supporting frame. Then feature multiplexing structure is utilized in VSR to make better use of the VQE information. Finally, a skip-attention structure is introduced in the reconstruction module to predict the high-quality video frame. Many experiments on benchmark datasets show that our method can achieve better performance both in quantitative and qualitative than the state-of-the-art (SOTA) two-stage approaches.
Highlights
Nowadays, video has become an essential part of digital network traffic — in 2019, video accounted for about 60% of global Internet traffic, and it continues to grow
This paper proposes a one-stage network: Feature Multiplexing Video Super-Resolution for compressed video (FM-video super-resolution (VSR)), which can directly recover high-quality and highresolution videos from low-resolution videos processed with high compression rates
We propose a one-stage video super-resolution network (FM-VSR) that can recover the high-quality, high-resolution videos from the low-resolution videos compressed with high compression rates
Summary
Video has become an essential part of digital network traffic — in 2019, video accounted for about 60% of global Internet traffic, and it continues to grow. Due to the limitation of video shooting conditions in practical applications, a portion of videos’ resolution is insufficient. After being transmitted under limited bandwidth conditions, the video quality will be damaged further, which seriously affects the visual experience. The development of video quality enhancement technology has been attracting people’s attention. As far as we know, the current deep learning technologies have achieved remarkable results in compressed video quality enhancement and video super-resolution, respectively. The existing video quality enhancement (VQE) and video super-resolution (VSR) can only solve a single problem in their respective fields. They cannot be directly applied to the restoration task of compressed low-resolution video in real life.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have