MANet: Improving Video Denoising with a Multi-Alignment Network
This work addresses noise reduction in videos, offering incremental improvements for applications like video processing and enhancement.
The paper tackles video denoising by proposing a multi-alignment network that generates multiple flow proposals and uses attention-based averaging to mimic non-local mechanisms, improving a baseline model by 0.2dB and reducing parameters by 47% with distillation.
In video denoising, the adjacent frames often provide very useful information, but accurate alignment is needed before such information can be harnassed. In this work, we present a multi-alignment network, which generates multiple flow proposals followed by attention-based averaging. It serves to mimic the non-local mechanism, suppressing noise by averaging multiple observations. Our approach can be applied to various state-of-the-art models that are based on flow estimation. Experiments on a large-scale video dataset demonstrate that our method improves the denoising baseline model by 0.2dB, and further reduces the parameters by 47% with model distillation. Code is available at https://github.com/IndigoPurple/MANet.