Detail-Revealing Deep Video Super-Resolution

    Abstract

    Previous CNN-based video super-resolution approaches need to align multiple frames to the reference. In this paper, we show that proper frame alignment and motion compensation is crucial for achieving high quality results. We accordingly propose a “sub-pixel motion compensation” (SPMC) layer in a CNN framework. Analysis and experiments show the suitability of this layer in video SR. The final end-to-end, scalable CNN framework effectively incorporates the SPMC layer and fuses multiple frames to reveal image details. Our implementation can generate visually and quantitatively high-quality results, superior to current state-of-the-arts, without the need of parameter tuning.

    Authors

    Xin Tao, Hongyun Gao, Renjie Liao, Jue Wang, Jiaya Jia, Kaustav Kundu

    Conference

    ICCV 2017

    Full Paper

    ‘Detail-Revealing Deep Video Super-Resolution’ (PDF)

    Uber ATG

    Comments