Video harmonization is an important step in video editing to achieve visual consistency by adjusting foreground appear-ances in both spatial and temporal dimensions.Previous methods always only harmonize on a single scale or ignore the inaccuracy of flow estimation,which leads to limited harmonization performance.In this work,we propose a novel architecture for video harmoniza-tion by making full use of spatiotemporal features and yield temporally consistent harmonized results.We introduce multiscale harmon-ization by using nonlocal similarity on each scale to make the foreground more consistent with the background.We also propose a fore-ground temporal aggregator to dynamically aggregate neighboring frames at the feature level to alleviate the effect of inaccurate estim-ated flow and ensure temporal consistency.The experimental results demonstrate the superiority of our method over other state-of-the-art methods in both quantitative and visual comparisons.
展开▼
