I-010 Video2Manga : A Comic-like Video Summary

概要

論文の詳細を見る
Video summarization and visualization plays an important role in recent years because we need efficient techniques for faster retrieval and easier understanding of video content. In this paper, we propose a "Video2Manga" system that can summarize and represent a video sequence as "Manga" style. Video key-frames first are extracted from video frames by using clustering-based method. In order to understand the video content easier and faster, we then perform Non-Photorealistic Rendering (NPR) for each key-frame to create an actual "Manga" style. The result shows that our system not only can keep the most significant information of video but also can make it easy to understand.
2010-08-20