3-7 単一映像フレームからの字幕領域検出手法(第3部門 メディア処理1)

抄録

We proposes a method for detecting display regions of open caption from single video frame. Proposed method scans an input image using small rectangle window to extract candidate regions, and then integrates the detected regions based on the area size and the edge density. Experimental results showed the effectiveness of the proposed method.