A Caption Detection Algorithm Based on Morphological Operation
一种基于形态学操作的新闻标题条检测算法

An algorithm of caption detection in news video is proposed in this paper. It makes use of space and time characters of the captions appearing in the video. On the basis of shot boundary detection, this method applies morphological operation and different filter strategies according to different cases. It achieves good results by experiments using MPEG-7 video testing data, especially caption detection in the case of gradual transition and heavy noises on the background.