Content-based news video story segmentation and video retrieval
Huayong Liu, Dongru Zhou · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2002
In this paper, we first define the basic structure of news video databases system, the system is developed for content-based broadcast news video browsing, and users can use this system to fmd the news story what they want to see quickly. And then we introduce the basic ideas of key techniques — text extraction and speech recognition, etc, which realize news video story segmentation and video retrieval. In this article, first, we use image analysis technology to detect the frames that contain the texts or the close captions, and we use the video OCR technology to detect the texts of these frames, which provides a good source of textual information for story segmentation and classification. The speech digital processing technology is also used to find the boundaries of the news video story and the speech recognition technology is used to extract the voice texts of the anchorperson and also used to classify the stories that are segmented from the news video. At last, we give typical experiment results.