Current advances in multimedia technology enable ease of capturing and encoding digital video. As a result, video data is rapidly growing and becoming very important in our life. It is because video can transfer a large amount of knowledge by providing combination of text, graphics, or even images. Despite the vast growth of video, the effectiveness of its usage is very limited due to the lack of a complete technology for the organization and retrieval of video data. To date, there is no "perfect" solution for a complete video data-management technology, which can fully capture the content of video and index the video parts according to the contents, so that users can intuitively retrieve specific video segments. We have found that successful content-based video datamanagement systems depend on three most important components: key-segments extraction, content descriptions and video retrieval. While it is almost impossible for current computer technology to perceive the content of the video to identify correctly its key-segments, the system can understand more accurately the content of a specificudvideo type by identifying the typical events that happens just before or after the key-segments (specific-domainapproach). Thus, we have proposed a concept of customisable video segmentation module, which integrates the suitable segmentation techniques for the current type of video. The identified key-segments are then described using standard video content descriptions to enable content-based retrievals. For retrieval, we have implemented XQuery, which currently is the most recent XML query language and the most powerful compared to older languages, such as XQL and XML-QL.
展开▼
