Abstract

Nowadays, digital video documents are stored and are grown in the number and the size. Therefore, it requires efficient management techniques that allow retrieving visual information in an efficient way. Traditional management by manual annotation and appropriate for a huge volume data as digital video dutu, so in this paper we represent method for automatic structural analysis digital video to generate the table of content (TOC) and the index table, storing digital video data by high-level features in small size, based on that structure we can enriched video data with semantic labels. We approach this problem by studying the decomposition of the video sequence into elementary segments, after that they are classified by hierarchical clustering algorithm and finally reducing hierarchical structure to generate T℃ and the Index Table. We experimented on the sports video, documentary video, the results showed that ToC and the Index Table contained useful information for retrieving visual information more efficiently. Based on this structure, we can: Retrieve visual information. Browse following the structure of video documents. Filter the content of video documents.