Video Taxonomy Identify Using Frame-Based Naive Similarity Finder Algorithm

Due to the advances in multimedia applications, large databases of videos require efficient methods that enable fast browsing and accessing the information pursued. Most of video data are stored in personal video recorders (PVRs) such as DVD recorders and hard disc recorders. We propose a video summarization approach for PVRs application, which is based on two-level repetitive information detection and content analysis. First the original video sequence is divided into shots and scenes, and key frames are extracted from these shots. Then it removes redundant video content in the shot level. Impact factors of scenes and key frames are defined, and parts of shots are selected to generate the initial video summary. Finally a repetitive frame segment detection step is used to remove redundant information in the initial video summary. With the two-level redundancy analysis procedure, this tool could remove almost all repetitive information.