Video indexing based on mosaic representations

摘要: Video is a rich source of information. It provides visual information about scenes. This implicitly buried inside the raw video data, however, and provided with cost very high temporal redundancy. While standard sequential form storage adequate for viewing in movie mode, it fails to support rapid access interest that required many emerging applications video. paper presents an approach efficient access, use manipulation data. The data are first transformed from their redundant frame-based representation, which scene distributed over frames, explicit compact scene-based each frame can be directly related. reorganization supports nonlinear browsing indexing provide interest. describes new set methods into sequence based on representation. These geometric dynamic contained complement more traditional content-based methods, utilize image appearance (namely, color texture properties) but considerably simpler achieve highly computationally efficient.