Moving Visual Representations of Video Objects for Content

This paper proposes object-based moving visual representations for quick browsing of video content. These representations are hierarchical, such that at the coarse level a sequence of alpha planes provides a moving representation of object shape and motion information for object contours. Alternatively, a 2D mesh representation provides a complete visual representation of object motion and shape. The nest level visual representation can be obtained by texture mapping onto the moving meshes. The paper also discusses trade-oos between each representation in terms of the amount of indexing information that needs to be stored, the robustness of the representation, and the accuracy of the representation. 1 Introduction EEective visual representation and summarization of content is a crucial supplement to metadata, keyword , and/or content-based indexing for quick browsing and searching of video databases. A set of selected key frames have been commonly used for content visualization with a minimal amount of data. An example is the \storyboard representation," in which time ordered thumbnail images are presented to provide a summary of the content. These representative key frames can be chosen at regular time intervals or one can be selected for each camera shot after temporal segmentation 1, 2]. Similarly, key frames can be presented in a \scene transition graph," which summarizes the visual content in a temporal story structure 3]. Another alternative is \video posters," which depict the visual summary of the video sequence in a