Video composition and retrieval

A significant increase in the amount of digital video has created a clear demand for specialized software that can manipulate video in an efficient and easy-to-use manner. This paper focuses on video composition-an area that is very important from the user´s viewpoint but has not been addressed in any depth. Traditionally, video composition is performed for the purposes of video production and analysis of video data. However, there are no software tools that provide the end user facility to compose, manipulate and query videos on an ad hoc basis. We propose a new method for video composition and retrieval, which presents many opportunities for structuring video segments. This enables users to compose and manipulate video on the basis of the objects that appear in them. Each video object represents a VOP (video object plane) as provisioned in MPEG-4. Various video editing operations, such as scaling, rotation, spatial and temporal segmentation, motion extraction, and camera operation detection, can be performed on video objects. The composed video objects may participate as part of a visual query in the VideoRoadMap system