Method for time segmentation of a video in video picture sequences and for the selection of key frames for the discovery of image contents, with the inclusion of a sub-shot - detection

摘要

Known process, a shot - detection (uninterrupted receiving a camera) and an assignment of a plurality of key frames to the shots by. A subsequent scenes recognition is based on the grouping of shots scenes. A scene, however, contains only one event in an environment. Since both can vary within a shots, is not every scenes - limit also at the same time a shot - limit. Furthermore, it is not sufficient each shot short, so that a secure location of different image contents is not ensured. In the method according to the invention therefore, in addition subs hots be introduced and defined in such a way that scenes - and shot - limits basically at the same time also sub-shot - limits are. Subs hots in addition, comprise only video images with a small change of the image content. In this way it can be reliably assured that each sub-shot for reliable recognition of bildinahlten always only has to be assigned precisely one key. Scenes can then be replaced by the grouping of successive subs hots accurately reflect will be described. The sub-shot - detection can be neer with the aid of various algorithms, for example, threshold value -, cluster - and / or movement - algorithm, be carried out.