| 摘要 |
The invention relates to a method for recording and playing back a sequence of events, in particular a sporting event, on a three-dimensional playing field, such as e.g. a soccer pitch, in which image data of the playing field is continuously generated by at least two cameras (2) and said image data is transmitted to a processing unit (3) having a ROI-sampling unit. The ROI-sampling unit selects algorithm-based relevant image regions of the particular camera (2), from which local feature data is then generated and subsequently relayed to a higher-level central evaluating unit (4). In the central evaluating unit (4) the local feature data for determining an object part is combined, in order by repetition of this step for local feature data in adjoining image regions of the cameras to construct a virtual 3D object (5) in a virtual 3D space (6). As a result, a real event that has been recorded by the cameras (2) is reconstructed in the virtual 3D space (6) using the constructed 3D objects (5) and finally displayed in the form of an image sequence. |
