A multi-view video synopsis framework

In this paper, a simple framework for multi-view video synopsis is introduced by combining both the benefits of video summarization and video synopsis. While video summarization includes the important actions of an original video, video synopsis, on the other hand, simultaneously displays multiple objects from different intervals of time. To create a multi-view video synopsis, the objects in the field-of-views of all the cameras are shown in a common background plane. All the moving objects are detected, prioritized based on their actions, and only important objects are included in the final synopsis video. The proposed framework is evaluated using two datasets and is compared with existing techniques, which shows a significant reduction in synopsis length with the proper inclusion of important objects.