Abstract: Objective Decomposition of a Geo-video stream presents the expression of video from spatial featureset.Although it has been studied widely,spatial relations underlying a scenario are not well under-stood.Here we use a method which takes advantage of`random graph theory to investigate the seman-tic knowledge in Geo-video,leading to correlation analysis of the target motion elements in a Geo-vide-o stream.We used the connections of target motion elements,both correlation and continuity,to leadto a structure in time series that reveals clues to the event development of the video stream.Further-more,it provides a method for the effective integration of semantic and campaign information.Ulti-mately,experimental results show that the method offers a better description of video elements thancan be achieved with existing schemes.In addition,the proposed method offers a new way of thinkingabout the semantic description of the geographic video scenarios.