中文摘要：

通过扩展Petri网定义，提出了一种监控视频事件时空关系和逻辑关系的描述方法，通过将语义级的查询事件映射成Petri网。再在Petri网推理过程中结合计算机视觉算法对场景运动目标行为的解释，实现了有关运动目标行为的事件抽取和相应监控视频片段的定位。

英文摘要：

With an extended definition of Petri net, this paper proposes a Petri net based representation scheme of the spatiotemporal relations and events in surveillance video. By mapping high-level semantic queries into a Petri net and inferring with interpretations of moving objects using low-level computer vision algorithms, events about activities of moving objects are inferred and the relevant video clip is well located.