期刊名称：International Journal of Innovative Research in Computer and Communication Engineering
印刷版ISSN：2320-9798
电子版ISSN：2320-9801
出版年度：2015
卷号：3
期号：2
出版社：S&S Publications
摘要：A numerous surveillance circumstances such as those involving a busy traffic scene, a busy rail station,or a shopping mall, various motions are involved. It is highly desirable to analyze the motion patterns and obtain somehigh-level interpretation of the semantic relations content. For example, in a video monitoring intersection, without anyprior knowledge about the traffic rules in the specific scene. A multilevel hierarchical motion pattern mining useful todiscover typical Color versions of one or more of the information. Two traffic scenes illustrate activities and trafficstates. Arrows and show single agent, multiple agent motion patterns while arrows grouped by the same colors showinteraction patterns vehicle behaviors and their dependencies involved in this scene and detect inconsistent motion forsecurity distress motion patterns involved in a complex dynamic scene usually are of a hierarchical nature; that is, atlow level, they consist of single agent motion, multiple agent patterns, which are combined at a higher level to forminteraction patterns.
关键词：Hierarchical motion pattern mining; semantic relations; video segmentation; visual surveillance
