期刊名称：International Journal of Electronics, Communication and Soft Computing Science and Engineering
印刷版ISSN：2277-9477
出版年度：2014
卷号：3
期号：Special
出版社：IJECSCSE
摘要：Networks of video cameras are being installed in manyapplications, e.g., surveillance and security, disaster response,environmental monitoring, etc. Currently, most of the datacollected by such networks is analyzed manually, a task that isextremely tedious and reduces the potential of the installednetworks. Tracking and activity recognition are two fundamentaltasks in this regard. For multitarget tracking in a distributedcamera network, the Kalman-Consensus algorithm can beadapted to take into account the directional nature of videosensors and the network topology. For the activity recognitionproblem, a probabilistic consensus scheme is derived whichcombines the similarity scores of neighboring cameras to come upwith a probability for each action at the network level.presents a distributed multicamera face tracking system suitablefor large wired camera networks. Instead, an efficient cameraclustering protocol is used to dynamically form groups of camerasfor in-network tracking of individual faces. The clusteringprotocol includes cluster propagation mechanisms that allow thecomputational load of face tracking to be transferred to differentcameras as the target move.
关键词：Activity recognition; camera networks; consensus;distributed image processing; tracking
