Distributed Online Localization of Wireless Camera-based Sensor Networks by Tracking Multiple Moving Objects
A typical wireless sensor network consists of a large number of sensor nodes densely deployed in the field, each equipped with wireless communication, sensing and computing capabilities with a limited power resource. Each node gathers its surrounding information according to its sensing modality, and shares the data by communicating with other nodes in the network. Since the data collected by each node is location-specific, obtaining the locations of nodes is one of the fundamental problems in sensor networks.  Consequently, there has been much effort in developing localization algorithms in wireless sensor network. These localization techniques, however, are not suited for camera sensors for two main reasons. First, the level of localization accuracy achieved is not sufficient for basic computer vision tasks. More important, the localization algorithms developed so far do not provide the orientation of a sensor, which is crucial for camera-based sensor networks.
Although the existing localization algorithms may be used to provide approximate locations of camera sensors, in order to obtain the precise positions and orientations of cameras that would be appropriate for basic computer vision tasks involving multiple cameras, an alternative localization technique is needed.  Thus, we have developed a distributed online localization algorithm that is suitable for wireless camera-based sensor networks.  Our system does not require any beacon nodes, but only utilize object features of moving objects in the scene extracted from image sequence. The algorithm is fully distributed, and the localization estimates can be improved throughout the course of network life as more object features are obtained. Video clips that show the simulations of our system using a graphical simulator are available below.
  •    Johnny Park
Henry Medeiros, Hidekazu Iwaki, and Johnny Park , “Online Distributed Calibration of a Large Network of Wireless Cameras Using Dynamic Clustering,” ACM/IEEE International Conference on Distributed Smart Cameras, 2008. (accepted)
The two video clips below show the localization process visualized by our simulation tool. Before viewing the videos, please see the table below so that you will fully understand the entire localization process.
Localization of 100 cameras in a room-like environment
Localization of 100 cameras in a street-like environment