Optimal sensor selection for video-based target tracking in a wireless sensor network

Optimized information processing in resource-constrained vision systems. From low-complexity coding to smart sensor networks

“…The task assignment methods of [Pahalawatta and Katsaggelos, 2004, Yao et al, 2010, Soro and Heinzelman, 2007 deal with limited computational and energy capacities with minor consideration for actual tracking accuracy in occluded and confusing environments (in contrast to our approach where we monitor the quality of the tracking through the suitability functions). In particular, [Yao et al, 2010] studies the uni-form distribution of computational load among the cameras such that the number of dropped objects is minimized and the system frame rate maintained.…”

“…In [Soro and Heinzelman, 2007], energy limitations of wireless camera nodes are integrated in the camera assignment in order to reduce the overall network power consumption. Camera assignment for tracking and localization has been studied in [Isler and Bajcsy, 2005, Pahalawatta and Katsaggelos, 2004, Denzler et al, 2003, Sommerlade and Reid, 2008, Snidaro et al, 2003, Gupta et al, 2007, Ercan et al, 2006. These methods will be discussed in more detail in Section 5.7.…”

“…The methods of [Pahalawatta and Katsaggelos, 2004,Denzler et al, 2003,Sommerlade and Reid, 2008,Snidaro et al, 2003] cannot effectively take occlusion into account -a frequent problem in tracking -without significant reformulation of the algorithms. In [Gupta et al, 2007] cameras are selected especially to avoid occlusion (and confusion -people being visible behind the target) in a localization task.…”

“…Most mentioned camera assignment approaches either mainly deal with limited computational and energy capacities [Yao et al, 2010, Soro and Heinzelman, 2007, Pahalawatta and Katsaggelos, 2004, or choose task quality as the primary criterion under the assumption of unlimited computational power [Gupta et al, 2007, Denzler et al, 2003, Sommerlade and Reid, 2008, Snidaro et al, 2003, Ercan et al, 2006. Other approaches focus on specific active vision systems [Bakhtari andBenhabib, 2007, Davis et al, 2007], while we aim at formulating a general solution.…”

“…In [Soro and Heinzelman, 2007], energy limitations of wireless camera nodes are integrated in the camera assignment in order to reduce the overall network power consumption. Pahalawatta et al [Pahalawatta and Katsaggelos, 2004] select a limited number of active sensors nodes for target tracking in order to maximize the network lifetime.…”