الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 101 |  |  |
| | | from | 101 | | |
Abstract
This dissertation presents the problem of counting and tracking multiple targets, whose motions are independent. The counting problem is investigated by estimating the number of targets. Estimation is done by using the dynamic counting techniques, which exploit the temporal and spatial dependencies of the targets. Meanwhile, the target tracking problem is approached by estimating the targets trajectories, through the implementation of an enhanced probability hypothesis density-based filter. The proposed target tracking approach enhances the probability hypothesis density-based filter, to include the dynamic counting techniques. This enhancement introduces the implementation and simulation of the enhanced dynamic counting probability hypothesis density (DC-PHD) approach. This paper investigates the implementation of four of the existing target tracking and counting algorithms:
1) ClusterTrack filter, which is based on particle filtering techniques. It explores multiple targets tracking in a one-dimensional environment.
2) A distributed energy efficient (DEE) algorithm that tracks a single target in a two-dimensional environment.
3) Multicolor particle filter (MCPF) technique, which tracks multiple targets in a two-dimensional environment.
4) Probability hypothesis density filter, upon which this work is based. It shows the originally proposed multiple target tracking and counting in a two-dimensional environment.
The dissertation shows the simulations of the DC-PHD, by exploring different environmental settings. The accuracy of the proposed algorithm is derived mathematically, where the disc sensing model is used. Simulations compare the performance of the proposed algorithm with the previously mentioned target tracking approaches. These comparisons show the efficiency of the proposed target counting and tracking technique, by using the disc sensing model of the sensor nodes.