Decentralized Bayesian Algorithm for Identification of Tracked Targets

摘要

The problem of identification of objects being tracked by a fully decentralizedsurveillance system is considered. A fully decentralized system is an interconnected network of intelligent sensors each with its own transducers and processing capabilities. Locally obtained results are communicated to the other sensors for further processing, allowing each sensor to achieve a global result. Such a system has many benefits in terms of modularity, speed, and survivability. In particular, this work relates to a real time multitarget surveillance system used to track targets (people and mobile robots) as they enter and move around a factory assembly room performing tasks. The sensors used on this system (CCD (Charged Coupled Device) cameras) reveal information about the targets that is sufficiently rich to allow them not only to be tracked but also identified as a person, a robot, etc. This identified information can be used as an man machine interface design and to facilitate situation assessment. The identification problem is defined and a centralized Bayesian algorithm for determining the identity of each target based on each sensor's information is developed. The algorithm is decentralized and the its performance compared to the centralized version. Results of an implementation of the algorithm working on real data from the surveillance system are presented.