Using Temporal Knowledge to Detect Person Movements with a Multisensor System

In recent years there has been increasing interest and research on multisensor applications, e.g. the ESPRIT project DIMUS deals with the observation of an underground station with a multisensor system. Our objective is the development of a robust multisensor integration method which can be used to detect person movements and their presentation in context of a certain multisensor configuration built in the DIMUS project. Spatially and timely distributed sensor information provided by the different sensors of the configuration is integrated in a four level logical sensor network. On each level of the network the logical sensors use temporal knowledge to perform their tasks. This includes the fault tolerant evaluation of physical sensor information and the fusion of spatially and timely redundant information. At the end of the integration process stands the robust detection of person movements.