Contextual and Human Factors in Information Fusion

Context and human factors may be essential to improve measurement processes for each sensor and the particular context in each sensor could be used to obtain a global definition of context in multisensor environments. The reality may be captured by human sensorial domain based only on machine stimulus and then generate a feedback which can be used by the machine, at its different processing levels, adapting its algorithms and methods accordingly. Reciprocally, human perception of the environment could also be modelled by context in the machine. In the proposed model, both machine and man take sensorial information from the environment and process it cooperatively until a decision or semantic synthesis is produced. In this work, we present a model for context representation and reasoning to be exploited by fusion systems. In the first place, the structure and representation of contextual information must be determined before being exploited by a specific application. Under complex circumstances, the use of context information and human interaction can help to improve a tracking system's performance (for instance, video-based tracking systems may fail when dealing with objects interaction, occlusions, crosses, etc.).