<title>Theoretical perspective in multisensor fusion</title>

Abstract

In the operation of an intelligent system, the role of multi-sensor integration and fusion can best be understood with reference to the type of information that the integrated multi-sensor can uniquely provide the system. The potential advantages gained through the synergistic use of this multi-sensor information can be decomposed into a combination of four fundamental aspects: the redundancy, complementarily, timeliness, and cost of the information. In this paper, the concept of fusion entropy is presented by generalizing the entropy theory, it is defined to be the posterior entropy of the fusion system. The validity of multi-sensor fusion is proved based on the transition of fusion entropy in information process; the result shows that the uncertainty in the output of a multi-sensor system is less than that in the output of a single sensor system. And then the rule to choose the multi-sensor is presented by the theorem.