Definitions

Data fusion exploits the synergy offered by the information originating from various sources. The operation of data fusion by itself is not new in any domain of application. For example, meteorologists predict weather for several tens of years. In remote sensing (i.e. Earth observation from spacecraft or aircraft), classification procedures are performed since long and are obviously relevant to data fusion. Data fusion allows formalizing the combination of these measurements, as well as to monitor the quality of information in the course of the fusion process. The formal framework for data fusion provides a better understanding of data fusion fundamentals and of its properties. Once established, such a framework permits a better description and formalization of the potentials of synergy between the available sources of information, and accordingly, a better exploitation of the data.

Data fusion research and development was conducted under a wide variety of systems, methods and names. Using recent words such as "data fusion", or "information fusion" translates the recent understanding that whatever the application domain, these synergistic approaches share common problems and common properties.

These common problems and common properties form a paradigm. Research in data fusion aims at exploring this paradigm. It expresses and clarifies the concept of data fusion and its properties. Definitions and terms of reference can be established that permits better co-operation between various domains because they share a common language. Research reveals the fundamentals in data fusion with respect to the fundamentals of the related sciences, e.g., physics, mathematics... It also expresses the properties of the data / information to be fused, of the methods for fusion, of the architectures, thus permitting better design, implementation and analysis of fusion processes. It is then easier to develop the most appropriate methods and algorithms, to monitor the quality throughout a process etc.

There are many advantages in using data fusion (E. Waltz and J. Llinas. Multisensor Data Fusion. Artech House, 1990)

* robustness and reliability. The system is operational even if one or several sources of information are missing or malfunctioning,
* extended coverage in space and time,
* increased dimensionality of the data space. It increase the quality of the deduced information; it also reduces the vulnerability of the system,
* reduced ambiguity. More complete information provides better discrimination between available hypotheses,
* providing a solution to the explosion of the information that is available today.

27-08-2001 - Copyright L. Wald, Armines / Ecole des Mines de Paris