Information Fusion in Signal and Image Processing: Major Probabilistic and Non-Probabilistic Numerical Approaches

Chapter 5

Information and Knowledge Representation in Fusion Problems

5.1. Introduction

In this chapter, we will briefly present the different modes for representing information and knowledge used in fusion, as well as how they are integrated into systems. Because numerical representations rely on the theories of probability, belief functions, fuzzy sets and possibility, they will be discussed again in greater detail in Chapters 6, 7 and 8. Knowledge-based systems, which can be used to structure information, knowledge and inference modes in order to combine them, will be presented only in broad strokes. They will not be discussed in detail in this book, but an example of a multi-agent system will be presented in Chapter 10. Symbolic approaches, as well as reasoning modes in different logics, will only be mentioned. They go beyond the scope of numerical fusion; however, their properties would deserve more attention in information fusion in signal and image processing.

5.2. Processing information in fusion

As we said in Chapter 1, we consider the word information in the broadest sense. Thus, the term information can be applied to any element that might be coded in order to be stored, processed or broadcast [DUB 01]. In signal and image processing, it often consists of information related to real worlds (observations, measurements, generic knowledge regarding real phenomena, etc.), but it can also consist of virtual worlds, in the expression of a user's goals and preferences.

This ...