Soft Data Fusion for Computer Vision.

The relevance of information fusion increases due to the complementary development of computer and sensory technologies. Information fusion basically attains the transformation of the information delivered by multiple sources into one representational form. Fuzzy systems research has generated a fruitful set of aggregation operators, which constitute a flexible alternative to operators customary used in information fusion. Among them is worth mentioning the fuzzy integral, which generalizes most of them. The fuzzy integral was proposed as a mathematical approach for the simulation of multi-criteria evaluation taking into account some cognitive aspects, since the process of multi-criteria integration undertaken by human beings is supposed to subsume the linear combination of the criteria. The successful employment of the fuzzy integral in real applications depends on the existence of methodologies for its automated parameterization. There is a lack of efficient methods with this purpose. The here presented dissertation attains two goals. First, it presents different computer vision applications, where the fuzzy integral is used as fusion operator. Furthermore, it analyzes different Soft Computing methodologies for the automated parameterization of the fuzzy integral, which allow the implementation of flexible automated systems with information fusion.