Classifier combination and score level fusion: concepts and practical aspects

This review article is about the classifier combination and score level fusion, which have received considerable attention in the research field over the past decade. The classifier combination covers research concerning the algorithms and methodologies that describe classifier combinations from a pattern recognition perspective. For the score fusion technique, this article addresses the fusion problem in those situations where the feature representations of the data are not available. These two concepts are discussed from the point of view of their mathematical formulations. Some practical aspects and applications are also treated.