Une nouvelle r\`egle de combinaison r\'epartissant le conflit - Applications en imagerie Sonar et classification de cibles Radar

TL;DR

This paper introduces a new evidence combination rule that proportionally distributes conflict among conflicting elements, improving decision accuracy in sonar and radar target classification applications.

Contribution

A novel combination rule for evidence theory that allocates conflict proportionally, with validation on sonar and radar data sets.

Findings

The new rule effectively manages conflict in evidence combination.

Improved classification accuracy in sonar and radar applications.

Comparison shows advantages over existing methods.

Abstract

These last years, there were many studies on the problem of the conflict coming from information combination, especially in evidence theory. We can summarise the solutions for manage the conflict into three different approaches: first, we can try to suppress or reduce the conflict before the combination step, secondly, we can manage the conflict in order to give no influence of the conflict in the combination step, and then take into account the conflict in the decision step, thirdly, we can take into account the conflict in the combination step. The first approach is certainly the better, but not always feasible. It is difficult to say which approach is the best between the second and the third. However, the most important is the produced results in applications. We propose here a new combination rule that distributes the conflict proportionally on the element given this conflict. We compare these different combination rules on real data in Sonar imagery and Radar target classification.