Confidence map of the fusion of N runs
When a map is generated by the fusion of N runs, there is no confidence map associated to this new classification. Unfortunaltely, there is no straightforward way of computing the confidence. In the case of the majority voting, the proportion of classifications voting for the majority class can be used as confidence. In the case of Dempster-Shaffer fusion, the degree of conflict between the decisions could be used.
This should be implemented in the corresponding OTB application by adding an optional confidence map as output (an issue has ben open here https://gitlab.orfeo-toolbox.org/orfeotoolbox/otb/issues/1584).
Edited by Jordi Inglada