The sonar images provide a rapid view of the seabed in order to characterizeit. However, in such as uncertain environment, real seabed is unknown and theonly information we can obtain, is the interpretation of different humanexperts, sometimes in conflict. In this paper, we propose to manage thisconflict in order to provide a robust reality for the learning step ofclassification algorithms. The classification is conducted by a multilayerperceptron, taking into account the uncertainty of the reality in the learningstage. The results of this seabed characterization are presented on real sonarimages.
展开▼
