Dissimmetric fusion of incomplete data for classification of underwater objects

摘要

To classify objects located in their environment, underwater mobile robots use sequential sensory data (sonar) . These pieces ofudinformation are imperfect, that means imprecise, uncertain and incomplete . Incompleteness is defined as the unavailability ofudsome parameters which makes some classification criteria impossible to compute and which delays the decisions . The paperudproposes to model data in the framework of possibility theory, and to apply fuzzy calculus to evaluate criteria even in the case ofudincompleteness . Results are sequentially fused by a dissymmetric combination process . The different dissymmetric fusion rules areudreviewed and a specific dissymmetric operator is proposed to solve the incompleteness problem .