Multi-sensor data fusion for underwater target recognition under uncertainty

The nonlinear, dynamic and random in underwater environment result in uncertainty in process of underwater target recognition. In order to exactly recognize underwater target type under uncertainty through lots of corrupted dynamic sensory information comes from different underwater sensors, we propose a dynamic information fusion framework, which is based on discrete dynamic bayesian network (DDBN) that provide a coherent and unified hierarchical probabilistic framework to represent, integrate and infer various target characteristics from dynamic sensory information of different modalities. The proposed framework can provide dynamic, purposive and sufficing information fusion particularly well suited to the underwater target recognition under uncertainty. Furthermore, we enhance inference efficiency and allow computation at various levels of abstraction suitable for underwater target recognition by distributed computation. Finally, The experimental results demonstrate the utility of the proposed framework in efficiently modeling and inferring dynamic events.