The effects of multipath on a bathymetric synthetic aperture sonar using belief propagation

This paper discusses the application of belief propagation to the reconstruction of bathymetric sonar images. Belief propagation is a powerful technique for probabilistic reasoning on Bayesian and Markov networks. It allows a priori information to be easily incorporated, in this case the expected seafloor height variation. We show that the algorithm works well without multipath interference but only a small improvement is obtained in the presence of multipath using a simple model. While a more sophisticated model incorporating multipath could be employed, the use of belief propagation would be computationally expensive.