Electrosensing of inhomogeneous targets

This paper addresses the electro-sensing problem for weakly electric fish in the case of inhomogeneous targets. It aims at providing a shape descriptor-based classification for inhomogeneous targets from measurements of the potentials on the skin of the fish. The approach is based on new invariants for the contracted generalized polarization tensors associated with inhomogeneous objects. The numerical simulations show that by comparing these invariants with those in a dictionary of precomputed homogeneous and inhomogeneous targets, one can successfully classify the inhomogeneous target.