ABSTRACT In this paper, the problem of acoustic wave scattering by obstacles in the two-dimensional inhomogeneous medium is considered. When the refractive index is a binary function, a hybrid method for reconstructing multiple impenetrable obstacles is proposed. The obstacle number is predicted based on the measured far-field data by using the neural network method. And then, it is used as the prior information of Bayesian method, the shape and position parameters of multiple obstacles are reconstructed using Markov chain Monte Carlo algorithm. The results of numerical experiments show that this method can effectively reconstruct the number, shape and position of obstacles in the inhomogeneous medium.
Read full abstract