Tracking targets from indirect through-the-wall radar observations

In this paper we address the practically important task of identifying and tracking moving objects (e.g., people) inside a building from through-the-wall radar data obtained outside the building. In order to solve this task, we combine modern regularization techniques for solving non-linear inverse scattering problems with a Kalman filter approach for tracking targets from (indirectly obtained) observations. A level set based shape reconstruction technique is employed in order to obtain accurate initializations for the Kalman filter iterations, and a sparsity regularized inverse scattering approach is used in the Bayesian analysis steps. Numerical simulations in 2D show for a proof-of-concept setup that this combined approach of filtering and regularized inversion is very promising for efficiently and accurately tracking moving objects (potentially) in real time.