Fitting tractable convex sets to support function evaluations (Q2046436)

The authors propose a new approach for dealing with the the geometric problem of estimating an unknown compact convex set from evaluations of its support function. It is based on convex optimization and does not present the shortcomings of the state of the art approaches. The new method is based on estimators that minimize the error over structured families of convex sets that are specified as linear images of some determinable sets. One of the main advantages of this approach is that convex sets parametrized in this way can be processed from a computational perspective, as one can efficiently minimize linear functionals over them. A geometric characterization of the asymptotic behavior of the new estimators is provided as well, while some numerical experiments highlight the utility of the newly proposed framework over the previous approaches.