Progress Report on the Project Automatic Target Recognition (N94-124) for SmallBusiness Innovation Research. Phase II

摘要

In this report, a real time optimal nonlinear filter is developed. This fastfilter is based on two techniques: (1) splitting of the convection and diffusion operators, and (2) tracking of important windows. Presented is an explicit scheme using the forward characteristic equation and the fast Gauss transform. The domain of interest is determined adaptively. Subroutines of the software code are briefly described and numerical results for 3-D problems are given.