Optimal Camera Placement and Trajectory Estimation for 3D Scenes (paper under submission)

Advisor: Prof. Joseph S.B. Mitchell

We propose a novel and intuitive approach to solve the guarding problem for simple polygons with holes. In our approach we try to find 'critical' regions in the polygon from which it can be covered. Each of these regions are then evaluated based on certain characteristics and a final set of such regions are selected which constitute the guard set. These guard regions identify the domain within which a point guard can be placed. This not only produces close to optimal guard numbers but also a region of flexibility for each guard within which it can reside. The algorithm has been extensively tested on various classes of polygons (random, orthogonal, with and without holes and also 3D) and has shown very promising results and remarkable performance. We also provide very fast GPU-based solutions to a discrete version of the AGP as well as the Watchman Route Problem (WRP). We address several variants of the AGP as well as the WRP like limited visibility guarding, partial and resource constrained coverage etc. Our approach has been the first to address a wide class of computational geometry problems related to the AGP for 2D and 3D domains.

#video