(The Eurographics Association, 2007) Xu, Fang; Mueller, Klaus; H.-C. Hege and R. Machiraju and T. Moeller and M. Sramek
There has been mounting evidence that the familiar Cartesian lattices, while convenient for signal processing and representation, are sub-optimal when it comes to signal fidelity. More suitable in this respect are optimal sampling lattices, such as the Hexagonal and Body Centered Cartesian (BCC) lattice, and recent work has employed these in the areas of volume rendering and image processing. In this paper we explore various applications of these lattices within the context of 3D Computed Tomographic Reconstruction, both in terms of the (2D) detector and the (3D) reconstructed object, and a theoretical analysis is provided. We combine this analysis with a practical application, that is, the use of these lattices within a real-time GPU-accelerated 3D reconstruction platform, in which performance is also of an immediate concern.