Voxel space automata: Modeling with stochastic growth processes in voxel space

Abstract

A novel stochastic modeling technique is described which operates on a voxel data base in which objects are represented as collections of voxel records. Models are "grown" from predefined geometric elements according to rules based on simple relationships like intersection, proximity, and occlusion which can be evaluated more quickly and easily in voxel space than with analytic geometry. Growth is probabilistic: multiple trials are attempted in which an element's position and orientation are randomly perturbed, and the trial which best fits a set of rules is selected. The term voxel space automata is introduced to describe growth processes that sense and react to a voxel environment. Applications include simulation of plant growth, for which voxel representation facilitates sensing the environment. Illumination can be effidently estimated at each plant "node" at each growth iteration by casting rays into the voxel environment, allowing accurate simulation of reaction to light including heliotropism.