Generation of Decomposition Hierarchies for Efficient Occlusion Culling of Large Polygonal Models

Abstract

Efficient handling of large polygonal scenes has always been a challenging task and in recent years, view-frustum and occlusion culling have drawn a lot of attention for accomplishing this goal. The problem of how to efficiently organize such scenes for fast image synthesis is widely neglected, although the answer heavily affects the overall performance. In this paper, we present three adapted algorithms for efficient scene decomposition and compare those with another already available algorithm for decomposing general polygonal models into a hierarchy of sub-models for an occlusion culling application. While the latter is available as a commercial product, the three other approaches introduce new algorithms for scene decomposition which achieve significant better results.