A dynamic hierarchy for large models based on adaptive segmentation and fast simplification algorithm

Recent advances in 3D scanning technology have made it feasible to generate highly detailed models that are hard to render at interactive frame rates due to the exceedingly large models. An efficient and fast dynamic hierarchy, based on the half-edge triangle mesh data structure and the new multiple choice technology for large models, is presented in this paper. It can perform adaptive segmentation and fast simplification with permitting preservation of surface sharp characters and ensure the level-of-detail (LOD) of two adjacent region be varied independently without creating holes or cracks. The natural hierarchy provided by this approach makes it particularly suited for applications such as progressive mesh and constructing LOD, and only spends a fraction of the time required for rendering to perform the error calculations and dynamic mesh updates.