Analysis of Interactive Editing Operations for Out-of-Core Point-Cloud Hierarchies

In this paper we compare the time and space complexity of editing operations on two data structures which are suitable for visualizing huge point clouds. The first data structure was introduced by Scheiblauer and Wimmer [SW11] and uses only the original points from a source data set for building a level-of-detail hierarchy that can be used for rendering points clouds. The second data structure introduced by Wand et al. [WBB+07] requires additional points for the level-of-detail hierarchy and therefore needs more memory when stored on disk. Both data structures are based on an octree hierarchy and allow for deleting and inserting points. Besides analyzing and comparing these two data structures we also introduce an improvement to the points deleting algorithm for the data structure of Wand et al. [WBB+07], which thus allows for a more efficient node loading strategy during rendering.