HyperBlock-QuadTIN: Hyper-Block Quadtree based Triangulated Irregular Networks

Terrain rendering has always been an expensive task due to large input data models. Hierarchical mul-tiresolution triangulation and level-of-detail rendering algorithms over regular structures of grid digital elevation models have been widely used for interactive terrain visualization. The main drawbacks of these are the large cost of memory storage required and the possible over-sampling of highresolution terrain models. Triangulated irregular networks (TIN) can reduce the amount of vertices at the expense of more complex and slower memory data access. We present a hyper-block quadtree based triangulated irregular networks approach, where the notion of vertex-selection is extended to block-selection. The hyper-block structure allows to store different pre-calculated triangulations. This reduces the vertex selection time per frame and removes the calculations needed to build the geometric rendering primitives (triangle strips) of the scene at the expense of a larger number of selected vertices. The presented approach shows a speed increment of 20% for high-quality terrain rendering with small screen-projection error thresholds.