Abstract

In a quadtree mesh, the refinement level of a quadrant is the number of times it has been recursively split into 4 new quadrants. During the refinement process of a quadtree mesh, it is possible to lose the balance of the mesh, meaning that difference between refinement levels of two neighbor quadrants is greater than one. The balance is necessary because for those type of meshes, a set of patterns have been designed to manage the transition between fine and coarse regions, producing thus, a congruent mesh. The balance can be maintained while refining or achieved afterwards. In a previous work, we have shown that the first option is faster. In order to efficiently do so, it is necessary to dynamically maintain information about the neighbors. Currently, the meshing algorithms are achieving this by using an approach based on quadrant edges for queries about the neighbor refinement levels. In this work, we propose an improvement to the process of storage and dynamic update of the information of the quadrants by designing a new data structure based on quadrant nodes rather than quadrant edges, which allows reducing the execution time and memory usage.