Data-Driven Space-Filling Curves

We propose a data-driven space-filling curve method for 2D and 3D visualization. Our flexible traverses the data elements in spatial domain way that resulting linearization better preserves features space compared to existing methods. achieve such coherency by calculating Hamiltonian path approximately minimizes an objective function describes similarity of values location neighborhood. extended variant even supports multiscale via quadtrees octrees. is useful many areas visualization including multivariate or comparative ensemble on regular grids visual analysis particle simulations. The effectiveness our evaluated with numerical comparisons techniques through examples datasets.