Delaunay Triangulation vs. Voronoi Diagrams: Key Differences Explained

Delaunay Triangulation vs. Voronoi Diagrams: Key Differences Explained

Overview

Delaunay triangulations and Voronoi diagrams are closely related geometric constructs that partition space based on a set of points. Both are fundamental in computational geometry and have widespread applications in mesh generation, spatial analysis, graphics, and scientific computing. This article explains what each structure is, how they relate, key differences, properties, common algorithms, and typical use cases.

Definitions

• Voronoi diagram: Given a set of sites (points) in the plane, the Voronoi diagram partitions the plane into regions where each region contains all locations closer to one particular site than to any other. Boundaries between regions are segments (or rays) equidistant to two sites; vertices are equidistant to three or more sites.
• Delaunay triangulation: For the same set of points, the Delaunay triangulation is a triangulation such that no point lies inside the circumcircle of any triangle. It maximizes the minimum angle over all possible triangulations, avoiding skinny triangles where possible.

Geometric relationship

• The Delaunay triangulation is the geometric dual of the Voronoi diagram: connect two sites with an edge in the Delaunay triangulation if and only if their Voronoi regions share a boundary. Voronoi vertices correspond to Delaunay circumcenters, and Voronoi edges are perpendicular bisectors of Delaunay edges.

Key differences

• Structure and elements
  • Voronoi: cells (polygons), edges, and vertices focused on regions of influence.
  • Delaunay: vertices (original sites), edges, and triangular faces forming a mesh.
• Purpose
  • Voronoi: describes nearest-neighbor regions and influence zones.
  • Delaunay: creates a quality triangulation for interpolation, meshing, and numerical methods.
• Duality
  • Voronoi and Delaunay are duals; one emphasizes proximity regions, the other connectivity.
• Local vs global properties
  • Voronoi directly encodes nearest-neighbor relationships; Delaunay emphasizes global triangulation quality (angle maximization).
• Degeneracies and handling
  • Collinear or co-circular point sets create degenerate cases: Voronoi cells may be unbounded; Delaunay may have non-unique triangulations. Robust implementations resolve ties with consistent predicates or symbolic perturbation.
• Complexity and size
  • Both structures have O(n) cells/edges/triangles in average planar cases; worst-case complexities remain linear in output size. Construction time is typically O(n log n) for efficient algorithms.

Important properties

• Empty circumcircle property (Delaunay): No site lies inside the circumcircle of any triangle.
• Max-min angle property (Delaunay): Among triangulations of a point set, Delaunay maximizes the minimum angle, producing fewer skinny triangles.
• Nearest neighbor (Voronoi): A site’s Voronoi cell contains all points whose nearest site is that site.
• Connectivity (Delaunay): Edges connect points that are mutual nearest neighbors under some directional constraints.

Common algorithms

• Voronoi
  • Fortune’s sweep line algorithm — O(n log n), constructs the Voronoi diagram directly.
  • From Delaunay dual — build Delaunay, then derive Voronoi by connecting circumcenters.
• Delaunay
  • Divide-and-conquer — O(n log n).
  • Incremental insertion with edge flips — average O(n log n).
  • Bowyer–Watson algorithm — robust incremental method used in practice.
  • Constrained Delaunay triangulation — allows matching domain boundaries or fixed edges.

Practical considerations and implementations

• Numerical robustness is crucial: use exact geometric predicates or well-tested libraries (e.g., CGAL, Triangle, Qhull) to avoid incorrect topology from floating-point errors.
• For large datasets or streaming data, incremental and parallel algorithms or spatial partitioning accelerate construction.
• Constrained Delaunay and weighted (regular) variants adapt the basic constructs to respect boundaries or incorporate weights (power diagrams correspond to weighted Voronoi).

Typical applications

• Voronoi diagrams
  • Nearest-neighbor queries and influence mapping
  • Facility location and service-area planning
  • Spatial interpolation and path planning
• Delaunay triangulations
  • Mesh generation for finite-element analysis and simulation
  • Surface reconstruction and terrain modeling
  • Interpolation (e.g., natural neighbor, barycentric coordinates) and computer graphics

When to use which

• Use a Voronoi diagram when you need explicit regions of nearest influence, e.g., locating service areas or computing additive distance fields.
• Use Delaunay triangulation when you need a high-quality triangulation for interpolation, meshing, or when you prefer connectivity among nearby points.

Example workflow

1. Start with a set of input points.
2. If you need region-based queries or nearest-site partitions, compute the Voronoi diagram (or derive it from Delaunay).
3. If you need a mesh for numerical simulation