Map3DProject — 3D Terrain Reconstruction from SRTM (.hgt) + Voronoi Meshing (Java / LWJGL)

Demo video: here.

---

What it does

1. Reads SRTM .hgt elevation tiles

   • Height samples are stored as 2-byte integers in a 1201×1201 grid (SRTM 3 arc-second tiles).
   • The tile name encodes its geographic bounds (e.g. N20E100.hgt covers 20°N–21°N and 100°E–101°E).
   • Elevation values can be mapped to colors (no-data areas remain transparent).

2. Computes a Voronoi diagram (Fortune’s algorithm)

   • Uses the Fortune sweep-line approach to split the plane into non-overlapping convex polygons.
   • Data structures:
     • Priority queues for site/circle events
     • A red–black tree to maintain the “beach line” (parabolic arcs)
   • This implementation is written in Java and was adapted from a C# codebase.

3. Builds a renderable surface (mesh)

   • For each Voronoi cell / site, determines the neighboring sites.
   • Sorts neighbors by angle around the site to build consistent triangles.
   • Produces a triangle mesh suitable for OpenGL rendering.

4. Renders in OpenGL (LWJGL)

   • GPU shaders are stored in res/shaders.
   • Meshes are optimized for rendering with OpenGL.

---

Project structure

• src/ — Java source code (Voronoi/Fortune algorithm, data loading, rendering)
• res/shaders/ — GLSL shaders
• data/ — input data (place .hgt tiles here)
• img/ — images/screenshots for documentation
• lwjgl_linux/ — LWJGL native binaries for Linux
• lwjgl_win/ — LWJGL native binaries for Windows
• out/ — build artifacts (recommended to exclude from Git)
• Map3DProject.iml — IntelliJ IDEA project file

---

Requirements

• A recent JDK (11+ recommended)
• IntelliJ IDEA (recommended, since the repo contains an .iml project)
• OpenGL-capable GPU/driver

Note: LWJGL native binaries appear to be bundled in lwjgl_linux/ and lwjgl_win/.

---

Quick start (IntelliJ IDEA)

1. Clone the repository.
2. Open the project in IntelliJ IDEA (it should pick up Map3DProject.iml).
3. Ensure src/ is marked as Sources Root.
4. Configure LWJGL:
   • Add LWJGL JARs to the project/module dependencies (if not already configured).
   • Set VM option to point to the native libraries directory:
     • Linux:

       -Djava.library.path=./lwjgl_linux
       

     • Windows:

       -Djava.library.path=./lwjgl_win
       

5. Find the entry point:
   • Search in src/ for a class containing:

     public static void main(String[] args)

   • Run that class.

---

Input data (.hgt)

1. Put your SRTM .hgt tiles into data/.
2. Update the code/config (where the tile name is selected) to point to your tile, e.g. N20E100.hgt.
3. Run the application.

---

How it works (high level)

• Elevation tile → height field

  • Reads the 1201×1201 grid of 16-bit samples.
  • Optionally maps elevation to color/alpha for visualization.

• Sampling / sites

  • Selects/uses a set of points (sites) to build a Voronoi diagram over the area.

• Voronoi (Fortune) → adjacency

  • Fortune sweep-line constructs Voronoi edges and neighbor relationships.

• Adjacency → mesh

  • For each site:
    • take neighbors
    • sort by polar angle
    • triangulate fan around the site
  • Emit triangle list for GPU

• Mesh → OpenGL

  • Upload vertex/index buffers
  • Render with GLSL shaders (res/shaders)

---

Credits

• Fortune’s algorithm implementation in Java (adapted from a C# codebase).
• SRTM elevation tiles are a NASA/USGS data product (see SRTM documentation for dataset licensing/usage).